text
stringlengths 0
330k
| metadata
stringlengths 459
10.9k
|
|---|---|
Rare variants analysis using penalization methods for whole genome sequence data
Background Availability of affordable and accessible whole genome sequencing for biomedical applications poses a number of statistical challenges and opportunities, particularly related to the analysis of rare variants and sparseness of the data. Although efforts have been devoted to address these challenges, the performance of statistical methods for rare variants analysis still needs further consideration. Result We introduce a new approach that applies restricted principal component analysis with convex penalization and then selects the best predictors of a phenotype by a concave penalized regression model, while estimating the impact of each genomic region on the phenotype. Using simulated data, we show that the proposed method maintains good power for association testing while keeping the false discovery rate low under a verity of genetic architectures. Illustrative data analyses reveal encouraging result of this method in comparison with other commonly applied methods for rare variants analysis. Conclusion By taking into account linkage disequilibrium and sparseness of the data, the proposed method improves power and controls the false discovery rate compared to other commonly applied methods for rare variant analyses.
Background
Despite success in detecting associations of common variants with complex traits (www.genome.gov/gwastudies/), it has proven difficult to elucidate a comprehensive picture of the genetic architecture of risk factor and disease traits without considering the effects of both rare and common variants via whole exome or genome sequencing. Decreasing costs and increasing quality have made discovery and genotyping of rare variants, which refer to variants with minor allele frequency (MAF) less than 0.05, more accessible across a large proportion of the genome and in large sample sizes. As a result of rapid expansion of human populations, there are very large numbers of rare variants segregating and these rare variants are relatively recent in origin [1,2]. Detecting genotype-phenotype associations and identifying novel loci having rare variant-phenotype associations are challenging since single-variant based statistical methods are inappropriate in this context due to the very large number of alleles and their low frequency. Furthermore, no or minimal effects of the majority of rare variants on a particular phenotype leads to a low signal-tonoise ratio and consequently underfitting with multiplevariant models. Hence, there is considerable interest in statistical methods that combine information across multiple variants, and thus reduce the cost of the large degrees of freedom in multivariate tests or adjustment for extensive multiple testing [3][4][5][6][7][8][9]. However simply combining information by pooling or collapsing does not take into account the direction of the variants' effects on a phenotype and alternative methods have been proposed that address this limitation (see, e.g. [10][11][12][13][14][15][16][17]). Furthermore, inclusion of large numbers of correlated variants may lead to overestimation.
Transitioning from common variant analyses to rare variant analyses creates three challenges related to sparse data [18]. First, within an individual personal genome, the number of sites that differ from the reference genome is small relative to the total number of bases. Second, sequence data, unlike array-based genotype data, contain a large number of rare variants. In fact, about half of the variant alleles in a study sample are seen only one or two times [19]. And third, only a small subset of the variable sites is expected to influence a given trait of interest, and the rest is expected to be neutral. This study presents a statistical and computational method tailored for sparse data and how it can be applied to whole genome sequence data to promote novel gene and rare variant discovery. We introduce a new method called Convex-Concave Rare variant Selection (CCRS), which includes both convex and concave penalization. We leverage the fact that rare variants data have low intrinsic dimensionality and are sparse. Hence, we project the variants into a full rank space with new coordinates in order to enhance information in new predictors comparing with original variants. We obtain these new coordinates using principal component analysis that includes a convex penalty to incorporate sparsity assumption. The CCRS improves the performance of sparse principal component (SPC) based method [20] in the context of rare variants analysis by selecting the components based on their degree of association with a complex trait which is appropriate for rare variant analysis. To this end, we use a concave penalized regression model to select the most promising variants while estimating their effect simultaneously.
Method
The CCRS method is applicable for all variants, but in this presentation, we focus on the analysis of rare variants because they pose special opportunities (i.e. large effects sizes) and challenges (i.e. sparsity). Assume we have detected and genotyped m rare variants X = (x 1 , . . . , x m ) in a sample of n individuals having a quantitative trait y = (y 1 , . . . , y n ) measured on each individual. In a typical whole exome or genome sequencing scenario, m is several orders of magnitude larger than n. To combat overdetermination, the typical analysis considers a subset of the variables at a time defined by physical proximity (e.g. a window) instead of functional characteristic (e.g. an annotated gene or enhancer element) because the vast majority of the rare variants are in noncoding regions in the genome. Interpretation of the results requires adjusting for multiple comparisons using accepted experiment-wise error or false discovery rate methods. Assume for the kth subset of X denoted by X k = {x jk } p j=1 where p < n, we have where = ( 1 , . . . n ) is an error vector, is an n × n diagonal matrix; α is the overall mean; T is an n × q covariate matrix, which includes non-genetic predictors such as age, sex and race; β k and θ are p-vector of genetic effects and q-vector of non-genetic effects, respectively. Although model (1) does not face the n p problem, the data lie in a lower-dimensional subspace due to dependency among rare variants [21,22] (i.e. linkage disequilibrium, LD) and coefficients are sparse because a large proportion of variants have small or no effects on the phenotype(s) of interest. Here, we introduce a new approach for rare variants analysis to address these two issues; LD and sparsity.
The CCRS approach
In rare variant analyses, the design matrix is more likely to be singular because of the LD structure in the population [21,22]. In addition due to low allele frequencies, there is little information about the association of each individual variant with a phenotype. Hence, applying a penalized regression model might not lead to identifying the true set of variants or genomic regions with nonzero effects. To bypass this difficulty, we project the genotype data into full rank space in order to reparameterize the regression model. Principal component analysis (PCA) is an appropriate tool for addressing collinearity and utilizes the low rank structure of the covariance matrix. One drawback of PCA is its lack of straight forward interpretability. However, in rare variant analyses each single variant is uninformative and there is a need to aggregate information in a region in order to identify association with the trait of interest. An issue of concern when applying PCA in the context of rare variants is that PCA may lead to new coordinates that include many non-influential variants due to sparseness. Accounting for such sparsity facilities identification of phenotype-influencing factors in each of the coordinates and also improves interpretability of the result because of the sparse loading matrix [23][24][25]. To accomplish this, we obtain a full rank approximation to the matrix X as X ≈ U n×r D r×r V r×p by imposing constraints on the columns of V and U similar to [20], where r is the rank of X, which is smaller than min(n, p); . a denotes L a norm; and D = {d j } r j=1 is a diagonal matrix of eigenvalues of the matrix X such that d 1 ≥ d 2 ≥ . . . ≥ d r ; v j and u j are the jth columns of V and U respectively. The L 1 norm penalization is equivalent to r | v ij |, where v ij is ijth entry of V, provides sparse principal components, UD. This is an optimization problem equivalent to maximizing u T j Xv j respect to u j and v j under constraint (2) and (3). This biconvex problem can be readily solved [26]. Therefore, we first fix u j and obtain v j when c is in the set of feasible solution We then obtain optimum solution of u j when u j 2 2 ≤ 1 and for j > 1, u j ⊥ u 1 , u 2 , . . . , u j−1 . The optimal value of c, which determines the level of sparsity, can be obtained through a cross validation approach [27,28].
By projecting data into a lower dimensional space, we reduce the number of predictors in the model to the rank of the design matrix, which increases the degree of freedom for hypothesis test and aggregates information into fewer predictor variables which helps alleviate one aspect of the low allele frequency challenge. These two features improve the power of identifying promising genetic regions influencing a phenotype of interest (see below).
In this context, it is not appropriate to select only the first few principal components as is usual in many applications, but rather we select the PCs based on their degree of association with the phenotype. To simultaneously measure the genotype-phenotype association and carry out variable selection, we consider a linear regression model including a concave penalization with loss function where Z = UD indicates a matrix of computed PCs with corresponding effect size γ , κ ∈ (0, 1) and regularization parameter ν ∈ R + . Without loss of generality, hereafter, we assume the overall mean is zero. This model is a form of Bridge regression and naturally yields sparse estimate for γ , in the sense that some of components of γ (κ,ν) , may be explicitly shrunk to zero [29,30]. The choice of κ < 1 leads to nonconcave minimization problems (see, e.g., [30][31][32] ) and provides a much sparser solution than the well-known penalized regression, lasso, with κ = 1 [33].
A simulation study
To evaluate the performance of the CCRS method, we randomly identified 1000 regions from a real whole genome sequence data set available from phs000668 study in dbGAP ( [URL]). Each region includes 50 variants (50,000 rare variants total) sequenced for 1456 individuals. Based on our experience, we have found that 50 variants are appropriate to capture the LD structure. As an example, Fig. 1 represents this LD structure for two regions of the genome.
We considered six different phenotypic effect scenarios (Table 1). We first randomly split the set of regions into two subsets to be influential regions and noninfluential regions. We then randomly selected 10 % of variants in each influential region to be causal variants with effect size +1 for Model-1 and Model-3 and with effect size ±1 for Model-2 and Model-4. In Model-5 and Model-6, the number of causal variants in a region is increased to 20 % of the total variants with different effect sizes randomly selected from U(0.5, 1) and {U(−1, −0.5), U(0.5, 1)}, respectively, where U denotes uniform distribution. Hence, we considered models with the same and also different effect directions.
To obtain a better understanding about the effect of LD on the result of the analysis, we selected variants based on their correlations. In Model-1 and Model-2, the causal variants are correlated with some neutral variants in their regions but in Model-3, Model-4 they are uncorrelated. For Model-5 and Model-6, both correlated and uncorrelated variants are selected (10 % of each).
In rare variants analysis, we are interested in identifying regions with significant effects on the phenotype corresponding to the following set of hypotheses for each region H 0 : ∀j γ j = 0 verses H 1 : ∃j s.t.γ j = 0. Table 1 Six genotype effect scenarios considered in simulation studies Model-1: 10 % of variants in influential regions are causal with effect size +1, while each one is correlated with some neutral variants in their region.
Model-2: 10 % of variants in influential regions are causal with effect size ±1, while each one is correlated with some neutral variants in their region. To test these hypotheses, we calculated the likelihood ratio of the selected model based on CCRS to the Null model, which does not include genotype variants in the model.
Model
We evaluated the performance of the CCRS method compared to four other commonly applied methods: Collapsing [8] denoted here as Col, CAST [3], SKAT-O [17] and sparse principal regression (SPC) [20]. The collapsing method generates a binary variable for each region to represent whether the minor allele is observed. It then tests the association between the traits level and the new binary variable I { j x j >0} through y = α + I { j x j >0} β + Tθ + regression model. The CAST method sums over all variants in the region and leads to y = α + j x j β + Tθ + . SKAT-O is a score based test, (y −μ) T P ρ (y −μ) when β k in (1) follows an arbitrary distribution with mean 0 and variance τ and pairwise correlation ρ between different β jk s. Here,μ is the predicted mean of y under H 0 , To examine the impact of significance level on the false and true discovery rates, we considered both α = 0.01 and 0.05 and calculated false discovery rate (FDR) and true positive discovery rate (TPR) defined as where F is the number of false positives; T is the number of true discoveries; R is the total number of significant regions; and M is the total number of regions.
To select the best model based on the CCRS method, we set ν = 0.01 and κ = 0. 5 The result of this simulation study shows that the CCRS performs better and more robust than other methods under a variety of genetic architectures, and it is much more prominent when the causal variants are not correlated with neutral variants in the region. Neglecting the presence of LD leads to overestimation of the overall effect of the regions. Although this overestimation might increase the power of detecting a region with some small effects that are correlated with some neutral variants,
Real data analysis
We analyzed sequencing data from the Atherosclerosis Risk in Communities (ARIC) study [35]. The data are described more fully in [19]. Briefly, 496 African-American individuals were whole genome sequenced at an average depth of 6.3-fold using an Illumine HiSeq 2000 and, after alignment, approximately 31 million high quality variants were called using SNPTools. We present here the result of an association analysis of rare and low frequency variants (MAF ≤ 0.05) with log transformed Apolipoprotein A1 levels (ApoA1). ApoA1 is a component of high density lipoprotein (HDL), which is associated with reduced risk of coronary heart disease [36,37]. The protein promotes lipid efflux, including cholesterol, from tissues to the liver for excretion [38].
The genotype data includes 949,986 rare variants that are mostly in noncoding regions in the genome [19]. Therefore, we used a sliding window approach to define physical proximity (window). There are approximately 38 thousand consecutive windows each including 50 rare variants and stepping 25 variants until the next window. Therefore, by design, the windows overlap and the results of consecutive windows are not independent. To detect associated regions potentially influencing plasma To define the threshold for statistical significance taking into account multiple hypothesis testing, we ran 100,000 permutation test over 100 windows. Based on this threshold, 10 −6 , we detected one significantly associated region by the CCRS method. Figure 5 shows the p-values of 80 windows around this region. All of the approaches except CAST and Collapsing test have a peak in this region. The figure shows that the CCRS maintains power for detecting phenotype-influencing region while keeping the p-value of the null or neutral regions small. This is an important property of CCRS that controls the false discovery rate.
The region contains the gene, FAM78B, which is expressed at high levels in myocytes, fibroblasts, endothelial cells. Little is known about the function of FAM78B. However, within the promoter for FAM78B, three binding sites for the transcription factor PPARG and two binding sites for the transcription factor HNF1A have been identified ( [URL] and [39]). Pi et al. [40] have shown a significant effect of PPARG on HDL and ApoA1; the major protein component of HDL [41]. PPARs are also expressed in the cardiovascular system such as endothelial cells, vascular smooth muscle cells and monocytes/macrophages (see for e.g. [42]).
Discussion
We have introduced a new approach, CCRS, for the analysis of whole genome sequence data in order to identify regions of the genome (e.g. genes or other functional motifs) influencing a phenotype of interest. The CCRS improves the power of identifying a set of variants associated with a phenotype by taking into account the sparseness and LD structure in the data. The CCRS applies a concave penalized regression method after projecting the sequence variants in a full rank space that is more informative via sparse principal component analysis. By applying sparse PCA, the CCRS aims to enhance the information in the predictors instead of reducing dimension as typical application of sparse PCA, which might increase risk of missing important variants in rare variants analysis. While the first step of analysis (sparse PCA) is an unsupervised method, it does not increase the FDR of the method in the second step of the analysis.
Although the CCRS method can be applied to both common and rare variants, the focus of this analysis was on rare variants because of the role of these variants on phenotype variation. The CCRS method also can be easily expanded to logistic regression and applied for case control studies. However, we investigated the CCRS performance for quantitative traits while the overwhelming majority of the literature focusing on case/control studies and there is a daunting need to develop methods for quantitative traits.
Using simulated data, we show that the FDR of the CCRS method is smaller than other commonly applied genomic region-based test methods while it has higher power of identification in most of the situations. Furthermore, the FDR of CCRS is smaller and robust to the LD structure in the region in comparison to the other methods. While the statistical test for rare variants are typically region-based test, there is risk of overestimation of overall effect of regions by neglecting the LD between causal and neutral variants in the region. Consequently, the risk of missing promising regions might increase through multiple hypotheses testing.
Penalized regression and other shrinkage methods that have been introduced for sparse data applications can correctly select nonzero coefficients under specific conditions [43,44]. Applying these approaches to largescale genome sequence applications that include correlated variants due to LD might not lead to a true set of selected variants with nonzero coefficients. Addressing this challenge is difficult in rare variant analyses because each individual variant by itself includes little information. To resolve this problem, the CCRS reparameterizes the model via PCA restricted with L 1 norm constraints to provide a full rank design matrix. Imposing L 1 penalization in PCA generates a sparse loading matrix that renders the analysis interpretable. The CCRS method efficiently incorporates information from low frequency variants by generating new predictors that are much more informative. The CCRS uses a concave penalized regression model to simultaneously select the most important PCs regarding their association with the phenotype of interest, but also to estimate their effect sizes. The zero effect sizes can be uniquely identified due to the use of full rank approximation of the design matrix. The advantage of the concave penalty term is that the rate of shrinkage gets smaller as the effect size increases. In other words, the CCRS not only has the property of parsimony, it also avoids shrinkage over large effect sizes. Thus, the CCRS maintains power for detecting phenotype-influencing regions while keeping the p-value of the neutral regions small.
As an example real data application, we used the CCRS method and genome sequence data to analyze plasma ApoA1 levels, and one region met the experiment-wise criterion for statistical significance. The region contains the gene, FAM78B, which is expressed at high levels in myocytes, fibroblasts, endothelial cells ( [URL]. proteinatlas.org/ENSG00000188859-FAM78B/tissue). In a real application, annotation of the non-coding regions should be integrated into the analysis, and replication in an independent sample would be the next step to consider it as a novel discovery.
Conclusions
Large-scale whole genome sequencing and high-powered computing are becoming more readily available and affordable. There is an emerging shift from sequencing and computing technologies toward study design, data processing algorithms, and statistical and informatics methods for extracting usable information from the very large amount of genome sequence data that are imminent. The CCRS method presented here for the first time is a practical, powerful and efficient method for taking into account the nature of whole genome sequence variation to identify regions of the genome influencing common complex risk factor phenotypes and diseases.\===
Domain: Biology Medicine Computer Science. The above document has
* 2 sentences that start with 'The CCRS improves the',
* 2 sentences that start with 'This is an',
* 2 sentences that start with ' Therefore, we',
* 3 sentences that start with 'The CCRS method',
* 2 sentences that end with 'rare variants analysis',
* 3 sentences that end with '1 , ',
* 2 sentences that end with 'neutral variants in the region'. It has approximately 3558 words, 160 sentences, and 32 paragraph(s).
|
[{"added": "2016-05-31T19:58:12.500Z", "created": "2015-12-01T00:00:00.000", "id": "11114549", "metadata": {"extfieldsofstudy": ["Biology", "Computer Science", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://bmcbioinformatics.biomedcentral.com/track/pdf/10.1186/s12859-015-0825-4", "pdf_hash": "e6fbd7ab20bfabe0c9a7818f1751eb34e174190a", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:3102", "s2fieldsofstudy": ["Biology", "Computer Science", "Medicine"], "sha1": "e6fbd7ab20bfabe0c9a7818f1751eb34e174190a", "year": 2015}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
The Molecular Genetic Architecture of Self-Employment
Economic variables such as income, education, and occupation are known to affect mortality and morbidity, such as cardiovascular disease, and have also been shown to be partly heritable. However, very little is known about which genes influence economic variables, although these genes may have both a direct and an indirect effect on health. We report results from the first large-scale collaboration that studies the molecular genetic architecture of an economic variable–entrepreneurship–that was operationalized using self-employment, a widely-available proxy. Our results suggest that common SNPs when considered jointly explain about half of the narrow-sense heritability of self-employment estimated in twin data (σg 2/σP 2 = 25%, h 2 = 55%). However, a meta-analysis of genome-wide association studies across sixteen studies comprising 50,627 participants did not identify genome-wide significant SNPs. 58 SNPs with p<10−5 were tested in a replication sample (n = 3,271), but none replicated. Furthermore, a gene-based test shows that none of the genes that were previously suggested in the literature to influence entrepreneurship reveal significant associations. Finally, SNP-based genetic scores that use results from the meta-analysis capture less than 0.2% of the variance in self-employment in an independent sample (p≥0.039). Our results are consistent with a highly polygenic molecular genetic architecture of self-employment, with many genetic variants of small effect. Although self-employment is a multi-faceted, heavily environmentally influenced, and biologically distal trait, our results are similar to those for other genetically complex and biologically more proximate outcomes, such as height, intelligence, personality, and several diseases.
Introduction
Economic variables such as income, education, and occupation are well-known to be related to health outcomes and longevity [1][2][3][4][5][6][7][8][9][10]. Specifically, there is a consistent inverse relation between indicators of socioeconomic status and cardiovascular disease [11]. For example, occupational choice is associated with the incidence of coronary heart disease among women [12]. Intriguingly, health outcomes, longevity, income, educational attainment, and occupational choice have all been shown to be partly heritable (see ref. [13] for complex diseases, refs. [14][15][16][17] for longevity, refs. [18][19][20][21][22] for education, refs. [23][24][25] for income, and refs. [26][27][28] for occupational choice). This suggests that the same genetic factors could be linked to socioeconomic status and health outcomes, or that indirect causal pathways from genetic variants to health outcomes exist that are mediated by individual behavior and the environment. For example, a potential mismatch between personal disposition and occupational choice may result in stress and decreased happiness, which have been shown to negatively affect (cardiovascular) disease incidence and longevity [29][30][31][32]. Therefore, knowledge about the specific molecular genetic architecture of socioeconomic variables and about the effects of mismatches between genetic predispositions and realized choices could yield important insights for epidemiology and public health policy. Unfortunately, most efforts to investigate the influence of genes on economic variables were until now limited to candidate gene studies that often failed to replicate later [33,34].
This study reports results from the first large-scale collaboration that studies the molecular genetic architecture of a specific economic behavior-entrepreneurship-using data from high-density SNP arrays. Entrepreneurship has been associated with poor health [35], increased stress [36], relatively low average incomes [37], but also with greater job and life satisfaction [38][39][40]. The analysis of entrepreneurship is complicated by the fact that it is a multi-faceted phenomenon [41]. Individuals may engage in entrepreneurial activity for a variety of reasons. For example, certain individuals may be motivated to pursue a business opportunity or to gain independence, whereas others may do so because of unemployment and a lack of viable alternatives in paid employment. Despite this complexity, empirical evidence suggests that entrepreneurship tends to run in families [42][43][44][45][46][47], and recent twin studies consistently estimate the heritability of this behavior to be on the order of 50% [26][27][28]. As these results suggest that entrepreneurship is partly influenced by genetic variation, specific markers that are associated with entrepreneurship should, in principle, exist. Research that is aimed at discovering these specific markers has thus far been limited to one candidate gene study. This study [48] found evidence for an association between a specific genetic variant in the DRD3 gene and entrepreneurship in a sample of n = 1,335. However, a more recent study [49] failed to replicate this association in three larger samples of n = 5,374, n = 2,066, and n = 1,925.
The molecular genetic architecture of entrepreneurship therefore remains largely unknown. A variety of alternative architec-tures could account for heritable variation. For example, there may be a small number of rare variants with strong effects, multiple common variants with small or modest effects, or some combination of these possibilities [50,51]. Therefore, we aimed to identify the molecular genetic architecture of entrepreneurship to facilitate a more sophisticated understanding of the nature of the associated heritable variation.
We use self-employment as a proxy for entrepreneurship in this study, which is the most widely available proxy for entrepreneurship. Self-employment is defined as having started, owned, and managed a business. Initially, we used a classical twin design to estimate the heritability of the tendency to engage in selfemployment. We performed this analysis to determine the comparability of our results with (1) estimates of previous twin studies, and (2) estimates from a novel method from molecular genetics. This recently described method [52] is used here to quantify the proportion of variance that is explained by common SNPs (and unknown causal variants that are in linkage disequilibrium with these SNPs) in the tendency to engage in selfemployment.
Furthermore, we performed a meta-analysis of genome-wide association studies (GWASs) of self-employment from sixteen studies to identify genetic variants that are robustly associated with self-employment. Together, these studies comprised 50,627 participants of European ancestry who are part of the Gentrepreneur Consortium [53,54]. This study is the first large-scale effort to identify common genetic variants that are associated with an economic variable. We also tested whether self-employment could be predicted out-of-sample solely using genotype data and the results of our meta-analysis.
Theoretical and empirical evidence from entrepreneurship research suggests that there may be differences between males and females with respect to the type of businesses they start. These differences also extend to individuals' motivations, goals, and resources [55][56][57][58][59] and exist because women face different-and typically more-barriers to entrepreneurship than men [60][61][62]. Therefore, we performed both pooled and sex-stratified analyses for all of our investigations.
Participating studies and self-employment measures
The analyses were performed within the Gentrepreneur Consortium [53,54], which included two out of the five studies that participate in the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) Consortium [63] (THI-SEAS), the UK Adult Twin Registry (TwinsUK), and the Cardiovascular Risk in Young Finns Study (YFS). The Swedish Twin Registry (STR) served as an in silico replication study, as genome-wide data were only available following the completion of the discovery stage.
The studies collected data regarding occupational status using questionnaires or interviews, from which self-employment status was distilled. Self-employment measures were defined in collaboration with the consortium leaders to minimize heterogeneity across participating studies. The cases were defined as individuals who were self-employed at least once, and the controls were defined as individuals who were never self-employed during their working life. However, for a number of studies, reliable data regarding work-life history were unavailable, possibly resulting in the inclusion of previously self-employed individuals in the control group. The details regarding the background and self-employment measures of each of the discovery studies and of the replication study are given in Table S1.
Ethics statement
All participating studies were approved by the relevant institutional review boards or the local research ethics committees, including the Icelandic National Bioethics Committee (VSN: 00-063), the Icelandic Data Protection Authority, and the Institutional Review Board for the National Institute on Aging (AGES); the Ethics Committee of the Medical Faculty of the University of Graz (ASPS); the Medical Ethics Committee at Erasmus University which approved the protocols for the ascertainment and examination of human subjects (ERF); the local ethics committee and data safety commissioner, the sampling design was approved by the federal data safety commissioner (GHS); the Ethics Committee for Epidemiology and Public
Genotyping, imputation, and quality control
The seventeen participating studies used a variety of commercially available SNP genotyping platforms to genotype their participants. Each study performed quality control of their genotypic data and imputed the genotypes of each participant to a common set of approximately 2.5 million SNPs from the HapMap CEU population. The exceptions to this were THI-SEAS, which only supplied results for directly genotyped SNPs, and HRS, which imputed to the 1,000 Genomes Project Phase I v3 panel. Prior to the meta-analysis, we performed parallel quality control of the association results for each study. SNPs were excluded on the basis of minor allele frequency (MAF,0.01 or MAF,0.05 if deemed necessary) and if the imputation quality (a measure of the observed variance divided by the expected variance of the imputed allele dosage from the imputation software output) was less than 0.4. Following these exclusions, approximately 2.4 million SNPs remained. Study-specific details regarding the genotyping, imputation, and quality control are given in Table S2.
Statistical analysis
Tetrachoric correlations were used to calculate self-employment correlations for MZ and DZ twin pairs. This analysis assumes a latent normally distributed tendency to engage in self-employment. We estimated the heritability of the tendency to engage in self-employment in the replication study using standard twin study methods, which were implemented in the program Mx [64]. Only complete twin pairs with data regarding self-employment status were included in the analysis and opposite-sex DZ twin pairs were excluded, resulting in a final sample size of 4,464 individuals. Specifically, for pooled males and females, males only, and females only, we fitted the three following nested models using the maximum likelihood approach on the raw data: (1) a model including an additive genetic effect, a shared common environment effect, and an individual-specific environment effect (the ACE model); (2) a model that included only an additive genetic and an individual-specific environment effect (the AE model); and (3) a model including only a common environment effect and an individual-specific environment effect (the CE model). For all of the samples, we controlled for a z-score of age by estimating agespecific thresholds. For the pooled sample, we additionally controlled for sex in a similar way.
We used the method that was recently developed by Yang et al. [52] to estimate the proportion of variance in the tendency to engage in self-employment that is explained by all of the common genotyped SNPs. The method is implemented in the GCTA software [65] and hinges on the assumption that in a sample of unrelated individuals, environmental factors segregate independently in the pedigree from the degree of genetic relatedness. In contrast to the twin study design, genetic relatedness is not inferred from the pedigree but is estimated directly from genome-wide SNP data. Under the assumption of no confounding by environmental variables, we can then estimate the accounted-for variance by relating the estimated genetic relatedness between pairs of individuals to their phenotypic correlation. The resulting estimate is actually a lower bound of the heritability that is estimated from classic twin and family studies. The reason for this is that twin and family studies capture the variation that is due to all of the additive causal variants, whereas the more recently developed method only captures the variants that are either directly genotyped or in linkage disequilibrium.
We used a combined sample of individuals from one of the discovery studies (RS-I) and the replication study (STR) to estimate the accounted-for variance. We restricted the sample from each study to individuals for whom data regarding selfemployment were available. Additionally, we included only one randomly selected individual from each family in the STR sample. A second round of quality control of the genotypic data was then performed for both studies. In the RS-I sample, we excluded 3,748 SNPs because they failed a test of Hardy-Weinberg equilibrium at p,1610 26 . We removed 24,993 SNPs with minor allele frequencies that were lower than 0.01 and another 6,665 due to data missingness greater than 5%. In total, 5,374 individuals and 561,466 autosomal SNPs were included in the analysis. In the STR sample, we removed two SNPs because they failed a test of Hardy-Weinberg equilibrium at p,1610 26 . Another 628 SNPs with a minor allele frequency lower than 0.01 were removed, as were two SNPs with data missingness greater than 5%. Therefore, 643,924 autosomal SNPs and 2,589 individuals were included in the analysis.
We then estimated the genetic relationships among 7,963 individuals in the combined sample from the 301,115 common autosomal SNPs. We dropped one of any pair of individuals with an estimated genetic relationship that was .0.025 while maximizing the remaining sample size to exclude the possibility of ascribing shared environmental effects to genetic effects and/or including the effects of causal variants not correlated with the genotyped SNPs but captured by the pedigree. The maximum relatedness in the remaining sample of 6,223 individuals therefore approximately corresponds to cousins two to three times removed [52].
Next, the linear mixed model y = m+g+e was fitted, where y is the binary phenotype, g the total additive genetic effect of the SNPs, and e is a residual effect. The restricted maximum likelihood (REML) was used to estimate the variance of the total additive genetic effect s g 2 of the SNPs by fitting the genetic relationships as the covariance structure. Because the analyzed phenotype is binary, s g 2 is the variance of the total additive genetics effects on the observed 0-1 scale. A latent normally distributed tendency to engage in self-employment was assumed when transforming the explained variance from the observed 0-1 scale to the latent scale using the transformation that is derived in the appendix of Dempster and Lerner [66]. For all of the analyses, we controlled for a z-score of age, study, and the first ten principal components of the genetic relationships of the combined sample. In the pooled sample, we also controlled for sex.
In addition to the Yang et al. [52] method, we employed a novel method developed by So et al. [67] that serves the same purpose, i.e., estimating the proportion of variance in the tendency to engage in self-employment that is explained by all of the common SNPs. However, in contrast to the Yang et al. [52] method, So et al.'s method does not require raw genotype data but attempts to recover the accounted-for variance from the meta-analysis results. Using PLINK [68], we restricted the meta-analysis results to SNPs that were present in the HapMap Phase II CEU panel (release 23a) and pruned those in strong linkage disequilibrium with other SNPs using a pairwise r 2 threshold of 0.25 in a window of 100 SNPs that slides in 25 SNP increments. After this procedure 172,742, 175,970, and 172,989 SNPs remained in the pooled males and females, males only, and females only sample, respectively. We used the Gaussian Kernel function, considered under the null-hypothesis of no association, and ran the simulation 500 times in each sample.
The genome-wide association analysis of self-employment was independently performed by each study according to a predefined analysis plan. The analyses were performed for pooled males and females, males only, and females only using an additive genetic model, controlling for age (#29 [reference]; 30-39; 40-49;$50) and sex in the pooled sample. To control for population stratification, the first four principal components of the genotypic data were also included if available. We provide details regarding the statistical analysis within each study in Table S2.
Following the association analyses, the genomic inflation factor l was calculated for each sample to quantify any remaining population stratification or cryptic relatedness. The lowest inflation factor was 0.989, and the highest was 1.156, although this latter value was for a study that did not include the first four principal components of the genotypic data in the analysis (Table S3). Genomic control [69] was applied in samples with inflation factors that were greater than one by adjusting the test statistics.
We next performed fixed-effect meta-analyses of the association results from the discovery studies for pooled males and females, males only, and females only using METAL software [70]. Although the phenotype was defined as self-employment in each participating study, we could not harmonize the exact wording of the question on which the self-employment measure was based. In addition, the connotations of self-employment may depend to some extent on the level of economic development and culture. This may lead to unobserved gene-environment interactions that could introduce additional noise in the GWAS results pooled across studies. We combined the association results using weighted z-scores that were based on the p-values and the direction of the effects. This method first computes a per-study signed z-score for each SNP based on its p-value and the effect direction. The zscores are then summed with weights that are proportional to the square root of the sample size of each study. Following the metaanalyses, only autosomal SNPs that were present in the Hapmap Phase II CEU panel (release 22, NCBI build 36) and in at least half of the contributing samples in each meta-analysis were retained prior to both reporting p-values and the creation of the Q-Q and Manhattan plots. We a priori set the genome-wide significance threshold to p,5610 28 . SNPs with p,1610 25 were considered suggestive and also carried forward to the replication stage. The heterogeneity of the test statistics between the studies was assessed using the I 2 metric [71,72] and Cochran's Q statistic [73].
Replication was attempted for significant and suggestive SNPs from each meta-analysis using an in silico replication study comprising 3,271 individuals. The association results for these SNPs were looked up in the replication study and meta-analyzed together with the discovery samples for pooled males and females, males only, and females only. To adjust for family relationships in the replication study, we performed family-based association tests implemented in the MERLIN software [74].
We used the discovery meta-analyses results to calculate genebased p-values using the VEGAS program [75]. The positions of the UCSC Genome Browser hg18 assembly were employed to assign SNPs to genes, which included regions that were 650 kb from the 59 and 39 UTRs.
For the prediction analyses, we followed the approach that was pioneered by The International Schizophrenia Consortium [76] and used the association results from the discovery meta-analyses to predict self-employment in the STR. Specifically, twelve overlapping sets of SNPs that were nominally associated in the discovery meta-analyses were created for different significance thresholds (p T ,0.01, p T ,0.05, p T ,0.1, p T ,0.2, p T ,0.3, p T ,0.4, p T ,0.5, p T ,0.6, p T ,0.7, p T ,0.8, p T ,0.9, and p T #1). These sets were used as inputs for score calculation in the STR. We restricted the STR sample to individuals for whom data regarding selfemployment were available and included only one randomly selected individual from each family, resulting in a final sample size of 2,589 individuals for the prediction analyses.
Prior to calculating the scores for each individual in the STR, we followed [76] and selected all of the autosomal SNPs, pruning those in strong linkage disequilibrium with other SNPs. This process was performed using a pairwise r 2 threshold of 0.25 in a window of 200 SNPs that slides in five SNP increments. Following this exclusion process, 135,823 SNPs remained. The PLINK [68] 'score' function was then used to calculate the total score for each individual in the STR. The score is defined as the sum of the number of score alleles, weighted by the estimated coefficients from the discovery meta-analyses, divided by the number of nonmissing genotypes. If an individual was missing a genotype, it was imputed as the mean genotype based on the score allele frequency in the STR. On average, the score was calculated from approximately 120,000 SNPs given that (1) the coefficients were only estimated for SNPs in the HapMap CEU population in the discovery meta-analyses, and (2) the overlap with the genotyped SNPs was not perfect. Lastly, we regressed self-employment onto the score using a logistic regression model. The variance that was explained by the score was estimated using the Nagelkerke pseudo-R 2 of the fitted model. We also calculated the area under the receiver operating characteristic curve (AUC) to evaluate the prediction accuracy.
Results
Heritability of self-employment and the degree of variance that is accounted for by common SNPs We used data from the Swedish Twin Registry (STR) and the classical twin design to estimate the heritability of the tendency to engage in self-employment. We computed the tetrachoric correlations between the tendencies to engage in self-employment within monozygotic (MZ) and dizygotic (DZ) twin pairs. Table 1 indicates that the correlations within the MZ twin pairs were consistently higher than within the DZ twin pairs for males only, for females only, and for pooled males and females. We note that the correlation within DZ twin pairs in the pooled sample was higher than for the DZ correlations in males and females when the two sexes are considered separately. This effect most likely results from imprecise estimation of the tetrachoric correlations due to the small number of cases. When we computed Pearson correlations, the pooled DZ twin pairs correlation was in between the male and female DZ twin pairs correlations. Applying Falconer's formula [77] to the correlations in Table 1, yields h 2 estimates of 0.39 for pooled males and females, 0.69 for males only, and 0.34 for females only.
A maximum likelihood approach was employed to estimate the relative contributions of the additive genetic (A), shared common environment (C), and individual-specific environment (E) components. This approach was performed using an ACE model and two nested submodels for pooled males and females, males only, and females only. Table 2 gives the estimates of the A component as 0.54 for pooled males and females, 0.67 for males only, and 0.38 for females only. The estimates of the C component were 0.01 for pooled males and females, 0.00 for males only, and 0.02 for females only. The A component was significant at the 95% confidence level for pooled males and females, and for males only, although the confidence intervals were very wide. This component was not significant for the females only analysis. However, the x 2 test for goodness-of-fit and Akaike information criterion indicated that the AE model was the best-fitting model in all samples. In this submodel, the estimate for the A component for females only did not change markedly compared to the ACE model but was significant at the 95% confidence level. The estimates of the A component for pooled males and females, and males only were 0.55 and 0.67, respectively; these results were significant.
The recently developed method by Yang et al. [52] was employed to estimate the degree of variance in the tendency to engage in self-employment that is explained by all of the genotyped autosomal SNPs in the GWAS datasets. The proportion of the explained variance was estimated for pooled males and females, males only, and females only. To maximize the power of the analysis, we used a combined sample of one of the discovery studies (Rotterdam Study Baseline [RS-I]) and the STR. We estimated that 25% (p = 0.032) of the variance in the tendency to engage in self-employment could be explained by the common genotyped autosomal SNPs for pooled males and females ( Table 3). The variance that could be explained for males only and for females only was 25% (p = 0.152) and 0% (p = 0.499), respectively. The estimates for males and females separately were not significantly different from one other. The fact that the variance that is explained was zero for females is most likely due to the very low number of female cases (n = 353) compared to the number of controls (n = 3,482). The estimation of the explained variance is therefore very imprecise. We also estimated the variance that was explained for pooled males and females, males only, and females only in the RS-I and the STR separately. The estimates were not significant because the standard errors of these estimates depend heavily on the sample size. However, considered in their entirety, the results were consistent with the estimates that we present for the combined RS-I and STR samples. Overall, the results for pooled males and females and for males indicated that the degree of variance in the tendency to engage in self-employment that is explained by all of the common autosomal SNPs simultaneously is only approximately half of the narrow-sense heritability that is estimated using the STR and the classical twin design. Furthermore, estimates using the method developed by So et al. [67] also provide non-zero estimates for heritability. Specifically, the accounted-for variance was 7% for pooled males and females, 21% for males only, and 15% for females only. However, confidence intervals and standard errors could not be calculated for these estimates because not all raw genotype data were available, prohibiting further interpretation of these results.
Meta-analyses of genome-wide association studies
We performed genome-wide association analyses of selfemployment using the data from sixteen discovery studies. These studies comprised 7,734 participants who had been self-employed at least once and 42,893 participants who did not report being selfemployed. Table 4 includes the descriptive statistics for the studies. The mean ages in the pooled samples of males and females ranged from 31 to 68.8 years, and the average age across all of the studies was 53.4 years. Following independent association analyses for each study, we performed a fixed-effect meta-analysis of the studylevel results for approximately 2.4 million SNPs using a pooled zscore approach.
The discovery meta-analysis Q-Q plot ( Figure 1A) did not indicate a strong deviation for the lowest p-values. However, no confounding issues related to population stratification, cryptic relatedness, or genotyping errors were detected, as no systematic deviation from the expectation under the null hypothesis of no association was observed [78]. As illustrated in the Manhattan plot ( Figure 2A), we observed twenty SNPs with 4.1610 26 #p,1610 25 (Tables 5 and S4). The SNP with the lowest p-value, rs6906622 (p = 4.10610 26 ), was located near the RNF144B gene, with most studies indicating that the minor allele increased the probability of being self-employed (Table 5).
We next attempted to replicate in silico the twenty suggestive SNPs in the STR (n = 3,271). Two of the twenty SNPs associated with self-employment were statistically significant at the 5% level in the replication study. However, the SNP effects were not in the same direction as in the majority of the discovery studies (Table S4), indicating that these SNPs were potential false positives. We then performed a combined meta-analysis of the discovery and replication studies. For all SNPs, the p-values were larger in the combined sample than in the discovery sample and did not reach genome-wide significance (Table S4).
The Q-Q plot for the male only meta-analysis ( Figure 1B) gave a certain degree of suggestive evidence of association; however, no evidence of population stratification, cryptic relatedness, or genotyping errors was observed, as only certain SNPs-those with particularly low p-values-deviated from their expectation under the null hypothesis of no association. The female only metaanalysis Q-Q plot ( Figure 1C) did not indicate a strong deviation for the lowest p-values and no evidence of population stratification, cryptic relatedness, or genotyping errors was observed. No SNPs reached genome-wide significance in the sex-stratified metaanalyses (Table 5), as can be observed in the Manhattan plots ( Figures 2B and C). The male meta-analysis resulted in 22 suggestive SNPs with p,1610 25 , and the female meta-analysis resulted in sixteen suggestive SNPs (Tables 5, S5, and S6). The top SNP in males, rs6738407 (p = 1.52610 27 ), was located in the HECW2 gene, and most studies reported that carrying the minor allele decreased the probability of being self-employed. The top SNP in females, rs2331548 (p = 1.93610 26 ), was located near the CBR4 gene, and most studies estimated that carrying the minor allele decreased the probability of being self-employed.
The replication strategy for the 38 suggestive SNPs from the sex-stratified meta-analysis that were carried forward into the replication stage was similar to that used for the meta-analysis replication of the pooled data. We performed an in silico replication study using the data from the STR. None of the SNPs reached nominal significance (p,0.05) in the replication study for males only (n = 1,409, Table S5) and females only (n = 1,862, Table S6). In addition, for the majority of the suggestive SNPs, the direction of the effect was not consistently in the same direction as was reported in the majority of the discovery studies, again indicating that these SNPs were potential false positives. We meta-analyzed the results from the sex-stratified discovery meta-analysis and the replication study in a combined meta-analysis. For males, five The genetic relationships were estimated from 301,115 directly genotyped autosomal SNPs that were available in both studies. All analyses controlled for age, study, and the first 10 principal components of the genetic similarity matrix of the combined sample of RS-I and STR. In the pooled sample we also controlled for sex. The results did not change markedly when 4 or 20 principal components were included; s g SNPs had lower p-values compared to the male discovery metaanalysis, although none reached genome-wide significance (Table S5). In the combined meta-analysis for females, we observed that one SNP, rs562487, had a smaller p-value in this combined metaanalysis; however, this SNP did not reach genome-wide significance (p = 4.01610 26 ; Table S6).
To identify novel genes that may be associated with selfemployment, we tested 17,697 genes for pooled males and females, 17,698 genes for males only, and 17,699 genes for females only, implying a significance level of p,2.8610 26 . None of the analyzed genes reached this predetermined significance level (Tables S8, S9, and S10). The gene with the lowest p-value was SLC15A3 for the pooled male and female analysis (p = 1.63610 24 ). For males only, the lowest p-value was for TMEM156 (1.61610 24 ), and for females only, the lowest p-value was for PCP4 (p = 4.70610 25 ).
We also sought to replicate the association that was reported by Nicolaou et al. [48] to exist between a common variant, rs1486011, which is located in the DRD3 gene, and the tendency to be an entrepreneur. The SNP was nominally significant in the discovery meta-analysis (p = 0.011; Table S11); however, most studies reported a positive effect of the C allele-opposite to that reported by Nicolaou et al. [48], corroborating the results from an earlier replication study [49]. We also sought to replicate this SNP in the sex-stratified discovery meta-analyses. In this analysis, we observed a certain degree of evidence for a positive effect of the C allele in males (p = 0.046; Table S11) but not in females (p = 0.112; Table S11).
Predicting self-employment from genotype data
We examined whether the results from the discovery metaanalyses could be used to predict self-employment in the replication study [76]. We pruned the set of autosomal SNPs to a subset of approximately 120,000 SNPs that are in approximate linkage equilibrium. In an initial prediction analysis, we included only the subset of these 120,000 SNPs that reached a 1% significance level. We calculated a predictive score for each individual in the replication study by determining, for each SNP, the product of the individual's number of effect alleles and the estimated regression coefficient from the discovery meta-analysis. This product was then summed across the included SNPs and divided by the number of included SNPs. We evaluated the predictive power of the SNPs by calculating the degree of variance in the tendency to engage in self-employment that was explained by the score and the area under the receiver operating characteristic curve (AUC). We repeated this prediction analysis eleven additional times, each time with a less stringent significance threshold required for a SNP to be included in the score. Hence, each time this analysis was performed, a larger subset of the 120,000 SNPs was analyzed.
For the pooled analysis of males and females (n = 2,589), the variance that was explained by the score reached a maximum of 0.184% when all SNPs were included (p = 0.039; Table S12). The scores for males only (n = 1,110) and for females only (n = 1,479) showed no evidence for association with self-employment (all p$0.144, Table S12). Furthermore, we did not observe a consistent positive relationship between the variance in the tendency to engage in self-employment that was explained by the score and the significance threshold p T (Figure 3).
Discussion
We present results from four methods of analysis, three of which are based on genome-wide molecular genetic data, to investigate the molecular genetic architecture of self-employment.
First, using a classical twin design, we report that 55% of the variance in the tendency to engage in self-employment is due to additive genetic effects, with higher heritability for males (67%) than for females (40%). Our estimates are in agreement with those of previous twin studies. These earlier studies suggested heritabilities of 48% in a sample of primarily female British twins [26] and of 38% in a sample of US twins [28]. In addition, Zhang et al. [27] estimated the heritability of current business ownership and selfemployment in a sample of Swedish twins and observed evidence of a significant additive genetic effect for females but not for males. Our results suggest significant heritability among males as well; however, the confidence intervals of the estimates are very wide for both our study and for that of Zhang et al. [27]. At least a portion of the differences between these two studies may be explained by imprecision and/or by the different samples and definitions of entrepreneurship that were used. Second, by applying a method that was recently developed by Yang et al. [52] to entrepreneurship, we estimate that approximately 25% of the variance in the tendency to engage in selfemployment (about half of the h 2 estimated in twin studies) could in principle be explained by the additive effects of common SNPs that are in linkage disequilibrium with the unknown causal variants. These results are in line with previous studies, which have estimated that common SNPs account for one-quarter to half of the narrow-sense heritability for height [52], intelligence [80,81], personality [51,82], several common diseases [83], schizophrenia [84], and recently for several economic and political preferences [22].
Several explanations may explain why the heritability estimate for self-employment using common SNPs is approximately half of the estimate that was obtained using the classical twin design. First, the causal variants may be in regions of the genome that are currently not covered by the available SNP arrays. Second, it is possible that the genotyped SNPs and the causal variants are not in complete linkage disequilibrium because, for example, the true causal variants have on average lower minor allele frequencies than the genotyped SNPs. Yang et al. [52] provide evidence for this in the case of human height. They estimated that 45% of the variance in height is accounted for by common SNPs, while the heritability of height is consistently estimated to be approximately 80%. The authors then developed a method that estimated the variance that was accounted for by common SNPs, assuming imperfect linkage disequilibrium between the genotyped SNPs and the unobserved causal variants. This method revealed that 84% of the variance in height, the complete heritability, could be explained by the causal variants. Twin and family studies do not suffer from this issue, as genetic relatedness is inferred from the expected relationships within the pedigree and include all of the additive genetic variation. Both of these explanations imply that the estimates that we obtained for self-employment using the more novel method are at the lower bounds of the heritability that is commonly estimated in twin and family studies. A third, alternative, explanation for the different results that were obtained using these techniques is that the twin-based heritability estimates are biased upwards because of, for example, genetic interactions [85] or a violation of the identical common environment assumption in twin studies [86].
Third, we perform the first meta-analysis of GWASs of an economic behavior (i.e., self-employment) using data from sixteen studies that together comprise approximately 50,000 participants. The discovery stage had 80% power to detect a variant at genomewide significance with a minor allele frequency of 0.25 and odds ratios of approximately 1.11 for pooled males and females, 1.15 for males only, and 1.17 for females only [87], assuming we had a non-noisy, harmonized measure of self-employment across studies. Yet, we do not identify genome-wide significant associations. This result suggests that there are no common SNPs for selfemployment with moderate to large effect sizes, thus placing an upper bound on the effect sizes of common SNPs that we can expect to exist. Gene-based tests for approximately 17,700 genes, including several candidate genes for entrepreneurship that have been previously suggested in the literature [48,79], do not reveal significant associations. In addition, we are unable to replicate a previously reported correlation, namely, rs1486011, a SNP that is located in the DRD3 gene. This common variant was identified by Nicolaou et al. [48], who reported its association with the tendency to be an entrepreneur. The non-replication of associations is common in candidate gene studies of human traits and behaviors. This failure to identify replicable associations is likely due to a combination of underpowered sample sizes (due to optimistic assumptions regarding plausible effect sizes) and publication bias [88]. Examples of non-replication of candidate genes studies on complex human traits include general intelligence [81], personality [89][90][91][92][93][94], and trust [95,96]. We therefore stress that caution is warranted when interpreting claims from candidate gene studies of SNPs or genes with strong effects on complex behavioral traits like self-employment.
Finally, we report that a genetic score that was estimated in our meta-analysis sample has only limited predictive power in our replication study. The variance that was explained by the score was always lower than 0.26%. However, this result does not contradict our finding that approximately half of the narrow-sense heritability can be explained by common SNPs. This latter heritability analysis uses the measured SNPs to estimate realized relatedness between individuals, and given the large number of SNPs in a dense SNP array, realized relatedness can be estimated fairly accurately. In contrast, estimating a strongly predictive score from a sample requires good estimates of the effects of individual SNPs. If our discovery sample was infinitely large, it would have been possible to precisely estimate all of the SNP effects and to obtain a score with the theoretically highest possible predictive power, as estimated using the Yang et al. [52] method. The smaller the discovery sample, the noisier the estimates of the individual SNP effects; therefore, the predictive power of the score will be lower [97,98]. Our estimates of the effects of the individual SNPs are still too imprecise to allow out-of-sample prediction with SNP data that would have practical utility. Together, our results demonstrate that common SNPs jointly account for a substantial share of the variance in the tendency to engage in self-employment (s g 2 /s P 2 = 25%). However, because we do not find specific SNPs in our large-scale meta-analyses of GWASs that examined self-employment, this heritability is not due to SNPs with moderate to large effects. A plausible interpretation of these results therefore appears to be that the molecular genetic architecture of self-employment is highly polygenic, implying that there are hundreds or thousands of variants that individually have a small effect and which together explain a substantial proportion of the heritability. We cannot rule out the possibility that rare genetic variants, or other, currently unmeasured, variants that are insufficiently correlated with the SNPs on the genotyping platforms, have large effects on an individual's tendency to be self-employed. However, if these genetic variants are rare, they would still not contribute a great deal to the population-based variance in self-employment, and large samples would still be required to identify these variants [51,83,99].
Our results are similar to those that have been reported for biologically more proximate human traits [51,52,[80][81][82] and diseases [76,83,84] for which a polygenic molecular genetic architecture has also been suggested. One implication of this similarity is that, with sufficiently large sample sizes, SNPs that are associated with self-employment-and possibly also other economic variables-can in principle be discovered, as has been the case for, e.g., height [100] and BMI [101]. However, a discovery sample of approximately 50,000 individuals is apparently still too small for a meta-analysis of GWASs on a biologically distal, complex, and relatively rare human behavior such as self-employment. A potential opportunity for future research are GWASs of endophenotypes such as risk preferences, confidence, and independence. The effect sizes of individual SNPs on these endophenotypes may be larger because of their greater biological proximity. However, these variables are difficult to measure reliably and not (yet) available in many genotyped samples.
Given the need for very large samples in meta-analyses of GWASs on complex traits, an important challenge of the present study was to identify a measure of entrepreneurship that is available in a sufficiently large sample. We opted to maximize the available sample size in this study and operationalized entrepreneurship as self-employment, which is also the most frequently used measure of entrepreneurship in the economics literature [102].
We included every study we were aware of in the analysis that included a measure of self-employment and which was willing to contribute data, although this approach necessitated that data from diverse populations (e.g., Eastern German self-employed individuals and US business owners) were pooled. The available measures of self-employment varied across studies, including different single-and multiple-item measures, data from standalone surveys, and data from repeated measures or retrospective employment histories of the participants. For a number of studies, this approach resulted in a lack of detailed and reliable data regarding work-life history. Substantial measurement error, especially with respect to the definition of the control group, was therefore unavoidable. Ideally, the control group would encompass only participants who had never been self-employed and who will never be self-employed. Such an analysis would have required data regarding the complete work-life history of participants and participants who had reached an appropriate age. However, only data regarding current employment status were available in the majority of the contributing studies. It is therefore possible that there was a certain degree of misclassification in the studies that included only single-item, single-response measures of self-employment, thereby adding noise to the phenotype definition and potentially reducing the statistical power with respect to association detection.
Statistical power may have also been reduced by heterogeneity within the case group, as this group comprised individuals who became self-employed for very different reasons. For example, certain individuals may have chosen self-employment because they had no viable alternatives in paid employment, whereas others may have done so because of their desire to pursue a business opportunity. The motivations, goals, and resources of these two groups of individuals are obviously very different, and the genetics underlying these various characteristics may likewise differ greatly. Unfortunately, more detailed information regarding the motivations, activities, and success of entrepreneurs was unavailable for most of the genotyped samples.
In general, GWASs face a practical trade-off between phenotype quality and sample size. Surprisingly, statistical power calculations suggest that studying a more noisy phenotype in a larger sample is often more likely to be successful than studying a perfect phenotype in a small sample. For example, assume that a common SNP exists with a minor allele frequency of 0.5 that increases the odds for all types of entrepreneurship by a factor of 1.13 on average (assuming 15% of the population are entrepreneurs and the data are population samples). The required sample size to detect this SNP with 80% power for a perfectly-measured outcome is approximately 30,000. Measuring entrepreneurship perfectly would require a lengthier survey that is administered more than once. Such a large genotyped sample with perfect measures of entrepreneurship does not currently exist. Smaller samples with perfect measures would be underpowered to detect the SNP. In contrast, if the available measures for entrepreneurship are noisy and have a test-retest reliability of only 0.6-which is typical for behavioral traits measured by brief surveys [103][104][105]280% power to detect this SNP requires a discovery sample of approximately 50,000 individuals. Thus, our study was wellpowered to detect effects of this magnitude even if there was substantial measurement error and noise in the data.
The results of our study have three implications for this future research agenda. First, the high share of variance in selfemployment that can be attributed towards interpersonal differences in common SNPs suggests that this research agenda is in principle feasible. Second, to investigate if and how genes that are related to economic variables influence medical outcomes, it will be necessary in the future to identify either the specific genetic variants that are underlying the heritability of economic variables (i.e., to investigate causal pathways from genes to medical outcomes), or to calculate genetic scores that have at least moderate out-of-sample predictive power (i.e., to investigate the medical consequences of a mismatch between genetic predisposition and economic outcomes). Even larger samples than what we had available in our present study will be needed to identify genome-wide significant SNPs and to estimate more accurate genetic scores for economic variables. Third, our results suggest that the effects of single SNPs on self-employment are likely to be very small. Given these effect sizes, statistical power calculations suggests that a research strategy that aims to maximize sample size by pooling data with slightly inaccurate measures of selfemployment is more likely to be successful than a research strategy that aims to collect perfect phenotype measures in a much smaller sample. If successful, this research could shed new light on the complex interaction of genes, environment, and personal choices on health and longevity. Table S8 Gene-based p-values for the top 25 genes associated with self-employment in the discovery metaanalysis for pooled males and females.
(DOC)
Table S9 Gene-based p-values for the top 25 genes associated with self-employment in the discovery metaanalysis for males only.
(DOC)
Table S10 Gene-based p-values for the top 25 genes associated with self-employment in the discovery metaanalysis for females only.
(DOC)
Table S11 Meta-analysis association results for SNP rs1486011 for pooled males and females, males only, and females only.
==
Domain: Biology Medicine Computer Science Economics
|
[{"added": "2016-11-08T18:56:27.780Z", "created": "2013-04-04T00:00:00.000", "id": "4000626", "metadata": {"extfieldsofstudy": ["Biology", "Computer Science", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1371/journal.pone.0060542", "pdf_hash": "233c72f8b58d4d1243d9a2714cbce30bee6c5759", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:4011", "s2fieldsofstudy": ["Economics"], "sha1": "854ef1c84764257eb22a238ee946bb8ccf32d8e9", "year": 2013}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Rapid Detection of Ebola Virus with a Reagent-Free, Point-of-Care Biosensor
Surface acoustic wave (SAW) sensors can rapidly detect Ebola antigens at the point-of-care without the need for added reagents, sample processing, or specialized personnel. This preliminary study demonstrates SAW biosensor detection of the Ebola virus in a concentration-dependent manner. The detection limit with this methodology is below the average level of viremia detected on the first day of symptoms by PCR. We observe a log-linear sensor response for highly fragmented Ebola viral particles, with a detection limit corresponding to 1.9 × 104 PFU/mL prior to virus inactivation. We predict greatly improved sensitivity for intact, infectious Ebola virus. This point-of-care methodology has the potential to detect Ebola viremia prior to symptom onset, greatly enabling infection control and rapid treatment. This biosensor platform is powered by disposable AA batteries and can be rapidly adapted to detect other emerging diseases in austere conditions.
Introduction
The 2014 Ebola virus outbreak is the largest in history, with widespread transmission in multiple West African countries and sporadic cases in Europe and North America [1][2][3][4]. The massive public health response has been limited, in part, by the inability to rapidly detect the presence of Ebola virus in potential patients living in remote areas [5]. The early symptoms of Ebola overlap with symptoms of endemic malaria and other febrile illnesses [6].
Rapid, point-of-care detection of the Ebola virus could enable early quarantine and halt future epidemics and pandemics [5,6]. While point-of-care nucleotide amplification tests exist [5], these are limited by the need for multiple reagents, refrigeration, and specialized personnel. Here, we present preliminary results of a Surface Acoustic Wave (SAW) biosensor with potential to detect Ebola virus in multiple, unprocessed sample types including blood, serum, saliva, and feces within 5-10 min.
Ebola virus is a class A select agent filovirus that was first identified in Zaire in 1976 and named after the River Ebola in Zaire [7]. Ebola outbreaks are sustained through person-to-person transmission from direct contact with infected people, bodily fluids, or contaminated clothes and linens [7]. Early detection of this highly contagious virus would allow for improved infection control measures and early treatment at specialized facilities. The current mortality rate from epidemic Ebola ranges from 40% to 90% and could be greatly reduced with early, point-of-care diagnostics [8,9]. Mathematical models suggest that the total direct costs of the present outbreak range from $82 million to $356 million in the three most effected countries; early diagnosis and treatment will also reduce aggregate healthcare costs [10].
Current approaches to Ebola identification include antigen-capture ELISA testing, IgM ELISA, RT-PCR, virus isolation, electron microscopy, and serologic testing for IgM or IgG antibodies. Quantitative measurement of the viral load also has prognostic importance, with a much higher case fatality rate in patients with viral loads over 10 million copies per milliliter [11]. Current tests for Ebola, especially PCR-based techniques, are generally limited by the need for added reagents, refrigeration, and the need for specially trained laboratory personnel. Lateral flow assays can provide rapid and fairly inexpensive qualitative results, but are unable to quantify viral load.
Our group has developed a point-of-care biosensor that employs surface acoustic waves for label-free pathogen detection without the need for any sample preparation [12][13][14]. This biosensor platform detects target antigens in the presence of confounding analytes and in various types of media, thus eliminating complicated sample preparation protocols. Such acoustic wave sensors can achieve limits of detection of <50 pg/cm 2 , which is an order of magnitude lower than detection limits for other marker-free systems such as optical surface plasmon resonance (SPR) and quartz crystal microbalance (QCM) devices [15].
With this sensor platform, viral detection is rapid (with acoustic wave signal detection within 2 min), and the entire detection protocol (from collecting the test sample to obtaining a positive or negative test result) requiring about 5-10 min. Such technological benefits would be ideal for rapid, point-of-care identification and diagnosis of the Ebola virus, especially in environments with minimal or overloaded infrastructure, such as in public health or emergency response situations. Here, we discuss initial modifications of the biosensor to demonstrate detection of Ebola and its potential utility to combat this recurrent epidemic.
Fabrication and Functionalization of the SAW Biosensor
The sensor chips were prepared using lithium tantalate (36°, y-cut, x-propagating LiTaO3) wafers by lithographic deposition and patterning of inter-digital transducers (IDT) and waveguide layers as previously described [12]. Briefly, cleaned LiTaO3 wafers were patterned with four IDT patterns with sets of transducers and delay lines using negative tone photoresist AZ2020 (AZ Electronic Materials, Branchburg, NJ, USA). Then, a metallization step was performed with 5000 Å aluminum using an electron-beam evaporator (Temescal, Wilmington, MA, USA). The wafer was placed in an acetone bath to lift off the photoresist and any excess aluminum, followed by an acetone spray as needed and by the following rinses in methanol, isopropyl alcohol, and deionized water. This process was repeated for ground metallization, bus lines, and contact pads. Next, a waveguide layer of 5000 Å silicon dioxide (SiO2) was deposited as a film on the wafer using lift-off plasma enhanced chemical vapor deposition (Oerlikon Versaline, Pfaeffikon, Switzerland). The oxide was coated with hexamethyldisilazane in a vacuum oven, and positive tone photoresist AZ4330 (AZ Electronic Materials) was used to form a photoresist mask on the wafer having exposed portions. The exposed portions of SiO2 was etched with reactive ion etching to access electrical contact pads. Finally, the resultant wafer was diced to form individual chips, in which photoresist from dies was removed by rinsing in acetone, methanol, and isopropanol.
Antibodies were individually patterned on sensing lanes as previously described [16]. Briefly, sensor chips were coated with toluene and 3-glycidyloxypropyl trimethoxysilane (90%/10%) in an oven at 60 °C for 1.5 h and then rebaked at 100 °C for 1 h. Each lane was coated with antibodies at a concentration of 10 µg/mL in phosphate-buffered saline (PBS). The following antibodies were used: mouse monoclonal antibodies (IgG2a isotype) specific for Ebola (AB-EB-MAB1, anti-Ebola virus monoclonal antibody 1; BEI Resources, Manassas, VA, USA); and mouse IgG1 antibody isotype control (F(ab')2 fragment) (ab37426, monoclonal isotype control; Abcam, Cambridge, MA, USA). Typically, two lanes were functionalized with antibodies specific for the target analyte, and two lanes were functionalized with isotype control antibodies; the latter are referred to as reference lanes.
Provenance and Handling of Ebola Virus Strain Zaire (Mayinga) Inactivated
The Ebola virus antigen sample (NR-31807; BEI Resources, Manassas, VA, USA) consists of inactivated and highly disrupted viral particles and was used as provided. Fully intact Ebolavirus is highly pathogenic and designated BSL-4 material. For research purposes, the supplier provides inactivated and disrupted Zaire (Mayinga) Ebola virus (BSL-1 material), which was prepared as follows. As stated by the supplier, Zaire (Mayinga) Ebola virus from infected Vero E6 pellets was suspended in 50 mM sodium borate, gamma irradiated (5 × 10 6 rads total dose) on dry ice, and sonicated. Culture cell debris were removed by centrifugation. The virus was confirmed non-viable (killed or inactivated) by inoculation of cell culture (10 days on Vero cells) followed by a second passage (10 days on Vero cells). The concentration of the Ebola virus antigen sample was provided in plaque-forming units (PFU/mL) by the supplier, indicating the concentration of viable viral particles prior to inactivation, and allowing for comparison across different animal models and experimental conditions [17].
The Ebola antigen samples were analyzed by scanning electron microscopy at the University of New Mexico Health Sciences Center Electron Microscopy Facility (Albuquerque, NM, USA) to roughly determine the size distribution of the degraded viral particles. Ebola samples were stored in aliquots of 50 µL in screw cap safety microtubes at −80 °C until ready for use.
When ready to be tested, the samples were thawed on ice and supplemented with phosphate buffered saline (PBS; 137.0 mM NaCl, 2.7 mM KCl, 10.0 mM Na2HPO4·2H2O, 2.0 mM KH2PO4, pH 7.4 with HCl). All Ebola antigen samples were prepared fresh for each experiment.
Virus Detection Using the SAW Biosensor
Detection of Ebola Zaire antigen was conducted in BSL-2 certified biosafety cabinets. Data expressed as phase shift (Δφ, expressed in degrees) were recorded with a custom acquisition program developed using Visual Studio (Microsoft) as previously described [12]. In brief, data from each IDT delay line were collected for each lane and acquired simultaneously as a function of time of acquisition. The biosensor supports a surface shear wave, and the IDT detects the wave by transducing the mechanical wave into an electrical signal. This electrical signal (expressed as voltage information) was converted to phase (φ, expressed in degrees) using the custom acquisition program. Data from reference lanes were subtracted from the data from Ebola test lanes at each time point to determine the specific Ebola signal. The data for quantitative measurement of specific Ebola particle detection refer to the phase shift (Δφ), which corresponds to the difference between the reference and Ebola signals that is measured continuously after addition of a viral antigen sample Ebola antigen samples of various concentrations were prepared in 100 µL of PBS and applied to the biosensor. Binding of viral antigens to the sensor surface resulted in a net phase shift of the signal Δφ (expressed in degrees on the y-axis) along the time plot (on the x-axis). While phase shift differences associated with specific Ebola signal are apparent and stable within seconds of sample addition, we continue measuring up to 5 min after sample addition to assure signal stability and reproducibility. All measurements represent multiplicates of at least three, and data points are represented as means ± standard error of the mean (SEM).
Results and Discussion
We present here the initial development and characterization of a biosensor for point-of-care, quantitative detection of the Ebola virus. This device enables log-linear, concentration-dependent detection of disrupted viral particles in the clinically relevant range. Overall sensitivity and limit of detection are dependent on the mass of the species that is bound to the sensor surface; this biosensor is likely to be orders of magnitude more sensitive for intact, infectious Ebola viral samples [15].
Design Elements of Surface Acoustic Wave Biosensor
The biosensor was adapted to allow for sensitive detection of the Ebola virus. The biosensor has a planar, piezoelectric substrate containing inter-digital transducers (IDTs) [12,13]. The piezoelectric substrate propagates horizontally polarized surface shear waves, and such waves are induced by applying an alternating voltage to the IDTs at a high frequency (generally between 80 and 400 MHz). The surface shear waves are characterized by a particular resonant frequency that is sensitive to changes on the sensor surface.
Specificity for the target bioagents is obtained by functionalizing the surface of the substrate (Figure 1a). Here, the surface of the piezoelectric substrate was sensitized with a monoclonal antibody specific for Ebola virus. A binding event causes a change in surface mass and results in a phase shift of the signal wave propagating across the sensor surface. Biochemical interactions occurring on the sensor surface can be quantified by measuring this change in phase shift (Δφ).
The overall schematic concept of the biosensor device for detecting the Ebola virus is shown in Figure 1. The biosensor includes a test lane including antibodies selective for the target bioagents (i.e., the Ebola virus in this study), as well as a reference lane including a control IgG1 antibody (Figure 1a). Measurements of Δφ were performed for both the test lane and the reference lane by use of an output interface device (connected to the SAW biosensor) and a laptop computer (Figure 1b). (b) The detection system for research use with fluidic housing surrounding the biosensor, an output interface device connected to the biosensor, and a laptop computer. The sensor is powered by disposable AA, and further miniaturization of the output interface is possible for field use. In addition, the fluidic housing and biosensor can be provided as a disposable module that can be detached from the output interface and then decontaminated prior to disposal.
Imaging Analysis of the Ebola Zaire Antigen Sample
Fully intact Ebolavirus is highly pathogenic and designated BSL-4 material. For this particular study, we employed inactivated and disrupted Zaire (Mayinga strain) Ebola virus (BSL-1 material), which is available for research purposes under non-BSL-4 conditions. Thus, transmission electron microscopy studies were conducted to identify the size distribution of the fragmented viral particles. Representative images are provided in Figure 2. Shown is a TEM image of a 1:20 diluted sample (with PBS), which includes a majority of particles that are less than 10 nm and the presence of larger aggregates between about 50 nm to 300 nm (Figure 2a). Rare filamentous structures, with diameter of about 30 nm (Figure 2b) were also noted. Intact Ebola virus has a filamentous structure with diameter of 80 nm and length greater than 950 nm [18]. TEM of the inactivated virus sample shows highly disrupted, fragmented particles that form aggregates (Figure 2a), rather than the elongated filaments observed in intact filovirus. When filamentous structures were observed (Figure 2b), the cross-section diameter was less than half of the diameter expected for intact filovirus [18]. While the predominant species in the inactivated virus sample are disrupted particles, the biosensor phase shift response is likely to be much greater for an infectious Ebola sample containing intact virus. Based on our previous study of HIV detection [16] and the relative masses of intact HIV and Ebola [18,19], we would estimate a 5-10× greater sensitivity for intact Ebola virus compared to intact HIV. As a first approximation, we assume a log-linear correlation between mass and phase shift. As particles of greater mass will provide a greater response, the sensor is likely to detect more massive, intact viral particles at lower concentration.
SAW Biosensor Detection of the Ebola Zaire Sample
In analogy to our previous report on the detection of HIV-1 and HIV-2 [16], we functionalized the SAW biosensor with monoclonal antibodies specific for Zaire (Mayinga) strain of Ebola Virus. Ebola virus antigens were detected in PBS solutions over a 2.5 log reference range at concentrations corresponding to 1.0 × 10 4 PFU/mL to 3.0 × 10 6 PFU/mL prior to virus inactivation. The lowest concentration is below the average viremia level of 3 × 10 4 RNA copies per mL observed in Ebola patients on the first day of disease symptoms [20].
Representative phase shift data are provided in Figure 3. The IgG control lane accounts for matrix effects and non-specific binding of potentially interfering species.
Detection of Ebola virus resulted in a concentration-dependent increase with Δφ values ranging from 0.20 ± 0.04 to 4.46 ± 0.86, corresponding approximately to 1.6 × 10 4 PFU/mL to 6.5 × 10 6 PFU/mL ( Figure 4). There was a log-linear relationship between viral load and Δφ for this concentration range of viral particles with a correlation coefficient R 2 of 0.92. The linear range and correlation coefficients compare favorably to those recently reported for qRT-PCR [21]. A limit of detection (LOD) of 1.9 × 10 4 PFU/mL was calculated for Ebola virus by linear regression and using the average background noise Δφ value of 0.31. These results indicate that the prototype SAW biosensor rapidly detects the Zaire strain of Ebola antigens in a defined buffer with detection limits below the average viremia level at onset of clinical symptoms.
Conclusions/Outlook
We have demonstrated an adaptable, label-free sensing system for the rapid detection of bioagents and show its use to detect the Ebola Zaire virus. We have previously shown the effectiveness of this system in detecting the Bacillus anthracis bacteria simulant, Bacillus thuringiensis [13]; the Coxsackie virus [12]; the Sin Nombre hantavirus [12]; the Francisella tularensis bacteria [22] and the Human Immunodeficiency Virus types 1 and 2 [16]. This study further extends the capability of this SAW biosensor platform to accommodate the rapid, label-free, and specific detection of the Ebola Zaire virus.
As fully intact Ebola virus is highly contagious, we used an inactivated virus in order to conduct this study under non-BSL-4 conditions. However, in a first responder or real world scenario, the sample would not necessarily have to be inactivated and could be tested directly, providing the sensor was contained in a controlled environment.
The SAW biosensor described here provides the basis for a rapid and specific response to Ebola outbreaks and other emerging diseases. This point-of-care sensor will provide rapid diagnosis and improved infection control, dramatically decreasing the human and economic costs of this disease. Further work will focus on portability and optimization for field use. A disposable cartridge can be employed to house the piezoelectric substrate and fluidic channels to guide the test sample to the substrate. The user interface can be readily simplified to provide a simple positive/negative result to a field worker on the front lines of the next epidemic.
Sample preparation is not required for this label-free sensing methodology, simplifying use in field conditions without centralized laboratories or refrigeration. While additional testing can be conducted to verify the detection limits for Ebola Zaire in complex media, such as blood, serum, or saliva [23], we believe that Ebola virus detection in such media will be guided by our past studies that show effective detection of other viruses in complex solutions (e.g., serum, plasma, river water, and sewage effluent [12,14,16]). In summary, the SAW biosensor is a versatile platform that shows promise to revolutionize rapid pathogen detection and enable early treatment in public health and emergency responses.
==
Domain: Biology Medicine Computer Science
|
[{"added": "2016-05-12T22:15:10.714Z", "created": "2015-04-01T00:00:00.000", "id": "789840", "metadata": {"extfieldsofstudy": ["Biology", "Computer Science", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1424-8220/15/4/8605/pdf", "pdf_hash": "60f9b30b0f80eb5d32264ddddb95158fa9066296", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:556", "s2fieldsofstudy": ["Medicine"], "sha1": "60f9b30b0f80eb5d32264ddddb95158fa9066296", "year": 2015}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
A long read mapping method for highly repetitive reference sequences
About 5-10% of the human genome remains inaccessible for functional analysis due to the presence of repetitive sequences such as segmental duplications and tandem repeat arrays. To enable high-quality resequencing of personal genomes, it is crucial to support end-to-end genome variant discovery using repeat-aware read mapping methods. In this study, we highlight the fact that existing long read mappers often yield incorrect alignments and variant calls within long, near-identical repeats, as they remain vulnerable to allelic bias. In the presence of a non-reference allele within a repeat, a read sampled from that region could be mapped to an incorrect repeat copy because the standard pairwise sequence alignment scoring system penalizes true variants. To address the above problem, we propose a novel, long read mapping method that addresses allelic bias by making use of minimal confidently alignable substrings (MCASs). MCASs are formulated as minimal length substrings of a read that have unique alignments to a reference locus with sufficient mapping confidence (i.e., a mapping quality score above a user-specified threshold). This approach treats each read mapping as a collection of confident sub-alignments, which is more tolerant of structural variation and more sensitive to paralog-specific variants (PSVs) within repeats. We mathematically define MCASs and discuss an exact algorithm as well as a practical heuristic to compute them. The proposed method, referred to as Winnowmap2, is evaluated using simulated as well as real long read benchmarks using the recently completed gapless assemblies of human chromosomes X and 8 as a reference. We show that Winnowmap2 successfully addresses the issue of allelic bias, enabling more accurate downstream variant calls in repetitive sequences. As an example, using simulated PacBio HiFi reads and structural variants in chromosome 8, Winnowmap2 alignments achieved the lowest false-negative and false-positive rates (1.89%, 1.89%) for calling structural variants within near-identical repeats compared to minimap2 (39.62%, 5.88%) and NGMLR (56.60%, 36.11%) respectively. Winnowmap2 code is accessible at [URL] in single-molecule sequencing technologies have inspired community e↵orts to produce high-quality human genome assemblies with accurate resolution of repetitive DNA. The complete, gapless, telomere-to-telomere (T2T) assembly of a human chromosome X is a recent breakthrough that involved assembling a 3.1 Mbp long centromeric satellite DNA array [1]. Similarly, a T2T assembly of human chromosome 8 spanned a 2.1 Mbp long centromere and the 0.6 Mbp long defensin gene cluster for the first time [2]. Such developments are steering genomics into an exciting new era where repeats that were previously thought intractable (e.g., segmental duplications, satellite, and ribosomal DNAs) will no longer remain out of reach. Genomic variants, expected to be in higher abundance within and near repetitive DNA, may be contributing to complex traits such as lifespan and disease [3][4][5]. PacBio and Oxford Nanopore (ONT) sequencing, due to their orders of magnitude longer read lengths than Illumina, can easily span many common duplications (e.g., LINEs) in the human genome. However, accurate long read mapping within > 100 kbp-sized repeats remains challenging.
Prior algorithmic developments for long read mapping have been crucial to resolving many repetitive sequences and complex variants. As such, several specialized methods have been published to improve long-read match seeding and extension [6][7][8][9][10][11][12][13][14][15][16][17]. The extension stage typically involves the optimization of a base-to-base alignment score which rewards matching bases while appropriately penalizing gaps and mismatches. However, these alignment scores do not always favor the correct loci in long near-identical repeats, because reads that include non-reference alleles will be penalized and their true loci may score worse than other copies of the repeat. Occurrence of this allelic bias (a.k.a. reference bias) and its e↵ect on estimates of variation and allele frequencies has been extensively discussed in the literature [18][19][20][21][22][23]. An analogous problem also occurs during genome assembly validation and polishing when reads are mapped back to a potentially erroneous draft assembly [1,24]. Compared to point mutations or short indels, structural variants (SVs) a↵ect more bases in the genome due to their larger size, and therefore, are bigger contributors to allelic bias. Most existing solutions to address this bias involve modifying the reference sequence, e.g., by adopting a graph-based representation which incorporates known genomic variation [20,[25][26][27]. While this remains a promising and complementary direction, here we seek to address allelic bias by developing a new long read mapping method that is robust to the presence of novel variation. In our proposed method, referred to as Winnowmap2, we introduce the concept of minimal confidently alignable substrings (MCASs), which are minimal-length read substrings that align end-to-end to a reference with mapping quality score above a user-specified threshold. Through MCASs, we can identify the correct mapping target of a read by considering the substrings that do not overlap non-reference alleles. In theory, the mapping quality of each substring quantifies the probability that it is correctly placed [28]. This framework draws advantage from paralog-specific variants (PSVs) [29,30], which allow us to di↵erentiate repeat copies from each other and identify the MCAS. We provide a formal definition of MCAS, an exact dynamic programming algorithm to compute them, as well as fast heuristics to scale this method to large mammalian genomes.
Winnowmap2 was empirically validated using both simulated and real human genome sequencing benchmarks. In both cases, we judge Winnowmap2 along with the currently available long read mappers by the downstream accuracy of SV calls produced by the SV caller Sni✏es [10]. The simulation uses SURVIVOR's SV benchmarking tool [31] which mutates a reference sequence (in our case, the first two completed human chromosomes 8 and X). Winnowmap2 alignments consistently enabled the most accurate SV calls using both ONT and PacBio HiFi data at varying coverage levels, when compared to other commonly used long read mappers. Using simulated HiFi reads from chromosome 8 (146 Mbp) sampled at 40x coverage, Winnowmap2, minimap2 and ngmlr achieved accuracy scores, i.e., false-negative and false-positive rates (FNR, FPR) of (0.09%, 0.18%), (3.36%, 0.93%) and (3.64%, 2.93%) respectively. Winnowmap2's improved handling of allelic bias was particularly evident within 7 Mbp of the most repetitive regions of chromosome 8, achieving (FNR, FPR) scores of (1.89%, 1.89%) in these regions compared to (39.62%, 5.88%) for minimap2 and (56.60%, 36.11%) for ngmlr, respectively. Winnowmap2 also achieved favorable accuracy on the Genome in a Bottle (GIAB) benchmark set [32], which excludes long > 10 kbp-sized repeats of the human genome.
Results
An overview of the Winnowmap2 algorithm. If an error-free read is simulated directly from a reference, then its correct mapping to that reference computed using a reasonable pairwise sequence alignment algorithm is naturally guaranteed to have the highest score. However, this guarantee does not hold if the same read is mapped to an alternate reference. Consequently, using a pairwise sequence alignment scoring system to judge the best mapping candidate is sub-optimal, and this is particularly true while mapping reads to highly repetitive sequences. Regardless of the type of scoring function used, e.g., with either a linear or an a ne gap penalty, the function would also penalize variant-induced di↵erences between the sequenced individual and the reference sequence. In cases where one of the repeat copies in a reference sequence contains a di↵erent allele from the sequenced individual, reads may achieve a better alignment score against an incorrect repeat copy ( Figure 1). An ideal scoring system should ignore non-reference bases when computing an optimal alignment, but these are typically unknown a priori.
Like most read mappers, Winnowmap2 follows a seed-and-extend workflow. The seeding step reuses Winnowmap's weighted minimizer sampling [14], which yields an accuracy improvement over the standard minimizer technique [33]. Winnowmap2's extend stage introduces a novel heuristic to tackle allelic bias. We split the extend stage into two steps. The first step involves identifying minimal confidently alignable substrings (MCASs) from each read to a reference. Informally, an MCAS at position i of a read refers to the minimum length read substring starting at the position i that achieves a 'unique' end-to-end alignment to a reference locus (see Methods for a formal definition). Here, uniqueness of an alignment is evaluated using a mapping quality (mapQ) score [28] that reflects the score gap between the best and second-best alignment candidates for a substring. Accordingly, an MCAS is valid if its alignment achieves a mapQ score above a user-specified threshold. A read can have as many MCASs as its length. By using MCAS alignments, read bases on either side of a variant can map uniquely to their correct reference loci as they can be scored independently from non-reference bases ( Figure 1).
Starting from any position of a read, minimal length of the MCAS is ensured by iteratively increasing substring length and checking whether its maximum scoring alignment to a reference satisfies the mapQ cuto↵. Suppose a read is sampled from a repetitive region, the frequency of PSVs at its correct mapping loci helps determine the length of an MCAS. The higher the number of PSVs, the smaller the length of the MCAS, because its mapQ cuto↵ will be satisfied at an earlier iteration with fewer aligned bases. Similarly, better raw read accuracy also leads to shorter MCAS Fig. 1. a. Illustration of allelic bias in near-identical genomic repeats. Paralog-specific variants (PSVs), indicated using colored '+' markers, are the variants that are unique to a specific repeat copy in an ancestral human genome. Mutations in the reference sequence are indicated using 'x' markers. Long reads can be mapped to an incorrect repeat copy if the best mapping is decided by pairwise sequence alignment score. b. MCAS alignments map to correct loci on the reference. An MCAS is a carefully selected substring of a read. By excluding non-reference alleles, this approach reduces allelic bias. c. MCAS computation is shown using a simple toy example. Starting from several positions in a read, we identify minimum-length substrings that can be uniquely mapped to a reference. Uniqueness of an alignment is determined by using its mapping quality score.
lengths, since more PSVs will be matched by a more accurate sequence. Shorter MCAS lengths help not only in terms of the runtime with fewer iterations spent but also in terms of accuracy, as MCASs are less likely to overlap non-reference bases.
Computing all MCAS alignments from a read in an exact manner can be computationally prohibitive (see Methods for complexity analysis). To bypass this issue, we rely on banded-alignment and mapQ scoring heuristics from minimap2 [12] to compute each MCAS. For the sake of e ciency, we avoid evaluating MCAS alignments from each consecutive position in a read. Rather, we identify MCAS alignments from a subset of positions that are equally spaced (e.g., 500 bp apart).
The final step in Winnowmap2 is to consolidate a read's MCAS alignments into a final alignment output. For various reasons (e.g. sequencing errors, approximation of mapQ computation, and complex sequence variants) some MCASs may be incorrectly mapped. During the consolidation step, we extract anchors that participate in each MCAS alignment, and re-execute the chaining and alignment extension algorithm by using the complete set of anchors to output a final alignment. A few anchors from false MCAS mappings are filtered out during the anchor chaining process. We will empirically show that the proposed strategy improves mapping accuracy in repetitive DNA while remaining highly scalable.
Advantage of Winnowmap2 illustrated using the -defensin gene cluster. We visualize the advantage of Winnowmap2 method by using the beta-defensin gene cluster on human chromosome 8 as an example. The 7 Mbp beta-defensin locus (chr8: 6,300,000-13,300,000) of the human genome is known to be a hotspot of copy-number variation [34]. In the sequenced CHM13 human cell line, this locus spans three large (> 500 kbp) segmental duplications [2]. To evaluate long read mapping accuracy at this locus, we simulated ONT reads from chr8 at 40x sequencing coverage by using NanoSim [35] (Methods). In addition, we artificially mutated chr8 by adding a 1 kbp deletion variant at position 12,000,000. This locus was chosen for our illustration as it overlaps with one of the three duplications. If mapped correctly, the 1 kbp simulated deletion in the reference should appear as a 1 kbp-long insertion in the overlapping read alignments.
Illustrating allelic bias Figure 2 shows an IGV visualization of primary alignments computed by Winnowmap2 and three other long read mapping tools NGMLR, minimap2 and graphmap. Among the four methods, Winnowmap2 achieved the expected mapping coverage in this region with most read alignments showing the expected insertion call. The other tools mapped fewer reads successfully, resulting in reduced coverage and poor read mapQ scores. When these alignments were used as input to Sni✏es, only Winnowmap2 alignments resulted in the true SV call. NGMLR, minimap2 and graphmap rely on pairwise sequence alignment scores across the full length of the read when choosing the best mapping target. Due to the large deletion penalty levied at the mutated (but correct) locus, the majority of reads were incorrectly mapped to the other two duplications. Among the three methods, NGMLR showed the least bias, but most of its correct alignments were associated with poor mapQ scores (< 10). A low mapQ score indicates a marginal alignment score di↵erence between the best and the second-best mapping candidate, and therefore, the read alignment may not be considered by the variant caller. This result illustrates the previously discussed limitation of using pairwise alignment scores to rank candidate alignments in genomic repeats. The use of MCASs in Winnowmap2 enabled correct read placements in this case. A few MCAS alignments computed by Winnowmap2 in this region are visualized as a dot-plot in Supplementary Figure S1.
Evaluation using a simulated benchmark and T2T human chromosomes. We simulated long reads, both HiFi (using PBSIM [37]) and ONT (using NanoSim [35]), at coverage levels of 20x and 40x from T2T assemblies of chromosome 8 (146 Mbp) and chromosome X (154 Mbp) respectively (Methods). To evaluate how well Winnowmap2 addressed allelic bias, we also simulated 1100 structural variants, including both indels (1000) and inversions (100) of size 1 kbp, in each reference chromosome sequence by using the SURVIVOR benchmarking tool [31]. Both the SV simulation and evaluation of variant sets against the ground truth were done using SUR-VIVOR (Methods).
We evaluated Winnowmap2, Winnowmap, minimap2 and NGMLR in this experiment to check their false-negative and false-positive rates (FNR, FPR), as well as runtime and memory requirements. The long read mappers produced SAM-formatted alignments, which were then fed to Sniffles [10] to compute SVs. A false negative indicates that a true SV is not supported by read alignments whereas a false positive indicates that a false SV is supported. As such, these statistics are good indicators of the correctness of read alignments. We also performed a de novo repeat annotation of each reference sequence (chr8 and chrX) by using Mashmap [38] to identify repetitive sequence intervals of length 10 kbp and identity 95% (Supplementary Figure S2). The identified repetitive intervals constitute a notable portion of the two chromosomes; 4.8% in chr8 and 6.9% in chrX. This allowed us to separately evaluate the accuracy of read mappers in near-identical repeats where Winnowmap2 is expected to perform better. c. d.
winnowmap2 winnowmap ngmlr minimap2 False positive rate (%) Fig. 3. False negative and false positive rates achieved by SV calls of four mapping methods: Winnowmap2, Winnowmap, minimap2 and NGMLR. The top two plots show accuracy statistics over T2T chromosomes 8 and X whereas the bottom two plots show the statistics within only the most repetitive intervals of these chromosomes. Winnowmap2 alignments enabled the most accurate Sni✏es SV calls with the least FNR and FPR scores. Note that y-axis scales di↵er in these plots. Figures 3a, 3b show the accuracy statistics of the four mapping tools. Winnowmap2 achieved the best FNR and FPR for both the HiFi and ONT read sets. Winnowmap2 FNR and FPR scores consistently stayed below 3% and 0.3% respectively. Most of these gains appear in repetitive sequences of the two chromosomes as evident from our accuracy evaluation within only the repetitive regions (Figures 3c, 3d). Winnowmap2 succeeds in addressing allelic bias in these regions by pre-serving good accuracy in complex repeats where the other tools struggle. These gains were made uniformly over all SV types-insertions, deletions and inversions that were simulated (Supplementary Table S1). When increasing coverage from 20x to 40x, FNR generally reduces for all methods as better sensitivity is naturally expected with higher sequencing coverage.
The Winnowmap2 implementation is optimized to run fast while using less memory ( Figure 4). As several substring alignments need to be identified from a single read, it requires execution of alignment routines several times rather than just once. In this experiment, minimap2 consistently used the least time, but Winnowmap2 is competitive as its runtime was consistently lower than NGMLR and roughly double minimap2's runtime. Evaluation using Genome in a Bottle benchmark. Evaluating mappers on real sequencing data is challenging without a known truth. The Genome in a Bottle (GIAB) Tier1 v0.6 benchmark set provides a high-quality characterization of SVs in the Ashkenazi cell line HG002 relative to the GRCh37 human reference. This call set encompasses 2.51 Gbp of the genome and includes 5262 insertions and 4095 deletions [32]. It excludes SVs overlapping segmental duplications and tandem repeats greater than 10 kbp. Nevertheless, this experiment was useful to validate that Winnowmap2 also achieves good mapping accuracy on real data within the commonly studied regions of the genome. Here we mapped three publicly available HG002 long read sequencing sets: HiFi (14-15 kbp library, 35x), ONT (Guppy 3.6.0, 35x) and ONT (Guppy 3.6.0, 50x) to GRCh37, and compared results with minimap2. Similar to our simulated benchmark, variants were called using Sni✏es. Winnowmap2 achieved slightly better precision and similar recall scores compared to minimap2 ( Figure 5), with similar runtime and memory requirements. We also observed that both Winnowmap2 and minimap2 achieved better SV accuracy using ONT data over HiFi with equal 35x coverage.
Discussion and Conclusions
Availability of long-range sequencing technologies makes it feasible to resolve large mega-base sized near-identical duplications in the human genome, a feat that was impossible to achieve using short reads alone [39][40][41][42][43]. These regions include recently diverged segmental duplications, ampliconic gene arrays, rRNA genes, and centromeres, all of which play important functional roles in the genome and all of which go largely unstudied by current variation analyses. As human reference gaps associated with these regions are progressively being resolved, this opens up the opportunity to expand the resolution of resequencing approaches. In this work, we highlighted that allelic bias becomes a major challenge for accurately mapping reads to repetitive reference sequences. This challenge a↵ects the accuracy of existing mappers because classic pairwise sequence alignment scoring schemes are not an ideal mechanism to identify the correct mapping target in a repetitive sequence. In Winnowmap2, we have implemented a new idea based on minimal confidently alignable substrings that can be mapped independently of non-reference bases, thus alleviating allelic mapping bias.
We highlighted the advantages of Winnowmap2 by demonstrating its superior downstream variant call accuracy compared to commonly used long read mappers. In particular, Winnowmap2 enabled notable gains in SV calling accuracy within the repetitive regions of human chromosomes. Although here we focused on structural variants, it is natural to expect that Winnowmap2's superior mapping accuracy will also benefit SNP and short indel variant calling. Prior studies have suggested high enrichment of SVs in repetitive regions that currently correspond to unresolved gaps in the human genome reference [3,44,5]. This underscores the importance of understanding how these regions di↵er between individuals.
Further algorithmic improvements will be needed to improve read alignment accuracy. In particular, it remains challenging to align bases precisely when multiple SVs are clustered in close vicinity. Our simulation made use of SURVIVOR, which simulates SVs at uniformly random positions in a reference sequence and could be an over-simplification of real data. In addition, read mappers and variant callers still remain limited in their ability to handle nested variation and other forms of complex rearrangements [45].
Methods
Minimal confidently alignable substring (MCAS). MCASs distinguish Winnowmap2 from previous read mapping methods. Prior to defining an MCAS, we formalize when we can confidently say that an alignment of a substring is correct. In practice, this confidence is derived using the score di↵erence between the best scoring alignment and other candidate alignments. The mapping quality (mapQ) score was originally defined to address this problem [28], but the existing mathematical definition is restricted to short reads because alignments were assumed to be ungapped. However, the majority of long read sequencing errors are indels and the longer reads are more likely to span structural variants. When allowing for indels, adjacent mapping loci in a reference can no longer be considered independent, as in prior models. Accordingly, we propose the following formulation.
Given a query string S and a reference R, the top scoring end-to-end (a.k.a. semi-global) alignment candidates of string S to R can be directly computed in O(|S| · |R|) time. From a pairwise alignment of S to R, we can identify the set of matched base positions between them. For instance, this set would include a tuple (i, j) if character S[i] is matched to character R[j]. We say that two alignment candidates do not overlap if and only if their corresponding sets are disjoint. We associate our confidence with the best-scoring alignment of string S to reference R if and only if its second-best non-overlapping alignment candidate has a score < ⌧ · opt, where opt refers to the optimal alignment score and ⌧ 2 (0, 1) is a user-specified parameter.
Let Q be a long read sequence. A minimal confidently alignable substring MCAS(i) of read Q refers to the shortest substring starting at position i that has a confident end-to-end alignment to reference R. For a given read Q, we seek MCAS(i) 8 0 i < |Q|, and their corresponding alignments. MCASs can have variable lengths and can overlap one another. Note that existence of MCAS(i) depends on whether it is possible to satisfy the confidence criteria. In the worst case where two repetitive regions lack any PSV (i.e., 100% identical duplicates), then a read sampled from either repeat copy will not contain an MCAS. The rationale of introducing the MCAS idea is to address allelic bias; whereas a non-reference SV allele will cause mis-alignment in the traditional approach, the MCASs are treated independently and those neighboring the SV will remain una↵ected ( Figure 6). better score better score alignment Fig. 6. Illustration of MCASs using a DP alignment scoring matrix. Similar to Figure 1, PSVs are shown using colored '+' markers. The left figure highlights the e↵ect of allelic-bias on read alignment scores. The score of a true alignment spanning non-reference alleles can be lower compared to the score of an incorrect alignment. On the right side, an MCAS of a read which does not span non-reference alleles can achieve correct and unique placement to a reference.
Considering the issue of allelic bias, it is also desirable to enforce a maximum length parameter for valid MCASs because long MCASs again become vulnerable to allelic bias. By default, we set the maximum length parameter to 8 kbp for HiFi reads and 16 kbp for ONT reads based on our experimental observations. As such, the maximum length of a valid MCAS is a constant. The lemma below summarizes the asymptotic complexity to compute MCASs. Proof. Assume any appropriate linear or a ne gap scoring function is being used. Denote p th character in read Q as Q[p] and a substring ranging from positions p to q as Q[p, q] with both ends inclusive. Consider the following algorithm to compute MCAS(i). Compute semi-global DP alignment of Q[i, i+j] to reference R while iterating the variable j from 0 to c. Here c is the maximum length allowed for a valid MCAS. Computing a row of the alignment scoring table requires O(|R|) time. As a new row is computed in an iteration, we need to check whether the best-scoring alignment satisfies the confidence criteria, i.e., whether its score compared to the second-best non-overlapping alignment exceeds by a user-specified threshold. To check this, we require the following additional steps.
Select the cell in the current row associated with the highest score. Next, we need to locate the maximum score achieved by a non-overlapping alignment. Assuming gap penalty is positive, then there can be at most O(j) cells in the DP row that are associated with overlapping alignments. As a result, if we consider cells in non-increasing order of alignment scores, we may need to check at most O(j) other cells. O(j) cells with the highest scores can be selected from the complete row in O(|R|) time using a partition-based selection algorithm. Assume that we also have scores of previous rows in memory. Then we can trivially check for overlap between a pair of alignments in O(j) time. Adding all the work, total asymptotic time spent per row is O(|R|+j 2 ). As total number of rows that may need to be computed is bounded by the constant c, total time spent to compute MCAS(i) remains O(|R|). Therefore, computing all MCASs requires O(|Q||R|) time. Asymptotic space complexity of the above algorithm is O(|R|).
u t An O(|Q||R|) time complexity resembles the complexity of DP-based alignment algorithms. As such, the exact algorithm does not o↵er desired scalability. In Winnowmap2, we make use of fast heuristics and make careful accuracy-performance trade-o↵s to address this. First, we perform the MCAS computation from a subset of equally spaced starting positions, e.g., after every 500 th base. Next, while computing an MCAS, we reduce the alignment search space by making use of known minimizer seeding and clustering ideas [12,14]. Starting from a small substring length, our iterative method exponentially grows the substring (rather than growing linearly). In each iteration, we check its mapping to reference R. This is done until the substring either satisfies the alignment confidence criteria or cannot be extended further. While computing each mapping, we rely on e ciently engineered anchor chaining, banded-alignment, and mapQ computation code from minimap2. In a way, the mapQ scoring heuristic in minimap2 approximates our definition for confidence assessment.
Heuristic to compute mapping quality. In Winnowmap2 implementation, we use the same heuristic as minimap2 to compute the mapping quality score of a read alignment. For completeness, we also mention it here. Once the anchors between a read and a reference are identified, minimap2 runs a co-linear chaining algorithm to locate alignment candidates. The chaining procedure ensures that alignment candidates use a disjoint set of anchors to prevent overlaps. To compute mapQ, minimap2 compares the anchor chaining score of the best-scoring chain relative to the secondbest. Suppose their scores are denoted as f 1 and f 2 respectively. Also, let m be the count of anchors chained along the best alignment. Minimap2 uses the following empirical formula to calculate mapQ score of the best alignment candidate: The above score is readjusted by minimap2 to fall within the range of 0 to 60. By default, we use mapQ cuto↵ of 5 in Winnowmap2 to mark an MCAS alignment as confident. This cuto↵ can be modified by users. In practice, a higher cuto↵ typically leads to longer MCASs, as expected. A lower cuto↵ increases the probability of an incorrect alignment to be considered as the best.
Consolidating several MCASs into a single alignment output. Once we compute MCAS alignments from a read, these need to be aggregated into a single alignment output. At this step, we extract anchors that were joined to form each MCAS alignment. Subsequently, the union of all anchor sets is passed to chaining and alignment routines to output the final set of best-scoring alignments. Typically, there are only a few anchors to process at this step, which does not require significant time.
Simulation and evaluation of structural variant calls. In our simulation benchmark, we made use of T2T chromosome assemblies for chromosome 8 (v9) and chromosome X (v0.7) that are available from [URL]13. SURVIVOR (v1.0.6) was used to simulate 1,100 SVs of length ranging from 50 bp to 1000 bp in each chromosome sequence. We also simulated PacBio HiFi reads as well as ONT reads using PBSIM (commit:e014b1) and NanoSim (v2.6.0) respectively. Command line parameters provided to these tools are listed in Supplementary Table S2. NanoSim requires real data for training its error model. Training was executed using a publicly available R10.3 Guppy 3.4.5 ONT sequencing data of the Escherichia coli K12 genome (ENA:PRJEB36648). PBSIM command line parameters were adjusted to achieve PacBio HiFi data characteristics with an indel error rate of about 1%. Supplementary Table S3 specifies the read length statistics. Long read mappers were tested using two sequencing coverage levels, 20x and 40x. In our mapping evaluation, we compared Winnowmap2 (v2.0), Winnowmap (v1.01), minimap2 (v2.17), ngmlr (v0.2.7) and graphmap (v0.5.2). Each mapper was executed using their recommended parameters and 28 CPU threads (Table S2). SV calling from BAM alignment file outputs was done using Sni✏es (v1.0.11). The SV call sets were evaluated using SURVIVOR against its own simulated ground truth. We also evaluated SV calling accuracy within repetitive reference intervals. For this, de novo repeat annotation of reference sequences was computed by using Mashmap (commit:ffeef4) to approximately identify all duplications of 10 kbp length and 95% identity. SV evaluation within the repeats was done by intersecting variant coordinates and repeat intervals using bedtools (v2.29.2) [46]. Evaluation using GIAB SV calls. We evaluated Winnowmap2 and minimap2 using the GIAB Tier1 (v0.6) SV call set [32] available for the HG002 human sample relative to the GRCh37 human genome reference. In this experiment, we utilized HG002 ONT and PacBio HiFi sequencing data [47,48] made available through the precision FDA site [URL]/10/. Sni✏es SV call sets were evaluated using SVanalyzer (v0.36).
==
Domain: Biology Computer Science
|
[{"added": "2020-11-05T09:11:02.699Z", "created": "2020-11-02T00:00:00.000", "id": "226263615", "metadata": {"extfieldsofstudy": ["Biology", "Computer Science"], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "https://www.biorxiv.org/content/biorxiv/early/2020/11/02/2020.11.01.363887.full.pdf", "pdf_hash": "cd9c3644b20b2485b23eb18824e82953183053a1", "pdf_src": "BioRxiv", "provenance": "peS2o-0028.json.gz:4806", "s2fieldsofstudy": ["Biology"], "sha1": "34884806d7f8a6fe0376f944e4cc261821053ea2", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
A protocol for adding knowledge to Wikidata, a case report
Pandemics, even more than other medical problems, require swift integration of knowledge. When caused by a new virus, understanding the underlying biology may help finding solutions. In a setting where there are a large number of loosely related projects and initiatives, we need common ground, also known as a “commons”. Wikidata, a public knowledge graph aligned with Wikipedia, is such a commons and uses unique identifiers to link knowledge in other knowledge bases However, Wikidata may not always have the right schema for the urgent questions. In this paper, we address this problem by showing how a data schema required for the integration can be modelled with entity schemas represented by Shape Expressions. As a telling example, we describe the process of aligning resources on the genomes and proteomes of the SARS-CoV-2 virus and related viruses as well as how Shape Expressions can be defined for Wikidata to model the knowledge, helping others studying the SARS-CoV-2 pandemic. How this model can be used to make data between various resources interoperable, is demonstrated by integrating data from NCBI Taxonomy, NCBI Genes, UniProt, and WikiPathways. Based on that model, a set of automated applications or bots were written for regular updates of these sources in Wikidata and added to a platform for automatically running these updates. Although this workflow is developed and applied in the context of the COVID-19 pandemic, to demonstrate its broader applicability it was also applied to other human coronaviruses (MERS, SARS, Human Coronavirus NL63, Human coronavirus 229E, Human coronavirus HKU1, Human coronavirus OC4).
Introduction
The COVID-19 pandemic, caused by the SARS-CoV-2 virus, is leading to a burst of swiftly released scientific publications on the matter (1). In response to the pandemic, many research groups have started projects to understand the SARS-CoV-2 virus life cycle and to find solutions. Examples of the numerous projects include outbreak.info (2), VODAN around FAIR data (3), CORD-19-on-FHIR (4) and the COVID-19 Disease Map (5). Many research papers and preprints get published every week and many call for more Open Science (6). The Dutch universities went a step further and want to make any previously published research openly available, in whatever way related to COVID-19 research (7).
However, this swift release of research findings comes with an increased number of incorrect interpretations (8) which can be problematic when new research articles are picked up by main-stream media (9). Rapid evaluation of these new research findings and integration with existing resources requires frictionless access to the underlying research data upon which the findings are based. This requires interoperable data and sophisticated integration of these resources. Part of this integration is reconciliation, which is the process where matching concepts in Wikidata are sought. Is a particular gene or protein already described in Wikidata?
Using a shared interoperability layer, like Wikidata, different resources can be more easily linked.
The Gene Wiki project has been linking the different research silos on genetics, biological processes, related diseases and associated drugs (10), creating a brokerage system between the research silos. The project recognises Wikidata as a sustainable infrastructure for scientific knowledge in the life sciences.
In contrast to legacy databases, where data models follow a relational data schema of connected tables, Wikidata ( [URL]/ ) uses statements to store facts (see Figure 1) (10)(11)(12)(13). This model of statements aligns well with the RDF triple model of the semantic web and the content of Wikidata is also serialized as Resource Description Framework (RDF) triples (14,15), acting as a stepping stone for data resources to the semantic web. Through its SPARQL endpoint ( [URL] ), knowledge captured in Wikidata can be integrated with other nodes in the semantic web, using mappings between these resources or through federated SPARQL queries (16). Automated editing of Wikidata simplifies the process, however, the quality control must be monitored carefully. This requires a clear data schema that allows the various resources to be linked together with their provenance. This schema describes the key concepts required for the integrations of the resources we are interested in the NCBI Taxonomy (17) , NCBI Gene (18) , UniProt (19) , the Protein Data Bank (PDB) (20) , WikiPathways (21) , and PubMed (22) . Therefore, the key elements for which we need a model include viruses, virus strains, virus genes, and virus proteins. The first two provide the link to taxonomies, the models for genes and proteins link to UniProt, PDB, and WikiPathways. These key concepts are also required to annotate research output such as journal articles and datasets related to these topics. Wikidata calls such keywords 'main subjects'. The introduction of this model and the actual SARS-CoV-2 genes and proteins in Wikidata enables the integration of these resources.
This paper is a case report of a workflow/protocol for data integration and publication. The first step in this approach is to develop the data schema. Within Wikidata, Shape Expressions (ShEx) are used as the structural schema language to describe and capture schemas of concepts (23,24). With ShEx we describe the RDF structure by which Wikidata content is made available. These Shapes have the advantage that they are easily exchanged and describe linked data models as a single knowledge graph. Since the Shapes describe the model, they enable discussion, revealing inconsistencies between resources and allow for consistency checks of the content added by automated procedures. The Semantic Web was proposed as a vision of the Web, in which information is given well-defined meaning and better enabling computers and people to work in cooperation (26). In order to achieve that goal, several technologies have appeared, like RDF for describing resources (15), SPARQL to query RDF data (27) and the Web Ontology Language (OWL) to represent ontologies (28).
Linked data was later proposed as a set of best practices to share and reuse data on the web (29). The linked data principles can be summarized in four rules that promote the use of uniform resource identifiers (URIs) to name things, which can be looked up to retrieve useful information for humans and for machines using RDF, as well as having links to related resources. These principles have been adopted by several projects, enabling a web of reusable data, known as the linked data cloud ( [URL]/ ), which has also been applied to life science (30).
One prominent project is Wikidata, which has become one of the largest collections of open data on the web (16). Wikidata follows the linked data principles offering both HTML and RDF views of every item with their corresponding links to related items, and a SPARQL endpoint called the Wikidata Query Service.
Wikidata's RDF model offers a reification mechanism which enables the representation of information about statements like qualifiers and references (see also [URL] ). For each statement in Wikidata, there is a direct property in the wdt namespace that indicates the direct value. In addition, the Wikidata data model adds other statements for reification purposes that allow enrichment of the declarations with references and qualifiers (for a topical treatise, see Ref. (31) ). As an example, item Q14875321 , which represents ACE2 (protein-coding gene in the Homo sapiens species) has a statement specifying that it has a chromosome ( P1057 ) with value chromosome X ( Q29867336 ). In RDF Turtle, this can be declared as: That statement can be reified to add qualifiers and references. For example, a qualifier can state that the genomic assembly ( P659 ) is GRCh38 ( Q20966585 ) with a reference declaring that it was stated ( P248 ) in Ensembl Release 99 ( Q83867711 ).
Specifying data models with ShEx
Although the RDF data model is flexible, specifying an agreed structure for the data allows domain experts to identify the properties and structure of their data facilitating the integration between heterogeneous data sources. Shape Expressions were used to provide a suitable level of abstraction. YaShE, [URL]/ , a ShEx editor implemented in JavaScript, was applied to author these Shapes (33).
This application provides the means to associate labels in the natural language of Wikidata to the corresponding identifiers. The initial entity schema was defined with YaShE as a proof of concept for virus genes and proteins. In parallel, statements already available in Wikidata were used to automatically generate an initial shape for virus strains with sheXer (34). The statements for virus strains were retrieved with SPARQL from the Wikidata Query Service (WDQS). The generated Shape was then further improved through manual curation. The syntax of the Shape Expressions was continuously validated through YaShE and the Wikidata Entity Schema namespace was used to share and collaboratively update the schema with new properties. Figure 3 gives a visual outline of these steps. Genomic information from seven human coronaviruses (HCoVs) was collected, including the NCBI Taxonomy identifiers. For six virus strains, a reference genome was available and was used to populate Wikidata. For SARS-CoV-1, the NCBI Taxonomy identifier referred to various strains, though no reference strain was available. To overcome this issue, the species taxon for SARS-related coronaviruses (SARSr-CoV) was used instead, following the practices of NCBI Genes and UniProt.
NCBI Eutils
The Entrez Programming Utilities (EUtils) is the application programming interface (API) to the Entrez query and database system at the National Center for Biotechnology Information (NCBI). From this set of services the scientific name of the virus under scrutiny was extracted (e.g. "Severe acute respiratory syndrome coronavirus 2").
Mygene.info
Mygene.info is a web service which provides a REST API that can be used to obtain up-to-data gene annotations. The first step in the process is to get a list of applicable genes for a given virus by providing the NCBI taxon id. The following step is to obtain gene annotations for the individual genes from mygene.info through [URL]/43740571 . This results in the name and a set of applicable identifiers (Figure 4).
UniProt
The annotations retrieved from mygene.info also contain protein identifiers such as UniProt, RefSeq and PDB, however, their respective names are lacking. To obtain names and mappings to other protein identifiers, RefSeq and UniProt were consulted. Refseq annotations were acquired using the earlier mentioned NCBI EUtils. UniProt identifiers are acquired using the SPARQL endpoint of UniProt, which is a rich resource for protein annotations provided by the Swiss Bioinformatics Institute. Figure 5 shows the SPARQL query that was applied to acquire the protein annotations.
Reconciliation with Wikidata
Before the aggregated information on viruses, genes and proteins can be added to Wikidata, reconciliation with Wikidata is necessary. If Wikidata items exist they are updated, otherwise, new items are created. Reconciliation is driven by mapping existing identifiers in both the primary resources and Wikidata. It is possible to reconcile based on strings, but this is dangerous due to the ambiguity of the labels used (37). When items on concepts are added to Wikidata that lack identifiers overlapping with the primary resource, reconciliation is challenging. Based on the Shape Expressions, the following properties are identified for use in reconciliation. The COVID-19 related pathways from WikiPathways COVID-19 Portal are added to Wikidata using the approach previously described (10). For this, a dedicated repository has been set up to hold the GPML files, the internal WikiPathways file format, The GPML is converted into RDF files with the WikiPathways RDF generator (39), while the files with author information are manually edited. For getting the most recent GPML files, a custom Bash script was developed ( getPathways.sh in the SARS-CoV-2-WikiPathways repository ). The conversion of the GPML to RDF uses the previously published tools for WikiPathways RDF (39). Here, we adapted the code with a unit test that takes the pathways identifier as parameter . This test is available in the SARS-CoV-2-WikiPathways branch of GPML2RDF along with a helper script ( createTurtle.sh ). Based on this earlier generated pathway RDF and using the Wikidataintegrator library, the WikiPathways bot was used to populate Wikidata with additional statements and items. The pathway bot was extended with the capability to link virus proteins to the corresponding pathways, which was essential to support the Wikidata resource. These changes can be found in the sars-cov-2-wikipathways-2 branch.
Scholia
The second use case is to demonstrate how we can link virus gene and protein information to literature. Here, we used Scholia ( [URL]/ ) as a central tool (13). It provides a graphical interface around data in Wikidata, for example, literature about a specific coronavirus protein (e.g. Q87917585 for the SARS-CoV-2 spike protein). Scholia uses SPARQL queries to provide information about topics. We annotated literature around the HCoVs with the specific virus strains, the virus genes, and the virus proteins as 'main topic'.
Semantic data landscape
To align the different sources in Wikidata, a common data schema is needed. We have created a collection of schemas that represent the structure of the items added to Wikidata. Input to the workflow is the NCBI taxon identifier, which is input to mygene.info (see Figure 3). Taxon information is obtained and added to Wikidata During this effort, which took three weeks, the bot created a number of duplicates.
These have been manually corrected. It should also be noted that for SARS-CoV-2 many proteins and protein fragments do not have RefSeq or UniProt identifiers, mostly for the virus protein fragments. ( [URL] ) and the articles that discuss them. Scholia takes advantage of the 'main subject' annotation, allowing the creation of "topic" pages for each protein.
For example, Figure 8 shows the topic page of the SARS-CoV-2 spike protein. Wikidata provides a solution. It is part of the semantic web, taking advantage of its reification of the Wikidata items as RDF. Data in Wikidata itself is frequently, often almost instantaneously, synchronized with the RDF resource and available through its SPARQL endpoint ( [URL] ). The modelling process turns out to be an important aspect of this protocol. Wikidata contains numerous entity classes as entities and more than 7000 properties which are ready for (re-)use. However, that also means that this is a confusing landscape to navigate. The ShEx Schema has helped us develop a clear model. This is a social contract between the authors of this paper, as well as documentation for future users.
Using these schemas, it was simpler to validate the correctness of the updated bots to enter data in Wikidata. The bots have been transferred to the Gene Wiki Jenkins platform. This allows the bots to be kept running regularly, pending on the ongoing efforts of the coronavirus and COVID-19 research communities. While the work of the bots will continue to need human oversight, potentially to correct errors, it provides a level of scalability and generally alleviates the authors from a lot of repetitive work.
One of the risks of using bots, is the possible generation of duplicate items. Though this is also a risk in manual addition of items, humans can apply a wider range of academic knowledge to resolve these issues. Indeed, in running the bots, duplicate Wikidata items were created, for which an example is shown in Figure 9. The Wikidataintegrator library does have functionality to prevent the creation of duplicates by comparing properties, based on used database identifiers. However, if two items have been created using different identifiers, these cannot be easily identified.
Close inspection of examples, such as the one in Figure 9, showed that the duplicates were created because there was a lack of overlap between the data to be added and the existing item. The UniProt identifier did not yet resolve, because it was manually extracted from information in the March 27 pre-release (but now part of the regular releases). In this example, the Pfam protein families database (42) identifier was the only identifier upon which reconciliation could happen. However, that identifier pointed to a webpage that did not contain mappings to other identifiers.
In addition, the lack of references to the primary source hampers the curator's ability to merge duplicate items and expert knowledge was essential to identify the
==
Domain: Biology Computer Science
|
[{"added": "2020-04-09T09:02:28.179Z", "created": "2020-04-07T00:00:00.000", "id": "215731495", "metadata": {"extfieldsofstudy": ["Biology", "Computer Science"], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "https://www.biorxiv.org/content/biorxiv/early/2020/06/04/2020.04.05.026336.full.pdf", "pdf_hash": "10787a01e9ce2b74c95f2e750b2f24b62533636e", "pdf_src": "BioRxiv", "provenance": "peS2o-0028.json.gz:2844", "s2fieldsofstudy": ["Computer Science"], "sha1": "a1e201db37976be471cdf0060e0b73d74c773b09", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Multi-Region Brain Stimulation Optimization Using Transcranial Direct Current Stimulation
Transcranial direct current stimulation (tDCS) is a type of noninvasive transcranial electrical brain stimulation. By optimizing the current distribution of each electrode on the scalp, the stimulation can be guided to a target brain region using a tDCS dense electrode array system. However, previous studies have yielded simple results using optimization schemes in single target stimulation cases. The detailed parameter settings for each optimization scheme and the associated simulation results have not been comprehensively assessed. In this study, we investigated parameter settings of optimization schemes in detail in both single target and multi-target cases. Two optimization schemes, minimum least squares (MLS) and maximum electrical field strength (ME), were examined in this study. MLS minimizes the squared errors between the expected electrical field and the estimated electrical field, whereas ME maximizes the electrical field strength in the target region. We constructed a five layer finite-element head model with 64 electrodes placed on the scalp according to the EEG 10/10 system for simulation. We evaluated the effects of stimulation using these two schemes under three conditions, 1) single target stimulation, 2) multi-target stimulation, and 3) multi-target stimulation under specific task activation, which shown that directly using MLS and ME scheme in multi-target stimulation case may lead to a wrong result. We also reported the improved results fixed by our proposed weighted MLS and weighted ME schemes which take detailed parameter settings into consideration. Our results indicate that the parameter settings in each optimization scheme greatly affected the final stimulation results, especially in the case of multi-target stimulation, and thus, indicate that the parameter settings of each optimization scheme should be carefully considered according to the expected stimulation mode. Our results also suggest that, by calculating the parameters through our proposed methods, the weighted ME and weighted MLS scheme can precisely distribute energy into each target brain region.
Transcranial direct current stimulation (tDCS) is a noninvasive brain transcranial 2 electrical stimulation method that has been widely used for many clinical and scientific 3 applications, such as epilepsy [1,2], Parkinson's disease [3,4], Alzheimer's disease [5,6], 4 depression [7,8], schizophrenia [9,10] and neuroscience research [11]. tDCS can be used 5 to induce long-lasting alterations of cortical excitability and activity by continuously 6 applying a constant low intensity current through scalp electrodes for a sufficient 7 duration of time [12,13]. Unlike transcranial magnetic stimulation, which generates 8 supra-threshold activation of brain neurons via short-lasting high intensity 9 electromagnetic currents, tDCS modulates spontaneous neural firing in the brain via 10 sub-threshold alterations of membrane potentials [12,13]. Thus, tDCS is a purely 11 neuro-modulatory intervention [13]. 12 In previous tDCS studies, a 0.5-2 mA tDCS stimulation is often applied directly to 13 the scalp via two 20-35 cm 2 electrode pads (one anode and one cathode). The anode is 14 placed on the scalp above the target region and the cathode is often placed on the scalp 15 above the contralateral region, neck, or another area [11][12][13][14]. This stimulation method 16 affects a large brain area between the anode and cathode [14,15]. In some cases, the 17 target region may not even be included in the actual stimulated region depending on 18 the locations of the electrodes. To overcome the drawbacks of this kind of large 19 electrode pads, numerous new tDCS electrode montages have been developed to 20 improve the guidance of the stimulation to the target region. These include 21 high-definition tDCS [16], and dense electrode array tDCS [17][18][19][20][21]. In high-definition 22 tDCS, the stimulation precision is improved by using different types of electrodes or 23 electrode placement. Saturnino et.al. investigated the effects of several types of 24 electrodes and electrode placement on the brain stimulation results, and recommended 25 that using a 4 × 1 ring electrodes placement or two focal center surround ring electrodes 26 can achieve better focality of stimulation [16]. In dense electrode array tDCS, N 27 electrodes or N pairs of electrodes are positioned on the surface of the whole scalp 28 according to a specific principle (e.g., Electroencephalogram (EEG) 10/20 or 10/10 29 system), and the stimulated target is located by using different current distributions on 30 these electrodes (i.e., giving each electrode a specific current). Dmochowski and Guler 31 pointed that when the current of each electrode is calculated using optimization 32 algorithms under certain safety constraints, the stimulation via dense electrode array 33 tDCS can be generally limited to the specific target region [17,18]. 34 When using optimization algorithms, the selection of an optimization scheme (i.e., maximum electrical field strength (ME) inside the target region [18]. In comparison 45 with the MLS scheme, the ME scheme can induce the maximum stimulation strength in 46 the target region and only needs to specify the direction of the expected stimulation. 47 However, these studies only examined optimization performance in the case of a single 48 target stimulation. The performance of the MLS and ME schemes in multi-target 49 stimulation scenarios is unknown. As individual brain functions often involve many 50 brain regions, simultaneously stimulating two or more brain regions may have more 51 impact on a specific brain function. Dmochowski proposed a MLS based optimization 52 scheme that uses the EEG recording to give neural sources which generate this EEG 53 recording a stimulation [20]. Although, this method can give neural sources a relatively 54 precise stimulation without given any additional parameter such as the strength and 55 direction of the expected stimulation, it can only apply to the situation that neural 56 sources which generate the recorded EEG need to be stimulated. Ruffini and his 57 colleagues first proposed the concept of multifocal transcranial current stimulation [21]. 58 In this concept, all brain voxels are stimulated under a specific stimulation plan 59 according to a given pattern from neuroimaging studies, such as a T-map in functional 60 magnetic resonance imaging or a apparent diffusion coefficient map in diffusion tensor 61 imaging [21]. However, this study only proposed a framework. The selection of an 62 optimization scheme, specific parameter settings of each optimization scheme, and their 63 effects on the optimization results were not given in detail.
64
Thus, in this study, we systematically investigated the optimization performance of 65 the MLS and ME optimization schemes when applied to brain stimulation via a 66 64-electrode dense array tDCS system, described previously [17]. We examined the 67 performance of the two schemes in three conditions: 1) single target stimulation 68 optimization, 2) multi-target stimulation optimization, and 3) stimulation optimization 69 under an activation by a specific task.
MATERIALS AND METHODS
Where E represents the electrical field distribution of the brain, V i represents the 80 stimulation potential of the i th electrode area, and N is the number of electrodes. Eq 1 81 indicates that, given each a potential V for each electrode, the potential anywhere in 82 the brain can be calculated by solving Laplace's equation.
T represents the potential of the nodes that are the vertices 92 of these elements, M is the total number of nodes, and K is the coefficient matrix of the 93 linear homogeneous equation system. The K is derived from the space coordinates and 94 corresponding conductivity of each node. Eq 2 not only simplifies the problem described 95 by Eq 1, but also markedly improves the speed at which solutions can be computed, 96 making FEM a popular simulation method in many fields. Solving the potential of each 97 node using Eq 2 enables the calculation of V, the corresponding electrical field E, and 98 the current density J of each node in brain [22].
99
Consider that N electrodes are applied onto the surface of scalp layer of a head stimulation with an arbitrary current magnitude is applied to each anode, the electrical 107 field distribution can be described as follows [17]: Eq 3 to Eq 6 imply that by calculating the distribution coefficient matrix A using FEM, 112 the electrical field distribution E can be calculated as a straightforward linear solution. 113 Thus, given a specific electrical field distribution E, we can use an optimization scheme 114 to calculate a set of stimulation currents I that can generate an E that will maximally 115 approximate E.
117
In this study, we investigated the performance of tDCS in a multi-object stimulation 118 scenario using two optimization schemes: MLS and ME scheme.
119
Minimum Least Squares Scheme 120 The MLS optimization scheme is widely used in many research fields. Given a specific 121 electrical field distribution E, the MLS optimization scheme is defined as follows [17]: Where || · || 2 represents the 2-norm of the vector, e 0 represents the expected electrical follows [22] to minimize the squared error under some specified safety constraints: September 12, 2019 5/24 As decribed in [17], if current satisfies this sum form of current constrains, then 132 |I i | ≤ I max , and | I i | ≤ I max
133
In FEM, the number of nodes M is usually very big (i.e., 10 6 -10 7 or greater). Thus, 134 direct optimization using Eq 8 can be both time and source consuming. Further 135 simplification of Eq 8 is need. The least squares can be extended as: and E T E is a constant. If we let W aa = A T A, w ea = E T A, and w ee = E T E, Eq 8 can 138 be simplified as: As A is already calculated using FEM, E is given, and W aa , w ea , and w ee can be 140 pre-calculated, optimization using Eq 10 can be completed in seconds.
142
In study [18], Guler stated that a MLS scheme should specify the strength and direction 143 of the expected electrical field E. In general, researchers tend to select the direction as 144 the normal or tangential direction [17,18,20]. However, the strength of the expect 145 electrical field e 0 is very important, as it will decide not only the stimulation strength of 146 E, but also the form of E as well (Fig 6). Thus, Guler proposed a new optimization 147 scheme, the maximum current density J inside the ROI, which only requires 148 specification of the direction of the expected electrical field E. As J = σE, where σ is 149 the conductivity of medium, the maximum current density inside the ROI is equivalent 150 to the maximum electrical field inside the ROI, i.e., ME The math formulation of the 151 ME scheme is defined as follows [18]: Where Brain-ROI ||σ(r)E(r)|| 2 2 dr represents the total power outside the ROI that, with 153 regards to safety considerations, should not be larger than p max . · represents the dot 154 product operation. Eq 11 shows that the goal of the ME scheme is to maximize the sum 155 of the electrical field in the target ROI under the safety constraint. According to 156 September 12, 2019 6/24 Guler [18], Eq 11 can also be simplified. The ROI (E(r) d(r))dr can be discretized as: 157 Where V m represents the volume of the mth elements. If we assume uniform 160 distribution of the nodes that construct elements in the brain, then V m can be 161 approximately treated as a constant. Thus, Eq 13 can be approximately represented as: 162 Similarly, Brain-ROI ||σ(r)E(r)|| 2 2 dr can be extended as: Eq 11 can be simplified as [18]: Also, W and Q can be pre-calculated. Optimization operations using Eq 18 can be also 166 completed in seconds.
167
Eq 10 and Eq 11 give the basic formulation of MLS and ME optimization scheme in 168 single target case. However, when we directly used these two schemes in multi-target (i.e. taking all target stimulation brain regions as a whole region), the simulation results 170 were not be the one we desired (Fig 5 & 6). To correctly guide stimulations to target 171 regions in the multi-targets case, we propose the weighted ME and weighted MLS 172 optimization scheme in next two subsections.
173
The Weighted ME Scheme 174 Using the ME scheme described in Eq 18, the W can be rewritten as, Thus, the basic weighted ME scheme can be defined as follows: Where α j represents the weight assigned to the j th ROI and satisfies α j > 0, To account for variations in the sizes of the ROIs, before applying the 178 expected weights to each ROI, the original weight of each ROI, which is mainly 179 proportional to the size of the ROI, should be equalized. Thus, the α j can be defined as 180 follows: Where 1/β j is the equalization coefficient of the j th ROI and S j should be calculated 182 according to specific requirements. Because the size of a ROI is not the only factor that 183 affects the original weight, for accurate calculation, β j can be redefined as: From Eq 22, for accurate calculation of β, single target optimization for each ROI using 185 the ME scheme should be computed first. As this can be time consuming, β as defined 186 in Eq 21 is sufficient for general use.
187
If the goal is to stimulate all ROIs with a similar strength, a strength distribution Where std(·) represents the standard deviation of a vector and s 0 > 0 is the maximum 190 standard deviation. The smaller the s 0 , the more similar the stimulation strength 191 among the ROIs. The main effect of Eq ?? is to constrain the differences between ROI 192 stimulation strengths to a relatively small range.
193
If primary electrodes that maintain the stimulation for each ROI need to be injected 194 with a similar current, the weighted ME scheme should be rewritten as: strength, an additional current distribution constraint should be added: and Where the S j needs to be calculated according to specific distribution requirements, and 203 in each ROI class (e.g., high stimulation class and low stimulation class), S j should be 204 From the MLS scheme described in Eq 10, E can be rewritten as: Thus, the weighted MLS scheme can be defined as follows: September 12, 2019 10/24 and white matter (WM). Briefly, brain tissues were first segmented into scalp, skull, 243 and brain areas using the brain extraction tool (BET) function. Then, the brain area 244 section was further segmented into CSF, GM, and WM using the FMRIB's automated 245 segmentation tool (FAST) function.
246
After segmentation, the five tissues masks were imported into simpleware software 247 ( [URL]/ simpleware.html) to generate the final head model (Fig 1). 248 We performed manual correction of the tissue masks using the ScanIP module. Then, this study are detailed in Table 1. stimuli, we predicted steady-state fields without much concern for the 270 charging/discharging effect of tissue capacitance (i.e., using quasistatic solution). The The ACC locates deeper in brain than PreCG. As shown in Fig 3, for a single target 279 case, the MLS scheme had higher focality (i.e., more concentrated energy distribution in 280 stimulation region) but lower intensity (i.e., lighter color in stimulation region) 281 compared with the ME scheme. The detailed performance characteristics are given in 282 Table 2. From Table 2, we can see that, when the depth of the stimulated region values gradually approximated those in the ME scheme when power constraints existed 292 outside the ROI (P Brain-ROI ≤ 1e-6 V 2 ·m). For the ROI on the grey matter on the 293 surface of the brain (e.g., PreCG), the performance of the MLS scheme was nearly equal 294 that of the ME scheme when the strength of the expected electrical field was set at the 295 maximum value (5.5 V/m, Fig 4E). The maximum value represents the maximum 296 strength to which the expected electrical field e 0 can be set. Values larger than this will 297 cause optimization results which do not satisfy the power constraint (i.e., 298 P Brain-ROI > p max ). For deeper ROIs in the brain (e.g., ACC), the MLS scheme had 299 better focality but lower intensity compared with the ME scheme when the strength of 300 the expected electrical field was set at the maximum value (1.7 V/m, Fig 4E).
3) When 301
the expected electrical field strength gradually increased to the maximum allowed value, 302 Person's correlation coefficient between the currents calculated by MLS and ME 303 schemes gradually increased to one (Fig 4C). These findings indicate that the the fifth column in Fig 7). Also, the α in Fig.6 were normalized. ROI was more concentrated than the one in weighted ME scheme, but the total 347 intensity level was lower than the weighted ME scheme (colorbar in figures). Thus, 348 similar with single target case, the weighted MLS scheme had better focality but a lower 349 intensity compared with the weighted ME scheme.
351
We used activation T-map generated from functional MRI data during a specific space 352 working memory task (SWMT) in 36 healthy humans as a reference for generating a 353 stimulation mode according to the activation strength. The SWMT activation T-map is 354 displayed in Fig 8 and the detailed activation ROI information is given in Table 4. We optimized two stimulation modes using weighted ME and weighted MLS scheme, 356 respectively. For the weighted ME scheme, we used a regional stimulation mode: each 357 ROI had a stimulation strength sum that was proportional to the corresponding peak T 358 value. The detailed parameters for the weighted ME scheme are shown in Fig 9 and the 359 results are shown in Fig 9 and Table 5. The derivation of parameters used in the The optimization results of a regional stimulation mode using weighted ME optimization scheme. The second stimulation mode directly used T values in the SWMT T-map as the 365 weight of each voxel's expected stimulation strength. We kept T values larger than 4.5 366 (p < 0.05, familywise error corrected). The others were set to zero and then normalized 367 using the maximum T value in the map. This stimulation mode was not suitable for the 368 weighted ME scheme, which was regional based. Before optimization, we performed 369 equalization according to the size of each ROI (e.g., number of nodes in the head 370 model). The detailed parameters are shown in Table 6. The result of the optimization is 371 shown in Fig 10 and Table 7 with e 0 selected as 1.5 V/m. We systematically investigated the optimization performance of two optimization 374 schemes, MLS and ME, using a 64 electrode dense array tDCS system. We first The MLS scheme had been widely used in many brain stimulation studies [17,20]. As 384 its definition can be decomposed into a simple form Eq 10, the processing time can be 385 very short. Dmochowski [17] pointed out the trade-off between intensity performance 386 and focality performance in the MLS scheme. From our investigations (Fig 3 & 4), we 387 found that changing the expected electrical field strength e 0 affected this trade-off.
388
Lower e 0 can increase the focality of stimulation while higher e 0 can increase the 389 intensity. If intensity and focality performance are both important, it may be preferable 390 to search the grid for the expected electrical field strength. However, this operation is 391 known to be time consuming.
392
As an improvement to the MLS scheme, the ME scheme could achieve the maximum 393 intensity performance that MLS scheme could be or might be reached (Fig 4) under 394 power constraints, given in Eq 18. The advantage of the ME scheme is that it is only 395 necessary to determine the expected stimulation direction, while for the MLS scheme, 396 the expected stimulation strength e 0 and direction must be determined. However, this 397 advantage prevented the trade-off between intensity and focality from being controlled, 398 leading to the best intensity performance but the worst focality performance when 399 compared with the MLS scheme (Fig 4). Thus, in the single target case, we recommend 400 the ME scheme for the best intensity performance, and for increased accuracy in 401 controlling the intensity and focality performance, we recommend the MLS scheme.
402
From the Eq 8 and Eq 11, it is the definition that makes MLS and ME scheme have 403 different performance in intensity and focality. The optimization problem of MLS 404 scheme is to minimize the mean square error between simulation electrical field and the 405 desired one. Thus, it can achieve the best focality performance. The optimization can achieve the best intensity performance.
408
Multi-Target Case 409 Normal brain function requires cooperation among many brain regions. Thus, 410 simultaneously applying stimulation to two or more brain regions may be more 411 meaningful in both therapeutic and research applications. We found that directly using 412 the MLS scheme or ME scheme (e.g., taking all target regions as a whole) for 413 multi-target optimization caused the 'overfitting' problem, i.e., the optimization 414 algorithm tended to optimize those regions that were larger in size or had a greater 415 impact on the cost function, and ignore other regions (the first column in Fig 5 and the 416 Fig 6). Thus, we used a weighted MLS scheme and weighted ME scheme to overcome 417 this problem. Our results shown that when applying appropriate weights to each ROI, 418 the weighted MLS scheme and weighted ME scheme could implement an arbitrary 419 regional based stimulation mode (the second column to the fifth column in Fig 5 and 420 Fig 7).
421
For the weighted ME scheme, besides applying weights to each ROI, additional the sum of the electrical field strength was very small compared to the values in Table 451 3. Because ROI 2 was much smaller than L SFG in Table 3 be improved, our proposed methods can still precisely distribute the energy into each 467 target brain region according to any given ratio (Table 5). Thus, our methods can 468 become a guideline for those studies or clinical treatments which need to stimulate two 469 or more brain regions simultaneously using tDCS. genetic algorithms to solve the optimization problem. However, we found that the 477 optimization results of these two algorithms are similar with interior-points method, and 478 more time will be cost by these two algorithms thanrunning the interior-points method 479 100 times. Thus, Future work will focus on the improvement of these two optimization 480 schemes and finding more effective algorithms to solve optimization problems.
481
Secondly, as shown in Fig 3, This study validates previous works in [17] and [18] in single target stimulation case 492 using MLS and ME optimization schemes, then adds upon their work, by investigating 493 the performance these two schemes in multi-target stimulation case. Our findings 494 suggest that directly using MLS or ME scheme in multi-target stimulation case (e.g.
508
We show the derivation of α in Fig 9. According to the specific stimulation parameters, 510 Eq 25 can be written as:\===
Domain: Biology Computer Science. The above document has
* 2 sentences that start with ' ≤ I',
* 2 sentences that start with ' 2 2 dr',
* 2 sentences that start with 'The detailed parameters',
* 2 sentences that end with 'the target region',
* 2 sentences that end with 'the expected electrical field E',
* 2 paragraphs that start with 'Transcranial direct current stimulation (tDCS)',
* 2 paragraphs that end with 'completed in seconds'. It has approximately 3865 words, 159 sentences, and 46 paragraph(s).
|
[{"added": "2019-09-19T09:13:11.296Z", "created": "2019-09-16T00:00:00.000", "id": "203370369", "metadata": {"extfieldsofstudy": ["Biology", "Computer Science"], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "https://www.biorxiv.org/content/biorxiv/early/2019/09/16/771345.full.pdf", "pdf_hash": "076060af0fc7821de2095bb5d5186cc8efac2643", "pdf_src": "BioRxiv", "provenance": "peS2o-0028.json.gz:8592", "s2fieldsofstudy": ["Computer Science"], "sha1": "209691b29dae9fba59cb442d095ce6998fc6de25", "year": 2019}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Recent advances of cell membrane-coated nanoparticles for therapy of bacterial infection
The rapid evolution of antibiotic resistance and the complicated bacterial infection microenvironments are serious obstacles to traditional antibiotic therapy. Developing novel antibacterial agents or strategy to prevent the occurrence of antibiotic resistance and enhance antibacterial efficiency is of the utmost importance. Cell membrane-coated nanoparticles (CM-NPs) combine the characteristics of the naturally occurring membranes with those of the synthetic core materials. CM-NPs have shown considerable promise in neutralizing toxins, evading clearance by the immune system, targeting specific bacteria, delivering antibiotics, achieving responsive antibiotic released to the microenvironments, and eradicating biofilms. Additionally, CM-NPs can be utilized in conjunction with photodynamic, sonodynamic, and photothermal therapies. In this review, the process for preparing CM-NPs is briefly described. We focus on the functions and the recent advances in applications of several types of CM-NPs in bacterial infection, including CM-NPs derived from red blood cells, white blood cells, platelet, bacteria. CM-NPs derived from other cells, such as dendritic cells, genetically engineered cells, gastric epithelial cells and plant-derived extracellular vesicles are introduced as well. Finally, we place a novel perspective on CM-NPs’ applications in bacterial infection, and list the challenges encountered in this field from the preparation and application standpoint. We believe that advances in this technology will reduce threats posed by bacteria resistance and save lives from infectious diseases in the future.
Introduction
Since its discovery in 1970s, antibiotics have been utilized to combat microbes and protect people from fatal infections. However, the overuse of antibacterial agents and the mutations in bacteria have caused significant drug resistance (Kurt Yilmaz and Schiffer, 2021;Saha and Sarkar, 2021). However, the process of developing novel antibiotics is extremely expensive and timeconsuming. The development of novel potent antibiotics is still lagging . Novel antibacterial agents emerge much slower than bacteria resistant to them. Bacterial infections are expected to cause 10 million deaths per year by 2050. Approximately $100 trillion can be spent on treating drug-resistant infections globally . Consequently, superbugs have emerged as a major health challenge (Church and McKillip, 2021). In a variety of scientific fields and in pharmaceutical companies, researchers have explored for novel antibiotics and therapeutic strategies to overcome bacterial resistance.
Aside from drug resistance, toxins pose a major challenge in the clinical management of bacterial infections (Escajadillo and Nizet, 2018;Keller et al., 2020). During the process of antibiotics killing bacteria, their exotoxins and endotoxins are rarely eliminated. The rupture of the bacteria may produce more endotoxins, further causing harm to the body . Moreover, a close relationship exists between the pharmacokinetics (PK) and pharmacodynamics (PD) properties of antibiotics. For example, meropenem exhibits time-dependent PD property. The time that free drug concentrations remain above the minimum inhibitory concentration (MIC) as a function of the dosing interval (%fT > MIC) has been shown to best predict antibacterial effect, in which MIC refers to the lowest concentration of drug capable of inhibiting bacterial growth (Pascale et al., 2019;Liu J. et al., 2021). Clinically, shortening the dosing interval, prolonging the infusion time, and increasing the dosage has been utilized to increase the %fT > MIC. However, a sole focus on optimizing drug dosing regimens offers limited results. Furthermore, pathogenic bacteria can form biofilms on the surfaces of medical devices or in infection sites. A biofilm's dense and adhesive structure can severely reduce the effectiveness of drug enrichment by preventing drug penetration (Gebreyohannes et al., 2019). Biofilms are also excellent for horizontally spreading antibiotic resistance genes (Bowler et al., 2020). Antibiotic resistance is aggravated by these issues, and high doses of antibiotics potentially causes organ damage. In summary, conventional antimicrobials are inadequate for the successful treatment of bacterial infections, particularly those caused by multidrug-resistant bacteria.
Nanodrug delivery systems (DDSs) can increase the retention time in the circulation, reduce the non-specific distribution, and enable targeted delivery of drug to the infection sites . Thus, the use of DDSs for treating deadly infections has recently been found to be a promising therapeutic possibility (Gupta et al., 2019;Makabenta et al., 2021). Combinations of nanomaterials and antibiotics may enhance therapeutic efficacy. As a result, they provide promising therapeutic options for combating bacteria and treating infectious diseases. Au, Ag, Cu, Fe and Ti-based DDSs have been developed for treating infectious diseases, but their clinical use is held back by safety concerns. The metallics may leak out and harm people as well (Guo et al., 2020). The development of smart nanomaterialbased delivery strategies, such as pH-activated, enzyme-activated, and bacterial toxin-activated DDSs, have attracted considerable interest (Canaparo et al., 2019). However, their sophisticated fabrication process has hindered their designs from entering the large-scale production stage. The body also has difficulty in metabolizing the complex material. Therefore, novel systems are urgently needed to overcome the above issues.
The use of nanoparticles (NPs) modified by natural cell membranes or synthetically produced membranes, known as "biomimetic nanoparticles, " has advanced the fields of drug delivery and received considerable attention in recent years . Cell membrane-coated nanoparticles (CM-NPs) combine some of the best features of both host and artificial nanoparticles (Zhen et al., 2019). Owing to their unique features, such as immunological invasion and enhanced targeting capacity, CM-NPs confer significant therapeutic and diagnostic value (Imran et al., 2022b). Another benefit of CM-NPs is their ability to preserve the intrinsic features and abilities of cell membranes. The biocompatibility of CM-NPs is good because they are considered by the body as parts of itself. Thus, since Zhang et al. 's report on cell membrane-coating technology in 2011, numerous membrane-mimicking nanoplatforms have been constructed for biomedical applications, with a primary emphasis on cancer therapy (Imran et al., 2022a).
It has also been reported that CM-NPs are capable of evading immune recognition, targeting pathogenic bacteria, neutralizing toxins, and delivering antibiotics for combating bacterial infections (Rao et al., 2020;Wang et al., 2020). Currently, the membranes of erythrocytes, platelets, macrophages, neutrophils, epithelial cells, bacterial and hybrid membranes have all been successfully applied for CM-NPs preparation . Due to their various constituents, such as membrane proteins, glycans, and lipids, cell membranes from various origins serve distinct functions. For example, nanoparticles coated with erythrocyte membranes have a longer circulation half-life time than those modified by polyethylene glycol (PEG) because they inherit the ability of red blood cells (RBCs) to circulate blood for an extended duration (Fan et al., 2020). White blood cells (WBCs) membrane-coated biomimetic nanoparticles are endowed with immune evasion and inflammatory chemotaxis functions . Moreover, Yingying Gan et al. propose the strategy of "fight bacteria with bacteria. " The potential of membrane vesicles (MVs) derived from bacteria as delivery systems to treat bacterial infection has been extensively researched as well (Gan et al., 2021). This review will provide an overview of the methods used in preparing CM-NPs, examining the obstacles and future directions faced by the field, and providing an outlook on the potential benefits to patients.
Preparation of cell membrane-coated nanoparticles
The preparation of CM-NPs mainly consists of three steps: membrane extraction, core nanoparticle synthesis, and fusion. All these steps are important for maintaining the intended functions of biomimetic nanoparticles .
Phospholipids and certain surface proteins constitute cell membranes. Membranes often have pivotal role in different types of biological processes. Extracting cell membranes requires two critical steps: membrane lysis and membrane purification . Both of the process require pinpoint accuracy and careful handling. Various membranes determine the specific extraction procedure. Cells without nuclei, such as mature RBCs and platelets, make membrane extraction a straightforward operation. After cells are separated from whole blood, the membranes are forcefully disrupted through hypotonic lysis or repeated freezing and thawing. Then, soluble proteins are separated through differential centrifugation, and nanovesicles are extruded into their final shapes. Obtaining membrane materials from eukaryotic cells, such leukocytes, requires elaborate methods. Isolating desired cells from the blood or tissues and culturing the cells are the initial steps. Then, cell membranes are isolated by removing nuclei and cytoplasm via hypotonic lysis, mechanically disrupting the membrane, and centrifugation of the pellet in a discontinuous sucrose gradient. Isoionic buffers clean the Frontiers in Microbiology 03 frontiersin.org membranes before sonication and extrusion through the membranes' porous polycarbonate matrices. In addition, MVs and exosomes are usually extracted by removing dead cells and cellular debris via centrifugation. Subsequently, evenly sized exosomes and bacterial MVs are obtained through ultrahigh-speed centrifugation Ma et al., 2022). Inner nanoparticles are crucial to biomimetic nanomaterial production because they carry antimicrobial components to sick tissues. Recently, the use of different materials encapsulated by cell membranes, such as mesoporous silica nanocapsules, metal-organic frameworks (MOFs), gold nanoparticles, nanogels, nanocrystals, and poly(lactic-co-glycolic acid) (PLGA) nanoparticles, have been explored and developed . The preparation processes vary by nanomaterial and delivery cargo requirement.
After cell membranes and inner core nanoparticles are obtained, CM-NPs undergo a membrane coating process on the surfaces of the nanoparticles. Current fusing techniques include membrane extrusion, ultrasonic fusion, electroporation, and co-incubation Imran et al., 2022a). Membrane extrusion and ultrasonic fusion are the two most often employed techniques. In membrane extrusion, mechanical forces enable nanoparticles and cell membranes to traverse membranes with varying pore diameters, allowing the coating of nanoparticles by cell membranes. However, despite being practical and effective, this technique is difficult to use in large-scale production. Similar to physical extrusion, ultrasound results in the spontaneous formation of a core-shell structure between membranes and nanoparticles with diminished material loss. However, resulting particles may exhibit large size heterogeneity and lack of homogeneity. The nanoparticles can be damaged by ultrasonography. A unique microfluidic electroporation strategy has recently been used for the production of membrane-coated nanoparticles. In this method, ingredients are combined in a Y-shaped or S-shaped conduit, where they are thoroughly mixed before electroporation. The method has been utilized to generate nanoparticles with high yields and high reliability after suitable tuning. Another approach for obtaining NP-containing exosomes is to incubate live cells with nanoparticles, and then induce secretion in a serum-free media. These procedures differ from the coating procedures described above, which entail purifying the cell membrane [!!! INVALID CITATION!!! ]. An ideal coating process should produce stable nanoparticles with consistent size and shape without altering the membranes' or core particles' functional qualities.
3. Cell membrane-coated nanoparticles for bacterial infection treatment 3.1. Red blood cell membrane-coated nanoparticles As a natural bio-membrane isolated from RBCs, the RBC membrane (RBCM) has attracted substantial interest as a potential drug carrier (Jin et al., 2018;Xia et al., 2019). Immunomodulatory indicators in the RBCM, such as CD47, sialic acid, peptides, and glycans, prolong the circulation of medicines by reducing absorption by macrophages (Gao et al., 2013;Chen et al., 2020). Liangfang Zhang et al. revealed that RBCM modification decreased the cellular uptake of PLGA nanoparticles by macrophages, and PLGA nanoparticles exhibited a longer half-life time than the widely applied longcirculating PEG-modified nanoparticles . Bacteria can create life-threatening toxins. Endotoxins and exotoxins are the major sources of bacterium-induced cytokine outbursts. Traditional detoxification methods rely on antigen-antibody specific binding, and each antibody is designed to neutralize a specific antigen . A challenge encountered in these methods is determining the structures of toxins and pathogenic microorganisms. Methods for neutralizing toxins disarm pathogens and directly ameliorate infection symptoms due to their destructive nature and crucial functions in pathogen processes (Zou et al., 2021). RBCs can be destroyed by various bacterial exotoxins, and RBC membrane-coated nanoparticles, sometimes known as"nanosponges,"can serve as decoy targets for several different bacterial toxins . Owing to the high affinity between the RBCM and exotoxins, nanosponges show great potential in neutralizing a wide variety of toxins. Additionally, the RBCM is considerably simple to isolate and purify because mature RBCs have no nuclei or other organelles. Thus, RBCM-modified nanoparticles fighting bacterial infections have drawn the attention of researchers. Representative RBCM-coated nanoparticles for bacterial infection treatment are listed in Table 1.
Biomimetic nanoparticles composed of ~100 nm PLGA nanoparticle core and RBCM were developed for infection induced by Group A Streptococcus (GAS). The constructed nanosponges sequestered streptolysin O (SLO), a well-defined virulence factor generated by the vast majority of GAS, and prevented GAS from damaging host cells, thus maintaining innate immune function and enhancing bacterial clearance. Neutrophils, macrophages, and keratinocytes were protected from SLO-mediated cytotoxicity after treatment with nanosponges. The topical use of biomimetic nanosponges reduced the lesion size and the counts of bacterial colony-forming units in mice with GAS necrotizing skin infection . The constructed nanosponges (the preparation process shown in Figure 1) significantly blocked the cytotoxic effects of pore-forming toxin β-hemolysin/cytolysin (β-H/C) of Group B Streptococcus (GBS). Nanosponge therapy prevented cellular death in lung epithelial cells and macrophages after exposure to GBS. In addition, increased GBS killing by neutrophil and lowered levels of IL-1β produced by macrophages were observed in response to GBS. In general, nanosponges exhibited a negative charge due to the nature of cell membranes, and could interact with cationic compounds (Koo et al., 2019). Zhang et al. mixed nanosponges with chitosan-functionalized PLGA nanoparticles possessing a similar size but opposite surface charge. After mixing, the two types of nanoparticles formed a stable three-dimensional network, dubbed a "nanosponge colloidal gel" (NC-gel). The gel preserved RBC-nanoparticles in the network without affecting their toxinneutralizing capacity. The obtained NC-gel, an injectable preparation, demonstrated remarkable antibacterial capacity as reflected by the significantly reduced skin lesions in mice subcutaneously infected by GAS .
Infections caused by bacteria typically result from released or secreted toxins. The process of toxin absorption by RBCM-coated nanoparticles seems like disarming the "enemy, " thus relieving clinical symptoms. However, the bacteria, the enemy without power, are not killed or destroyed. The strategy of "disarming" and "killing" the enemy simultaneously may be a means for treating infections . In another study, RBCM-coated PLGA nanoparticles were used for tedizolid phosphate delivery. The prepared nanoformulation displayed excellent compatibility and accelerated the healing rate of mice with MRSAinfected wound .
As a system for biomimicry, RBCM-coated nanoparticles effectively counter pore-forming toxins (PFTs) produced by different types of bacteria. Without inducing the development of antibiotic resistance, photodynamic therapy (PDT) shows great promise in treating multidrug-resistant microorganisms. Based on these developments, an RBCM-coated PFT-responsive nanobubble system was developed for the on-demand release of therapeutic gases and therapy of bacterial infections. RBCM coating allowed nanobubbles to bind and neutralize several types of PFTs, thus preventing toxinmediated cytotoxicity from affecting healthy cells. By forming pores on nanobubble surfaces, therapeutic cargoes can be rapidly released Frontiers in Microbiology 05 frontiersin.org within the microenvironments of bacterial infections, thereby significantly improved antibacterial capacity. In O 2 delivery, supplied oxygen increases PDT via the generation of reactive oxygen species after laser irradiation. Moreover, additional therapeutic gases, such as nitric oxide, can be delivered and released in a bacterial toxindependent manner ( Figure 2A; Zhuge et al., 2022). This research provided a general strategy for the on-demand delivery and release of therapeutic gases and for improving precision in the treatment of bacterial infections. Extremely diverse microorganisms produce gelatinase . Biocompatible and degradable gelatin nanoparticles (GNPs) can serve as optimal drug delivery vehicles for the microenvironment-responsive release of encapsulating antimicrobial agents. In addition to PFT-responsive nanosytems, RBCM is applied to bacterium-responsive biomimetic nanosystems. For example, core-shell supramolecular GNP coated with RBCM was constructed for adaptive and "on-demand" Van delivery. Given that the core consisted of cross-linked GNPs, the encapsulating Van was released in a microenvironment-responsive manner. The shell RBCM coating served as a mask that inhibited clearance by the immune system during antibiotic delivery (Li et al., 2014). The RBCM absorbed bacterial exotoxins, thus alleviating the symptoms of bacterial infection (Zhuge et al., 2022). This design illustrated a novel antibiotic delivery system for treating bacterial infection at a low drug dose. However, the therapeutic effect of the constructed nanosystems in vivo has not been investigated. In another study, Ange Lin et al. reported an RBCM-coated GNP for delivering Ru complex-modified selenium nanoparticles (Ru-Se NPs). The developed nanoformulation Ru-Se@GNP-RBCM showed the following advantages: (1) the RBCM exerted immune-evading and toxin-clearance capacities; (2) the degradation of GNP by gelatinase facilitated the responsive release of Ru-Se NPs in infectious microenvironments; (3) the nanosystem exerted synergistic antibacterial activity and real-time in vivo imaging capacity, allowing the accurate monitoring of the treatment procedure (shown in Figure 2B; Lin et al., 2019). The presented results confirmed the hypothesis. However, the process of preparing the nanosystem is complicated, and additional characterization is needed for confirming successful synthesis. Additionally, its effectiveness against bacteria is not greater than that of Van. Nevertheless, the reported biomimetic nanoparticle Ru-Se@GNP-RBCM still exhibited remarkable potential as a biological antibiotic replacement.
The generation of biofilms impedes the entry of antibiotics and deactivates them, consequently developing bacterial resistance (Malaekeh-Nikouei et al., 2020). RBCM-NPs displays potential in combating biofilm as well. For example, a novel biomimetic nanosystem, RBCM-NW-G, composed of RBCM, nanoworm (NW) particles and gentamicin was developed by Ran et al. The obtaining nanoparticles retained the capacity of RBCM with a longer blood circulation time and good biocompatibility. AuAg and polydopamine in the nanoworm exerted photothermal effect under near infrared excitation. Near infrared excitation triggered the release of antibiotic and Ag + into the microenvironment. The increased loacal temperature damaged the biofilms. Subsequently, antibiotic and Ag + could effectively penetrate the biofilm and achieve excellent bactericidal activity. The research provides a novel means for combating bacterial biofilm infections (Ran et al., 2021).
White blood cell membrane-coating nanoparticles
WBCs, also known as leukocytes, are immune cells that help the body fight off infections, engulf and digest foreign invaders, heal damaged tissues, and ward off illnesses . WBCs, such as macrophages, neutrophils, T cells, and natural killer cells, play crucial roles in bacterial infection and inflammation . A sentinel monocyte or macrophage recognizes endotoxins released by bacteria during sepsis. Septic shock or death may be cause by these cells activating or potentiating downstream inflammation cascades. Dynamic and varied functions have served as the basis for constructing WBC membrane-coated nanoparticles (WBC-NPs) (Mohale et al., 2022). With wide bio-interfacing properties, WBC-NPs mimic the broad biofunctional properties of source cells and can be used therapeutically. WBC-NPs can recognize the pathogen-associated molecular patterns of harmful bacteria; this ability enables them to realize targeted antibiotic delivery effectively. The benefits of WBC-NPs have led to their investigation as a medication delivery carrier for bacterial infection treatment. Table 2 lists some examples of the WBC-NPs' uses in treating bacterial infections.
Macrophages are crucial immune cells identifying infections and responding to them. By producing complicated receptors (especially Toll-like receptors), which bind to microbial molecular patterns, (Robinson et al., 2019;Galli and Saleh, 2020). Changes in receptor expression and dimerization occur in response to macrophage activation by various microorganisms (Rosenberg et al., 2022). Additionally, when macrophages are cultured with certain bacteria, the expression of recognition receptors on bacterial membranes increases dramatically (Sica et al., 2015). Therefore, macrophage membranes' unique capacity to identify microorganisms has served as an inspiration for the construction of macrophagecoated nanoparticles. Wang et al. reported a S. aureus-pretreated macrophage membrane-coated gold-silver nanocages (Sa-M-GSNC).
Their results demonstrated that S. aureus pretreatment increased the adhesion capacity of the system to therapeutically relevant bacteria effectively. Apart from heat generated by GSNC under laser light irradiation, the constructed Sa-M-GSNC killed the bacteria and significantly reduced the bacterial counts in vitro and in vivo (Wang et al., 2018). Clinically, intravascular catheters, mechanical ventilation equipment, hemodialysis machines, and other medical devices are all potential entry points for bacteria (Cecconi et al., 2018). The survival percentage of septic patients can be greatly improved by using antibiotics immediately after diagnosis (Font et al., 2020). Given the rising prevalence of multidrug-resistant bacteria, extracorporeal cleansing devices are increasingly attractive (Kang et al., 2014;Didar et al., 2015). Current extracorporeal blood purification systems rely on the adsorption or adherence of pathogen-associated molecular patterns and cytokines released by bacteria. Their therapeutic efficacy is inconsistent and unclear (Kang, 2020). Liu et al. reported a microfluidic device equipped with an interconnected nanowired silicon (Si) capture surface. To use macrophages' inherent blood compatibility and ligand-receptor binding ability, they coated the membranes onto Si nanowire surfaces. Upon stimulation by S. aureus or Escherichia coli, the macrophages developed low negative zeta potentials, which allowed them to capture nonspecific bacteria. Additionally, the particular bacterial capture was aided by Toll-like receptors in bacterially activated membrane coatings on nanowired surfaces, which are missing in nonactivated membrane coatings. These two features, along with the maintenance of fluidity in activated membrane coatings, were responsible for the broad spectrum and high capture efficiency of all ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Escherichia coli) panel pathogens, which are regarded as the most threat to people health (Figure 3; . Sepsis, caused by systemic bacterial infection, is a potentially fatal condition with widespread consequences (Landersdorfer and Nation, 2021). As the standard treatment for sepsis, antibiotic therapy encounteres challenges of drug-resistant bacteria (Dugar et al., 2020;Asner et al., 2021). Cao et al. reported a macrophage membrane-coated MOF system for delivering antimicrobial gene LL37 and fighting sepsis. LL37 can be delivered specifically to macrophages by coating macrophage membranes. This method facilitated the production of antimicrobial peptides in a continuous fashion. The constructed system significantly increased the survival rates of immunosuppressed septic mice infected with MRSA via effective gene therapy and sequester of inflammatory cytokines (Figure 4; Cao et al., 2022), demonstrating the potential of macrophage membrane coating in gene therapy. Soracha et al.
reported a biomimetic nanoparticle composed of PLGA core and macrophage membrane surface. The developed nanoparticles were able to bind to endotoxins, absorb them, and prevent them from eliciting an immunological response. In addition, the nanoparticles mimicking macrophages captured proinflammatory cytokines and prevented them from amplifying the sepsis cascade. In a mouse E. coli bacteremia model, the obtained nanoparticles lowered the levels of proinflammatory cytokine, limited bacterial dissemination, and increased the survival rate (Thamphiwatana et al., 2017). Similarly, magnetic composite nanoparticles with osteoconductive Ca 3 (PO 4 ) 2 and antibacterial TiO 2 were phagocytosed into macrophages for the preparation of membrane-coating nanoparticles and bone infection treatment. The obtained system displayed excellent properties in recognizing and absorbing bacteria, toxins, and inflammatory cytokines, thus exerting good antibacterial capacity in vitro and in vivo .
In addition to delivering genes, recognizing bacteria, and absorbing endotoxins, WBC-NPs can serve as carriers for antibiotic delivery. For example, bacteria "hide" in mammalian cells to avoid being killed by antibiotics and the host immune system (Weiss and Schaible, 2015). When S. aureus enter macrophages, the killing mechanisms of antibiotics might be inactivated, and this effect allow intracellular habitation and spread of infections (Peyrusson et al., 2020;Pidwill et al., 2020). Li et al. prepared a binary antimicrobial nanoparticle (ANP) consisting of triclosan and ciprofloxacin. Encapsulation within macrophage membranes improved the stability of the ANP . The obtained macrophage membraneencapsulated ANP, named Me-ANP, were superior to nonencapsulated ANP and clinically used ciprofloxacin in killing S. aureus in mice with peritoneal infection. Me-ANP were more effective than ciprofloxacin in eradicating organ infections caused by the spread of infected macrophages through the bloodstream in mice. The findings suggested a viable approach for therapeutic application in humans to tackle chronic infections.
Neutrophils play vital roles in fighting invading pathogens and can express more than 30 receptors capable of sensing inflammatory mediators (Németh et al., 2020). Through various mechanisms, including phagocytosis, degranulation, neutrophil extracellular traps, and reactive oxygen species-mediated damage, neutrophils can neutralize infectious threats (Petri and Sanz, 2018). Nanoparticles coated by neutrophil membranes are the essential components of WBC-NPs. For example, bioinspired microrobots that can actively move in biological fluids have attracted substantial interest for biomedical applications (Aziz et al., 2020). Zhang et al. attached neutrophil membrane-coated polymeric nanoparticles delivering ciprofloxacin to natural microalgae to obtain microrobots. The microrobots demonstrated quick speed in simulated lung fluid, homogeneous distribution, minimal clearance by alveolar macrophages, and excellent tissue retention duration after intratracheally administrated. Mice with acute Pseudomonas aeruginosa pneumonia receiving microrobots treatment exhibited a significant decrease in bacterial burden and remarkable increase in survival rate . Neutrophil membrane-coated MOF loaded with glucose oxidase and chloroperoxidase were applied for the therapy of subcutaneous infection caused by S. aureus in mice. The developed nanoparticles, when injected intravenously, greatly decreased local bacterial load and wound size in comparison to uncoated MOF nanoparticles .
Platelet membrane-coated nanoparticles
In the bloodstream, platelets-tiny and anucleate cell fragmentsabound. Platelets are traditionally responsible for regulating blood clotting and the integrity of blood vessels. However, recent studies have shown that platelets serve as sentinel effector cells in the onset of bacterial infection (Jahn et al., 2022). Interactions between platelets and innate immunity play a significant role in responses to pathogens because platelets can sense and react to dangerous signals, and guide leukocytes to inflammation, damage, or infection sites (Yadav et al., 2019). Platelets can affect the recruitment and activation of immune effector cells by direct or indirect means (Palankar et al., 2018). In an intravascular infection, the first and most numerous cells to accumulate are platelets. Additionally, platelets can kill microorganisms by secreting antimicrobial peptides, such as defensins, cathelicidins, and kinocidins (Pircher et al., 2018;Kunde and Wairkar, 2021). Aggregated platelets can capture microbes and prevent their spread. Surface moieties specific to platelets mediate immunological evasion, subendothelial adhesion, and pathogen contacts (Imran et al., 2022a). Overall, platelets perform functions apart from inducing blood clotting and play a major role in regulating hosts' immunological and inflammatory responses. In addition, platelet membrane-coating nanoparticles (PLT-NPs) can remarkably reduce macrophage uptake and complement activation and demonstrate platelet-like functions even after camouflaging, thereby solving the problem of plasma protein absorption on nanomaterial surfaces. Thus, esearch into PLT-NPs in infectious diseases has attracted widespread interest (as shown in Table 3; Kunde and Wairkar, 2021).
Among the most common opportunistic gram-positive bacterial pathogens, S. aureus (especially MRSA), lead to a wide range of illnesses, such as sepsis, bacteremia, and endocarditis (Cheung et al., 2021). The production of toxins is the driving force behind S. aureus pathogenesis. Platelets express disintegrin and metalloproteinase domain-containing protein 10, a receptor of α-toxin, on surface membranes (Surewaard et al., 2018). Thus, platelets are the targets of S. aureus as well. Kim et al. constructed biodegradable nanoparticles consisting of PLGA cores and biomimetic platelet membranes for blocking the cytotoxicity of S. aureus and protecting the body from lethal systemic infection. By inhibiting platelet damage caused by S. aureus toxin secretion, the nanoparticles stimulated platelet activation and acted as bactericides. In a similar fashion, the nanoparticles counteracted S. aureus-secreted toxin-induced macrophage damage. Thus, MRSA-induced neutrophil extracellular trap release was reduced, and the bactericidal, oxidative burst, and nitric oxide generation of macrophages were bolstered. Therapy with the developed nanoparticles decreased bacterial counts in the blood and reduced mortality in MRSA-induced systemic infection in mice. The findings offered demonstrative evidence of the therapeutic efficacy of PLT-NPs in neutralizing toxin, providing cytoprotection, and enhancing resistance to infection (Kim et al., 2019). Schematic illustration of the preparation process of Na-m-SiNWs and Ba-m-SiNW. Ba-m-SiNWs exhibit more enhanced pathogen capture than Na-m-SiNWs. Reproduced with permission Copyright© 2021, Wiley-VCH). Na-m-SiNWs, silicon nanowires coated by nonactivated macrophage membranes; Ba-m-SiNWs, silicon nanowires coated by bacterially activated macrophage membranes.
Frontiers in Microbiology 09 frontiersin.org In another study, 2-methylimidazole and silver nitrate were used in preparing a nanosilver MOF for Van loading. Ag-MOF-Vanc were encapsulated by nano-sized platelet vesicles to obtain PLT@Ag-MOF-Vanc. PLT@Ag-MOF-Vanc outperformed free Van against a panel of typical clinical pathogens. The mechanisms of PLT@Ag-MOF-Vanc killing bacteria is multifaceted and involves disturbance of bacterial metabolism, catalysis of reactive oxygen species generation, disruption of cell membrane integrity, and inhibition of biofilm formation. The platelet membrane allowed PLT@Ag-MOF-Vanc to attach to MRSA and infectious sites. In a mouse model of MRSA pneumonia, PLT@ Ag-MOF-Vanc showed a strong anti-infective effect that was significantly more effective than free Van and did not cause any evident harm .
Phage treatment potentially mitigates antibiotic resistance (Kortright et al., 2019). The short half-lives of bacteriophages in the bloodstream are the largest obstacles to their widespread use. Jin et al. reported a blood circulation-prolonging peptide (BCP1) with an RGD motif binding to integrins; the peptide improved circulation by interacting with platelets (Jin et al., 2021). Researchers have developed biomimetic phage-platelet hybrid nanoparticles (PPHNs), aiming to extended the use of BCP1 in bacterial infection. The obtained PPHNs with a particle size of ~350 nm showed sustained antimicrobial capacity and increased blood retention time to combat E. coli infection in vivo, demonstrating more effective antibacterial properties prophylactically and therapeutically when compared with BCP1 (Jin et al., 2021). This investigation on hybrid membrane further expanded the use of platelet membrane and provided a novel strategy for applying phage-based nanoparticles to bacterial infection treatment.
Bacterial membrane-derived nanoparticles
Bacterial MVs are particles with diameters of 20-400 nm and are secreted by bacteria (Sartorio et al., 2021). Gram-negative and grampositive bacteria can secret MVs into the extracellular environment. MVs from gram-positive bacteria are called extracellular vesicles (EVs), and those from gram-negative bacteria are outer membrane vehicles (OMVs) (Toyofuku et al., 2019). MVs carry a diverse range of immunogenic antigens and pathogen-associated molecular patterns essential for regulating host immune responses, and allow the immune system to respond either broadly or specifically (Behrens et al., 2021). Integrating gold nanoparticles with E. coli-secreted OMVs provides nanovaccine capable of eliciting significantly increased immune responses in comparison to intact E. coli OMVs (Gao et al., 2015), indicating that the immunomodulatory function of antigens on the OMV surface are unaffected by the membrane-coating technology (Brown et al., 2015). Immunostimulatory property and proteoliposome nanostructure make MVs attractive candidates for vaccines or delivery systems against bacterial infections (Sartorio et al., 2021). Examples of bacterial membrane-derived nanoparticles' applications are shown in Table 3.
OMVs-based nanodrug delivery systems have been reported. Wu et al. coated mesoporous silica nanoparticles (MSNs) encapsulating rifampicin (Rif) with OMVs isolated from E. coli. Owing to their homotypic targeting activity, OMVs significantly enhanced MSN absorption in E. coli but not in gram-positive S. aureus. Compared with free Rif with only modest bactericidal activity, Rif@MSN@OMV with excellent biocompatibility totally eradicated bacteria at a comparable Rif concentration. Moreover, Rif@MSN@OMV increased the survival rates of mice with peritonitis induced by E. coli and significantly reduced the bacterial load in the intraperitoneal fluid and related organs (shown in Figure 5; . The premature release of encapsulated antibiotics was inhibited, and the internalization process by the same bacteria was significantly enhanced by membrane coating. In theory, this method increases the susceptibility of gram-negative bacteria to medications that have been historically ineffective. Most complications of S. aureus bacteremia result from the pathogen's ability to survive within host phagocytes, particularly macrophages. Eliminating the intracellular S. aureus is fundamental for clinical success (Foster, 2005). In addition to research focusing on macrophage membrane-coated nanoparticles , the use of EVs in bacterial infection treatment has been explored. Gao et al. coated antibiotic-loaded nanoparticles (NP-antibiotic) with EVs secreted by S. aureus to prepare an active targeted delivery nanosystem. NP-antibiotic@EV served as a "Trojan horse, " bringing antibiotics into infected macrophages and subsequently killing intracellular S. aureus. The results demonstrated that NP-antibiotic@EV is as effective as encapsulated antibiotics or more effective despite the prolonged release behavior. After intravenous administration to mice bearing S. aureus, NP@EV achieved active and targeting distribution in organs suffering from metastatic infections. Moreover, NP-antibiotic@EV enabled the encapsulated antibiotic to exert remarkably enhanced anti-infection capacity .
Dental caries is a chronic, progressive, and devastating disease caused primarily by bacterial infection cariogenic biofilms. Cariogenic bioflms are difficult to treat because bacterial pathogens like Schematic illustration of the application of macrophage membranecoated metal-organic framework (MOF) for pLL37 delivery (MMD-LL37) and sepsis therapy. The constructed MMD-LL37 can sequester pro-inflammatory cytokines, realize the targeted delivery of plasmid, generate antibacterial LL37, and eradicate bacteria in circulation and those hidden inside cells. Reproduced with permission (Cao et al., 2022;Copyright© 2022, American Chemical Society).
Frontiers in Microbiology 10 frontiersin.org Streptococcus mutans reside in a self-producing matrix of extracellular polymeric substances (EPS). The penetration and retention of antibiotics into biofilms is severely deficient. To solve this problem, membrane derived from Lactobacillus acidophilus was utilized to coat PLGA nanoparticles encapsulating triclosan (TCS), in which L. acidophilus was capable of inhibiting the colonization and bioflm formation of S. mutans via the coaggregation with S. mutans. The resulting biomimetic nanoparticles, LA/TCS@PLGA-NPs, have the following advantages: (1) By viture of the native properties of L. acidophilus, LA/TCS@PLGA-NPs can adhere to S. mutans and hinder the formation of S. mutans' biofilm; (2) LA/TCS@PLGA-NPs can incorporate into the biofilm, and serve as a depot for sustained antibiotic release to prevent the spread of S. mutans biofilm (Weng et al., 2022). This study offers novel insights into CM-NPs for combating bacterial biofilms and associated illnesses.
Other cell membrane-coated nanoparticles
Membranes from RBC, WBC, PLT, and bacteria have been combined with nanomaterials. The procedure prevents clearance by the immune system and extends retention time, but whether the prepared biomimetic nanomaterials can specifically target specific cells remains a pressing issue (Pang et al., 2019a). Hybrid cell membranes have recently attracted attention because they potentially improve the functionality of biomimetic nanoparticles . For example, a polymyxin B (PMB)-modified RBC-biomimetic hybrid liposome (P-RL) was developed for antivirulence therapy of E. coli infection. The interaction between PMB and E. coli membrane contributed to the attachment and anchoring of P-RL to E. coli. The fusion of the RBCM and the modified PMB allowed P-RL to effectively neutralize endotoxins and exotoxins (Table 3; Jiang et al., 2021). In another study, a hybrid of RBC membrane and PLT membrane was applied to functionalize nanorobots for the simultaneous elimination of bacteria and toxins (Esteban-Fernández de Ávila et al., 2018). As well-known antigen-processing cells, dendritic cells (DCs) play a crucial role in triggering innate and acquired immunological response after bacterial recognition via Toll-like receptors (Adib-Conquy et al., 2014;Bieber and Autenrieth, 2020). Hou et al. used S. aureus-pretreated DC membrane (SM) to coat CuFeSe 2 nanocrystals that are highly efficient in converting light into heat. The resulting complex, SM@NC, efficiently bind to S. aureus and exhibited stealthy immune evasion and targeted delivery to the infectious site ( Figure 6A). The membrane coating remarkably enhanced the antibacterial capacity of the CuFeSe 2 nanocrystals. In combination with near infrared irradiation, the intravenous administration of SM@ NC significantly reduced the bacterial colonies in infected tibia (Table 3). These findings provided proof of concept for the therapeutic application of pathogen receptor membrane-coated nanoparticles to the treatment of infectious diseases (Hou et al., 2021).
Helicobacter pylori infection is a primary cause of pepticulcer illness, inflammatory gastritis, and gastric cancer, creating a global healthcare burden (Fischbach and Malfertheiner, 2018). A combinational therapy including a proton pump inhibitor and antibiotics, such as clarithromycin and amoxicillin, has been widely recommended for H. pylori infection treatment (de Brito et al., 2019). However, increasing drug resistance often results in therapy failure. Indeed, novel and effective strategy is urgently needed. Inspired by the cell membrane-coated nanoparticles, Angsantikul et al. developed gastric epithelial cell membrane-coated nanoparticle for delivering antibiotics against H. pylori based on the specific binding between H. pylori and gastric epithelial cell ( Figure 6B). With the feature of sharing the same antigens on surfaces with the original AGS cells, the obtained nanosystem can naturally adhere to H. pylori bacteria. AGS-NPs demonstrated specific accumulation on bacterial surface. The clarithromycin (CLR)-encapsulated AGS/NP exhibited markedly improved antibacterial efficacy compared with free CLR and nontargeted nanoparticles. These findings highlighted the potential of employing the membrane of native host cells significantly related to the specific pathogens to functionalize nanocarriers .
Genetically engineered cell membranes have also been applied to coat nanomaterials and used in bacterial infection treatment. Genetic membrane engineering was utilized to endow HEK293T cells with the capacity of expressing specific antibody MEDI4893, a monoclonal antibody (MAb) that specifically neutralizes alpha-toxin of MRSA (Kong et al., 2016). Subsequently, MAb-piloting nanovesicles (ANVs) were applied to encapsulate sonosensitizer (Pang et al., 2019a). The extremely active antibody-toxin interaction made the ANVs more effective in capturing toxins than traditional passive neutralizing toxin capacity endowed by natural RBCM. Sonosensitizers, when activated by ultrasound, produce reactive oxygen species that efficiently kill bacteria and accelerate virulence clearance (Pang et al., 2019b). By eradicating bacteria and neutralizing pathogenic toxins simultaneously, the reported nanomaterial potentially prevents and controls infection caused by multidrug-resistant bacteria. Moreover, when guided by an antibody, a nanocapturer may precisely pinpoint MRSA infection and differentiate it from benign inflammation. As the first of its kind, this novel approach combines antibacterial sonodynamic therapy with antivirulence immunotherapy, opening the door to antibiotic-free nanotheranostics that can effectively combat multidrug-resistant bacterial illnesses.
Extracellular vesicles (EVs) released by cells are nano-sized MVs that have proven to be highly effective in the development of biomimetic nanoplatforms (Herrmann et al., 2021). In comparison to EVs made from mammalian cells, plant-derived edible EVs with low immunogenicity are environmentally friendly, sustainable, and amenable to large-scale production. Combining ginger-derived EVs and Pd-Pt nanosheets, Qiao et al. yielded a biomimetic nanoplatform (EV-Pd-Pt) for synergistic bacteria and biofilm elimination. Gingerderived EVs helped EV-Pd-Pt stay in the blood longer without being cleaned by the immune system and accumulate at infection sites. EV-Pd-Pt got access to the interior of bacteria via an EV lipiddependent means. Remarkably, electrodynamic and photothermal therapies mediated by EV-Pd-Pt nanoparticles exhibited synergistic benefits and great efficiency in elimination of S. aureus biofilms (Qiao et al., 2022).
Discussion: Perspectives and challenges
Most nanomaterials require surface modification before application in biomedical settings, but conventional approaches have been shown to be unsatisfactory (Nemani et al., 2018). Based on the many roles of the cell membrane, cell membrane coating is a soughtafter alteration to bestow nanomaterials with excellent biological Schematic illustration of (A) the preparation process and (B) the application of OMV-coated biomimetic nanodelivery system Rif@ MSN@OMV in bacterial infection. OMV modification achieved homotypic targeting capacity and enhanced uptake, thus exerting strong antibacterial activity. Reproduced with permission . These CM-NPs can be used for various purposes, including neutralizing toxins, evading clearance by the immune system, targeting specific bacteria, delivering antibiotics, and regulating immune responses . The recent advances in applications of CM-NPs in bacterial infection are listed in Tables 1-3. Owing to the strategy's adaptability, various membranes and nanoparticle cores can be combined to serve a variety of therapeutic goals. Through genetic engineering, cell membranes can acquire abilities not found in their source cells (Pang et al., 2019a). These methods provide hope for the integration of numerous antimicrobial components into multifunctional CM-NPs by the simultaneous expression of multiple proteins on the cell membrane surface and for effective therapy against bacterial infection. Furthermore, antibiotic-free therapeutic strategy attracts considerable interest in the context of post-antibiotic era. Compounds of natural origin with antibacterial capacity extracted from plant, animal, bacteria and so on, are of tremendous scientific interest as potential therapeutic tools (Álvarez-Martínez et al., 2020). For example, tannins, found in diverse plants such as mimosa, chestnut, quebracho, display good antimicrobial and anti-biofilm effects against various bacteria, and are potential alternatives to conventional antibiotics. Mesoporous silica, hyaluronic acid (HA), gelatin, and chitosan nanocarrier were utilized for delivering tannins (Farha et al., 2020). The nanoparticles can be further modified by cell membrane to enhance the antibacterial capacity of tannins. CM-NPs may be also used in conjunction with photodynamic, sonodynamic, and photothermal therapies to overcome the challenges posed by the photosensitizers' and sonosensitizers' potential instability in circulation and lack of targeted distribution (Lin et al., 2019;Pang et al., 2019a;Chen et al., 2021). In addition, by optimizing the fabrication of antibiotic-loaded CM-NPs, we may combine the antibiotics' unique PK/PD indices to boost their anti-infective efficiency by regulating drug release behavior. Currently, the application of CM-NPs in bacterial infection mainly focuses on MRSA and E. coli. Massive research is required to confirm their effectiveness (A) Construction of dendritic cell (DC) membrane coated CuFeSe2 nanocrystals and their therapeutic application with NIR in bacterial infection (TLR, SM@CuFeSe 2 and NIR referred to toll-like receptor, pathogen-pretreated membrane-coated CuFeSe 2 and near-infrared, respectively). Reproduced with permission (Hou et al., 2021;Copyright© 2021, American Chemical Society). (B) Schematic illustration of the developed of gastric epithelial cell membrane-coated PLGA nanoparticles' application in targeted antibiotic delivery for treating Helicobactor pylori infection. Reproduced with permission Copyright© 2018, Wiley-VCH).
Frontiers in Microbiology 13 frontiersin.org in combating multidrug-resistant gram-negative bacteria, such as Klebsiella pneumoniae, Acinetobacter baumannii, and P. aeruginosa. Despite the numerous benefits of CM-NP for bacterial infection, their use still faces numerous obstacles. First, from the point of view of CM-NP preparation, the uniformity of CM-NPs can be affected by the donor cell type. Medicinal efficacy may vary from batch to batch because of the unique characteristics of cell membranes at different stages of growth during the cell cycle. The whole process of preparing CM-NPs needs to be conducted in a sterile setting. If a desired product is contaminated by bacteria, infection symptoms and diseases will worsen after administration . Overcoming these obstacles should be possible with the imminent implementation of efficient workflow and appropriate quality control assays (Imran et al., 2022a). Second, the insufficient supply of numerous cells and the inefficiency of cell membrane extraction procedures have posed obstacles to largescale production (Le et al., 2021). Third, from the application standpoint, unwanted side effects may be induced by the membrane modifications necessary to create multifunctional CM-NPs. The overuse of CM-NPs can cause or exacerbate inflammation via interaction with the immune system and results in the release of pathological mediators . Nanoparticles with maximal simplicity and functionality should be further explored.
Conclusion
In conclusion, we briefly summarized the recent advances in applications of cell membrane-coated nanoparticles in bacterial infection. Therapeutic biomimetic nanosystems consist of cell membranes, and nanoparticles constitute an exciting field for bacterial infection treatment. We expect that advances in this technology will improve methods for treating bacterial illnesses and reducing threats posed by bacteria resistant to antibiotics despite the abovementioned obstacles.
Author contributions
YS contributed to the conceptualization and wrote the original draft. XZ reviewed and edited the manuscript. JH, SM, KL, and JC performed literature research and organized the database. XX supplied guidance and revised the draft. XL and XW provided the funding support and revised and edited the draft. All authors contributed to the article and approved the submitted version.
==
Domain: Biology Engineering
|
[{"added": "2023-02-19T16:20:14.691Z", "created": "2023-02-17T00:00:00.000", "id": "257015103", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": null, "oa_url": null, "pdf_hash": "2e62687b11d03a4b23106c2bed07d4265c21534c", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:4831", "s2fieldsofstudy": ["Biology", "Engineering"], "sha1": "5f94293361942d4e2a6fc07783b0cb61f9aba4c4", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Isolation and Characterization of Extracellular Vesicles from Gastric Juice
Simple Summary Gastric cancer (GC) is one of the most common cancers and the fifth leading cause of cancer-related deaths worldwide. The steadily growing interest in secreted extracellular vesicles (EVs) is related to their ability to carry a variety of biologically active molecules, which can be used as markers for liquid noninvasive diagnosis of malignant neoplasms. For these applications, blood is the most widely used source of EVs. However, this body fluid contains an extremely heterogeneous mixture of EVs originating from different types of normal cells and tissues. The aim of this study was to assess the possibility of using gastric juice (GJ) as an alternative source of EVs since it is expected to be enriched in vesicles of tumor origin. We validated the presence of EVs in GJ using transmission electron microscopy (TEM), nanoparticle tracking analysis (NTA) and western-blot analysis of exosomal markers, showed for the first time the feasibility of their isolation by ultracentrifugation and demonstrated the prospect of using GJ-derived EVs as a source of GC miRNA markers. Abstract EVs are involved in local and distant intercellular communication and play a vital role in cancer development. Since EVs have been found in almost all body fluids, there are currently active attempts for their application in liquid diagnostics. Blood is the most commonly used source of EVs for the screening of cancer markers, although the percentage of tumor-derived EVs in the blood is extremely low. In contrast, GJ, as a local biofluid, is expected to be enriched with GC-associated EVs. However, EVs from GJ have never been applied for the screening and are underinvestigated overall. Here we show that EVs can be isolated from GJ by ultracentrifugation. TEM analysis showed high heterogeneity of GJ-derived EVs, including those with exosome-like size and morphology. In addition to morphological diversity, EVs from individual GJ samples differed in the composition of exosomal markers. We also show the presence of stomatin within GJ-derived EVs for the first time. The first conducted comparison of miRNA content in EVs from GC patients and healthy donors performed using a pilot sampling revealed the significant differences in several miRNAs (-135b-3p, -199a-3p, -451a). These results demonstrate the feasibility of the application of GJ-derived EVs for screening for miRNA GC markers.
Introduction
Extracellular vesicles are phospholipid bilayer membrane-enclosed particles of different sizes and intracellular origin secreted by various cells.
The process of EV formation is closely related to the selection and loading of biologically active molecules, which are further transported into recipient cells or interact with plasma membrane molecules. This ultimately leads to epigenetic changes and alterations in intracellular signaling [1]. Although EV secretion is shown for almost all cell types, a large body of evidence suggests that cancer cells release higher amounts of EVs compared to non-malignant cells [2,3]. Moreover, the EV secretion seems to increase with cancer progression and the disease stage [4][5][6]. Furthermore, given the high stability of EV cargo molecules as well as the detection of vesicles in all body fluids, EVs are considered a promising source of cancer markers [7][8][9].
Gastric cancer is one of the most common cancers worldwide and is responsible for over one million new cases in 2020 and an estimated 769,000 deaths, ranking sixth for incidence and fifth for mortality globally [10]. GC represents a highly heterogeneous group of diseases, 95% of which are adenocarcinomas [11]. Despite the progress in surgery and therapy techniques, the 5-year survival rate of GC patients is still low, mostly due to the low rate of early diagnosis [12]. A high proportion of advanced cancer stages as well as high morphogenetic variability of GC determine the requirement of additional markers for early diagnosis and differential diagnosis. Secreted extracellular vesicles are of growing interest in terms of screening for new molecular markers for noninvasive liquid cancer diagnosis.
Blood (plasma or serum) is the most commonly used body fluid for EV analysis. EVs from peritoneal lavages and ascitic fluids are also used to search for gastric cancer markers [13][14][15]. It should be noted that blood as a source of EVs for the study of cancer markers has several limitations, the main ones being the extremely high overall heterogeneity of vesicles and the low proportion of EVs of tumor origin [16,17]. In addition, blood is enriched with particles of non-vesicular origin with sizes and other physical characteristics (such as lipoprotein complexes) similar to EV, which contaminate EV preparations [18].
Peritoneal fluid probably contains a higher proportion of vesicles of tumor origin, although this body fluid is more suitable as a source of markers for prognosis assessment or tumor staging than for diagnosis. GJ appears to be a promising source of EVs for the search for diagnostic markers of GC since it can be obtained from both healthy people and cancer patients, and in the case of GC, it should be enriched with vesicles of tumor origin. To our surprise, EVs from GJ have been largely unexplored. The presence of EVs in GJ has been shown in two early studies [19,20] in which, however, EVs were not sufficiently characterized according to ISEV (International Society for Extracellular Vesicles) guidelines. In 2019, Kagota et al. clarified the existence of EVs in GJ [21]. The authors of the latter study developed a significantly modified protocol since they failed to isolate EVs using the standard ultracentrifugation method. This protocol contained two additional steps, including a rather complicated preprocessing procedure and the binding of ultracentrifuged vesicles to microbeads. Accordingly, the morphology of Evs has not been properly studied.
The objectives of this study were to test whether EVs can be isolated from GJ using the standard ultracentrifugation technique; to determine the concentration of GJ-derived EVs in the obtained preparations; to characterize the obtained EVs by size and morphology and to evaluate the presence and composition of various exosomal markers according to the criteria recommended by ISEV; to compare the content of certain miRNAs in EVs from GC patients and healthy subjects through a pilot sample.
We demonstrate the feasibility of the ultracentrifugation method for the isolation of EVs, including exosome-like vesicles, from GJ. The high yield of EVs allows for performing further analysis of their molecular cargo. Notably, the content of proteins commonly used as exosomal markers, such as CD9, Alix, TSG-101, Flot-2, varied enormously in EVs obtained from different individuals. Moreover, the size spectra of CD9(+) and CD9(−) EV samples differed significantly, indicating the existence of distinct subtypes of vesicles present in GJ. Interestingly, the majority of the EV samples contained stomatin, a member of the SPFH family, previously suggested as a new EV marker [22]. First-time analysis of miRNA content in GJ-derived EVs from GC patients and non-cancer individuals revealed significant upregulation of miR-135b-3p and miR-199a-3p and downregulation of miR-451a.
Clinical Specimens and Patient Consent
Gastric juice samples were obtained from gastric cancer patients (intermediate to high-grade adenocarcinoma, GC patients, N = 7; clinical and morphological characteristics are shown in Supplementary Materials, Table S1) and patients with no history of cancer (control group, N = 6) using a OLYMPUS GIF H-185 diagnostic videogastroscope at the beginning of the endoscopy. All patients were food-starved for 12 h and water-starved for 6 h before manipulation. Samples were received from the Endoscopy Department of the N. N. Blokhin National Medical Research Center of Oncology. Written informed consent was sought and obtained from all participants in accordance with the N. N. Blokhin National Medical Research Center of Oncology Ethics Committee guidelines.
Sample Processing
The initial volume of GJ samples ranged from 2 to 5 mL. The obtained samples were diluted with 5 mL of ice-cold PBS (#70011-044, Gibco, Grand Island, NY, USA) just after collection and processed within 2 h at 4 • C. Further steps of sample processing were performed according to the method described by Théry et al. for purifying exosomes from viscous fluids [23] with slight modifications. After brief vortexing, samples were centrifuged at 800× g for 20 min and at 2000× g for 30 min using an A-4-81 rotor (Eppendorf Centrifuge 5810R, Eppendorf AG, Hamburg, Germany) to remove cells and loose cellular mucosal debris. It is noteworthy that sometimes white-yellowish flakes remained in the non-transparent supernatant obtained. They were sedimented alongside large particles at 12,000× g using an F-34-6-38 rotor (Eppendorf Centrifuge 5810R) for 1 h after four-times dilution with ice-cold PBS. After these steps, transparent supernatants were frozen at −80 • C until further steps of EV isolation.
Isolation of EVs
We followed the protocol for EV isolation from viscous fluids by differential ultracentrifugation described by Théry et al. [23] and Caby et al. [24] with slight modifications. Thawed supernatants were diluted with ice-cold PBS to a final volume of 35 mL and transferred to ultracentrifuge tubes (#326823, Beckman Coulter, Brea, CA, USA) to perform a first ultracentrifugation round at 110,000× g (4 • C) for 3 h using an SW-28 swinging bucket rotor (k factor 245.5; Beckman Coulter). The obtained pellets (containing mostly small EVs) were resuspended in 5 mL of ice-cold PBS (Gibco), transferred to small ultracentrifuge tubes (#326819, Beckman Coulter) and centrifuged again at 110,000× g (4 • C) for 90 min using an SW-50.1 swinging bucket rotor (k factor 154.5; Beckman Coulter). The final cleared pellets were resuspended in 120 µL of ice-cold PBS and aliquoted in Protein LoBind tubes (#0030108434, Eppendorf AG, Hamburg, Germany) for NTA, TEM, protein analysis and RNA extraction. Aliquots were frozen in liquid nitrogen and stored at −80 • C for further analysis.
Particle Size Distribution and Quantification
Size distribution and concentration of EVs were determined by NTA using a NanoSight LM10 HS instrument equipped with a NanoSight LM14 unit with on-board temperature control (Malvern Panalytical Ltd., Malvern, UK), LM 14C (405 nm, 65mW) laser unit and high sensitivity camera with a Scientific CMOS sensor (C11440-50B, Hamamatsu Photonics, Hamamatsu City, Japan). Six 60 sec videos were recorded for two independent replicates, generating 12 individual measurements for each sample. Further processing was performed as we described previously [25].
Transmission Electron Microscopy
EV's morphology analysis was performed using a JEM-1011 transmission electron microscope (JEOL, Ltd., Akishima, Japan) operating at 80 kV according to the protocol described in Skryabin et al. (at least 10 fields of view per sample) [25].
Immunoblotting and Antibodies
The concentration of total protein in EV samples and cells lysed in RIPA buffer was determined using the NanoOrange ™ protein quantitation kit (#N6666, ThermoFisher Scientific, Eugene, OR, USA) according to the manufacturer's recommendations using a SpectraMax M5e microplate reader (Molecular Devices, LLC., San Jose, CA, USA). Immunoblotting was performed according to the previously described procedure [25] with the differences that 5 µg of total protein was applied to SDS-PAGE and proteins were visualized with SuperSignal ™ West Femto Maximum Sensitivity Substrate (#34095, Ther-moFisher Scientific, Rockford, IL, USA). The following primary and secondary antibodies and dilutions were used: anti-Alix
RNA Isolation, Reverse Transcription and Quantitative Real-Time PCR
RNA from EVs was isolated using the Total Exosome RNA and Protein Isolation Kit (#4478545; ThermoFisher Scientific, Vilnius, Lithuania) according to the manufacturer's protocol. RNA was eluted from the last column with 60 µL of nuclease-free water and stored at −80 • C until further analysis. Concentration, size distribution and percentage of small RNA were analyzed by Agilent 2100 Bioanalyzer using Small RNA Kits (Agilent Technologies, Santa Clara, CA, USA).
Stem-loop RT-PCR for miRNA quantification was performed according to the method described by Chen et al. [26]. RNA concentration was measured using a NanoDrop™ ND-1000 Spectrophotometer (ThermoFisher Scientific, Wilmington, DE, USA), and 6 ng of exosomal RNA was used for stem-loop RT-PCR according to the previously described procedure [25].
Statistical Analysis
Based on the NTA-measured particle size and concentration, values of mean, mode, percentile data (10th and 90th), standard deviation, and confidence interval were calculated using Wolfram Mathematica ver. 11 ((Wolfram Research, Champaign, IL, USA) software. Student's t-test and analysis of variance (ANOVA) were used for the comparison of groups. p-values lower than 0.05 were considered statistically significant. For statistical analysis, we used the statistical software package GraphPad Prism ver. 8.0.0 package for MS Windows, engineering-mathematical package Wolfram Mathematica ver. 11. The package MS Excel 2016 for MS Windows was used for plotting graphs.
Characterization of EVs Isolated from Gastric Juice
GJ samples were collected from patients with gastric cancer (N = 7, clinical and morphological characteristics are shown in Supplementary Materials, Table S1) and individuals without a history of cancer (non-cancer patients, N = 6) during a routine esophagogastroduodenoscopy procedure. EVs were isolated using an ultracentrifugation-based method with slight modifications (see Materials and Methods). The size and concentration of EVs were assessed by NTA ( Figure 1A-C). The levels of various exosomal markers were tested by immunoblotting (Figure 2A). EV size and morphology have been visualized by TEM ( Figure 2B).
All preparations obtained from GJ contained a high number of "cup-shape" particles corresponding in size and morphology to EVs under the TEM analysis.
The mean size of particles assessed by NTA varied from 92 to 178 nm, with modes from 52 to 141 nm in different individual EV preparations. The mean size of EVs and median over the entire sampling were 149 (SD 29) and 133 nm (SD 37), correspondingly ( Figure 1C). According to NTA data, the concentration of EVs varied from 10 11 to 10 13 particles per mL (the average EV concentration over the entire sampling was 5.12 × 10 12 particles/mL). The mean size of the EVs in preparations obtained from GC patients and individuals without a history of cancer (non-cancer patients) had no significant differences (Student's t-test, p > 0.05). Furthermore, no relationship between vesicle characteristics and tumor grade was observed (p > 0.05).
To confirm the presence of exosomes in EV preparations, we further studied the exosomal markers in EV preparations by immunoblotting. In accordance with ISEV guidelines [27], several proteins from different functional classes and with different intracellular compartmentalization were analyzed. The selected markers included tetraspanin CD9, the tumor susceptibility gene protein 101 (TSG-101) and Alix, known members of the ESCRT (endosomal sorting complex required for transport)-dependent pathway of exosome biogenesis; flotillin-2, a structural and functional component of membrane microdomains, as well as stomatin, which we proposed previously as a new exosomal marker [22]. PCNA was used to confirm the absence of cellular proteins of non-vesicular origin in EV preparations. Cell lysate of the gastrointestinal stromal tumor cell line, GIST-T1, and GC tissue lysate were used for the comparison of protein levels in EVs and cells. All proteins were analyzed in a single experiment, making it possible to compare the ratio of studied proteins in different EV preparations. The results of the analysis showed high variability in the composition of exosomal markers among individual EV preparations. In particular, we observed CD9-positive EVs devoid of all or some of the other markers; CD9-negative samples enriched with other markers, and EVs containing all the markers studied ( Figure 2A). To confirm the presence of EVs in samples with different combinations of exosomal markers, Figure 2B shows EV images from TEM analysis of the respective samples.
To find out whether there are differences between vesicles containing different sets of exosomal markers, we compared their size distributions. We found that the mean size of vesicles, as well as mode value (the particle size that appears most often in a set of data values) for CD9-positive and CD9-negative EVs, had significant differences (p < 0.01) ( Figure 1D). Thus, the average size and the mode value for CD9(+) EVs were 165 and 106 nm, and for CD9(−) EVs-116 and 58 nm, respectively. It is likely that EVs with different compositions of exosomal markers, in particular CD9(+) and CD9(−) vesicles, represent different subtypes of EVs, which may or may not be present in various combinations in individual GJ samples.
Remarkably, in addition to the full-length CD9 protein present in both cell lysates and most EV samples, a lower molecular weight protein was observed in EVs exclusively. We speculate that it corresponds to a fragmented CD9, resulting from the proteolytic activity of gastric proteases such as pepsin or gastricin, which could lead to the cleavage of tetraspanin domains located on the outer side of the EV membrane. Table S1) and non-cancer individuals (N1-N6). Full Western blot images can be found in Figure S1. The PCNA protein was used to confirm the absence of cellular proteins of non-vesicular origin in EV preparations. Protein lysates of GIST-T1 cells (Cntrl 1) and GC tissue (Cntrl2) were used as molecular weight controls and to compare levels of proteins in cells and EVs. Two bands of CD9 protein correspond to full-size form ( Notably, we showed for the first time the presence of stomatin protein in almost all GJ-derived EV samples. This finding confirms our previously published data linking this protein to the biogenesis of EVs [22]. Stomatin has never been studied in EVs except for EVs originating from blood cells. Another interesting observation is the presence of vesicles with non-canonical morphology, including elongated and multilayered ones ( Figure 3). Although, it cannot be ruled out that this shape is caused by distortions resulting from TEM analysis. Vesicles of similar morphology have been shown previously in several studies [28][29][30].
3.2. miR-135b-3p, miR-199a-3p and miR-451a Are Differently Presented in EVs from GJ of Gastric Cancer Patients and Non-Cancer Individuals Data from the Agilent 2100 Bioanalyzer revealed the wide spectrum of small RNAs presenting in GJ-derived EVs, including a peak of about 23 nucleotides corresponding to microRNAs. The percentage of microRNA in total small RNA varied drastically, reaching the maximum of 43%, as shown in Figure 4A. The relative expression of miR-451a, miR-199a-3p, miR-135b-3p (* p < 0.05); miR-204-3p and miR-135b-5p (p > 0.05) in EVs of GC patients (Tumor) and non-cancer individuals (Normal) from RT-qPCR data. Gene expression data were normalized to miR-23a. Fold change (FC) was determined using the ∆∆Ct method. miRNA levels in EVs from GJ of GC patients and non-cancer individuals were further analyzed by stem-loop RT-qPCR. Several miRNAs were selected for the initial study based on literature data, including hsa-miR-135b-3p, hsa-miR-135b-5p, hsa-miR-199a-3p, hsa-miR-204-3p, hsa-miR-451a, hsa-miR-16-5p and hsa-let-7b-5p. Sequences of primers used for individual miRNA analysis are shown in Supplementary Materials, Table S2. According to RT-qPCR data, one of the most equally expressed miRNAs in all EV samples was miR-23a-3p. As the levels of miR-16-5p as well as let-7b-5p, often used for miRNA normalization, varied significantly between EV samples, we used miR-23a-3p as a reference to assess miR-135b-3p, miR-135b-5p, miR-199a-3p, miR-204-3p and miR-451a (Ct values for each sample are shown in Supplementary Materials, Table S3). The difference between studied miRNA levels in the compared groups (fold change) was calculated using the ∆∆Ct method. We found miR-135b-3p and miR-199a-3p to be significantly upregulated (4.29-and 3.97-fold increase, respectively) and miR-451a to be downregulated (3.78-fold decrease) in EVs from GC patients compared to the control group (p < 0.05) ( Figure 4B). The level of miR-204-3p showed no significant difference between compared groups. Surprisingly, there were also no significant differences in miR-135b-5p expression, although upregulation of this miRNA in gastric cancer has been shown repeatedly. However, it should be noted the wide scattering of Ct values for this miRNA in EV specimens.
These results indicate a different composition of miRNAs in EVs obtained from GJ of GC patients and non-cancer individuals. However, the data on the differential expression of certain miRNAs obtained in this initial study should be interpreted with caution due to the strong variability in EV composition identified here. Further studies are needed to understand the reasons for this heterogeneity as well as to determine differences in the molecular composition of distinct subpopulations of vesicles characterized by different sets of exosomal markers.
Discussion
Exosomes and microvesicles belong to the secreted EVs, a heterogeneous group of cellderived membrane structures. EVs are involved in intercellular communication through the exchange of cargo biomolecules, consisting of proteins, lipids, metabolites and various types of nucleic acids. They are present in almost all body fluids and participate in multiple physiological and pathological processes mediating epigenetic regulation of gene expression and alterations in intracellular signaling [31,32].
The results of numerous studies indicate that EVs contribute to the malignant phenotype and the survival of primary tumor cells, regulating the processes of proliferation and apoptosis [33,34]. EVs also participate in tumor spread by enhancing the migratory and invasive activity of cancer cells and stimulation of angiogenesis [35][36][37][38]. In addition, EVs secreted by tumor cells are involved in the processes of the tumor-stroma interaction, reorganization of tissue microenvironment, pre-metastatic niche formation, and reprogramming of immune cells to evade anti-tumor immunity [7]. Repeatedly described similarities in molecular signatures, including proteome and transcriptome associations, between parental cells and secreted EVs have highlighted the potential of tumor-derived EV molecules as promising liquid biopsy markers for cancer diagnosis and monitoring.
The importance of EVs in the pathogenesis of gastric cancer has been confirmed by numerous studies on both experimental models [39,40] and clinical specimens [41]. For instance, several studies have shown that GC-derived exosomes promote tumor cell proliferation and invasion through activation of PI3K/Akt and MAPK/ERK-dependent signaling pathways [42,43]. Exosomes have also been shown to be involved in Treg cell formation through TGF-β1 activity and contribute to lymphogenic metastasis of gastric carcinoma [44,45]. Several data indicate that exosomes participate in mesothelialto-mesenchymal transition and promote peritoneal metastasis, a primary metastatic route in advanced GC [46][47][48]. Zhang et al. demonstrated that exosomes promote GC liver metastasis through the delivery of EGFR and rearrangement of the liver microenvironment [49].
The association of EV molecular composition with tumor malignancy and cancer progression has been repeatedly shown for many cancer types, including GC [50][51][52][53][54]. Furthermore, based on deep sequencing data, differences in miRNA expression profiles in exosomes derived from GC stem-like cells and differentiated GC cells were shown [55]. Numerous studies have focused on the identification of EV markers for the diagnosis and prognosis of gastric cancer [56,57]. Based on the differences in proteomic profiles of exosomes from the serum of GC patients and healthy controls, down-regulation of TRIM3 protein was suggested as a biomarker for GC diagnosis [58]. Similarly, a decrease in exosomal gastrokine-1 level has been shown to be associated with gastric cancer [59]. Among the miRNAs proposed as GC markers were miR-101, which has been significantly reduced in both exosomes and plasma of GC patients [60], and miR-23b, the level of which in plasma-derived exosomes was associated with recurrence and progression of GC [61]. Wang et al. identified a panel of serum exosomal miRNAs, including miR-19b-3p, miR-17-5p, miR-30a-5p, and miR-106a-5p for GC diagnosis [62]. Several studies demonstrate the potential significance of certain long noncoding RNAs such as LINC00152 and HOTTIP in relation to GC diagnosis and prognosis [63,64].
The above examples demonstrate the high potential of exosomes as a source of GC biomarkers. However, it is worth noting the low convergence of the data with respect to the specific molecules identified in the various studies. In addition to differences in methodological approaches to the isolation of EVs and analysis of their molecular composition, the inconsistency of the data can be explained by the choice of blood (plasma or serum) as a source of EVs. Blood is an extremely heterogeneous body fluid in terms of EV composition, in which the vast majority of EVs are produced by blood cells, immune cells and epithelial cells of different histogenesis, while only a very small percentage of vesicles are of tumor origin [17]. In addition, the composition of exosomes in the blood differs rather unpredictably according to a variety of factors, including gender, age, lifestyle and many other parameters [16].
In some studies, malignant ascites and peritoneal fluid have been used as a source of EVs. Such body fluids might contain a higher proportion of tumor-derived EVs and thus more fully reflect tumor-associated changes in EV composition. At the same time, this approach seems more effective for detecting prognostic markers and assessing recurrence than for diagnostic tasks. For example, several exosomal miRNAs from peritoneal fluid have been shown to be associated with peritoneal metastasis, including four miRNAs (miR-21-5p, miR-92a-3p, miR-223-3p and miR-342-3p) that were elevated and miR-29 family members that were decreased in patients with peritoneal metastases [13]. Based on the analysis of exosomes from GC malignant ascites, peritoneal lavage fluids, and conditioned media of GC cell lines, miR-21 and miR-1225-5p were identified as potential prognostic biomarkers of peritoneal metastasis [14]. In addition, reduced expression of exosomal miR-29 family in peritoneal fluid has also been shown as a predictor of peritoneal recurrence of GC [15].
GJ appears to be a very suitable source of EVs for the task of searching for diagnostic markers of GC, as the expected proportion of EVs originating from tumor cells and microenvironmental cells should be significantly higher in this body fluid compared to the circulation. In addition, unlike blood, GJ should not contain ribonucleoproteins and lipoprotein complexes, which almost inevitably contaminate EV preparations. Surprisingly, GJ-derived EVs have been hardly investigated so far, with the exception of the few above-mentioned studies [19][20][21].
We confirmed the presence of EVs in GJ and showed that they can be isolated by ultracentrifugation-based techniques. The obtained EVs were characterized according to ISEV recommendations [27], including the size and morphology of vesicles (determined by NTA and TEM), as well as the expression of exosomal markers belonging to different functional protein groups with different intracellular localization. TEM analysis revealed remarkable morphological heterogeneity of GJ-derived EVs. Particularly, unlike the EVs we observed in other sources, such as ascitic fluid, blood plasma, uterine lavage, or cell-conditioned media [22,65,66], preparations from GJ contain vesicles, very similar in shape to those previously described in several papers, such as so-called "double", "tubular" and "multilayered" vesicles [28][29][30]. It cannot be ruled out that such shapes are artifacts resulting from overlapping vesicles or other distortions caused by technical problems. Further studies using cryo-EM and other imaging techniques are needed to clarify this issue.
Another interesting observation is the high variability in the content of exosomal markers in GJ-derived EVs. Thus, tetraspanin CD9 was detected in 9 out of 13 samples; cytosolic proteins, including TSG-101 and Alix (components of the ESCRT complex)in 12 and 7 out of 13 EV preparations, respectively; the membrane protein flotillin-2 (a component of lipid microdomains)-in 10 out of 13 EV preparations. It is noteworthy that the indicated proteins were present in almost all combinations. That means, apparently, that EVs isolated from GJ by ultracentrifugation consist of different subpopulations of vesicles characterized by a distinct set of exosomal markers.
Variability in the composition of exosomal markers among different EV populations, presented in both body fluids and cell culture media, has been shown repeatedly. It is still not entirely clear what exactly accounts for these differences, and the data from various studies in this regard are rather contradictory. For instance, comparative proteomic analysis has identified four subcategories even in the category of small vesicles, namely: sEVs coenriched in CD63, CD9, and CD81 tetraspanins and endosome markers; sEVs devoid of CD63 and CD81 but enriched in CD9; sEVs devoid of CD63/CD9/CD81; and sEVs enriched in serum-or extracellular matrix-derived factors [67]. In contrast to these findings, another study states that sEVs bearing CD9 and CD81 with little CD63 correspond to ectosomes, whereas others bearing CD63 with little CD9 were qualified as exosomes [68]. Using Rab27a inhibition to modulate exosome secretion, Bobrie et al. showed the existence of at least two distinct populations of sEVs, the secretion of which was differently dependent on Rab27a, that is, the Rab27a-dependent subpopulation containing CD63, TSG-101, Alix and Hsc70, and the Rab27a-independent one enriched in CD9 and Mfge8 [69]. In contrast, based on a comparison of the protein content of EVs from 60 cell lines, it was shown that only CD81, Alix, and HSC70 were present across all samples, while other proteins, including CD63, CD9, TSG-101, syntenin-1, and flotillin-1, were present in at least two-thirds of the samples [70].
Such heterogeneity in the data can be attributed both to the natural heterogeneity of the vesicles and to the diversity of isolation methods, which may result in the enrichment of preparations with different subpopulations of vesicles [71] or particles of non-vesicular origin [72]. The association of exosomal marker composition with different vesicle subpopulations is also confirmed here by the revealed correlation between vesicle size and the presence of CD9. We suggest that CD9(−) vesicles may represent a distinct population of smaller EVs.
Another noteworthy feature is the presence of two forms of CD9 revealed by immunoblotting. The first one of 24 kDa corresponds to a full-length protein and is present in both cell lysates and in EVs, while the other is a lower molecular-weight (approximately 20 kDa) protein of unknown origin. We have not previously observed CD9 of this size in EVs from other origins, including blood plasma, ascites, aspirates and flushes from uterine cavity, culture medium, etc. [22,25]. We hypothesize that the truncated CD9 results from the proteolytic activity of gastric juice proteases, such as pepsin or gastricsin, which may lead to cleavage of the tetraspanin's extracellular domains located on the outer surface of the vesicle membrane.
Among other proteins presented in GJ-derived EVs we found stomatin. Stomatin, similarly to flotillins, belongs to the SPFH (stomatin/prohibitin/flotillin/HflK/C) protein family and colocalizes with flotillins in lipid microdomains (also called lipid rafts) [22,73,74], highly dynamic liquid-ordered subdomains enriched in sterols, sphingolipids and glycosphingolipids and involved in the compartmentalization of signaling and transport processes on the plasma membrane [75]. Although several data indicate that lipid rafts play an important role in the formation of intraluminal vesicles (intracellular precursors of exosomes) [76], stomatin has not previously been studied in relation to EVs. In our previous study, we first showed the presence of stomatin in exosomes produced by epithelial cancer cells (lung, breast and ovarian cancer cells) as well as in EVs from various body fluids, including blood plasma, ascitic fluid and uterine flushes. Based on the high content of stomatin in EVs of various origins and its enrichment in exosomes, we proposed this protein as a promising exosomal marker [22]. Here, we observed stomatin in almost all EV samples, confirming its ubiquitous presence in EVs of various origins and indicating its association with EV biogenesis.
Next, we confirmed the presence of various miRNAs in GJ-derived EVs by RT-qPCR. Several microRNAs were selected for the pilot analysis based on literature data, including miR-135b-3p, miR-199a-3p, miR-204-3p, miR-451a. Levels of miR-135b-3p were found to be significantly higher in EVs from GC patients compared to non-cancer individuals. This result confirms numerous data on the tumor-promoting role of miR-135b, both miR-135b-3p and miR-135b-5p forms in GC carcinogenesis. miR-135b is involved in epithelialmesenchymal transition, proliferation, apoptosis, migration, angiogenesis, and anticancer immunity through the regulation of a number of intracellular signaling pathways, including MAPK and PI3K/AKT cascades, FOXO, Wnt, and TGFβ-dependent pathways and several others [77][78][79][80][81]. The involvement of miR-135b in GC progression is also mediated through its intercellular transmission within the exosomal cargo [82,83]. It has also been shown that the expression level of miR-135b in plasma of GC patients is generally higher than in healthy individuals. Moreover, an increase in miR-135b has been proposed as a prognostic marker [81] as well as a diagnostic marker of GC [84]. This oncomiR was also among seven miRNAs identified as robust biomarkers for GC by a bioinformatic integrated analysis of differentially expressed miRNAs from five microarray datasets in the Gene Expression Omnibus database. In this study, seven miRNAs were filtered from fourteen primary miRNAs using the validation set of The Cancer Genome Atlas Program database [85].
Among the miRNAs identified in the same study was miR-204, the downregulation of which in GC has been suggested as a marker for early diagnosis of GC. At the same time, the results of our study showed no significant differences in miR-204-3p levels between GJ-derived EVs from GC and non-cancer samples. miR-199a-5p, according to the same study, has been classified as one of the most frequently altered in cancer but was categorized as a microRNA "with unclear expression changes in GC tissues compared to normal tissues or adjacent normal tissues". Indeed, the deregulation of miR-199a-5p in tumors and its involvement in carcinogenesis has been shown in many in vitro and in vivo studies [86]. In gastric cancer tissues, miR-199a-3p expression was shown to be upregulated in 69.2% of patients [87]. Consistent with these results, we found an increased content of miR-199a-3p in EVs from patients with GC compared to non-cancer patients. In the same samples, we found a significant decrease in miR-451a, indicating that this miRNA may act as a tumor suppressor. The negative role of this miRNA in cancer progression is well established. For example, Streleckiene et al. found that miR-451a is markedly deregulated and displays tumor-suppressive activity in GC through regulation of the PI3K/AKT/mTOR signaling pathway [88]. Su et al. showed decreased miR-451 expression in the GC tissues and cell lines and reported that downregulation of miR-451 tended to be positively correlated with lymphatic metastasis, advanced clinical stage, and shorter overall survival in patients with GC [89]. Shen et al. demonstrated a correlation of low miR-451 expression with tumor stage, lymphatic metastasis, and overall survival in patients with GC and suggested the downregulation of miR-451 as a diagnostic and prognostic biomarker in GC [90]. Similar results evidencing the prognostic significance of miR-451a have also been reported based on the investigation of tumor tissues and the clinicopathological features of 180 patients with GC [91].
Conclusions
In conclusion, we have shown for the first time that exosome-like EVs can be isolated from GJ by ultracentrifugation in amounts sufficient for further analysis of their molecular cargo, including miRNA composition. Analysis of exosomal protein markers revealed differences in size between CD9(+) and CD9(−) EV populations, indicating the existence of distinct subtypes of EVs in GJ. The results of the analysis of a pilot sampling of EVs showed a significant increase in miR-135b-3p and miR-199a-3p, as well as a decrease in miR-451a levels in EVs from GC patients compared to non-cancer individuals. The observed changes indicate for the first time the difference in the content of miRNAs in EVs present in GJ of GC patients and healthy individuals. Thus, EVs derived from GJ are a promising source of miRNA markers of gastric cancer.
Supplementary Materials: The following supporting information can be downloaded at: [URL]: //www.mdpi.com/article/10.3390/cancers14143314/s1, Figure S1: Full Western Blot images from Figure 2A, Table S1: Clinical and morphological characteristics of GC patients and NTA data on EV size distribution and concentration, Table S2: Sequences of reverse transcription primers used; sequences of RT-qPCR primers and TaqMan ™ probes used, Table S3: Average Ct values of hsa-miR-199a-3p, hsa-miR-204-3p, hsa-miR-451a, hsa-miR-23a-3p, hsa-miR-16-5p, has-let-7b-5p obtained by RT-qPCR in EVs isolated from gastric juice of patients suffering from gastric cancer and non-cancer individuals. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patients to publish this paper.
Data Availability Statement:
The data presented in this study are available in the article and Supplementary Materials.
==
Domain: Biology Engineering
|
[{"added": "2022-07-10T15:07:44.790Z", "created": "2022-07-01T00:00:00.000", "id": "250403838", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2072-6694/14/14/3314/pdf?version=1657199934", "pdf_hash": "8918b9593f7e489f33c1740f76229ca67bfe289c", "pdf_src": "MergedPDFExtraction", "provenance": "peS2o-0028.json.gz:5596", "s2fieldsofstudy": ["Biology", "Engineering"], "sha1": "c45c430c960a7398a2afc72ffe5d9057fe0f23a1", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
The Arabidopsis ortholog of rice DWARF27 acts upstream of MAX1 in the control of plant development by strigolactones.
Strigolactones (SLs) are carotenoid-derived plant hormones that regulate shoot branching, secondary growth, root development, and responses to soil phosphate. In Arabidopsis (Arabidopsis thaliana), SL biosynthesis requires the sequential action of two carotenoid cleavage dioxygenases, MORE AXILLARY GROWTH3 (MAX3) and MAX4, followed by a cytochrome P450, MAX1. In rice (Oryza sativa), the plastid-localized protein DWARF27 (OsD27) is also necessary for SL biosynthesis, but the equivalent gene in Arabidopsis has not been identified. Here, we use phylogenetic analysis of D27-like sequences from photosynthetic organisms to identify AtD27, the likely Arabidopsis ortholog of OsD27. Using reverse genetics, we show that AtD27 is required for the inhibition of secondary bud outgrowth and that exogenous application of the synthetic SL GR24 can rescue the increased branching phenotype of an Atd27 mutant. Furthermore, we use grafting to demonstrate that AtD27 operates on a nonmobile precursor upstream of MAX1 in the SL biosynthesis pathway. Consistent with the plastid localization of OsD27, we also show that AtD27 possesses a functional plastid transit peptide. We demonstrate that AtD27 transcripts are subject to both local feedback and auxin-dependent signals, albeit to a lesser extent than MAX3 and MAX4, suggesting that early steps in SL biosynthesis are coregulated at the transcriptional level. By identifying an additional component of the canonical SL biosynthesis pathway in Arabidopsis, we provide a new tool to investigate the regulation of shoot branching and other SL-dependent developmental processes.
Strigolactones (SLs) are a relatively new class of plant hormones that control multiple facets of plant growth and development. SLs were originally identified from root exudates as organic compounds that stimulated the germination of particular parasitic weeds (Cook et al., 1966;Xie et al., 2010) and were more recently found to promote the symbiosis between plants and mycorrhizal fungi (Akiyama et al., 2005;Dor et al., 2011). The most widely characterized role of SL in plant development is in the inhibition of shoot branching (Dun et al., 2009a;Domagalska and Leyser, 2011). In addition to enhanced shoot branching, SL-deficient and SL response mutants also exhibit reduced stature, suppressed development of the cambium ring in the main stem (secondary growth), and enhanced lateral and adventitious root development (Stirnberg et al., 2002;Arite et al., 2007;Agusti et al., 2011;Kapulnik et al., 2011;Rasmussen et al., 2012). It is also likely that SLs form part of the signaling system used for responding to available soil nutrients. In wild-type plants, SL content is greatly enhanced under phosphate deficiency, which promotes lateral root and root hair lengthening and mycorrhizal associations, while suppressing aerial growth via reduced shoot branching (Umehara et al., 2010;Kapulnik et al., 2011;Kohlen et al., 2011;Ruyter-Spira et al., 2011).
SLs are carotenoid-derived terpenoid lactones. Only four enzymes have been attributed to this biosynthetic pathway, and only three of these have been described in Arabidopsis (Arabidopsis thaliana; Xie et al., 2010). These are encoded by the MORE AXILLARY GROWTH genes MAX1, MAX3, and MAX4. MAX3 and MAX4 are carotenoid cleavage dioxygenases (CCD7 and CCD8, respectively) that act in the chloroplast to cleave a carotenoid substrate. MAX1, a cytochrome P450, acts downstream on a mobile intermediate that can traverse a graft union. In Arabidopsis, MAX3 and MAX4 are largely expressed in the vasculature and function in the shoot and root. MAX3 and MAX4 transcripts are regulated both by auxin and by long-and short-distance feedback systems that depend on SL signaling and 1 This work was supported by the Australian Research Council (grant nos. DP1096717 to S. M. S. and DP110100997 to C. A. B.) and by a University of Western Australia-University of Queensland Bilateral Research Collaboration award (to S. M. S. and C. A. B.). * Corresponding author; e-mailThe author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Christine A. Beveridge ([email protected]).
[C] Some figures in this article are displayed in color online but in black and white in the print edition. [W] The online version of this article contains Web-only data. [OA] Open Access articles can be viewed online without a subscription.
www.plantphysiol.org/cgi/doi/10.1104/pp. 112.196253 response (Bennett et al., 2006;Brewer et al., 2009;Hayward et al., 2009). Based on impaired SL response in corresponding mutants of Arabidopsis and rice (Oryza sativa), two proteins have been identified as necessary for this process: the F-box protein AtMAX2/ OsD3 (Gomez-Roldan et al., 2008;Umehara et al., 2008) and the a/b-fold hydrolase family protein AtD14/OsD14 (Arite et al., 2009;Waters et al., 2012). Relative to wild-type plants, the SL response mutants produce elevated amounts of SL, presumably due to impaired feedback on SL biosynthesis (Umehara et al., 2008;Arite et al., 2009;Kohlen et al., 2011). Accordingly, MAX3 and MAX4 gene expression is greatly enhanced in max2 mutants . In rice, a plastid-targeted protein, DWARF27 (D27), is required for SL biosynthesis (Lin et al., 2009). As with other SL mutants in rice, the d27 mutant has a highly tillered and dwarf habit, although the phenotype of d27 is intermediate between the wild type and other characterized SL-related mutants. Like MAX3 and MAX4 SL biosynthesis genes, D27 is expressed in the vasculature. D27 also has strong expression in axillary buds (Lin et al., 2009). Recent in vitro evidence suggests that D27 is a b-carotene isomerase that converts all-trans-b-carotene into 9-cis-b-carotene, which is then further processed by CCD7 and CCD8 into a SL precursor termed carlactone (Alder et al., 2012). If this biochemical activity is also the case in planta, it is likely that D27 acts upstream of MAX1 in the control of the biosynthesis of a mobile shoot-branching signal. The availability of an equivalent d27 mutant in a species amenable to grafting would allow a more direct test of this hypothesis. Moreover, it will allow an extended genetic, biochemical, and physiological analysis of SL biosynthesis and its regulation. In this study, we identify the Arabidopsis ortholog of OsD27 and characterize it with respect to its SL response, expression profile, and regulation and use grafting to place it within the SL biosynthesis pathway.
At1g03055 Is Orthologous to OsD27
The Arabidopsis genome encodes three proteins with superficial similarity to rice OsD27. To identify the likely ortholog of OsD27, we performed a phylogenetic analysis of D27-like proteins from land plants, green algae, and cyanobacteria. The eukaryotic sequences broadly grouped into three distinct clades, two of which (clades 1 and 2) only contained representatives from land plants, while the third (clade 3) also contained members from chlorophyte algae and diatoms (Fig. 1). With three cyanobacterial proteins serving as an outgroup, the phylogeny suggests that there was a single D27-like gene copy in the common ancestor of the chlorophyte algae and the land plants. Subsequently, the first of two gene duplication events took place in the lineage leading to the land plants, giving rise to clades 1 and 2. The long branches within clade 3, indicative of long periods of sequence divergence, further suggest that this group of proteins arose prior to, and is functionally distinct from, clades 1 and 2. Intriguingly, several gene duplication events have occurred independently within the individual algal lineages, leading to extensive paralogy within clade 3.
The clades specific to land plants, clades 1 and 2, both had strong Bayesian posterior support, and the two groups together formed a superclade that was also strongly supported. Clades 1 and 2 each contained a representative sequence from the moss Physcomitrella patens, indicating that each group derived from a second gene duplication event that occurred during the early evolution of the land plants. Clade 1 contained OsD27, its ortholog in Medicago truncatula , and an Arabidopsis protein encoded by At1g03055. Notably, two other Arabidopsis proteins with similarity to OsD27, encoded by At1g64680 and At4g01995, unambiguously were assigned to clades 2 and 3, respectively. Thus, we concluded that At1g03055 is the likely Arabidopsis ortholog of OsD27. Nevertheless, the presence of other D27-like proteins in land plant genomes is indicative of potential genetic redundancy.
The AtD27 transcript comprises seven exons, and the exon length and structure are highly conserved with the MtD27 and OsD27 mRNAs, further supporting orthology between the three genes (Fig. 1B). The AtD27 protein is predicted to be 264 amino acids long and shares 48% pairwise identity with OsD27 (excluding the predicted plastid transit peptides; Supplemental Fig. S1). At present, the D27 protein family does not possess any annotated InterPro domains and belongs to a "domain of unknown function" family (Pfam: DUF4033) of 91 eukaryotic and eubacterial sequences, which are widely represented in Figure 1A. However, the rice D27 protein binds a nonheme iron cofactor, which presumably is required for redox-based activity as a b-carotene isomerase (Lin et al., 2009;Alder et al., 2012).
AtD27 Is Required for the Regulation of Axillary Branching
A search of available germplasm collections for mutations in AtD27 revealed three possible T-DNA insertion alleles among the GABI-Kat collection (Kleinboelting et al., 2012). Of these, two were confirmed to carry a T-DNA insertion within the At1g03055 locus. A T-DNA copy resides within the fifth exon in GK-134E08 and in the seventh exon in GK-774D06 ( Fig. 2A). Reverse transcription (RT)-PCR analysis showed that both alleles result in the accumulation of incomplete AtD27 transcripts, consistent with the predicted T-DNA location (Fig. 2B). However, only one of these two alleles resulted in a clear mutant phenotype: in plants homozygous for the GK-134E08 insertion, adult plants exhibited a substantial increase in axillary rosette branches relative to wild-type controls, while GK- Figure 1. At1g03055 encodes the Arabidopsis ortholog of rice D27. A, Bayesian inference phylogeny produced from an alignment of D27-like protein sequences from land plants, green algae, and cyanobacteria. Topology support for nodes is given by posterior probability. Terminals are labeled with the genus name followed by a sequence identifier, as detailed in Supplemental Table S2. The tree is rooted on the cyanobacterial clade, based on their ancestral relationship to chloroplasts in photosynthetic eukaryotes. B, Exon structure of AtD27, MtD27, and OsD27 mRNAs. A block arrow denotes each exon. Numbers indicate the length of each exon in bp. Protein-coding sequences are indicated by gray fill, and white fill denotes noncoding untranslated regions. The two nucleotides on either side of each exon-exon interface are indicated, with splice junctions denoted by a forward slash (/). [See online article for color version of this figure.] 774D06 homozygotes had a normal branching pattern (Fig. 2, C and D). On the basis of its mutant phenotype, we named the GK-134E08 allele d27-1 (hereafter referred to as d27 for simplicity). The T-DNA insertion in d27 disrupts the predicted protein at Leu-189, comparable to the nonfunctional protein produced in Osd27, which is disrupted after Gly-187 (Supplemental Fig. S1; Lin et al., 2009). These data suggest that AtD27 is necessary for the regulation of axillary bud outgrowth and that the extreme 39 end of the transcript is nonessential for the function of the resulting protein.
To confirm that the d27 lesion was responsible for the increased branching phenotype, we complemented the mutant with a cDNA encoding the predicted D27 protein, expressed under the control of the cauliflower mosaic virus 35S promoter. We isolated four independent transgenic lines, in which the transgene segregated in a ratio consistent with a single Mendelian locus and was expressed approximately 40-fold higher than in the wild type at the transcriptional level (Supplemental Fig. S2). In all four lines, the number of axillary branches and overall plant height were not significantly different from wild-type levels, fully complementing the d27 phenotype (Fig. 3, A and B). We performed feeding experiments with GR24, a synthetic SL, to evaluate whether the increased branching phenotype of d27 likely results from impaired SL biosynthesis, as is the case in rice (Lin et al., 2009). When grown in culture vessels on medium containing GR24, d27 plants produced significantly fewer branches than those grown on control medium (Fig. 3C). The SL-deficient mutant max3-11 exhibited a similar response, although its branching phenotype was more severe. In contrast, the SL-insensitive mutant Atd14 showed no response at all to GR24, as demonstrated previously . A similar response to GR24 was seen with d27 and max3-9 plants grown hydroponically (Supplemental Fig. S3). Together, these data demonstrate that D27 is essential for the regulation of axillary branching and that branching inhibition in d27 is limited by SL content rather than SL response.
D27 Acts Upstream of MAX1 in the SL Biosynthetic Pathway
Grafting wild-type rootstocks to max1, max3, or max4 scions restores a wild-type branching pattern to the mutant shoots (Booker et al., 2005). This outcome results from the graft transmission of root-synthesized SL to the shoot presumably via the transpiration stream (Foo et al., 2001;Kohlen et al., 2011). Conversely, grafting with these lines shows that SL biosynthesis also occurs in the shoot, as wild-type shoots have wild-type branching phenotypes even where grafted with mutant rootstocks. To verify that d27 shoots are similarly responsive to a graft-transmissible signal, we generated reciprocal grafts between wildtype ecotype Columbia (Col-0) and d27 and found that wild-type rootstocks could suppress the increased branching phenotype of d27 shoots (Fig. 4). A mutant rootstock had no effect on the phenotype of a wildtype shoot, showing that D27 expression in the shoot is sufficient to inhibit branching (Fig. 4).
It has been shown previously that MAX1 operates downstream of MAX3 and MAX4 in SL biosynthesis, because a max1 rootstock can restore a wild-type branching pattern to max3 and max4 shoots, but not vice versa (Booker et al., 2005). These findings have also demonstrated that a compound intermediate between MAX3/MAX4 and MAX1 is subject to longdistance transport from the root to the shoot. Likewise, the substrates of MAX3 and MAX4 are nonmobile (or are unable to access MAX3/MAX4 in the plastid of a remote cell), because a max3 rootstock fails to complement a max4 shoot, and vice versa (Booker et al., 2005). To investigate the relative position of D27 within the MAX pathway, we generated reciprocal grafts between d27 and either max1 or max4. A max1 rootstock was able to fully restore the branching phenotype of a d27 shoot, demonstrating that MAX1 operates downstream of D27 (Fig. 4). Interestingly, as for max4 discussed below, the branching of max1 scions grafted to d27 rootstocks was reduced to that of d27 self-grafts (Fig. 4). This suggests that d27 may generate a small amount of SL, or a substance that can partly compensate for SL deficiency, and is consistent with the relatively mild branching phenotype of d27 compared with max1, max3, and max4 (Figs. 3C and 4; Supplemental Fig. S3).
In contrast to graft combinations involving d27 with the wild type or max1, reciprocal grafts between d27 and max4 failed to restore branching to wild-type levels in either mutant shoot ( Fig. 4; Sorefan et al., 2003;Hayward et al., 2009). This situation is similar to grafts between the carotenoid cleavage dioxygenase mutants of Arabidopsis (max3 and max4) and pea (Pisum sativum; rms1 and rms5), which, unlike grafts with wild-type partners, are unable to show branching inhibition (Morris et al., 2001;Booker et al., 2005). As discussed below, this is consistent with D27 acting in the plastid on a plastid-localized, nonmobile substrate. Unlike grafts between max3 and max4, max4 rootstocks caused a slight enhancement of branching in d27 shoots and d27 rootstocks caused a slight reduction of branching in max4 shoots. As d27 has an intermediate branching phenotype compared with max4 ( Fig. 4), d27 presumably has elevated graft-transmissible hormone content compared with max4. In summary, these grafting data are consistent with D27 functioning upstream of MAX1 in the biosynthesis of SL.
AtD27 Is Localized to the Plastid
Given the apparent nonmobility of the D27 precursor and the biochemical function of D27, it is likely that the Arabidopsis D27 protein is spatially restricted to the plastid. To determine the likely cellular location of D27, we generated a chimeric construct encoding the full-length D27 protein fused to the N terminus of GFP. Following transient expression in onion (Allium cepa) epidermis, the GFP fluorescence fully overlapped with the red fluorescence from the coexpressed plastid-specific marker (Fig. 5, A-D). No GFP fluorescence was detected beyond the plastid, suggesting that the full-length D27 protein was strictly localized to this organelle. This is in accord with plastid localization of OsD27 (Lin et al., 2009). The D27 protein encodes a predicted N-terminal plastid transit peptide of 47 amino acids (Supplemental Fig. S1). To test whether this presequence is sufficient to direct plastid localization, we generated a translational fusion between the first 50 amino acids of D27 and GFP and subjected onion epidermal cells to transient transformation as before. Again, the fluorescence signals from both the GFP and the plastid-specific marker fully overlapped (Fig. 5, E-H), indicating that D27 possesses a bona fide plastid transit peptide. Together, these data suggest that D27 functions in the plastid, consistent with the demonstrated cellular location of OsD27 in rice and its anticipated activity on a nonmobile precursor.
D27 Transcripts Are Subject to Auxin and Feedback Control D27, MAX3, and MAX4 transcript abundance from public microarray data revealed potential differential expression intensities throughout the plant, with a relative deficit of D27 in root tissue (Supplemental Fig. S4). Quantitative RT-PCR analysis of Col-0 tissues confirmed that D27 transcripts were more abundant in shoot tissue than in roots of 3-week-old wild-type plants, while MAX3 and MAX4 transcripts accumulated most strongly in the hypocotyl and roots (Fig. 6A). In contrast, MAX1 transcript levels varied little over the tissues examined. Consistent with their role in SL response, we found that MAX2 and AtD14 transcripts were evenly expressed across the shoot, hypocotyl, and root (Fig. 6A). Differential spatial expression of D27 compared with MAX3 and MAX4 then led us to investigate whether D27 was under similar feedback control as MAX3 and MAX4.
In all max mutants, levels of MAX3 and MAX4 transcripts in the hypocotyl are 6-to 10-fold higher than in the wild type, indicative of feedback up-regulation . This feedback process is dependent on a combination of auxin and nonauxin Figure 5. The N terminus of Arabidopsis D27 is sufficient for plastid targeting. A and E, Epifluorescence micrographs of onion epidermis transiently transformed with plasmids encoding a full-length D27-GFP protein fusion (A) or the first 50 amino acids of D27 fused to the N terminus of GFP (E). B and F, Plastids are positively identified by coexpression of a protein fusion between the full-length Rubisco small subunit from pea (SSU) and RFP. C and G, The merged images confirm colocalization of the two fluorescent signals. Insets in A to C provide magnification and show clustering of plastids around the cell nucleus. D and H, Brightfield micrographs of the same field of view to provide the cell outline. Bars = 50 mm (main images) and 20 mm (insets).
signals. SL-related auxin responses require the AUXIN RESISTANT1 (AXR1) protein, as axr1 auxin response mutants have reduced MAX3 and MAX4 transcript levels, at least in cauline internodes and hypocotyls of 5-week-old plants . In hypocotyls of 3-week-old prebolting max2 and max4 mutants, D27 transcripts were significantly up-regulated relative to the wild type (Fig. 6B). However, these changes were typically half the increase observed for MAX3 transcripts, which were over 10-fold and 5-fold up-regulated in a max2 and max4 background, respectively (Fig. 6B). Likewise, while MAX3 transcripts Figure 6. D27 transcripts are sensitive to auxin and feedback regulation. A, Accumulation of SL-and auxin-related transcripts in shoots (leaves and stem above the cotyledons), hypocotyls, and roots of wild-type plants grown hydroponically for 3 weeks. Transcripts were normalized to TIP41L (At4g34270) and scaled to the value in the shoot sample. Data are means 6 SE (n = 3 pools of 8-10 plants). B, MAX3 and D27 transcript accumulation in hypocotyls of soil-grown, prebolting plants of the indicated genotypes, as measured by quantitative RT-PCR. Transcripts were normalized to the combined expression of three ACTIN genes (At3g18780, At5g09810, and At1g49240) and scaled relative to the expression level in Col-0. Data are means 6 SE (n = 4 pools of approximately 20 plants). C, Auxin-dependent regulation of MAX3, MAX4, and D27 transcript levels in hypocotyls of 17-d old, soil-grown wild-type plants. Treatments are as follows: IAA addition (in a lanolin ring around the top of the hypocotyl); decapitation (decap); decapitation and IAA addition (in lanolin added to the stump); addition of the auxin transport inhibitor NPA (in a lanolin ring around the top of the hypocotyl); and combined addition of IAA and NPA (together in a lanolin ring around the top of the hypocotyl). Treatment time was 4 h. Data are means 6 SE (n = 3 pools of approximately 20 plants per pool), normalized to ACTIN and scaled as above. D, SL-and auxin-related gene expression in rootstocks of reciprocal grafts between the Col-0 wild type (WT) and d27-1. For each shoot/rootstock combination, hypocotyl and upper root tissue was harvested below the graft union. Data are means 6 SE (n = 3 pools of approximately 4 plants per pool), normalized to ACTIN and scaled as above. Values for D27 transcripts in d27-1 mutants were omitted. They were 13.8-6 0.8-fold higher than in Col-0, but this difference is not biologically meaningful because the mutation may affect transcript processing and/or stability. The quantitative PCR primers for D27 are located upstream of the T-DNA insertion site. Asterisks denote significant differences from Col-0 (* P , 0.05, ** P , 0.01, one-way ANOVA).
were 5-fold down-regulated in axr1 mutants, D27 transcripts were only moderately (albeit significantly) less abundant. Consequently, although the magnitude of expression changes for D27 are less than for MAX3, the response is similar and therefore may involve common regulatory components.
Levels of MAX3 and MAX4 transcripts in the hypocotyl decrease in response to removal of the shoot tip (decapitation) by decreasing expression levels ). This presumably reduces SL levels in order to allow buds below the decapitation site to grow and compensate for the missing shoot tip. A proportion of the decrease in gene expression is due to auxin depletion in the hypocotyl after removal of the shoot tip (the shoot tip being the main source of auxin for the plant). We checked D27 gene expression in hypocotyls after various treatments known to modulate auxin content: auxin (indole-3-acetic acid [IAA]) addition; decapitation; decapitation followed by IAA addition; addition of the auxin transport inhibitor N-1-naphthylphthalamic acid (NPA); and combined addition of IAA and NPA. D27 expression showed similar trends to that of MAX3 and MAX4 across these treatments, although, as above, the magnitude of the response was less (Fig. 6C). Classical auxin-responsive genes, IAA1, IAA5, and PIN1, also responded in a similar manner in these samples, but with varied magnitudes (Supplemental Fig. S5). The only difference between trends among the treatments for the classical auxin response genes compared with the SL genes was that exogenous IAA added to NPAtreated plants fully restored and indeed overcompensated for the NPA, whereas the SL genes were not fully restored back to control levels in this IAA+NPA treatment. This raises the possibility of auxin and nonauxin effects of NPA regulation of MAX3, MAX4, and D27 gene expression.
In pea, there is evidence that feedback regulation of MAX3 and MAX4 orthologs can occur over long distances in a graft-transmissible and localized manner (Foo et al., 2005;Johnson et al., 2006;Dun et al., 2009b). In Arabidopsis, a wild-type shoot can suppress the elevated MAX3 and MAX4 expression in a max2 rootstock ), indicative of a downwardly mobile feedback signal. Nevertheless, in Arabidopsis, the local effect of max2 in the rootstock is greater than that of the long-distance signaling from shoots. In similar experiments in pea, the effect of the scion appears greater than the local effect of the rootstock (Foo et al., 2005;Johnson et al., 2006). In contrast to SL response mutants, grafts between the wild type and SL-deficient mutants in pea show a minor effect of the scion on gene expression in the rootstock, possibly due to the restoration of SL content by the wildtype graft partner. To investigate the scenario for d27, we analyzed gene expression in the hypocotyl and rootstock of plants reciprocally grafted with the wild type (Fig. 6D). In d27 self-grafts, MAX3 and MAX4 transcripts were up-regulated, consistent with feedback. (D27 transcripts were also highly overexpressed, but this may be primarily a result of misregulation of a nonfunctional transcript.) As with wild-type rootstocks of pea grafted to mutants in SL biosynthesis (Dun et al., 2009b), expression in the root was not significantly affected by the genotype of the shoot in either reciprocal graft in Arabidopsis. As in pea, this may be because wild-type rootstocks can restore branching, and presumably the long-distance feedback signal, to normal levels in d27 shoots. Notably, in grafts involving d27, the three auxin-responsive transcripts did not exhibit a statistically significant difference from wild-type self-grafts, suggesting that auxin transport and/or signaling is not substantially affected in d27 mutants. This outcome may be due to the relatively moderate phenotype of d27.
DISCUSSION
The identification of AtD27 as the ortholog of OsD27 is supported by phenotypic, physiological, and genetic data. From a genetic point of view, we identified the sequence with greatest similarity to OsD27 ( Fig. 1; Supplemental Fig. S1), obtained homozygous T-DNA insertion Atd27 mutant lines (Fig. 2), and complemented this mutant phenotype to the wild type with overexpression of AtD27 ( Fig. 3; Supplemental Fig. S2). Like the d27 mutant in rice, Atd27 affects the number of axillary shoots and plant height (Fig. 3). Interestingly, the Atd27 mutant is also similar to Osd27 in the sense that both mutants are intermediate in phenotype compared with their respective wild types and other characterized SL mutants (Fig. 3). As expected, Atd27 responds to GR24 supplied through the roots ( Fig. 3; Supplemental Fig. S3).
Our phylogenetic analysis demonstrates that AtD27 belongs to a wider family of proteins that originated early in the evolution of photosynthetic eukaryotes. Although D27-like proteins are found throughout cyanobacteria and green algae, two clades are strictly specific to land plants. Notably, CCD7 (MAX3) and CCD8 (MAX4) form monophyletic clades in land plants, and putative orthologs of both CCD7 and CCD8 are present in Chlamydomonas reinhardtii and other photosynthetic eukaryotes (Ledger et al., 2010;Vallabhaneni et al., 2010;Proust et al., 2011). Thus, the enzymatic machinery necessary for the early steps of SL biosynthesis may have been present in the ancestors of the land plants. Interestingly, all land plants so far examined possess homologs of the proteins required for SL response (i.e. MAX2 and D14), but so far, no homologs have been found in other photosynthetic eukaryotes, including the streptophyte algae, the closest living relatives of the land plants (Waters et al., 2011. Although this finding could reflect incomplete genome sequence data for the streptophyte algae, it is possible that SL signaling coincided with the evolution of the land plants and the concurrent development of complex body plans. As carotenoids are essential pigments associated with light harvesting and photoprotection in oxygenic photosynthetic organisms, carotenoid metabolism is a fundamental requirement for photosynthesis and likely the original function of CCD7-, CCD8-, and D27-like proteins. As multicellularity evolved, we propose that gene duplications permitted the acquisition of novel gene function, and a separate pathway for carotenoid breakdown was co-opted for signaling purposes. Recently, SLs were shown to have a dramatic role in gametophytic development of the moss P. patens. Loss of PpCCD8 activity leads to severely compromised SL profiles and impaired control of protonemal branching, which in turn limits the ability to sense neighboring colonies (Proust et al., 2011). As the P. patens genome also encodes a D27 ortholog, multiple D14 homologs, and a single MAX2 ortholog, it appears that the SL biosynthesis and signaling system is conserved among land plants. The availability of complete genome sequences for the streptophyte algae, combined with an assessment of their ability to produce SLs, will help in determining the origin of SL signaling.
We have contributed further to the understanding of the D27 protein by demonstrating that the Arabidopsis version of D27 functions upstream of MAX1 in the production of SL-related signal(s) and displays similar feedback control compared with other genes encoding plastid-localized SL biosynthesis proteins. AtD27 controls a graft-transmissible signal because branching can be inhibited in Atd27 mutant shoots by grafting to wild-type rootstocks (Fig. 4). Whereas max1 mutant rootstocks retain this inhibitory function, max4 rootstocks do not. This is clear evidence in support of D27 acting upstream of MAX1. On the other hand, the grafting data do not permit the ordering of D27 with respect to MAX4, presumably because the carotenoid-derived substrates of D27 and MAX4 are unable to leave the plastid (Alder et al., 2012). Similarly, the relationship between MAX3 and MAX4 could not be determined via grafting (Booker et al., 2005). Instead, biochemical studies have shown that MAX3/CCD7 and MAX4/CCD8 likely act sequentially in carotenoid cleavage (Schwartz et al., 2004). The precise functions of MAX3 and MAX4 were only recently elucidated following the identification of D27 as a b-carotene isomerase (Alder et al., 2012). In vitro, the combination of purified D27, CCD7, and CCD8 is sufficient to convert b-carotene into carlactone, while exogenous carlactone is sufficient to rescue the high-tillering phenotype of rice d27, ccd7, and ccd8 mutants (Alder et al., 2012). Thus, there are at least three enzymatic steps in the plastid before the generation of a mobile SL precursor (Fig. 7). These findings are consistent with our grafting studies. At present, it is unclear whether carlactone itself can exit the plastid and mobilize and whether further modifications by subsequent enzymes are required for the transport of the branching signal from root to shoot.
Like the d27 mutant in rice, which is substantially SL deficient, the Arabidopsis d27-1 shows a weaker phenotype than other branching mutants in the MAX/ D/RMS pathway. Our grafting studies support the notion of residual bioactive branching inhibition capacity in d27, because d27 rootstocks can partly inhibit branching in max1 and max4 scions down to the level of d27 control plants (Fig. 4). Based on our phylogenetic analysis, it is possible that another D27-like protein can perform the isomerization of all-transb-carotene in the absence of functional D27. Although the function of At1g64680 is unknown, the predicted protein has been localized to the chloroplast envelope and/or thylakoids by liquid chromatographytandem mass spectrometry analysis (Ferro et al., 2010), while its closest homolog in rice, encoded by Os08g0114100, is weakly predicted to be plastidic by TargetP software (Emanuelsson et al., 2007). Alternatively, nonenzymic isomerization may provide a small amount of 9-cis-b-carotene substrate for MAX3/ CCD7, allowing only limited levels of SL to be synthesized in the d27 mutant (despite enhanced MAX3 and MAX4 expression). However, the rice d27 mutant does not contain any detectable SL, either by liquid chromatography-mass spectrometry or a bioassay based on Orobanche seed germination (Lin et al., 2009), which may suggest that any residual SLs are below the limit of detection in d27. On the other hand, max1 and max4 mutants, which exhibit a strong branching phenotype, still contain detectable levels of orobanchol . This contradiction means that the basis for the relatively mild branching phenotype of d27 in both rice and Arabidopsis is currently unclear. Therefore, detailed biochemical analyses of the d27 Figure 7. Position of D27 within the SL biosynthesis and signaling pathway. Enzymatic steps are denoted with solid arrows, transport steps with dashed arrows, and signaling steps with thick arrows. AtD14 may be a SL receptor, a hydrolytic enzyme that processes SL into a further active compound, or a combination of both; for a discussion, see Scaffidi et al. (2012). [See online article for color version of this figure. ] mutant, intact or grafted, compared with other mutant and wild-type lines may be useful in evaluation of the levels and profiles of endogenous SLs and their potential correlations with phenotypic responses.
Interestingly, despite the complete rescue of d27 shoots by grafting to wild-type rootstocks, the expression of D27 is not greatest in the rootstock. Indeed, its expression in roots and hypocotyls of 3-week-old plants is greatly and slightly reduced, respectively, compared with shoots (Fig. 6A). This regulation is a point of difference with MAX3 and MAX4, which are both highly expressed in the hypocotyl. In a further contrast, MAX3 is relatively weakly expressed in roots compared with shoots, while MAX4 is again highly expressed. The expression patterns that we observed are fully consistent with meta-profiles derived from multiple microarray experiments (Supplemental Fig. S4). Recently, it was found that two GRAS-type transcription factors, NSP1 and NSP2, regulate D27 expression and SL levels in Medicago and rice . In Medicago, the nsp1 mutant does not produce any detectable SL while the nsp2 mutant exhibits a specific deficiency in didehydro-orobanchol, and these deficiencies correlate with a decrease in MtD27 transcripts in both mutants. Interestingly, MtCCD7 and MtCCD8 expression was unaffected in nsp1 and nsp2 mutants , suggesting that the carotenoid cleavage steps in SL biosynthesis are under distinct transcriptional control from the initial isomerization step involving D27. This finding is consistent with our data from Arabidopsis. Although it is unclear whether MtD27 transcription is directly regulated by NSP1 and NSP2, the conservation between rice and Medicago of this regulatory system implies that a similar mechanism might operate in Arabidopsis and beyond. The identification of putative NSP1 and NSP2 orthologs in Arabidopsis (Zhu et al., 2006) opens the door for reverse genetics to address this question.
In contrast to the differences in relative expression among different tissues, the SL biosynthesis genes D27, MAX3, and MAX4 show similar trends in response to auxin-related treatments and in different mutant backgrounds. Although D27 transcripts show lower magnitudes of response across all experiments, they show similar trends to other SL biosynthesis transcripts. D27 expression is enhanced in the hypocotyl in SL-deficient or SL response mutant backgrounds, and it appears to be up-regulated by auxin. The auxin response is indicated by reduced expression in axr1 mutants and in response to auxin addition or depletion via decapitation and NPA treatment. As shown from grafted plants, d27 mutants also show enhanced expression of MAX3 and MAX4, which is again consistent with SL deficiency causing the up-regulation of these genes. Restoration of SL content in the mutant shoots by wild-type rootstocks means we cannot easily test whether a component of this feedback is graft transmissible . However, the majority of the feedback induced by d27 is likely to be local, since d27 rootstocks cause high expression of MAX3 and MAX4 similar to mutant self-grafts even where the scion is the wild type.
The regulation by auxin of a third protein in the SL pathway further supports the hypothesis that SLs act downstream of auxin in the control of shoot branching (Brewer et al., 2009). However, again, very small increases in the expression of PIN1 and IAA5 in d27 mutants (Fig. 6D) indicate enhanced auxin signaling or content, which could be indicative of enhanced auxin transport as observed in Osd27 and several other branching mutants (Lin et al., 2009;Crawford et al., 2010;Domagalska and Leyser, 2011). We have recently modeled long-distance auxin transport in the stem, and these data, in addition to data on auxin transport under high auxin content, demonstrate that wild-type plants do not have reduced auxin transport capacity relative to SL mutants (Brewer et al., 2009;Renton et al., 2012). Rather, in those experiments, the enhanced auxin transport seen in max3 and max4 mutants was more likely to be due to enhanced uptake of the exogenously supplied radiolabeled auxin to the top of the long-distance polar auxin transport stream (Renton et al., 2012).
In summary, d27 represents a new genetic tool for the elucidation of shoot branching mechanisms and their regulation in Arabidopsis. Together with MAX3 and MAX4, D27 acts in the plastid on the production of carlactone, an intermediate in the SL pathway, prior to MAX1. These genes encoding plastid-localized proteins, although expressed somewhat differently at the level of different plant parts, appear to carry conserved properties in terms of the regulation of their expression, particularly by the plant hormone auxin.
Unless otherwise stated, Arabidopsis (Arabidopsis thaliana) plants were grown under fluorescent lamps emitting 100 to 120 mmol photons s 21 m 22 (intensity at the rosette level) in a 6:1:1 mixture of peat, vermiculite, and perlite or in University of California mix with added 60% vermiculite. Plants from the Col-0 background were grown in a 16-h-light/8-h-dark photoperiod and a 22°C light/16°C dark temperature cycle, or in constant 22°C. Ws-4 was initially grown for 2 weeks in 14 h of light/10 h of dark and constant 18°C, in order to compensate for the extreme early flowering in Ws-4 and to match their flowering time with Col-0.
Grafting and Branching Assays
Grafting of Arabidopsis seedlings was performed as described in Brosnan et al. (2007) and as described in Brewer et al. (2009). Treatment of plants with GR24 in Phytatrays was performed as described (Brewer et al., 2009). Treatment of plants grown in hydroponic medium has been described previously .
IAA Treatments
Prebolting Ws-4 plants were treated with a lanolin ring around the top of the hypocotyl containing 2.5% dimethyl sulfoxide and 2.5% ethanol (control) with 10 mM IAA and/or 10 mM NPA (an auxin transport inhibitor). Decapitation involved removing the apical meristem and some small leaves/flowers from the shoot apex of each plant. Hypocotyl tissue was harvested 4 h post treatment. Three replicates were used for each genotype, with approximately 20 plants used per replicate.
Generation of 35S:AtD27 Plants
RNA was isolated from 7-d-old whole seedlings using the Qiagen RNeasy procedure. cDNA was generated from 1 mg of total RNA using Invitrogen SuperScript III. The At1g03055.1 coding sequence was first amplified using primers MW350 59-AAAAAAGCAGGCTATGAACACTAAGCTATCAC-39 and MW351 59-CAAGAAAGCTGGGTCTAATGCTTCACACCGTAGC-39 (translation initiation codon and stop codon, respectively, in boldface) using a proofreading DNA polymerase (Phusion; New England Biolabs). Approximately 250 pg of the PCR product was then reamplified using Gateway attB adapter primers 59-GGGGACAAGTTTGTACAAAAAAGCAGGCT-39 and 59-GGGGACCACTTTGTACAAGAAAGCTGGGT-39 (overlapping regions with MW350 and MW351, respectively, are underlined). The resulting PCR product was cloned into pDONR207 via Gateway-based recombination, and positive clones were confirmed by sequencing. The full-length coding sequence was then transferred to pMDC32 (Curtis and Grossniklaus, 2003) via recombination, generating the final 35S:D27 binary vector. The d27-1 mutant was then transformed by the floral dip method. Transgenic seedlings were selected by growth in continuous light for 7 d on one-half-strength Murashige and Skoog medium containing 25 mg mL 21 hygromycin B.
Transient Transformation and Localization Studies
The full-length D27 coding sequence was reamplified from pDONR207-D27 using primers MW350 and MW353 (59-CAAGAAAGCTGGGTTAT-GCTTCACACCGTAGC-39), in which the native stop codon was deleted. The region of D27 encoding the plastid transit peptide was amplified using primers MW350 and MW352 (59-CAAGAAAGCTGGGTTCTCTTTTG-CAGCCTTGAG-39). Both PCR products were then reamplified with the universal Gateway attB primers described above and cloned into pDONR207. Positive clones of each were then recombined with pMDC83 (Curtis and Grossniklaus, 2003) to generate clones expressing N-D27-mGFP6-C protein fusions. For colocalization, the full-length cDNA of the small subunit of Rubisco from pea (Pisum sativum) was fused to the N terminus of red fluorescent protein (RFP;Carrie et al., 2009).
Plasmids (5 mg) were precipitated onto 1-mm gold particles and biolistically transformed into onion (Allium cepa) epidermis as described previously (Thirkettle-Watts et al., 2003). Images were obtained with an Olympus BX61 epifluorescence microscope equipped with GFP (U-MGFPHQ) and RFP (U-MRFPHQ) filters and manipulated with Olympus cell^R software.
RT-PCR and Quantitative RT-PCR Analysis
Total RNA was extracted using the RNeasy (Qiagen) or NucleoSpin RNA Plant (Macherey-Nagel) procedure. For small tissue samples (Fig. 6D), a modified TRIzol (Invitrogen) procedure was used with subsequent RNeasy cleanup. Contaminating DNA was removed with Turbo DNA-free (Ambion) or DNase (Macherey-Nagel or Qiagen), and RNA was quantified with a NanoDrop 1000 spectrophotometer.
In the case of Figure 2B, Figure 6A, and Supplemental Figure S2C, cDNA was generated from 0.5 mg of total RNA in a 10-mL reaction using the iScript cDNA Synthesis kit (Bio-Rad). Quantitative RT-PCR was performed on a Roche LC480 using LightCycler 480 SYBR Green Master Mix (Roche) in 5-mL reactions. Cycle conditions were as follows: 95°C for 10 min; then 45 cycles of 95°C for 20 s, 60°C for 20 s, and 72°C for 20 s; followed by melt-curve analysis.
Crossing point values were calculated under high confidence. For each biological replicate, two technical replicates of each real-time PCR were examined, and the mean crossing point value was used to calculate expression relative to an internal reference gene using the formula (E gene ) 2Cp_gene /(E ref ) 2Cp_ref , where E is the primer efficiency and Cp represents the crossing point. Primer efficiencies were determined in separate runs using serial dilutions of pooled cDNA.
Phylogenetic Analysis
Protein sequences with significant homology to D27 were identified with BLASTP searches of GenBank protein databases using the rice (Oryza sativa ssp. japonica) D27 amino acid sequence as a query ( [URL]). Over 90 sequences were identified with a highly significant E-value score (E 210 or smaller). To reduce the complexity of the analysis, 62 sequences were selected on the basis of wide taxonomic coverage while also providing sufficient support for each clade. Sequences were further screened for duplication and truncation. Full-length sequences were then aligned using MAFFT ( [URL]:// mafft.cbrc.jp/alignment/software) using the default settings. The alignment was conservatively trimmed using PFAAT ( [URL]) to remove regions of poor homology, including putative N-terminal plastid transit peptides that show poor conservation. MrBayes version 3.1.2 (Ronquist and Huelsenbeck, 2003) was used to infer Bayesian trees (random start tree, four chains of temperature 0.2 in each of four independent runs, WAG substitution matrices (Whelan and Goldman, 2001), and four discrete categories of g-distribution substitution rate). Computational analysis was performed using CIPRES Science Gateway version 3.1 ( [URL]/), and phylograms were generated using Dendroscope version 2.7.4 ( [URL]. informatik.uni-tuebingen.de/software/dendroscope).
Statistical Analysis
For comparisons between multiple treatments and a control, one-way twosided ANOVAs (Dunnett's t test) were performed as described in the figure legends. The overdispersed data shown in Figure 3C were analyzed using the more appropriate nonparametric Mann-Whitney U tests. When transcript expression data spanned several orders of magnitude (Fig. 6A), data were log transformed prior to analysis to control for large differences in variance between groups. Statistical analysis was performed with SAS Enterprise Guide 4.3.
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number NM_202019.1 (AtD27). Protein accession numbers are listed in Supplemental Table S2.
Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Alignment of selected D27 orthologs from angiosperms.
Supplemental Figure S5. Response of auxin-related transcripts to IAA, decapitation, and NPA treatments.
Supplemental Table S1. Primers used for quantitative RT-PCR.
Supplemental Table S2. List of protein sequences used in phylogeny reconstruction.
==
Domain: Biology Environmental Science Medicine Engineering
|
[{"added": "2018-04-03T02:32:44.787Z", "created": "2012-05-22T00:00:00.000", "id": "23621082", "metadata": {"extfieldsofstudy": ["Biology", "Engineering", "Medicine"], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "http://www.plantphysiol.org/content/plantphysiol/159/3/1073.full.pdf", "pdf_hash": "bc41e4d0be362587681c2e64a90e4f1ff146447a", "pdf_src": "Highwire", "provenance": "peS2o-0028.json.gz:534", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "2bf777a08879d297ff890edfc15c69f76bf63503", "year": 2012}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
This below document has 2 sentences that start with 'ferdowsii Mirshamsi, Zamani and Marusik', 2 sentences that start with 'The new species', 2 sentences that end with 'as the female of O', 2 sentences that end with 'Zamani et al', 2 sentences that end with 'is similar to O', 2 sentences that end with 'that of O', 3 paragraphs that start with 'The specific epithet refers to', 2 paragraphs that start with 'In the habitus, the new', 2 paragraphs that start with 'Known only from the type'. It has approximately 1759 words, 110 sentences, and 37 paragraph(s).
===
New species and records of Oecobius Lucas, 1846 (Araneae: Oecobiidae) from Iran and Azerbaijan
ABSTRACT New data on the oecobiid spiders of the genus Oecobius Lucas, 1846 occurring in Iran and Azerbaijan are provided. Five species are described as new to science: O. dariusi sp. n. (♀; Alborz and Tehran ‒ northern Iran), O. melanocephalus sp. n. (♀; Lorestan, Razavi Khorasan, Semnan, and Tehran ‒ south-western, northern and north-eastern Iran; hitherto described as the female of O. ferdowsii Mirshamsi, Zamani and Marusik, 2017), O. naxuanus sp. n. (♂; Nakhchivan ‒ western Azerbaijan), O. pasargadae sp. n. (♂♀; Fars ‒ south-central Iran), and O. zagros sp. n. (♂; Kermanshah ‒ western Iran). Furthermore, O. navus Blackwall, 1859 is reported in Iran for the first time, and new distribution records are provided for O. putus O. Pickard-Cambridge, 1876 and O. nadiae (Spassky, 1936). [URL] Lucas, 1846, the largest genus of the spider family Oecobiidae, has an almost global distribution and currently comprises 90 species, as well as one fossil species from Dominican amber (Dunlop et al. 2020;WSC 2023). Although it is the most species-rich oecobiid genus, no species-group divisions have been proposed for it, except for the species found in the Canary Archipelago, where the genus has the highest species diversity -41 species (Wunderlich 1987). These species were categorised into three species groups, with just one, the navus group, being non-endemic to the archipelago. Although a worldwide revision of this genus has never been undertaken, it is relatively well studied within specific regions, particularly in the Nearctic (Shear 1970), as well as areas in the Mediterranean region, e.g. the Canary Islands, Madeira, and the East Mediterranean (Wunderlich 1987(Wunderlich , 1992(Wunderlich , 1995)). Wunderlich has described 49 species (WSC 2023), which accounts for over half of the presently recognised species of Oecobius. Only four of his described species have been later synonymised. Besides the Mediterranean region, only Iran has been the subject of several publications dealing with Oecobius (e.g. Zamani et al. 2017;Zamani and Marusik 2018;Zamani and Bosselaers 2020), resulting in six species currently known from this country (including two endemics). However, the fauna of the adjacent Caucasus remains relatively poorly studied, with only five species currently known from the region, all of which have wide distributions (Otto 2022). In this paper, we contribute to the knowledge of the diversity of this genus in Iran and the Caucasus by providing descriptions of five new species and recording one species that is new to Iran.
Material and methods
Photographs were obtained using an Olympus Camedia E-520 camera attached to an Olympus SZX16 stereomicroscope, and a JEOL JSM-5200 scanning electron microscope at the Zoological Museum of the University of Turku. Digital images of different focal planes were stacked with Helicon Focus™ 8.1.1. Illustrations of the vulvae were made after digesting tissues off in a 10% potassium hydroxide (KOH) aqueous solution. Leg segments were measured on the dorsal side. Measurements of legs are listed as: total length (femur, patella, tibia, metatarsus, tarsus). All measurements are given in millimetres. The map (Figure 9) was prepared using SimpleMappr (Shorthouse 2010).
Etymology
This species is named after Darius I, who is more commonly known as Darius the Great. He was the third King of Kings of the Achaemenid Empire and reigned from 522 BCE until his death in 486 BCE. During his rule, the empire reached its territorial peak.
Diagnosis
In the shape of the epigyne, the new species resembles O. latiscapus Wunderlich, 1992, known only from Tenerife. Both species have a wide 'scape' (Sc), a character unknown in the congeners. The new species can be distinguished by the heavily sclerotised tip of scape (vs not sclerotised) and by having many arched wrinkles in the anterior half and three posteriorly to the 'scape' (vs numerous straight wrinkles posteriorly from the 'scape'), and a few straight wrinkles anteriorly from the 'scape' (compare Figure 2A-C and Wunderlich 1992, fig.283). The vulva of O. latiscapus has never been illustrated, and therefore cannot be compared with that of the new species.
Etymology
The specific epithet refers to the characteristic black colouration of the carapace in this species.
Diagnosis
The new species differs from all its congeners occurring in the 3A, B; epigynal plate oval, ca 1.3 times wider than long; with distinct, as long as wide 'scape' (Sc), and pair of 'hoods' (H), twice wider than 'scape'; sclerotised capsules (Cs) round, separated by ca 3 diameters; receptacles (Re, poorly distinct in Figure 3B) oval, transversal, each shorter than half of epigynal plate width.
Comments
This species was initially described as the female of O. ferdowsii Mirshamsi, Zamani and Marusik, 2017. This is herein considered a mismatch: not only are there striking differences in the colouration pattern between the two species (compare Zamani et al. 2017, fig.2A, B and 3A, B), but the results of ongoing research on Central Asian Oecobiinae also indicate that O. ferdowsii should be classified within a separate genus. The females of this currently undescribed genus exhibit a completely different conformation of epigyne and vulva compared to those of O. melanocephalus sp.n.
Comments
This species differs from the congeners occurring in the region by the dark lateral patches on clypeus, three pairs of sublateral dots on carapace (indistinct in some specimens), and the presence of dots on legs (but not annulations). We noted that it has a kind of retrolateral tibial apophysis (arrowed in Figure 4A), which has not been previously documented.
Etymology
The specific epithet refers to Naxuana, which is the name given to Nakhchivan in Ptolemy's Geography and in the works of other classical authors.
Diagnosis
In the habitus, the new species is similar to O. dariusi sp.n., but differs by having no annulations on the legs (compare Figure 1A and B). In the general conformation of the male palp and particularly the shape of radical apophysis, it is similar to O. rhodiensis
Distribution
Known only from the type locality in Nakhchivan Autonomous Republic, Western Azerbaijan (Figure 9).
Etymology
The specific epithet refers to the capital of the Achaemenid Empire under Cyrus the Great, which is nowadays an archaeological site approximately 90 km north-east of Shiraz; noun in apposition.
Diagnosis
In the habitus, the new species differs from all of the congeners occurring in the region by lacking a carapace pattern in combination with lacking leg annulations. Its male can be easily distinguished from the congeners occurring in the region by having a tegular bump (Tb), almost equally long longitudinal terminal (Ta) and radical (Ra) apophyses (vs different in length or not elongate), and by having a kind of tooth (Rt) on the tip of radical apophysis. The female of the new species differs from all other species known in the region by having an arch-shaped notch of the plate (Pc), with the plate bearing copulatory openings (vs plate with extension ('scape') bearing copulatory openings).
Etymology
The specific epithet is a noun in apposition, referring to a long mountain range in Iran, northern Iraq and south-eastern Turkey, in which the type locality of the new species is situated.
Diagnosis
In the general conformation of the male palp and particularly the shape of radical apophysis, the new species is most similar to O. ilamensis Zamani, Mirshamsi and Marusik, 2017, a species so far known only from Ilam Province in western Iran (Zamani et al. 2017). The male palp of the new species differs from that of O. ilamensis by having straight, spine-like subterminal apophysis (St) (vs with bent tip) and abrupt tip of terminal apophysis (Ta) (vs finger-shaped) (compare Figure 7D and Zamani et al. 2017, fig. 1C).
Distribution
Known only from the type locality in Kermanshah Province, western Iran (Figure 9). On the label, the locality is listed as Lorestan Province, while the coordinates refer to another locality in Ilam Province. We consider Dizgaran in Kermanshah to be the correct type locality.
Discussion
Based on the results of this paper, 11 species of Oecobius are known from Iran. Among these, seven species exhibit a similar conformation of the copulatory organs and can be considered to fall within the same species group. These species are as follows: O. cellariorum (Dugès, 1836) It is important to note that while two of these species are currently known only from males and two are known only from females, we do not believe that they are conspecific, due to their different colouration patterns and distribution range. To verify this, however, it is necessary to collect material of both sexes for these four nominal species. The species newly described from Azerbaijan in this paper, O. naxuanus sp.n., also belongs to this species group, and it represents the first endemic species of this genus from the Caucasus (Otto 2022).
The taxonomy of the group comprising O. ferdowsii will be addressed in a separate publication. Prior to this study, O. ferdowsii was thought to have a widespread distribution in Iran, with material collected from several localities in the south-western, northern, and north-eastern parts of the country. However, our research revealed that the females previously attributed to this species were actually misidentified and are therefore described in this paper as a new species (O.melanocephalus sp.n.). Consequently, O. ferdowsii is now confidently known only from Razavi Khorasan Province in northeastern Iran. The record of this species from Kazakhstan (Fomichev 2022) also appears to be based on a misidentification, likely resulting from the lack of illustration of the ventral aspect of the male palp in the original description.
During the preparation of this paper, we discovered an error in two of our previous works involving male specimens of Oecobius (Marusik et al. 2015;Zamani et al. 2017), and also in Fomichev (2022). Despite using the terminology proposed by Baum (1972), we overlooked that he had depicted the right palps rather than the standard left ones. This oversight led to inaccuracies in identifying the structures of the male palp. Consequently, in those papers, what is originally identified as the radical apophysis is, in fact, the terminal apophysis, and vice versa. Similarly, the designation of the embolus should be attributed to the subterminal apophysis, and vice versa.
Zamani and Marusik, 2018odiensisKritscher, 1966, by having a uniformly dark carapace. The new species differs from O. rhodiensis by dark sternum and venter of abdomen (vs light). The epigyne of the new species is somewhat similar to that of O. fahimiiZamani and Marusik, 2018in the similar position of the sclerotised capsules, but differs by a relatively longer posterior part of 'scape' (Sc), 'hoods' (H) longer than scape is wide, and round sclerotised capsules (Cs) (vs 'scape' wider than 'hoods', sclerotised capsules oval; see Figure3A-D).
==
Domain: Biology Environmental Science
|
[{"added": "2023-11-22T16:45:50.018Z", "created": "2023-10-16T00:00:00.000", "id": "265347925", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://www.tandfonline.com/doi/pdf/10.1080/00222933.2023.2272356?needAccess=true", "pdf_hash": "ea60ac7afc522e05bb8fbae86736424144ba2e0e", "pdf_src": "TaylorAndFrancis", "provenance": "peS2o-0028.json.gz:3432", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "d793e4f93dca0d82db9a2fc2f1fccf30dd512889", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Molecular phylogeny and taxonomic position of Macrobrachiumlanchesteri (De Man, 1911), with descriptions of two new species from Thailand (Decapoda, Caridea, Palaemonidae)
Abstract Macrobrachiumlanchesteri (De Man, 1911), a translucent freshwater prawn has a wide distribution range throughout mainland Southeast Asia. A high morphological variation and genetic divergence between different geographical M.lanchesteri populations in Thailand have peculiarly extended the uncertainty of species boundaries and blended confusingly with several Macrobrachium species. To clarify these circumstances, broad sample examinations of the morphological variation, including topotype specimens, and phylogenetic reconstruction based on the concatenated mitochondrial dataset (16s rRNA and COI genes) were performed. Broad morphological examination of M.lanchesteri has shown congruency with phylogenetic analyses by revealing prominent lineages of M.lanchesteri sensu stricto and two new sibling lineages with interspecific variation between 6.48–8.76% for COI and 3.06–4.23% for 16S. Descriptions of two new species, named herein as M.panhai Chaowvieng & Siriwut, sp. nov. and M.rostrolevatus Chaowvieng & Siriwut, sp. nov. are provided. Morphological investigation of rostral form suggested plasticity in M.rostrolevatus populations showing the morphological trait associated with their habitat preferences. Furthermore, phylogenetic positions of the three taxa affirmed the hidden diversity of Thai freshwater Macrobrachium fauna correlated with the river network in the Mekong and Chao Phraya basins, Thailand. The genetic data and distribution records obtained in this study may also assist future river conservation plans as well as the sustainable management of freshwater prawn diversity.
Introduction
Palaemonid freshwater prawns of genus Macrobrachium Spence Bate, 1868 have shown high species richness comprising 271 species worldwide (WoRMS 2023). This genus has a broad geographical distribution and is commonly found in the Oriental Region of Asia (De Grave et al. 2008). Several Macrobrachium ZooKeys 1190: 163-193 (2024), DOI: 10.3897/zookeys.1190.113898 Apisara Chaowvieng et al.: Molecular phylogeny and morphology of two Macrobrachium prawns from Thailand species demonstrate economic impacts, serving as protein resources and for utilisation in ornamental fish aquaculture (Cai et al. 2004;Wowor et al. 2004). According to its remarkable species richness and diversifications of aquatic and terrestrial invertebrate faunas in Indochina, the intensive fauna exploration and historical biogeography using both morphology and genetics were reinvestigated systematically in several taxa such as river prawns, bivalves, land snails and millipedes (De Bruyn et al. 2014;Pholyotha et al. 2021;Jeratthitikul et al. 2022;Likhitrakarn et al. 2023). Indochinese Macrobrachium prawns have gained attention recently, especially in the context of taxonomy and systematics (Cai et al. 2004;Wowor et al. 2004;Hanamura et al. 2011). Several molecular taxonomic studies have been verified nominal species and consequently supplemented the taxonomic account of some cryptic Macrobrachium prawns (Mar et al. 2018;Siriwut et al. 2020;Jurniati et al. 2021;Saengphan et al. 2021). Additionally, the DNA barcoding and molecular delimitation methods were implemented to clarify the taxonomic boundaries of several Macrobrachium species. Moreover, the phylogenetic positions of several species have been addressed some morphological complexity groups based on barcode gap distance threshold (Siriwut et al. 2021).
Currently, 34 species have been documented in Thailand (Cai et al. 2004;Cai and Vidthayanon 2016;Saengphan et al. 2018Saengphan et al. , 2019Saengphan et al. , 2020Saengphan et al. , 2021;;Siriwut et al. 2020Siriwut et al. , 2021)). Two major river basins, the Chao Phraya and the Greater Mekong, have been discussed as being significant hotspots for native Macrobrachium faunal diversity (Cai and Ng 2002;Hanamura et al. 2011). Some Thai Macrobrachium species have been reported to show narrow distribution within these basins, such as M. chainatense Saengphan, Panijpan, Senapin, Laosinchai, Ruenwongsa, Suksomnit & Phiwsaiya, 2019 which was only found in Central Thailand, and M. spelaeus Cai & Vidthayanon, 2016 that live in stygobiotic habitats. Contrastingly, some widespread species have also been documented about their distribution occupancy crossed inland basins and some insular territory of Southeast Asia, such as M. sintangense (De Man, 1898), and M. dienbienphuense Dang & Nguyen, 1972(Cai et al. 2004;Wowor et al. 2004;Hanamura et al. 2011). For this reason, freshwater faunas in Thailand and neighbouring countries are capable linkage in terms of species composition, reaching an occurrence data of coexistence and cryptic species according to the connection of the river network (Hanamura et al. 2011;Siriwut et al. 2020).
A small translucent and common M. lanchesteri (De Man, 1911) dominantly occupies all river basins throughout mainland Southeast Asia with scattered distribution records from Malaysia, Singapore, Indonesia; it has even expanded northward to South China (Wowor and Choy 2001;Cai and Ng 2002;Cai et al. 2004). This species was originally found in southern Thailand and was diagnosed as having a straight and short rostrum not exceeding the scaphocerite and slender, thin second pereiopods (Kemp 1918;Holthuis 1950). The lectotype designation and morphological study of M. lanchesteri by Chong and Khoo (1988) advocated diagnostic character variation, particularly on rostral structure and body size variation in male regarding sexual dimorphism. Additionally, M. lanchesteri was mentioned with an argument on taxonomic boundary with some other congeners such as M. peguense (Tiwari, 1952), M. kistnense (Tiwari, 1952), and M. tiwarii Jalihal, Shenoy & Sankolli, 1988. Moreover, M. lanchesteri also blended confusingly with the juveniles of several species such as M. idae (Heller, 1862) and M. lar (Fabricius, 1798) (Lanchester 1902;Kamita 1966).
Previous phylogenetic and population genetic studies of M. lanchesteri in Thailand have also detected high genetic diversity, both between and within populations (Reingchai et al. 2009;Khanarnpai et al. 2019;Siriwut et al. 2021). Moreover, the possible existence of cryptic species within several Macrobrachium species in Thailand under traditional morphological discrimination criteria was reported, including M. lanchesteri, based on DNA barcode delimitation thresholds (Siriwut et al. 2021). The lack of intensive collection from different river basins impeded comprehensive genetic and morphological information that would contribute to taxonomic boundary clarification and phylogenetic relationships of M. lanchesteri and other native species in this area. To elucidate the taxonomic confusion of several coexistent translucent Macrobrachium prawns, the integration of traditional morphological identification and molecular analysis could be investigated concurrently. Therefore, this study aimed to clarify the taxonomic boundaries of M. lanchesteri in Thailand by broad-scale sampling and reconstruct the phylogenetic relationships with various related translucent species based on COI gene and 16S rRNA markers, which have been used extensively to investigate the phylogenetic relationships between crustaceans (Costa et al. 2007;Pileggi and Mantelatto 2010;Castelin et al. 2017;Jamaluddin et al. 2019;Rossi et al. 2020). This study will contribute to elucidate the taxonomic status of M. lanchesteri s. str.and its closely related species as well as assist economical freshwater prawn management in the future.
Sample collection and preparation
Prawn specimens were collected from various freshwater basins in Thailand. Live specimens were photographed to document body coloration using a Nikon D5300 camera with a micro-Nikkor 105 mm f/2.8 IF-ED Macro Lens. Prawns were gradually euthanised following the protocols approved by the Mahidol University-Institute Animal Care and Use Committee (MU-IACUC) under approval number MUSC66-026-656. Specimens were preserved in 95% ethanol and stored into a container for further morphological examination and molecular analysis. Voucher specimens were deposited at the Chulalongkorn University Museum of Zoology, Bangkok, Thailand (CUMZ) and Mahidol University Museum of Natural History, Department of Biology, Faculty of Science, Mahidol University, Thailand (MUMNH). Traditional identifications were carried out based on previous taxonomic studies of Macrobrachium species: Lanchester (1902), Holthuis (1950), Chong and Khoo (1988), Cai and Ng (2002), Cai et al. (2004), Wowor et al. (2004), andHanamura et al. (2011). The morphological variation of prawn specimens was observed and illustrated under a stereomicroscope. A list of abbreviations used in the descriptions is given as follows: Fin (finger), Pal (palm), Car (carpus), Mer (merus), Che (chela), Dac (dactylus), Pro (propodus), cl (carapace length), rl (rostrum length). All morphological characters were measured using Dinocapture software v. 2.0 and reported in millimetres.
DNA extraction and PCR protocol
All prawn specimens used for molecular analysis in this study are listed in Table 1. Genomic DNA was extracted from pleonal muscle tissue by using DNA extraction kits (NucleoSpin Tissue kit: MACHEREY-NAGEL). Genomic DNA quality was evaluated and visualised by gel electrophoresis and a UV illuminator. Two mitochondrial genes, 16S rRNA and cytochrome c oxidase subunit I (COI), were amplified. Three sets of primer such as 16Sa-L (5' CGC CTG TTT ATC AAA AAC AT 3') and 16Sbr-H2 (5' CTC CGG TTT GAA CTC AGA TCA 3') following Palumbi (1996) for 16S gene, LCO1490 (5'GGT CAA CAA ATC ATA AAG ATA TTG G 3'; Folmer et al. (1994), MacroNancy (5' GCG GGT AGR ATT AAR ATR TAT ACT TC 3'; Siriwut et al. (2020), HCOoutout (5' GTA AAT ATA TGR TGD GCTC 3';Schulmeister et al. (2002) for COI were used in this study. PCR was performed using T100 TM thermal cycler (BIO-RAD) with a gradient temperature function. The PCR profile consisted of the following steps: 94 °C for 5 min as an initial step followed by 34 cycles 94 °C for 30 sec for denaturing, 45-49 °C for 40 sec, 72 °C for 15 sec for extension, and final extension at 72 °C for 10 min. PCR products were run by 1% agarose gel electrophoresis stained with SYBR Safe illuminant (Invitrogen, USA). The purified products were sent for sequencing by a commercial company (Macrogen and Bioneer, Korea) using an Applied Biosystems automatic sequencer.
Phylogenetic analyses
Sequences were aligned and corrected using the ClustalW algorithm in MEGA 11 (Tamura et al. 2021). All sequences have been registered and deposited in Gen-Bank database under accession numbers OR575072-OR575118 for COI and OR578642-OR578698 for 16S (Table 1). The voucher specimen locality of each species used in molecular analysis is illustrated in Fig. 1. The DNA dataset for phylogenetic analyses was assembled including ten deposited COI sequences of Macrobrachium species in GenBank database. To depict the clade of M. lanchesteri sensu De Man (1911), topotype sequences were selected as representative indicators. Macrobrachium villosimanus (Tiwari, 1949) was used as the rooting outgroup.
Phylogenetic trees were constructed using maximum likelihood (ML) and Bayesian inference (BI) methods throughout the online CIPRES Science Gateway server (Miller et al. 2010). The concatenated dataset of two markers with the partitioned file for nucleotide substitution model fit was prepared in Kakusan 4 (Tanabe 2007). ML tree was visualised in RAxML v. 8.2.12.(Stamatakis 2014). The GTR+G model was set as the model for all gene partitions with 1,000 bootstrap replicates performed to verify tree topology and clade support. BI tree was estimated using MrBayes v. 3.2.7 (Ronquist et al. 2012). Markov chain Monte Carlo (MCMC) was configured as 10,000,000 generations of the sampling process; the first 25% of obtained trees were discarded as burn-in. Finalised trees were estimated for the consensus tree topology. The annotation and illustration of clade and branch length were performed in Figtree (Rambaut 2010). Node posterior probabilities of 0.95 were considered statistically significant for BI, and bootstrap support values greater than 70 were considered highly supported for ML (Huelsenbeck and Hillis 1993;Larget and Simon 1999). Pairwise genetic distance of intra and interspecific of each gene dataset was calculated using the p-distance method in MEGA 11 (Tamura et al. 2021).
Molecular phylogeny and genetic divergence
Forty-seven sequences of partial COI and 57 sequences of partial 16S genes were successfully amplified and obtained (Table 1). COI sequence contained 627 bp with 417 bp of conserved sites, 210 bp of variable sites and 202 of parsimony informative sites.16S sequence contained 554 bp with 373 bp of conserved sites, 181 bp of variable sites and 154 bp of parsimony informative sites. The proportional range of genetic variations in M. lanchesteri species complex and other Macrobrachium species were revealed by p-distance. Inter and intraspecific variations ranged from 15.12-20.68%for COI, 8.6-16.18%for 16S and 0.9-5.79%for COI and 1.08-3.19%for 16S, respectively. Both ML and BI trees based on 1,181 bp concatenated dataset of the COI and 16S gene fragments revealed the six Macrobrachium species as monophyletic groups with strong statistical support values (Fig. 2). Clade C comprised all M. sintangense sequences. Phylogenetic tree also showed that M. rosenbergii (De Man, 1879) is closely related to M. lanchesteri species complex clade, forming clade D. The genetic distance between M. rosenbergii and M. lanchesteri species complex clade was 15.12% for COI and 8.6% for 16S. In the clade E, Macrobrachium lanchesteri species complex was divided into three monophyletic groups with high statistical supports for both ML (100) and BI (1). The interspecific variation ranged from 6.48-8.76%for COI and 3.06-4.23%for 16S. The intraspecific variation also ranged from 0.92-2.27%for COI and 0.7-2.23%for 16S. In the results of this study, clade H was shown as M. lanchesteri based on the topotype sequences assembled. The monophyletic group of M. lanchesteri s. str.herein represents two subclades, lower Isthmus of Kra (Clade I) and upper Isthmus of Kra populations. Macrobrachium panhai sp.nov.(Clade J) was nested as a sister clade of M. lanchesteri s. str.with sufficient support in ML ( 74), but partial support in BI (86). Macrobrachium rostrolevatus sp.nov.(Clade F) was separated from congeneric members of M. lanchesteri species complex and all samples in this clade were strictly distributed inside freshwater basins on the Khorat Plateau, i.e. the Mun, Chi and Songkhram Rivers.
Composite description. Rostrum (Fig. 4B). Straight or slightly convex proximally and upward distally. Rostrum length exceeding end of antennular peduncle and slightly shorter than scaphocerite. Dorsal margin with 6-10 teeth including 1-3 teeth distally with small gap from rest. Postorbital margin with one or two teeth, reaching to one-fourth of carapace length. Ventral margin with 1-6 teeth, starting from middle to distal margins. Short setae present between rostral teeth.
Cephalon (Fig. 4B). Well-developed eye. Ocular beak without laterally expanded tip. Cornea longer and broader than stalk. Postantennular carapace margin rounded. Cornea osculum longer than stalk. Antennular peduncle longer than wide with fine setae, basal segment short, second segment shorter than third segment. Stylocerite projection sharp, reaching beyond basal segment. Antennal spine sharp situated below orbital margin. Hepatic spine slightly larger than antennal spine, positioned posteriorly and lower than antennal spine. Scaphocerite with straight margin, distolateral tooth sharp and not reaching end of lamella. Epistome bilobed (Fig. 4C). Branchiostegal suture starting from carapace margin to behind hepatic spine. Carapace surface smooth.
First pereiopods. Long and slender, reaching end of scaphocerite. Fingers as long as palm, tips with fine setae. Series of setae present at anterior inner part of palm. Carpus slightly longer than merus. Distal articulation of carpus with series of fine setae. Ischium shorter than merus. Scattered setae present on all segments.
Second pereiopods (Fig. 4D). Long and slender, similar in form and exceeding scaphocerite. Fingers subcylindrical covered with scattered setae. Palm 1.1-1.4×longer than fingers. Fingers with translucent razor edge present anteriorly and one or two tiny teeth on proximal quarter of cutting edges. Tip of fingers crossed and covered by fine setae (Fig. 4E). Carpus cylindrical shape and articulation margin expanded. Carpus 1.3-1.5×longer than chela. Merus subcylindrical. Carpus 1.1-2× longer than merus. Scattered short setae present on all segments.
Fourth and fifth pereiopods. Long and slender, exceeding scaphocerite. Propodus of fourth pereiopods with 5-10 pairs of spines distributed along its length, 2× longer than dactylus. Propodus slightly longer than merus. Ischium shorter than merus. Propodus with fine setae at distal articulation. Scattered short setae present on all segments. Propodus of fifth pereiopods with 7-13 pairs of spines distributed along its length and fine setae at distal articulation. Propodus 2× longer than carpus. Propodus as long as merus. Scattered short setae present on all segments.
Thoracic sternum. Fourth and fifth thoracic sternites with transverse plate. Sixth and seventh thoracic sternites smooth. Eighth thoracic sternite with or without acute median process.
Pleon. Smooth. All pleonal sternites with transverse ridge. First and second pleonal sternites usually with small median process. Third and fourth pleonal sternites smooth. Fifth pleonal sternite with triangular ridge. Preanal carina present, obtuse ridge developed without spine or setae. Ventral margin of pleural tergum with small setae.
Telson (Fig. 4G). Tapered posteriorly, protruding point on middle margin with lateral spines and few fine setae. Inner spines longer than outer spines. Dorsal surface with two pairs of small spines similar in size.
Remarks. The specimen collected in this study generally agrees with the original description in Lanchester (1902), and a subsequent description of the lectotype provided by Chong and Khoo (1988). Previous studies reported that male specimens tended to display the sexual dimorphism with a large body size, tomentose fingers, and minute spinules on all segments (except fingers) of second pereiopods. In this study, only one large male specimen, collected from Loei Province, Thailand, exhibits this characteristic. Typically, both male and female specimens possess fine setae on fingers and scattered setae on surface of second pereiopods. Furthermore, this study also observed two variable characters occurring on the second pereiopods. Firstly, the proportional length and form of second pereiopods were found to be variable in specimens from Krabi population. Their second pereiopods are shown to be prominently long and robust, similar to those of M. sintangense (a common riverine species). The palm margin is laterally inflated and slightly shorter than fingers, and the chela slightly longer than the carpus. Additionally, Chong and Khoo (1988) reported the presence of two tiny teeth on the basal portion of cutting edges of fingers in M. lanchesteri as a diagnostic character. In this study, one or two tiny teeth were present on the cutting edges of fingers and vary among M. lanchesteri populations. Historically, M. lanchesteri was noted to resemble several other species including M. idae, M. peguense (see under remarks of M. panhai sp.nov.), M. sankollii Jalihal, Shenoy & Sankolli, 1988, M. unikarnatakae Jalihal, Shenoy & Sankolli, 1988, and M. sintangense. Further phylogenetic relationships and phylogenetic placement of aforementioned taxa should be tested to elucidate and verify their taxonomic identities.
Macrobrachium lanchesteri has a wide distribution across mainland Southeast Asia and southern China. This species can live in various freshwater ecosystems by inhabiting aquatic vegetation in stagnant freshwater habitats such as ponds, lakes, and paddy fields. Diagnosis. Rostrum straight proximally and slightly upward distally. Rostrum length reaching beyond end of antennular peduncle and exceeding the scaphocerite. Rostral formula: 8-12/3-6 teeth including two or three distal teeth with small gap separate from rest. Carapace smooth. Epistome bilobed. First pereiopods reaching end of scaphocerite. Second pereiopods thin and long, similar in form and equals in length, exceeding scaphocerite. Fingers covered with scattered setae, slightly shorter than palm. Translucent razor edge present anteriorly between fingers and no teeth on inner side of cutting edges. Carpus cylindrical shape and articulation margin expanded. Carpus 1.5× longer than chela. Merus subcylindrical. Carpus 1.5× longer than merus. Third pereiopods thin and long, reaching end of scaphocerite. Dactylus curved distally with short setae. Propodus 2× longer than dactylus. Propodus with three or four pairs of spines and fine setae present scarcely on articulation margin. Propodus 2× longer than carpus. Sixth to eighth thoracic sternites smooth. First and second pleonal sternites with small median process or smooth. Third and fourth pleonal sternites smooth. Fifth pleonal sternite with triangular ridge. Uropodal diaeresis with inner movable spine slightly longer than outer angle.
Macrobrachium panhai
Composite description (holotype in parentheses). Rostrum (Fig. 5B). Straight or proximal convex and slightly distal upward. Rostrum length exceeding end of antennular peduncle and slightly exceeding scaphocerite (rl 7.32 mm). Dorsal margin with 8-12 (10) teeth including two or three (3) teeth distally with small gap from rest. Postorbital margin with one or two (1) teeth, reaching one-third of carapace length. First dorsal tooth positioned slightly behind hepatic spine. Ventral margin with 3-6 (4) teeth, starting from middle to distal margin. Short setae present between rostral teeth.
Cephalon (Fig. 5B). Eye well developed. Ocular beak without laterally expanded tip. Cornea longer and broader than stalk. Postantennular carapace margin rounded. Cornea osculum longer than stalk. Antennular peduncle longer than wide, with fine setae. Basal segment short, second segment shorter than third segment. Stylocerite projection sharp, reaching beyond basal segment. Antennal spine sharp, situated below orbital margin. Hepatic spine slightly larger than antennal spine, positioned posteriorly and lower than antennal spine. Scaphocerite with straight margin, distolateral tooth sharp and not reaching end of lamella. Epistome bilobed (Fig. 5C). Branchiostegal suture beginning at carapace margin to behind hepatic spine. Carapace surface smooth (cl 5.76 mm).
First pereiopods. Thin and long, reaching end of scaphocerite. Fingers as long as palm, tips with fine setae. Series of setae present on anterior inner part of palm. Carpus slightly longer than merus. Distal articulation of carpus with series of fine setae. Ischium shorter than merus. Scattered setae present on all segments.
Third pereiopods (Fig. 5F). Thin and slender, reaching end of scaphocerite. Dactylus short and curved distally. Propodus 2× longer than dactylus. Propodus with three or four pairs of spines along inferior-lateral margin and fine setae at distal articulation, 2× longer than carpus. Ischium shorter than carpus. Scattered short setae present on all segments.
Fourth and fifth pereiopods. Long and slender, exceeding scaphocerite. Propodus of fourth pereiopods with 3-6 ( 4) pairs of spines distributed along its length, 2.5× longer than dactylus. Propodus as long as merus. Ischium shorter than merus. Propodus with fine setae at distal articulation. Scattered short setae present on all segments. Propodus of fifth pereiopods with 4-8 pairs of spines distributed along its length and fine setae at distal articulation. Propodus 2.5× longer than carpus. Propodus as long as merus. Scattered short setae present on all segments.
Thoracic sternum. Fourth and fifth thoracic sternites with moderately transverse plate without median process, and seventh thoracic sternite smooth. Eighth thoracic sternite usually smooth.
Pleon. Smooth. All pleonal sternites with transverse ridges. First and second pleonal sternites with or without small median processes. Third and fourth pleonal sternites smooth. Fifth pleonal sternite with triangular ridge. Preanal carina present, obtuse ridge developed without spine or setae. Ventral margin of pleural tergum with small setae.
Telson (Fig. 5G). Tapered posteriorly, protruding point on middle margin with lateral spines and few fine setae. Inner spines longer than outer spines. Dorsal surface with two pairs of small spines, similar in size.
Uropods (Fig. 5G). Uropodal diaeresis with inner movable spine, as long as or slightly longer than outer angle. Exopods longer than endopods.
Etymology. The specific name panhai is dedicated to Prof. Dr. Somsak Panha, a taxonomist from Faculty of Science, Chulalongkorn University, Thailand well known for his remarkable contributions and endorsement to the study of invertebrate fauna in Thailand.
Distribution. This species is distributed in the Chao Phraya and Mekong River Basins, Thailand.
This new species also differs from M. peguense sensu Tiwari (1952) by processes of rostral formula 8-12/3-6 teeth (vs 6-9/2-4 teeth in M. peguense). Second pereiopods had palms shorter than half of carpus (vs palm slightly more than half of carpus in M. peguense). Propodus of third pereiopods are 2× longer than dactylus (vs 3 in M. peguense). Dorsal surface of telson is without depression (vs longitudinal depression in M. peguense). Movable spine at uropodal diaeresis is slightly longer than outer angle (vs movable spine is shorter in M. peguense). Cai and Ng (2002) also mentioned that the egg size can be used to distinguish M. peguense and M. lanchesteri group (1.15-1.5 × 1.6-2.1 mm and 0.6-0.7 × 0.8-1 mm, respectively). Currently, the distribution range of M. peguense was found only from Myanmar. Diagnosis. Rostrum long and thin, proximal half straight and uplifted distal half. Rostrum length reaching beyond end of antennular peduncle and prominently exceeding scaphocerite. Rostral formula: 6-11/4-9 teeth including 2-4 teeth distally with large gap from rest. Apical teeth usually present with trifid. Carapace smooth. Epistome bilobed. First pereiopods reaching end scaphocerite. Second pereiopods thin and long, similar in form and length, exceeding end of scaphocerite. Fingers covered with scattered setae with translucent razor edge present anteriorly between fingers and one tooth on proximal quarter of cutting edges. Palm 1.25× longer than fingers. Carpus cylindrical shape and articulation margin expanded. Carpus 1.5-2× longer than chela. Merus subcylindrical. Carpus 1.5× longer than merus. Third pereiopods thin and long, slightly exceeding scaphocerite. Dactylus curved distally with short setae. Propodus 2× longer than dactylus. Propodus with 3-6 pairs of spines distributed along its length and fine setae at its articulation. Propodus 2× longer than carpus. Sixth to eighth thoracic sternites smooth. First and second pleonal sternites with small median process. Third and fourth pleonal sternites smooth. Fifth pleonal sternite with triangular ridge. Uropodal diaeresis with inner movable spine slightly longer than outer angle.
Cephalon (Fig. 6B). Eye well developed; ocular beak without laterally expanded tip. Cornea longer and broader than stalk. Postantennular carapace margin rounded. Cornea osculum longer than stalk. Antennular peduncle longer than wide, with fine setae. Basal segment short, second segment being shorter than third segment. Stylocerite projection sharp, reaching beyond basal segment. Antennal spine sharp, situated below orbital margin. Hepatic spine slightly larger than antennal spine, positioned posteriorly and lower than antennal spine. Scaphocerite with straight margin, distolateral tooth sharp and not reaching end of lamella. Epistome bilobed (Fig. 6D). Branchiostegal suture starting from carapace margin to behind hepatic spine. Carapace surface smooth (cl 7.14 mm).
First pereiopods. Long and slender, reaching end of scaphocerite. Fingers as long as palm, tips with fine setae. Series of setae present at anterior inner part of palm. Carpus slightly longer than merus. Distal articulation of carpus with series of fine setae. Ischium shorter than merus. Scattered setae present on all segments.
Fourth and fifth pereiopods. Long and slender, exceeding scaphocerite. Propodus of fourth pereiopods with 4-7 (5) pairs of spines distributed along its length, 2× longer than dactylus. Propodus slightly shorter than merus. Ischium shorter than merus. Propodus with fine setae at distal articulation. Scattered short setae present on all segments. Propodus of fifth pereiopods with 4-10 pairs of spines (holotype damaged) distributed along its length and fine setae at distal articulation. Propodus 2.5× longer than carpus. Propodus as long as merus. Scattered short setae present on all segments.
Thoracic sternum. Fourth and fifth thoracic sternites with moderately transverse plate. Sixth to eighth thoracic sternites usually smooth.
Pleon. Smooth. All pleonal sternites with transverse ridge. First and second pleonal sternites with or without median process. Third and fourth pleonal sternites smooth. Fifth sternite with triangular ridge. Preanal carina present, obtuse ridge developed without spine or setae. Ventral margin of pleural tergum with small setae.
Telson (Fig. 6H). Tapered posteriorly, protruding point on middle margin with lateral spines and few fine setae. Inner spines longer than outer spines. Dorsal surface with two pair of small spines, similar in size.
Uropods (Fig. 6H). Uropodal diaeresis with inner movable spine, as long as or slightly longer than outer angle. Exopods longer than endopods.
Etymology. The specific epithet rostrolevatus is from the Latin compound words rostro, for rostrum, and levatus, referring to lifted.
Distribution. This species is distributed in freshwater basins of Khorat Plateau, Northeast Thailand.
Remarks. Macrobrachium rostrolevatus sp.nov.differs from M. lanchesteri s. str.based on the presence of single tooth on movable and fixed fingers of second pereiopods (vs 1 or 2 teeth on movable and fixed fingers in M. lanchesteri), movable spine at uropodal diaeresis slightly longer than the outer angle (vs shorter than outer angle in M. lanchesteri), and the presence of 3-6 pairs of spines on propodus of third pereiopods (vs 4-8 pairs of spines in M. lanchesteri). This new species also differs from M. villosimanus sensu Tiwari (1949) and M. rosenbergii sensu De Man (1879) by having 6-11/4-9 rostral teeth (vs 12-14/7-10 rostral teeth in M. villosimanus; 9-13/10-15 rostral teeth in M. rosenbergii). The second pereiopods are smooth and covered with fine setae (vs spinules in entire cheliped, movable finger densely pubescent and fixed finger sparsely pubescent in M. villosimanus; coarse velvet hairs on movable
Discussion
Morphological and genetic analyses revealed three distinct lineages (prior assumption as geographical variation of M. lanchesteri), which are recognised herein as M. lanchesteri s. str., M. panhai sp.nov., and M. rostrolevatus sp.nov. Previously, the taxonomic identity of M. lanchesteri s. l. was investigated based on the morphological examination of and reinvestigation of type specimens (Lanchester 1902;Chong and Khoo 1988). In this study, the clarification of species boundaries and phylogenetic positions were supplemented by molecular analyses. The phylogenetic position of M. lanchesteri is closely related to M. rosenbergii, although some morphological characteristics might appear similar to the M. sintangense species group. Current observation noted that a juvenile of M. sintangense and M. lanchesteri were morphologically overlapping. Ecologically, they commonly co-exist in several habitats such as riverbanks and lentic reservoirs in mainland Southeast Asia. Their life histories were supposedly influenced by a convergent evolutionary mechanism (Wowor et al. 2009), the same example as noted in other species with abbreviated larval development (ALD) such as Macrobrachium species: M. platycheles Ou & Yeo, 1995, M. sundaicum (Heller, 1862), and M. malayanum (Roux, 1935) (Murphy and Austin 2005). The independent lineages of ALD species were hypothesised as evidence of multiple invasions of marine ancestors (Liu et al. 2007;Murphy and Austin 2005;Wowor et al. 2009). To elucidate the effect of environmental conditions and feeding preferences altering morphological characteristics among coexisting species, comprehensive materials along an environmental gradient could be investigated. Additionally, M. rosenbergii showed distinctiveness in both morphological characters and a reproductive strategy different from M. lanchesteri. The life cycle of M. lanchesteri is completed typically in freshwater as opposed to M. rosenbergii, which had larval development and egg hatching occurring in brackish water. The close phylogenetic relationship between M. lanchesteri and M. rosenbergii seem to potentially derive from a common ancestor through evolutionary divergence processes.
The evidence of genetic divergence and composition differences in Thai invertebrate population are often documented between the lower and upper Isthmus of Kra regions. This evidence was sparsely seen in M. lanchesteri s. str. The same patterns of genetic divergence correlated to subregional populations were also detected in the widespread M. spinipes (Schenkel, 1902). This species shows a wide distribution range in the Indo-Australasian region due to a historical event during the last glacial maximum (De Bruyn and Mather 2007;Ng and Wowor 2011). Currently, the geographical distribution of M. lanchesteri in Southeast Asia seems to possibly include the introduction by human activities, particularly from local fishery-related activities such as in Sabah and Brunei Darussalam (Ng 1994;Wowor and Choy 2001). Thai M. lanchesteri s. str.failed to show a strong subregional pattern despite widespread distribution records, and a similar pattern was also observed in some freshwater gastropods collected from different parts of Thailand (Saijuntha et al. 2021). This might be the consequence of the commercial trade of aquatic plants in Thailand that accidentally introduced freshwater gastropods throughout the area. Contrastingly, M. rostrolevatus sp.nov.has a narrow distribution range and a dense population specifically found in the sub-basins of the Songkhram, Chi, and Mun rivers on the Khorat Plateau.
However, a comprehensive survey of the adjacent sub-basins along the Lower Mekong River Basin should be implemented to affirm its geographic range.
Macrobrachium prawns exhibit a vast variation of morphological characters, with several species demonstrating sexual dimorphism and morphological plasticity (Holthuis 1950;Dimmock et al. 2004;Short 2004). These phenomena increased the uncertainty of species boundaries and the complication of taxonomic discrimination criteria for various Macrobrachium species groups. Recent studies have employed tools, including molecular identification using mitochondrial gene datasets, to clarify and resolve taxonomically ambiguous situations (Liu et al. 2007;Carvalho et al. 2013;Castelin et al. 2017;Rossi et al. 2020;Saengphan et al. 2021). In this study, the mitochondrial genes 16S and COI showed potential to be useful for taxonomic clarification between closely related taxa and revealed the existence of cryptic species, as in the cases of M. panhai sp.nov.and M. lanchesteri s. str. Although M. panhai sp.nov. shares morphological characteristics with M. lanchesteri, genetic differentiation falls within the delimitation gap suggested by Siriwut et al. (2021). For this reason, the delimitation threshold based on inter-and intraspecific variations of Macrobrachium species would be considered an additional tool for cryptic fauna exploration and delineation of morphologically ambiguous groups of Macrobrachium prawns.
Macrobrachium rostrolevatus sp.nov.has different forms of rostrum that appear to be associated with habitat preference. The long and upcurved rostrum is prevalent in lentic habitats i.e., ponds and lakes, whereas the shorter and straight rostrum is dominant in lotic habitats like river tributaries. This rostral shape variability may indicate phenotypic plasticity, similar to observations in M. australe (Guérin-Méneville, 1838) and members of the genus Caridina H. Milne Edwards, 1837, where rostral shape is influenced by water current speed. In an area with fast-flowing current, the long rostrum can be more fragile and impede movement whereas the shorter, more robust, and straight rostrum might better resist the strong water current (Zimmermann et al. 2011;Mazancourt et al. 2017). Moreover, the variation in morphological traits influenced by environment was also found in M. australiensis Holthuis, 1950, an endemic Australian freshwater prawn and M. nipponense (De Haan, 1849), a widespread species in Taiwan (Dimmock et al. 2004;Chen et al. 2015). This study provided additional evidence that the diagnostic characters of Macrobrachium can be influenced by the environment. Therefore, morphological identification alone should be implemented carefully, especially for species with high morphological variability (Liu et al. 2007;Siriwut et al. 2020). The integration of other molecular markers such as nuclear markers and morphometric analysis could be used to further enhance the accuracy of taxonomic identification and phylogenetic relationships of Macrobrachium in the future.
Figure 1 .
Figure 1. Distribution map of three Macrobrachium species in Thailand. A colour symbol indicates the locality of specimen used in phylogenetic analyses. A transparent symbol indicates the locality of specimen examined based on morphology. Equivalent symbols, whether coloured or not, indicate the same species.
Figure 2 .
Figure 2. Phylogenetic tree based on a concatenation of COI and 16S genes. Nodes of a phylogenetic tree marked with a black circle indicate statistical support from both ML and BI (≥ 70 bootstrap values and ≥ 0.95 posterior probability scores). A white circle indicates statistical support for either ML or BI. An asterisk indicates the topotype in M. lanchesteri and holotype in the new species.
Figure 5 .
Figure 5. Morphological characteristics of Macrobrachium panhai sp.nov.(A, F ovigerous female paratype MUMNH MP00303 B-E, G-H ovigerous female holotype CUMZ MP00302) A lateral view B carapace C epistome D second pereiopod E teeth between fingers F third pereiopod G uropod and H movable spine at uropodal diaeresis. Scale bars: 1 mm.
Figure 6 .
Figure 6. Morphological characteristics of Macrobrachium rostrolevatus sp.nov.(A, B, D-I ovigerous female holotype CUMZ MP00323 C ovigerous female specimen MUMNH MP00338.1)A lateral view B carapace C rostral variation D epistome E second pereiopod F teeth between fingers G third pereiopod H uropod and I movable spine at uropodal diaeresis. Scale bars: 1 mm.
Table 1 .
Locality and GenBank accession numbers of specimens used in phylogenetic analyses.
==
Domain: Biology Environmental Science
|
[{"added": "2024-01-31T16:08:17.574Z", "created": "2024-01-29T00:00:00.000", "id": "267329458", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://zookeys.pensoft.net/article/113898/download/pdf/", "pdf_hash": "5dbf1ee29818b9190236b6eda467f25ee98d574c", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:1916", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "5569a3a9f98396477410f37c10745586683b0428", "year": 2024}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Investigating the Influence of Varied Light-Emitting Diode (LED) Wavelengths on Phototactic Behavior and Opsin Genes in Vespinae
Simple Summary This research aims to explore the phototactic behavior and key opsin genes associated with Vespinae. The results showed that the two species, Vespula germanica and Vespa analis, exhibited varying photophilic rates under different wavelengths of light, suggesting that light wavelength significantly affects their phototactic behavior. Additionally, the opsin genes of the most aggressive hornet, Vespa basalis, have been sequenced. There are only two opsin genes, one for UV light and the other for blue light, and Vespa basalis lacks long-wavelength visual proteins. However, they exhibit peak phototaxis for long-wavelength light and instead have the lowest phototropism for UV light. This suggests that the visual protein genes have a complex regulatory mechanism for phototactic behavior in Vespinae. Our findings provide a sound theoretical basis for further investigation of visual expression patterns and phototactic mechanisms in Vespinae. Abstract The phototactic behavior of insects is commonly used to manage pest populations in practical production. However, this elusive behavior is not yet fully understood. Investigating whether the opsin genes play a crucial role in phototaxis is an intriguing topic. Vespinae (Hymenoptera: Vespidae) are a common group of social wasps that are closely associated with human activities. Efficiently controlling wasp populations while maintaining ecological balance is a pressing global challenge that still has to be resolved. This research aims to explore the phototactic behavior and key opsin genes associated with Vespinae. We found significant differences in the photophilic rates of Vespula germanica and Vespa analis under 14 different light conditions, indicating that their phototactic behavior is rhythmic. The results also showed that the two species exhibited varying photophilic rates under different wavelengths of light, suggesting that light wavelength significantly affects their phototactic behavior. Additionally, the opsin genes of the most aggressive hornet, Vespa basalis, have been sequenced. There are only two opsin genes, one for UV light and the other for blue light, and Vespa basalis lacks long-wavelength visual proteins. However, they exhibit peak phototaxis for long-wavelength light and instead have the lowest phototaxis for UV light. This suggests that the visual protein genes have a complex regulatory mechanism for phototactic behavior in Vespinae. Additionally, visual protein sequences have a high degree of homology among Hymenoptera. Despite the hypotheses put forward by some scholars regarding phototaxis, a clear and complete explanation of insect phototaxis is still lacking to date. Our findings provide a strong theoretical basis for further investigation of visual expression patterns and phototactic mechanisms in Vespinae.
Introduction
Phototaxis, the phenomenon wherein insects respond to light stimuli by moving towards (positive phototaxis) or away from (negative phototaxis) a source of light, is a behavior that has evolved as an inherent characteristic in insects [1]. This behavior is functionally adaptive as it regulates light exposure and aids in spatial orientation [2]. It influences various aspects of insect life, including mating, feeding, life cycle, and reproductive modes [1][2][3]. The response of insects to light is influenced by the wavelength of light [4]. Extensive research in recent years has explored the phototactic behavior of different insect species, with a particular focus on pests [5,6]. Hymenoptera, a group that includes bees and wasps, exhibits sensitivity peaks in the visual spectrum at 340 nm, 430 nm, and 535 nm, indicating a predominantly trichromatic visual system [7,8]. For instance, the honeybee (Apis mellifera) is sensitive to light at 344 nm, 436 nm, and 556 nm [9]. Phototactic behavior is also influenced by light intensity and is often used to assess an animal's ability to discriminate between different levels of light intensity [10,11]. The underlying mechanisms of phototaxis are not hardwired and can be modulated by various external factors, such as light, wavelength, intensity, sex, developmental stages, environmental conditions, and biological rhythms [12][13][14][15][16][17]. Insect phototactic behavior is also subject to modulation during biological rhythms, with different species exhibiting peak light-up periods at specific times [18,19]. Scholars have proposed several hypotheses to explain insect phototaxis, such as light orientation, biological antennae, light interference, and light stress hypothesis [20][21][22][23]. This research aims to further investigate the intricate nature of insect phototaxis, shedding light on its underlying mechanisms and contributing to our understanding of this behavior.
Vision is an important physical sense in insects, facilitating foraging, mating, and defense against predators [24]. The ability of insects to recognize different spectra and produce corresponding visual responses depends critically on the visual pigments present in the photoreceptors inside the compound eye. These pigments consisted of photosensitive chromophores and optic proteins [25]. In most insects, there is only one form of photosensitive chromophore, and its structure is relatively simple. Therefore, the diversity of spectral recognition in insects relies solely on the sequence of optic proteins. Opsin, a G-proteincoupled receptor characterized by the presence of seven transmembrane domain structures, is the primary visual protein [26]. Visual proteins can be classified into seven subfamilies based on their specific functions and molecular structures: neuropsin, encephalopsin, retinal photoisomerase, Go-coupled retinal proteins, Gq-coupled retinal proteins, and vertebrate visual and non-visual retinal proteins [26]. Optic proteins can be further classified into non-visual opsins and visual opsins based on their roles in visual imaging. Opsins can be categorized as rhodopsin or visinin based on their molecular properties, which differ in the amino acid sequence at residues 122 and 189 [27]. Previous studies have shown that insect opsins can be divided into three clades based on the spectral peaks of the pigments: long-wavelength-sensitive opsin (LW opsin, peak absorbance 500-600 nm), blue-sensitive opsin (blue opsin, peak absorbance 400-500 nm), and ultraviolet-sensitive opsin (UV opsin, peak absorbance 300-400 nm) [28]. Insects often adjust their spectral sensitivity range to match the spectrum of their environment, leading to duplication, loss, and variation in the opsin genes. For example, certain insects in the Coleoptera order have lost optic proteins that are sensitive to short-wavelength light [29]. Variations, losses, and duplications of optic protein genes have also been observed in species such as Drosophila, mosquitoes, and butterflies [30]. Research on the function of the opsin genes has become important. UV opsin is important for flower recognition by honeybees [31] and is involved in the migratory orientation of the monarch butterfly Danaus plexippus [32]. In experiments with psyllid Diaphorina citri, reducing the expression of three opsin genes resulted in reduced phototactic efficiency to UV, green, and blue lights [33].
The subfamily Vespinae, belonging to the family Vespidae of Hymenoptera, are social insects present worldwide and attack flies, moths, locusts, cicadas, and beetles. Vespinae are often regarded as pests due to the threats they pose to human production, health, and life, such as attacking humans, preying on bees, destroying fruits, and contaminating food [34]. Currently, pest management has traditionally relied on integrated pest management, which utilizes sex pheromones and phototactic behavior, which is a sustainable approach in production practice [35,36], While research has yielded results on the phototactic behavior of insects and pest control strategies based on it, the phototactic behavior of wasps in the subfamily Vespinae has been understudied compared to other taxa.
In this study, we aimed to investigate the phototactic behavior of two Vespinae species, Vespula germanica (nest underground) and Vespa analis (nest on the tree). We analyzed and compared their phototactic behavior at six different times of the day using a range of LED lights with distinct colors or wavelengths. Our main objective was to identify patterns and gain insights into the phototactic behavior of Vespinae. Additionally, we fully sequenced and analyzed the UV and blue light opsin genes of V. basalis (the most aggressive species of hornets). Using bioinformatics techniques, we sequenced the deduced amino acid sequences of the opsin genes. Combining with the hymenopteran data online (NCBI) (Table S9), a phylogenetic tree was constructed, which revealed both the conservation of the opsin genes and the phylogenetic relationships among the species.
Monitoring Phototactic Behavior
The phototaxis monitoring system (Figure 1 [37,38]. It included a phototactic reaction chamber (light zone), a static chamber, and a dark chamber (dark zone). The system was made of transparent Plexiglas and measures 30 cm × 23 cm × 25 cm. The phototactic chamber and the lightavoidance chamber were positioned at 90 • to ensure minimal interference between both. A circular pathway with a diameter of approximately 6 cm was provided between the static and phototactic chambers and the light-avoidance chambers. Above each chamber was a circular opening, approximately 10 cm in diameter, for the release and retrieval of insects. The bottom and top walls were covered with opaque cardboard to minimize exposure to light from the other chambers.
and life, such as attacking humans, preying on bees, destroying fruits, and contaminating food [34]. Currently, pest management has traditionally relied on integrated pest management, which utilizes sex pheromones and phototactic behavior, which is a sustainable approach in production practice [35,36], While research has yielded results on the phototactic behavior of insects and pest control strategies based on it, the phototactic behavior of wasps in the subfamily Vespinae has been understudied compared to other taxa.
In this study, we aimed to investigate the phototactic behavior of two Vespinae species, Vespula germanica (nest underground) and Vespa analis (nest on the tree). We analyzed and compared their phototactic behavior at six different times of the day using a range of LED lights with distinct colors or wavelengths. Our main objective was to identify patterns and gain insights into the phototactic behavior of Vespinae. Additionally, we fully sequenced and analyzed the UV and blue light opsin genes of V. basalis (the most aggressive species of hornets). Using bioinformatics techniques, we sequenced the deduced amino acid sequences of the opsin genes. Combining with the hymenopteran data online (NCBI) (Table S9), a phylogenetic tree was constructed, which revealed both the conservation of the opsin genes and the phylogenetic relationships among the species.
Monitoring Phototactic Behavior
The phototaxis monitoring system (Figure 1) was designed based on Huang et al. (2023) and Jiang et al. (2023) [37,38]. It included a phototactic reaction chamber (light zone), a static chamber, and a dark chamber (dark zone). The system was made of transparent Plexiglas and measures 30 cm × 23 cm × 25 cm. The phototactic chamber and the light-avoidance chamber were positioned at 90° to ensure minimal interference between both. A circular pathway with a diameter of approximately 6 cm was provided between the static and phototactic chambers and the light-avoidance chambers. Above each chamber was a circular opening, approximately 10 cm in diameter, for the release and retrieval of insects. The bottom and top walls were covered with opaque cardboard to minimize exposure to light from the other chambers. A white LED light of the light source with a spectral wavelength ranging from 360 to 635 nm (Xuzhou Aijia Electronic Technology Co., Ltd.(Jiangsu, China)) was placed along one side of the glass with a 5 W 12 V lithium-ion battery LED light board. We customized 14 different models with wavelengths ranging from 360 to 365 nm (UV) and 380 to 385 nm (ultraviolet light), 400 to 410 nm (violet light), 420 to 430 nm (blue-violet light), 440 to 445 nm (dark blue light), 460 to 475 nm (blue light), 490 to 505 nm (faint green light), 515 to 525 nm (bright green light), 530 to 545 nm (green light), 550 to 565 nm (green-yellow light), 570 to 590 nm (yellow light), 600 to 610 nm (orange-red light), 625 to 635 nm (red light), and 400 to 840 nm (composite white light that mimics sunlight). During the operation of the LEDs, the temperature varied by ±1 • C. The LED lights were tested with a Polaroid polarizer to confirm that LEDs do not produce polarized light. To measure illumination intensity and spectrum, we used an Avos V3 photometer to more precisely measure the light intensity inside the monitoring cages.
Behavioral Response of Vespula germanica and Vespa analis to Different Spectral Wavelength Light
To explore the phototactic behavior characteristics of wasps, we conducted phototropic experiments on male and female Vespula germanica and Vespa analis. The period from 7:00 to 19:00 was selected based on their overall activity pattern. Each experiment lasted for 2 h, comprising a 1 h dark reaction phase, a 45 min light reaction phase (with a light intensity of 160 lx), and a 15 min subsequent treatment phase [39]. At 7:00, 30 female wasps were placed in a static chamber. The circular pathway connecting the phototactic and light-avoidance chambers was then blocked, and the entire experimental setup was covered with a black cloth to conduct the dark reaction for 1 h. Following the dark reaction, the light source and circular pathway were activated, and the experimental setup remained covered with black cloth to prevent interference. The light reaction was then carried out for 45 min. Then, the number of insects in the phototactic and light-avoidance chambers was counted for 15 min. Finally, the chambers were wiped with alcohol pads and cleaned. Following this, we placed 30 test insects into the resting chamber for the next experiment. This process was repeated until the experiment concluded at 19:00. Each experimental setup could conduct six sets of experiments per day. The light reaction periods were from 8:00 to 8:45, 10:00 to 10:45, 12:00 to 12:45, 14:00 to 14:45, 16:00 to 16:45, 16:00 to 16:45, and 18:00 to 18:45, with each light type being treated for four consecutive days. A total of 24 replications were conducted. Phototactic behavioral response rate (%) = (number of wasps in the reaction chamber/total number of wasps) × 100. All experiments were repeated thrice; photophobic rate percentage (%) = (number of light-avoidance reaction chamber wasps/total number of wasps) × 100; and percentage of active wasps (%) = (number of wasps in phototactic and light-avoidance chambers/total number of wasps) × 100.
Cloning of Opsin Genes
The most aggressive hornet, Vespa basalis, being collected from Qianyang (Baoji, Shaanxi), was selected for cloning its opsin genes. Their compound eyes, being dissected soon, were frozen in liquid nitrogen for the following experimental steps. Total RNA was isolated using a Trizol extraction kit No. B511321 (Sangon Biotech, Shanghai, China), following the manufacturer's protocol. A NanoDrop 2000 spectrophotometer (Hitachi, Tarrytown, New York, NY, USA) was used to perform spectroscopic quantification. Primers were designed based on the cDNA sequences (Table 1). The following reagents were added to 0.2 mL PCR tubes: 5 µL RNA sample, 1 µL random primer, 1 µL ddH 2 O, 70 • C warm bath for 5 min. Ice bath for 2 min; centrifugation: 2.0 µL 5X First-Strand Buffer, 0.5 µL 10 mmol dNTP, 0.25 µL RNase inhibitor, 0.25 µL Reverse Transcriptase, 10.0 µL total volume, 42 • C warm bath for 60 min, and 72 • C warm bath for 10 min. Then, the PCA reaction was performed. The results were observed through 1% agarose gel electrophoresis. The PCR products were recovered and sent to Sangon Biotech (Shanghai) Co., Ltd. for the complete sequencing of the DNA.
Primer Name
Primer Sequence 5
Phototactic Rhythmicity of Vespula germanica at Different Times
The phototactic rates of Vespula germanica varied significantly across the six time periods for each light, indicating a rhythmic behavior. V. germanica exhibited varying phototactic rates across different wavelengths of light, with active periods of different lengths. The lowest photophilic rates were mainly concentrated in the morning, while the highest photophilic rates appeared entirely in the noon and the late afternoon. The lowest and highest photophilic rates appeared at the same or similar periods under different wavelengths, but the specific phototactic trends and rates were quite different. V. germanica exhibited varying phototactic strengths for different wavelengths of light, as evidenced by the considerable variation in specific phototactic trends and rates, despite the lowest and highest rates occurring at similar times. Additionally, the phototactic trends under neighboring wavelengths of light were similar (Figure 2 and Table S1).
The photophilic and photophobic rates were not significantly different most of the time under both UV lights, with significant differences only at mid-day, late afternoon, or afternoon hours. Under violet, blue-violet, blue, and yellow light, photophilic and photophilic rates were significantly different except in the morning. Significant differences in photophilic and photophobic rates over the whole spectral range of green light (490-565 nm) were mainly observed at mid-day, late afternoon, and evening, which is a unique photophilic rhythm in the face of green light, i.e., photophilic shifted from a downward trend to an upward trend in the afternoon, suggesting that a second peak of photophilic may be formed in the evening and onwards. The differences in photophilic and photophobic rates within the two red light types and the composite light types were significant, except for morning and evening, suggesting that the phototactic activity of V. germanica to long-wavelength light may only be active in the morning to afternoon hours. Overall, photophilic and photophobic rates had the highest and most significant differences among the 14 light types at mid-day and late afternoon, whereas there were mostly no significant differences in the morning and evening (Figure 2 and Table S1). The photophilic and photophobic rates were not significantly different most of the time under both UV lights, with significant differences only at mid-day, late afternoon, or afternoon hours. Under violet, blue-violet, blue, and yellow light, photophilic and photophilic rates were significantly different except in the morning. Significant differences in photophilic and photophobic rates over the whole spectral range of green light (490-565 nm) were mainly observed at mid-day, late afternoon, and evening, which is a unique photophilic rhythm in the face of green light, i.e., photophilic shifted from a downward trend to an upward trend in the afternoon, suggesting that a second peak of photophilic may be formed in the evening and onwards. The differences in photophilic and photophobic rates within the two red light types and the composite light types were significant, except for morning and evening, suggesting that the phototactic activity of V. germanica to long-wavelength light may only be active in the morning to afternoon hours. Overall, photophilic and photophobic rates had the highest and most significant differences among the 14 light types at mid-day and late afternoon, whereas there were mostly no significant differences in the morning and evening (Figure 2 and Table S1).
By analyzing the photophobic rate of the different periods, we found that V. germanica exhibited low photophobic rates at different times of the day, with a generally smooth trend. It was only significantly different in the UV light (380-385 nm), green light, and composite light (Table S2). Although there were significant differences in the photophobic rates of the three light types, most of them remained at a low level. Therefore, it is reasonable to assume that V. germanica does not exhibit a significant photophobic rate (Figure 2).
The analysis of the total activity rate of V. germanica at different times all day showed significant differences in the active rates (Figure 2 and Table S3). The analyses indicated that the activity rate and change trend were similar to that of the photophilic rate. This was because the photophobic rate is generally low or even zero, and the trend of change was mostly smooth without significant fluctuations (Tables S4 and S6). As V. germanica did not exhibit photophobic behavior, its activity was determined by the photophobic rate By analyzing the photophobic rate of the different periods, we found that V. germanica exhibited low photophobic rates at different times of the day, with a generally smooth trend. It was only significantly different in the UV light (380-385 nm), green light, and composite light (Table S2). Although there were significant differences in the photophobic rates of the three light types, most of them remained at a low level. Therefore, it is reasonable to assume that V. germanica does not exhibit a significant photophobic rate (Figure 2).
The analysis of the total activity rate of V. germanica at different times all day showed significant differences in the active rates (Figure 2 and Table S3). The analyses indicated that the activity rate and change trend were similar to that of the photophilic rate. This was because the photophobic rate is generally low or even zero, and the trend of change was mostly smooth without significant fluctuations (Tables S4 and S6). As V. germanica did not exhibit photophobic behavior, its activity was determined by the photophobic rate (Table S5).
The Phototactic Behavior of Vespula germanica under Different Wavelengths of Light
We classified 14 types of light based on their wavelengths. Vespula germanica had the lowest phototropism rate of about 20% when it came to sensitivity to UV light. Meanwhile, green, red, and white light have phototropism rates of about 30%, which were significantly higher than that of UV light. Blue light had the highest phototropism rate at 38.4%, indicating a stronger preference compared to the other types of light. In conclusion, the sensitivity and tendency of V. germanica to different wavelengths of light varied, and a bimodal response curve emerged, with violet light and yellow light being the most preferred, with an average phototactic rate of 47.0% and 44.1%, respectively (Figure 3 and Table 2). Vespula germanica demonstrated notable variations in both photophilic and photophobic rates across all 14 light sources. The photophilic rate exhibited a much stronger response compared to the photophobic rate. Specifically, a significant difference was observed between the photophilic rate and the overall activity rate only under green light. The overall activity rate of V. germanica was primarily influenced by the photophilic rate, while the photophobic rate remained relatively weak across most light conditions. Notably, significant disparities were observed between the photophobic rate and the overall activity rate for all 14 light sources. These findings align with the results obtained from comparing the photophilic and photophobic rates (Figure 2 and Table 3). Vespula germanica exhibited weak responses to 14 different light sources, with a generally low photophobic rate. The photophobic response was less pronounced compared to the photophilic response, and the curve exhibited a gentler slope. However, there was a significant variation in the photophobic rate among different wavelengths. The curve displayed multiple peaks, with three prominent peaks observed in blue (460-475 nm), green (530-545 nm), and orange-red (600-610 nm) light. It is important to note that the highest photophobic rate was observed under green light, while the photophilic rate was at its lowest (Figure 2). The difference in the photophobic rate can primarily be attributed to the considerably higher photophobic response to green light compared to other light wavelengths.
The total activity rate of V. germanica and the photophilic rate exhibited a high degree of consistency. The photophobic rate was only higher in the green light, which prevented V. germanica's total activity rate curve from dipping as much as the photophilic rate curve. This led to the formation of a distinct result in the formation of a new trough (Figure 3).
Vespula germanica demonstrated notable variations in both photophilic and photophobic rates across all 14 light sources. The photophilic rate exhibited a much stronger response compared to the photophobic rate. Specifically, a significant difference was observed between the photophilic rate and the overall activity rate only under green light. The overall activity rate of V. germanica was primarily influenced by the photophilic rate, while the photophobic rate remained relatively weak across most light conditions. Notably, significant disparities were observed between the photophobic rate and the overall activity rate for all 14 light sources. These findings align with the results obtained from comparing the photophilic and photophobic rates (Figure 2 and Table 3). The asterisk (*) indicates the significance level of the test result. The symbol "ns" represents non-marked differences, whereas a single * and **** represent statistically marked differences at 5% and 0.01%, respectively.
Phototactic Rhythmicity of Vespa analis at Different Times
By studying the phototactic rhythm of Vespa analis, it was found that the phototactic rate of V. analis was significantly different in six time periods under each type of light, indicating that the phototactic behavior of V. analis had a rhythmic nature. V. analis had different phototactic rhythms under different wavelengths of light, the phototactic period varied from long to short, and the photophilic rate was always high from morning to evening under some lights, which indicated that the phototactic period of V. analis might include more periods. Then, the lowest photophilic rate occurred mostly in the morning, similar to Vespula germanica, but the highest photophilic rates were not concentrated in the mid-day and late afternoon as in V. analis but occurred in the morning and evening. The minimum and maximum photophilic rate of V. analis in different wavelengths of light occurred at similar times, but the specific phototactic tendency and photophilic rate varied considerably, reflecting that V. analis had different phototactic strengths for different lights (Figure S1 and Table S1).
Analysis of the total activity rate of V. analis in different periods showed significant differences in the six time periods. The trend was consistent with the phototactic rate, as the photophobic rate of V. analis was generally low or even zero. The phototactic rates of V. analis exhibited no significant difference across the six time periods of the 14 light types, except for the mid-day period under the composite light. Furthermore, when the significance of the difference between the photophobic rate and the total activity rate of V. analis was compared, we found that the photophobic rate and total activity rate of V. analis were significantly different in most of the periods. In conclusion, it is further demonstrated that the variation in the total activity rate of analis is mainly dependent on the photophilic rate (Figure S1).
The analysis of the photophobic rate of V. analis revealed that the photophobic response was much weaker than the phototactic response at different wavelengths, and in some cases, there was no photophobic response at all. The photophobic rate varied significantly across different wavelengths, and the photophobic rate curve had multiple peaks. There were six sensitive peaks located in the ultraviolet, blue-violet, microgreen, green, and composite light regions. The highest photophobic rate was observed in the ultraviolet regions. The significance of the difference in photophobic rate mainly stems from these two types of light. It is important to note that the photophobic rate was highest while the photophilic rate was lowest in UV light with a wavelength of 380-385 nm. The results indicated that the total activity rate of V. analis is mainly dependent on the photophilic rate, as evidenced by the consistent curves observed for both parameters across all 14 light species tested (Figure S1).
By analyzing the significance of the difference between the different periods for all monochromatic lights, it was found that the phototactic rate of V. analis was significantly different between the different periods for blue light, bright green light, and mixed light (Table S2). Comparative analyses of the photophobic rate of three light sources at different times of day showed that under blue light, it had the highest photophobic rate in the morning with a significant difference; under bright green light, it had the highest photophobic rate in the morning with a significant difference; and under compound light, it had the highest photophobic rate in the mid-afternoon. Although there were some periods of significant difference between the photophobic rates of the three lights, the photophobic rates were generally low and stable for most of the time (Figures 3 and S1).
Analysis of the significance of differences in photophilic and photophobic rates of V. analis at different times of the day showed that under ultraviolet light (360-365 nm), there were significant differences in photophilic and photophobic rates at every time of the day except the afternoon. Under ultraviolet light (380-385 nm), orange-red light, and red light, there were significant differences in photophilic and photophobic rates at each time of the day except in the morning and afternoon. Under violet, blue-violet, dark blue, blue, and greenish light, photophilic and photophobic rates were significantly different for all six time periods. Under bright green, yellow, and composite light, there was a significant difference in all periods except morning. Under green and green-yellow light, significant differences were found only in the morning, mid-day, and late afternoon. Overall, significant differences in photophilic and photophobic rates were found mainly in the morning and evening (Figure S1 and Table S4).
Phototactic Behavior of Vespa analis under Different Wavelengths of Light
The total activity rate of Vespa analis varies greatly throughout the day. Not all phototactic behavior data from each period can adequately reflect the phototactic behavior characteristics of V. analis. Therefore, we selected three of the six time periods with the highest total activity rate of V. analis to investigate the phototactic behavior of different wavelengths of light (Figure S1). It was found that V. analis exhibited phototactic responses to all 14 types of light. However, there were significant differences in the photophilic rate between wavelengths, suggesting that V. analis had varying degrees of sensitivity and preference for different wavelengths of light. In this study, it was found that V. analis had the lowest photophilic rate of 14.2% when exposed to ultraviolet light. White light had a photophilic rate of less than 30%, which was significantly higher than that of ultraviolet light. Violet and yellow light showed a further increase in preference compared to the previous light. Blue, green, and red lights were the most preferred lights by V. analis, with an average photophilic rate of about 40% (Table 2).
In conclusion, the comparison of photophilic, photophobic, and the total activity rate of V. analis in different wavelengths of light revealed significant differences in 14 light types (Table 3). The photophilic behavior of V. analis was much stronger than its photophobic behavior. When comparing the phototactic rate and the total activity rate of V. analis, a significant difference was found only under 380-385 nm UV light (Table S5). The total activity rate of V. analis was mainly dependent on the photophilic rate. The photophobic behavior was weak in most light conditions. The comparison between the photophobic rate and the total activity rate of V. analis revealed significant differences across all 14 lights (Table S6).
Physico-Chemical Characterization of Opsin Genes in Vespa basalis
The opsin genes Vb-BL and Vb-UV were obtained from Vespa basalis and submitted to the GenBank database. The Vb-BL and Vb-UV fragments contain 1775 and 1641 bases, with an open reading frame of 1137 and 1122 bases that encode a polypeptide comprising 378 and 373 amino acids. The predicted isoelectric point of the opsin genes was 8.40 and 8.02, and the Molecular weights were 42.98 and 41.45 kDa, respectively (Table S7). The functional sites of Vb-BL and Vb-UV all contained an N-glycosylation site, a protein kinase C phosphorylation site, a casein kinase II phosphorylation site, an N-myristoylation site, and a visual pigments (opsins) retinal binding site; however, Vb-UV also contains G-protein coupled receptors family 1 signature (Table S8).
Predictive Analysis of the Structure of the Opsin Genes in Vespa basalis
The transmembrane topologies of Vb-BL and Vb-UV were predicted. Vb-BL's amino acid sequence was found to have seven transmembrane topologies located at positions 58-81, 95-115, 131-152, 172-194, 219-241, 283-307, and 317-339. Four topologies are from the outside to the inside, and three are from the inside to the outside (Figure 4B). The amino acid sequence of Vb-UV has seven transmembrane topologies located at positions 49-72, 85-109, 122-143, 162-186, 211-235, 275-275, and 275-298, respectively. Four of these transmembrane topologies are oriented from the outside in, while the remaining three are oriented from the inside out (Figure 4A).
The 3D structure models of Vb-BL and Vb-UV of V. basalis were constructed using (Template) 6i9k.1. A (2.1 Å) as a template with GMQE values of 0.63 and 0.67, QMEAN values of −4.64 and −4.51, and sequence identities of 33.33% and 40.82% (Figure 5). The possibility of the modeled 3D structural models of Vb-BL and Vb-UV was assessed by Ramachandran plots. This study indicates that the stereo conformations and dihedral angles of the 3D structural models of Vb-BL and Vb-UV meet the requirements of the two dihedral angles Ψ and Ω distributions (Figure 5), suggesting that the constructed 3D structures are reliable. The 3D structure models of Vb-BL and Vb-UV of V. basalis were constructed using (Template) 6i9k.1. A (2.1 Å) as a template with GMQE values of 0.63 and 0.67, QMEAN values of −4.64 and −4.51, and sequence identities of 33.33% and 40.82% (Figure 5). The possibility of the modeled 3D structural models of Vb-BL and Vb-UV was assessed by Ramachandran plots. This study indicates that the stereo conformations and dihedral angles of the 3D structural models of Vb-BL and Vb-UV meet the requirements of the two dihedral angles Ψ and Ω distributions (Figure 5), suggesting that the constructed 3D structures are reliable. The cloned V. basalis has two visual proteins, one short-wavelength and one longwavelength. The phylogenetic tree for opsins from two groups, each with 22 insect species, showed that opsins were clustered with Vb−BL and Vb−UV opsins, respectively, and species within the same taxa (genus, family, and order) were also well clustered. The opsins of V. basalis were highly homologous to other vespids and were most closely related to its congeners (viz. V. velutina, V. crabro, and V. mandarinia) (Table S9). This suggests that opsin is conserved during evolution. Significantly, within the family Vespidae, the species of subfamily Vespinae (including the genera Vespa and Vespula) separated with the subfamily Polistinae (including the genus Polistes) clearly in long-wavelength opsins while mixed in short-wavelength opsins. Thinking of a few poor bootstrap value numbers in the tree, the phylogenetic relationships of these proteins were only to some degree consistent with traditional classification, indicating the limited reliability of the clustering relationships in the phylogenetic tree (Figure 6). The cloned V. basalis has two visual proteins, one short-wavelength and one longwavelength. The phylogenetic tree for opsins from two groups, each with 22 insect species, showed that opsins were clustered with Vb−BL and Vb−UV opsins, respectively, and species within the same taxa (genus, family, and order) were also well clustered. The opsins of V. basalis were highly homologous to other vespids and were most closely related to its congeners (viz. V. velutina, V. crabro, and V. mandarinia) (Table S9). This suggests that opsin is conserved during evolution. Significantly, within the family Vespidae, the species of subfamily Vespinae (including the genera Vespa and Vespula) separated with the subfamily Polistinae (including the genus Polistes) clearly in long-wavelength opsins while mixed in short-wavelength opsins. Thinking of a few poor bootstrap value numbers in the tree, the phylogenetic relationships of these proteins were only to some degree consistent with traditional classification, indicating the limited reliability of the clustering relationships in the phylogenetic tree (Figure 6). Among them are long wavelengths (croci) and short wavelengths (blue). The red color indicates that these are newly sequenced genes.
Discussion
The photophobic rate of Vespula germanica and Vespa analis was consistently lower than the photophilic rate, suggesting a weak aversion to light. This phenomenon may be attributed to the circadian rhythms of Vespula species. While not all Vespula species have identical activity patterns, they generally initiate their activities early in the morning, peak in and out of the hive around 11:30 a.m., and experience a significant decrease in activity in the evening (between 5:00 p.m. and 8:00 p.m.) [34,46]. The circadian rhythms of V. mandarinia, V. basalis, and V. velutina in Shaanxi Province, China, are generally similar, with activity commencing around 5:00 a.m. in the summer, increasing from 7:00 a.m. and then dramatically decreasing after returning to the nest at around 6:00 p.m. [47]. In September, V. orientalis in Israel exhibited increased activity in the early morning, with frequent movements in and out of the nest from 11:00 to 19:00, followed by a rapid decline in activity [48]. Overall, the influence of different periods on the phototactic behavior of V. germanica and V. analis aligns with their respective circadian rhythms.
Currently, a comprehensive explanation for insect phototactic behavior remains elusive. Although various hypotheses have been proposed [16,21,49], they are limited to specific types or classes of insect phototactic behavior. They do not fully elucidate the underlying mechanisms of dynamic phototactic behavior in insects. The study of insect phototactic behavior has progressed with advancements in biotechnology. Through scanning electron microscopy of the compound eye, retinal potentials, and other studies, many species's light-sensitive peaks and the basic structure of the insect compound eye have been elucidated [50,51]. Molecular biology-based investigations of visual genes in insects hold promise for unraveling insect phototactic behavior [52,53].
The phototactic responses of V. germanica and V. analis varied across 14 different light wavelengths, indicating their sensitivity variations. For instance, V. germanica exhibited the highest sensitivity to violet light, while V. analis was most sensitive to red-orange light. However, their maximum phototactic rate reached only about 50%. Notably, they displayed low sensitivity to ultraviolet light, and their photophobic behavior was insignificant. Research has shown that Hymenoptera insects possess a trichromatic visual system, with spectral sensitivity peaks at 340 nm, 430 nm, and 535 nm [8]. This indicated the presence of UV-sensitive, blue-light-sensitive, and long-wave-sensitive retinoblast genes in insect eyes. For example, Aphidius gifuensis is sensitive to UV light at 330-340 nm, green light at 490 nm, and blue light at 530 nm [54]. Apis mellifera has three phototactically sensitive bands at 344 nm, 436 nm, and 556 nm, corresponding to their trichromatic visual system. However, unlikely honeybees and Vespinae possess only two retinoblast genes [9,55].
In general, insect retinoid genes are classified into three categories: ultraviolet-sensitive opsin, blue-sensitive opsin, and long-wave-sensitive opsin [56]. However, the cloning procedure of V. basalis yielded only two opsin genes, blue-sensitive opsin (Vb-BL) and ultraviolet-sensitive opsin (Vb-UV) [9,51], which represents a departure from the general situation [57,58]. Under the trichromatic color vision principle, insects with only two photoreceptor genes, such as hornets, would be expected to perceive a color-poor world, which is often referred to as color blindness. However, the absence of a gene for long-wavelength photoreceptor protein does not imply that wasps cannot perceive long-wavelength light. Conversely, the presence of a gene for a specific photoreceptor protein does not necessarily indicate that wasps exhibit a preference for that wavelength. Our study provides empirical evidence that supports this view; for example, the phototropic behavior of V. germanica and V. analis demonstrated that both species exhibited a phototropic peak in long-wavelength light, with the lowest phototropic rate observed for ultraviolet light. This suggests the potential for a more intricate regulatory mechanism underlying the influence of the visual protein gene on the phototactic behavior of wasps.
Insects have evolved visual systems that are adapted to their specific habitats. Many insects, such as Helicoverpa armigera, Ostrinia furnacalis, and Chilo suppressalis, exhibit clear positive phototactic behavior towards light [1,59] and some negative phototaxis (avoidance of light) [18]. As a result, light traps have been widely utilized in integrated pest management (IPM) for predicting and controlling various insect pests [60]. Diurnal insects have developed intricate visual systems that utilize ambient light for navigation and color vision [25,61]. Vision plays a crucial role in visual perception, circadian rhythm regulation, and pupil response in insects [62]. A detailed analysis of opsins, light-sensitive proteins, in insects is of great importance for understanding the mechanisms behind phototactic behavior and for enhancing the efficiency of light traps in attracting target insect pests.
Conclusions
This study provides valuable insights into the phototactic behavior of Vespinae species and their associated opsin genes. The findings indicate that the photophilic rates of Vespula germanica and Vespa analis are influenced by their circadian rhythms, with lower photophobic rates suggesting a weaker aversion to light. The observed variations in phototactic responses across different light wavelengths suggest species-specific sensitivities. V. germanica exhibited high sensitivity to violet light, while V. analis showed greater sensitivity to red-orange light. However, both species displayed low sensitivity to ultraviolet light. Therefore, this study follows a previous study highlighting the presence of a trichromatic visual system in Hymenoptera insects, with spectral sensitivity peaks at ultraviolet, blue, and long-wave wavelengths. These findings align with previous research on other insect species and reinforce the importance of retinoblast genes in mediating phototactic behavior. The research also emphasizes the potential of molecular biology-based investigations to further unravel the mechanisms underlying insect phototaxis. Insects exhibit diverse phototactic behaviors that are influenced by their ecological niche and behaviors. The compound eyes of insects play a vital role in their light responsiveness. Light traps utilizing positive phototaxis have proven effective in integrated pest management strategies. Understanding the visual systems and opsins in insects can enhance the efficiency of such traps in attracting target insect pests.
Figure 1 .
Figure 1. Apparatus for monitoring the phototactic response of Vespula germanica and Vespa analis: (A) light-sheltered reaction chamber (dark area); (B) picking up and releasing insects around the
Figure 1 .
Figure 1. Apparatus for monitoring the phototactic response of Vespula germanica and Vespa analis: (A) light-sheltered reaction chamber (dark area); (B) picking up and releasing insects around the mouth; (C) LED light; (D) pathway for light or/and wasps; (E) phototactic chambers (light zones with internally mounted light sources); and (F) waiting room.
Figure 4 .
Figure 4. Analysis of opsin genes in V. basalis: (A) predicted transmembrane topology models of Vb-BL and Vb-UV and (B) the predicted secondary structure of Vb-BL and Vb-UV.
Figure 4 .
Figure 4. Analysis of opsin genes in V. basalis: (A) predicted transmembrane topology models of Vb-BL and Vb-UV and (B) the predicted secondary structure of Vb-BL and Vb-UV.
Figure 5 .
Figure 5. The 3D structures and Ramachandran plots of Vb−BL and Vb−UV in V. basalis: (A) the 3D structure of Vb−BL; (B) Ramachandran plot of Vb−BL; (C) the 3D structure of Vb−UV; and (D) Ramachandran plot of Vb−UV.
Figure 5 .
Figure 5. The 3D structures and Ramachandran plots of Vb−BL and Vb−UV in V. basalis: (A) the 3D structure of Vb−BL; (B) Ramachandran plot of Vb−BL; (C) the 3D structure of Vb−UV; and (D) Ramachandran plot of Vb−UV.
Figure 6 .
Figure 6. A maximum-likelihood (ML) tree of the opsin genes based on amino acid sequences. The other opsins represent the Hymenoptera species. Values above the nodes represent bootstrap values. Among them are long wavelengths (croci) and short wavelengths (blue). The red color indicates that these are newly sequenced genes.
Table 2 .
Photophilic, photophobic, and total activity rate of Vespula germanica and Vespa analis under different wavelengths of light.
Table 3 .
Comparison with phototactic behavior of Vespula germanica and Vespa analis under different wavelengths of light.
==
Domain: Biology Environmental Science
|
[{"added": "2024-05-26T15:34:57.495Z", "created": "2024-05-23T00:00:00.000", "id": "270036903", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2076-2615/14/11/1543/pdf?version=1716465057", "pdf_hash": "edc72b09bba92e9f122ff54a8bbef9e53dcb245a", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:684", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "1728d76d4823eee0935432a4e0ec6896e614e53f", "year": 2024}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
The genome sequence of the Autumnal Rustic, Eugnorisma glareosa (Esper, 1788)
We present a genome assembly from an individual male Eugnorisma glareosa (the Autumnal Rustic; Arthropoda; Insecta; Lepidoptera; Noctuidae). The genome sequence is 631.0 megabases in span. Most of the assembly is scaffolded into 30 chromosomal pseudomolecules, including the Z sex chromosome. The mitochondrial genome has also been assembled and is 15.39 kilobases in length. Gene annotation of this assembly on Ensembl identified 19,768 protein coding genes.
Background
The Autumnal Rustic, Eugnorisma glareosa, is a medium-sized noctuid moth with a wingspan of about 32-38 mm. Its forewings usually exhibit dorsally crisp black markings on a light grey background, with two short black dashes or dots at the forewing base. The inner margins of the faintly outlined orbicular and reniform stigmata are in the form of conversely placed black arrow or axe heads. The former marking sometimes appears with adjacent black dots and often shows a diffuse rufous-grey or pinkish subterminal band. The forms in south-east England are more orange-brown and much darker forms are found in the Shetlands. As the vernacular name suggests, the adult moth emerges late in the Palaearctic Autumn, flying in the UK from August to October (Randle et al., 2019), and the moth overwinters as a small larva (Waring et al., 2017).
The Autumnal Rustic has a preference for heathland, moorland or other types of open country such as downs and shingle beaches. The larva feeds on various low growing plants such as heathers and bedstraws as well as scrub birch and sallow (Waring et al., 2017). Genera fed on include Calluna, Galium, Hieracium, Lactuca, Plantago, Poa, Rumex, and Salix (FUNET, 2023). The adult nectars on flowers such as heathers.
E. glareosa is generally fairly common and widespread in the western Palaearctic only, from northern and eastern Ireland (poorly represented in the west) to southern Scandinavia to Spain and the northern Mediterranean; avoiding Italy, but has relatively few records for eastern Europe (GBIF Secretariat, 2023).
Populations in the UK have declined severely since 1970 (Randle et al., 2019); between 1968 and 2006 reductions in Rothamsted trap numbers of at least 94% were documented with an annual change decline of about 7% (Conrad et al., 2006).
Based on mitochondrial DNA, E. glareosa shows a single cluster on BOLD, BOLD:AAE3673 (21/08/2023) with southern Mediterranean exemplars slightly (up to 1.79%) divergent from the others; south-eastern English exemplars show identity with central European populations. The genus Eugnorisma Boursin, 1946 was partly revised by Varga et al. (1990) and these authors considered a previously suggested relative Protexarnis McDunnough, 1929 as morphologically convergent, and they also considered a possible relationship based on genital characteristics with Eugraphe Hübner,1821, Paradiarsia McDunnough, 1929, andXestia Hübner, 1818. Surprisingly, the nearest neighbouring species of E. glareosa on BOLD is the dissimilar looking Double Dart Graphiphora augur, which is a mere 3.1% or more pairwise divergent in COI-5P (21/08/2023); other species of the genus Eugnorisma, including ones with similar wing patterns, are more distant. In this context the type species of the genus Eugnorisma is the Central Asian Graphiphora insignata Lederer 1853 (whereas Graphiphora Ochsenheimer, 1816 is currently considered monobasic, containing only G. augur). The genome sequence should therefore be useful in resolving the phylogenetic relationships of Eugnorisma using multiple loci. The data may also be useful to investigate genes that are potentially selected to influence forewing colouration in relation to darker resting substrates in northern Scotland and the Shetlands (see (Kettlewell & Berry, 1969), and may also be relevant to the study of the evolution of melanism.
Genome sequence report
The genome was sequenced from one male Eugnorisma glareosa (Figure 1) collected from Beinn Eighe (57.63,. A total of 42-fold coverage in Pacific Biosciences singlemolecule HiFi long reads was generated. Primary assembly contigs were scaffolded with chromosome conformation Hi-C data. Manual assembly curation corrected 12 missing joins or mis-joins and removed 3 haplotypic duplications. The final assembly has a total length of 631.0 Mb in 47 sequence scaffolds with a scaffold N50 of 22.0 Mb (Table 1). Most (99.86%) of the assembly sequence was assigned to 30 chromosomal-level scaffolds, representing 29 autosomes and the Z sex chromosome. Chromosome-scale scaffolds confirmed by the Hi-C data are named in order of size (Figure 2-Figure 5; Table 2). While not fully phased, the assembly deposited is of one haplotype. Contigs corresponding to the second haplotype have also been deposited. The mitochondrial genome was also assembled and can be found as a contig within the multifasta file of the genome submission.
Metadata for specimens, spectral estimates, sequencing runs, contaminants and pre-curation assembly statistics can be found at [URL] acquisition and nucleic acid extraction
A male Eugnorisma glareosa (specimen ID NHMUK014451776, individual ilEugGlar6) was collected from Beinn Eighe National Nature Reserve, Scotland, UK (latitude 57.63, longitude -5.35) on 2021-09-09, using a light trap. The specimen was collected and identified by David Lees (Natural History Museum) and dry frozen at -80°C.
DNA was extracted at the Tree of Life laboratory, Wellcome Sanger Institute (WSI). The ilEugGlar6 sample was weighed and dissected on dry ice with tissue set aside for Hi-C sequencing. Thorax tissue was disrupted using a Nippi Powermasher fitted with a BioMasher pestle. High molecular weight (HMW) DNA was extracted using the Qiagen MagAttract HMW DNA extraction kit. HMW DNA was sheared into an average fragment size of 12-20 kb in a Megaruptor 3 system with speed setting 30. Sheared DNA was purified by solidphase reversible immobilisation using AMPure PB beads with a 1.8X ratio of beads to sample to remove the shorter fragments and concentrate the DNA sample. The concentration of the sheared and purified DNA was assessed using a Nanodrop et al., 2013) and uses these annotations to select the final mitochondrial contig and to ensure the general quality of the sequence.
Table 3 contains a list of relevant software tool versions and sources.
Genome annotation
The BRAKER2 pipeline (Brůna et al., 2021) was used in the default protein mode to generate annotation for the Eugnorisma glareosa assembly (GCA_947578425.1) in Ensembl Rapid Release.
Wellcome Sanger Institute -Legal and Governance
The materials that have contributed to this genome note have been supplied by a Darwin Tree of Life Partner. The submission of materials by a Darwin Tree of Life Partner is subject to the 'Darwin Tree of Life Project Sampling Code of Practice', which can be found in full on the Darwin Tree of Life website here. By agreeing with and signing up to the Sampling Code of Practice, the Darwin Tree of Life Partner agrees they will meet the legal and ethical requirements and standards set out within this document in respect of all samples acquired for, and supplied to, the Darwin Tree of Life Project.
Further, the Wellcome Sanger Institute employs a process whereby due diligence is carried out proportionate to the nature of the materials themselves, and the circumstances under which they have been/are to be collected and provided for use. The purpose of this is to address and mitigate any potential legal and/or ethical implications of receipt and use of the materials as part of the research project, and to ensure that in doing so we align with best practice wherever possible. The overarching areas of consideration are: • Ethical review of provenance and sourcing of the material
Andrea Battisti
University of Padova, Padova, Italy I do not have major concerns about the manuscript, only a few comments in the attached PDF file which can be viewed here.
I congratulate the authors for the quality of the work that will contribute to the knowledge on the Lepidoptera genomic science and make it possible to compare the species with the few others for which a full genome is available.
Fahad Alqahtani
King Abdulaziz City for Science and Technology, Riyadh, Saudi Arabia In "The genome sequence of the Autumnal Rustic, Eugnorisma glareosa (Esper, 1788)," the authors have achieved a highly complete genome assembly at the chromosome level for a male Autumnal Rustic, Eugnorisma glareosa (Esper, 1788).
The genome assembly was reconstructed using two sequencing technologies: Pacific Biosciences SEQUEL II and Hi-C Illumina. The completeness of the genome assembly was assessed using BUSCO analysis, which indicated a genome size of 631 megabases, with 99% of common genes in the lepidoptera_odb10 present.
There is just one minor comment that should be addressed: 'Pretext' in the Genome Assembly section should be corrected to 'PretextView' to match the name in Table 3.
Is the rationale for creating the dataset(s) clearly described? Yes
Are the protocols appropriate and is the work technically sound?Yes
Are sufficient details of methods and materials provided to allow replication by others? Yes
Are the datasets clearly presented in a useable and accessible format?Yes Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Bioinformatics I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
Figure 2 .
Figure 2. Genome assembly of Eugnorisma glareosa, ilEugGlar6.1:metrics. The BlobToolKit Snailplot shows N50 metrics and BUSCO gene completeness. The main plot is divided into 1,000 size-ordered bins around the circumference with each bin representing 0.1% of the 631,006,646 bp assembly. The distribution of scaffold lengths is shown in dark grey with the plot radius scaled to the longest scaffold present in the assembly (48,685,131 bp, shown in red). Orange and pale-orange arcs show the N50 and N90 scaffold lengths (21,954,397 and 15,478,121 bp), respectively. The pale grey spiral shows the cumulative scaffold count on a log scale with white scale lines showing successive orders of magnitude. The blue and pale-blue area around the outside of the plot shows the distribution of GC, AT and N percentages in the same bins as the inner plot. A summary of complete, fragmented, duplicated and missing BUSCO genes in the lepidoptera_odb10 set is shown in the top right. An interactive version of this figure is available at [URL] 5 .
Figure 5. Genome assembly of Eugnorisma glareosa, ilEugGlar6.1:Hi-C contact map of the ilEugGlar6.1 assembly, visualised using HiGlass. Chromosomes are shown in order of size from left to right and top to bottom. An interactive version of this figure may be viewed at [URL] Review Current Peer Review Status: Version 1
Darwin Tree of Life (DToL) project. All raw sequence data and the assembly have been deposited in INSDC databases. Raw data and assembly accession identifiers are reported in Table 1. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Is the rationale for creating the dataset(s) clearly described? Yes Are the protocols appropriate and is the work technically sound? Yes Are sufficient details of methods and materials provided to allow replication by others? No Are the datasets clearly presented in a useable and accessible format? Yes Competing Interests:
No competing interests were disclosed.
have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
==
Domain: Biology Environmental Science
|
[{"added": "2023-09-22T15:10:15.265Z", "created": "2023-09-19T00:00:00.000", "id": "262094674", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.12688/wellcomeopenres.19987.1", "pdf_hash": "6005cfb9cd396df2a708e6ce1f9f0bf8f67ec897", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:7921", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "1c06fa27c3ecd86f2e9003edf65f08f2af238216", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
High-Throughput Viability Testing of Microbial Communities in a Probiotic Product Using Flow Cytometry
: There is growing scientific and commercial interest in multi-species probiotic products due to their potential benefits in maintaining gut health. Determining the viability of probiotic microorganisms in these products is essential to ensure that they confer maximal health benefits. The gold standard for enumerating probiotic viability is the plate count method. However, this may be inaccurate for enumerating mixed probiotic populations, with recognised limitations including difficulty measuring metabolically active yet unculturable, very slow growing microbes, microencapsulated, enteric coated microbes, or multi-strain formulations that require differing growth media. Here, we developed a flow-cytometry-based approach using SYTOX TM Green dye to assess the viability of probiotic microorganisms in a multi-species, fibre-containing probiotic product and compared this to the traditional plate count method. This method was suitable for enumerating both total bacterial cells and the viable cell fraction in the complete product mixture, and could also be used to assess how stressors, such as gastric digestion and exposure to bile acids, affect bacterial cell viability. Flow cytometry measurements routinely detected higher viable cell counts than plate counting. This work demonstrates that flow cytometry assays can be established as a suitable method for rapid enumeration of viable cells in complex, multi-species probiotics.
Introduction
The gut microbiome plays a crucial role in human metabolism and in maintaining the integrity of the gut mucosal layer and immune homeostasis [1]. Specific alterations in the composition and functions of the gut microbiome have been associated with various diseases [1]. Therapeutic modulation of the gut microbiome offers promising strategies to improve and maintain host health through prevention and treatment of diseases. Probiotics are studied and marketed for their ability to alter the gut microbiome and potential health-promoting benefits [2]. Consequently, the global market for probiotics has grown rapidly in the past few years [3,4]. Probiotics are defined as "live microorganisms, which, when administered in adequate amounts, confer a health benefit to the host" [5]. These probiotic microorganisms are generally consumed either in the form of capsules, powder, or tablets [6], or are contained within whole foods such as yogurt and milk [7,8], with the latter accounting for the largest share of the market [9].
Currently available probiotic products are diverse in their formulations and in the microbial strains they contain. Some products contain microorganisms belonging to a single genus, such as yogurt containing only Lactobacillus species [10], while other products contain mixed microbial populations including representatives of more than one genus, e.g., Lactobacillus, Bifidobacterium, and Streptococcus [2]. Lactic acid bacteria belonging to Lactobacillus and Bifidobacterium genera represent groups commonly included in probiotics [11]. In addition to bacterial strains, some probiotic products also contain yeasts such as Saccharomyces cerevisiae and Saccharomyces cerevisiae var.boulardii which have been shown to improve human gastrointestinal health [12,13]. Probiotic products can additionally contain certain additives that are included to support viability and help stimulate the growth of probiotic bacteria in the gut. Among the most common additives are prebiotics, which are non-digestible food ingredients that are utilised as substrates by selective intestinal microflora, including externally applied probiotics [2,14]. Examples of prebiotic additives include oligosaccharides, such as lactulose, oligofructose, galacto-oligosaccharides, and polysaccharides, such as inulin [2,15].
For probiotic microbes to confer their beneficial effects, they need to be viable and have the capability to survive and maintain their properties during manufacturing processes and storage conditions [15,16]. Ideally, once ingested, these viable cells will reach the colon, adhere to the intestinal mucosa, colonise the gut [17,18], and become a crucial effector of host-microbial interactions [19]. There are also site-specific interactions where probiotics can confer a benefit at a certain point of the gastrointestinal (GI) tract, and this benefit can differ if the probiotic is viable, dead, or degraded entirely [20]. Therefore, the viability of microorganisms is an important property of probiotic products and accurate enumeration is paramount to assessing and predicting the efficacy of products [21]. The current industry standard for enumerating probiotic viability is the traditional culturebased plate count method, where the product is spread onto solid microbiological media in petri dishes, then incubated at growth-permissive temperatures for some period of time, following which visible colony forming units are counted. However, this method is limited in its abilities to accurately enumerate viable cells, particularly from products that contain multiple species. Firstly, plate counting is reliant on the ability of bacterial cells to form colonies on the specific microbiological media under the set incubation conditions [22] and is recognised to underestimate viable microbes by excluding viable but non-culturable cells (VBNCs) [23]. Bacteria in the VBNC state exhibit metabolic activity and are considered to be live but are unable to grow on culture medium [24]. They enter this state to resist stress and it has been suggested that stress associated with probiotic product processing and formulation may result in some proportion of cells entering the VBNC state [25]. Another limitation of the plate count method is the assumption that one microbial cell will give rise to a single colony [25]; however, there is a risk of clumping in probiotic dry powders where multiple cells may give rise to a single colony, resulting in an underestimation of viable cells in products. Multi-species probiotic products containing bacterial and yeast strains require selective media [25] and additional additives (cysteine, antibiotics, etc.) to enumerate different sets of microorganisms within the mixture [19,25]. Furthermore, competition among mixed microbial populations for nutrients can favour the growth of certain microorganisms over others, with those that are slow growing most likely to be under-considered by culture-based enumeration techniques [25].
Non-culture-based techniques such as PCR-based methods (qPCR and digital PCR) and flow cytometry have been used previously to assess viable microbial cells from culture isolates and simple microbial mixtures [26,27]. Application of qPCR-based methods for mixed-species products have some drawbacks as routine enumeration techniques requiring validated primer sets for different members of the mixture and are limited in their capacity to determine absolute rather than relative abundances without the use of spiked-in synthetic standards [28]. For studies involving cross-domain mixtures, such approaches also must consider two different taxonomic marker genes, running separate reactions to amplify both the 16S rRNA gene in prokaryotes and the ITS regions in eukaryotes [28]. Furthermore, there are chances of PCR-related biases being introduced, for example, primer biases, and the complications of data interpretation due to varying 16S rRNA copy numbers in different bacteria. Flow-cytometric-based techniques have been shown to overcome many of the issues encountered with plate counting and those associated with PCR-based enumeration and have been successfully used to assess probiotic viability in products containing relatively simple mixes of bacteria [23,29,30].
Flow cytometry is a multi-parametric single cell analysis technique that identifies and enumerates cells from both homogeneous and heterogenous populations [22] based on the shape and size of cells [19]. Flow cytometry can be used to enumerate viable microbial cells by applying this technique in tandem with cell staining using established viability dyes such as SYTOX TM Green, a nucleic acid stain that indicates bacterial cell membrane integrity status, to help discriminate live cells from dead cells [23,[29][30][31]. Other nucleic acid stains, such as propidium iodide (PI), SYTO, and combinations of nucleic acid dyes including SYTO 9/PI [4,23], and dyes that can discriminate based on metabolic activity such as 5-cyano-2,3-ditolyl tetrazolium chloride (CTC), are also used to assess the viability of bacterial cells using flow cytometry [32]. In some instances, flow cytometry is already used as an alternative for probiotic viability assessment [33]. The International Organization for Standardization and the International Dairy Federation have suggested the use of flow cytometry as the standard method for the enumeration of active lactic acid bacteria and other probiotic microorganisms from dairy products [30,33]. However, uptake of flow cytometry for product assessment is still limited within the broader probiotic industry. Furthermore, studies involving single probiotics species comparing flow cytometry and traditional plate counting have reported a strong agreement between the numbers of dead/live cells enumerated using both methods [34,35]. In contrast, examinations of probiotic bacterial blends containing both Lactobacillus spp.and Bifidobacterium spp.have reported a higher number of viable cells when assessed using flow cytometry compared to plate counting [23,29], demonstrating a discrepancy in viable cell counts observed between the two methods when more than one bacterial species or genera are enumerated. Given the increasing numbers of probiotic products with multiple species of bacteria, yeast, and other active ingredients such as prebiotic fibres, there is a clear and timely need to develop protocols to better utilise high-throughput techniques, such as flow cytometry, to enumerate the viable cell counts of probiotic products.
Here, we developed a flow cytometric approach using a single viability dye to accurately and rapidly analyse and quantify viable microbial cells from heterogenous bacterial and yeast populations from a complex commercial probiotic product (Factors Group Australia). This product contains multiple bacterial strains including representatives of Lactobacillus, Bifidobacterium, and Streptococcus genera, as well as strains of the yeast Saccharomyces cerevisiae, prebiotics, and other plant-based ingredients. As a proof-of-principle, we then applied this protocol to enumerate viable cells following simulated GI tract digestion of the probiotic product, comparing the results obtained from flow cytometry with those derived from standard plate count methods.
Factors Group Australia Multi-Species Probiotic Product
The multi-species probiotic product was provided by the manufacturer Factors Group Australia. Each capsule (transparent hard shell) contained a freeze-dried mixture of the live bacterial strains Bifidobacterium longum subsp.infantis, Lactobacillus acidophilus, Lacticaseibacillus casei (formerly Lactobacillus casei), Lactobacillus gasseri, Lacticaseibacillus paracasei subsp.paracasei (formerly Lactobacillus paracasei), Lactiplantibacillus plantarum (formerly Lactobacillus plantarium), Lactobacillus reuteri, Lacticaseibacillus rhamnosus (formerly Lactobacillus rhamnosus), Ligilactobacillus salivarius subsp.salivarius (formerly Lactobacillus salivarius), and Streptococcus thermophilus. Additionally, the capsules contained yeast strains including Saccharomyces cerevisiae and Saccharomyces cerevisiae (boulardii), as well as plant and fungal extracts. Probiotic capsules containing only the lyophilised bacterial mixture without addition of yeast or other active ingredients were also provided.
Sample Processing
Each capsule was processed by dissolving capsule contents in 50 mL of 0.20 µm filtered phosphate buffered saline (PBS, Oxoid, Sydney, Australia), applying vortexing to ensure a homogenous mixture was generated. For flow cytometry assays where prebiotic fibre removal was necessary, the above-described protocol was carried out and the resulting solution was then passed through a 100 µm cell strainer (FalconA ® , In Vitro Technologies, Sydney, Australia) to filter out larger particles such as plant material. The samples were then 1:10 serially diluted in PBS and the technique appropriate dilutions were used for enumeration via plate count and flow cytometry.
Enumeration and Sequencing of Viable Cells by the Plate Count Method
Bacterial cells within the Factors Group Australia probiotic product were enumerated by spread plating serially diluted capsule contents onto de Man, Rogosa and Sharpe (MRS, Merck, Sydney, Australia), MRS L-cysteine agar and Streptococcus thermophilus agar solid culture media. For each assay, three separate capsules (as biological replicates) were independently prepared as detailed in Section 2.2 and appropriate dilutions were spread plated onto agar plates in triplicate. The MRS and Streptococcus thermophilus agar plates were incubated aerobically at 37 • C for 72 h, while MRS L-cysteine agar plates were incubated anaerobically at 37 • C for 72 h. The plate count results were expressed as colony forming units/capsule (CFU/capsule). Visibly identifiable colonies were enumerated only from plates containing 30-300 colonies per plate. The 16S rRNA gene (using 27F 5 -AGAGTTTGATCMTGGCTCAG-3 and 1492R 5 -TACGGYTACCTTGTTACGACTT-3 primers) of individual colony forming units presented on MRS L-cysteine agar and Streptococcus thermophilus agar plates were sequenced at Macrogen, Korea. The resulting nucleotide sequences were analysed using the basic local alignment search tool nucleotide (blastn) against the National Library of Medicine (NLM) NCBI database, considering the percent identify score when determining the taxonomy assignment.
Generation of Dead Cell Control Populations for Flow Cytometric Analysis
To locate the position of dead cells in the product on flow cytometric plots, a membrane compromised (dead) cell population was generated by exposing an aliquot of the diluted samples to heat at 70 • C for 30 min (water bath incubator, Thermo Fisher Scientific, Sydney, Australia). The resulting cell suspension served as the cell membrane compromised (dead) control population in subsequent viability assessment assays run with the SYTOX TM Green stain [36,37].
Cell Viability Staining for Flow Cytometric Analysis
Bacterial cell viability assays were carried out using SYTOX TM Green nucleic acid dye (Thermo Fisher Scientific, Sydney, Australia) which stains only membrane compromised cells [32], thus distinguishing membrane compromised (dead) from membrane intact cells (viable cells). Briefly, the diluted probiotic suspension samples along with the dead cell control population and filtered PBS negative controls were stained with SYTOX TM Green at a final concentration of 1 µM. These were incubated in the dark for 15 min at room temperature. The filtered PBS control was used to detect and measure any background non-specific fluorescence. All flow cytometry analyses were performed on three separate capsules with and without SYTOX TM Green stain (stained cells and unstained cells), with each biological replicate analysed using three technical replicates.
Flow Cytometric Data Acquisition and Analysis
The samples were analysed on a CytoFLEX S (Beckman Coulter, Brea, CA, USA) flow cytometer using the CytExpert software (Beckman Coulter, Brea, CA, USA) for data analysis. The following settings were used: gain for forward scatter (FSC)-100, side scatter (SSC)-100, and FITC (green fluorescence detector)-40, while the threshold for FSC was set to 1500 and that for SSC to 500 (the flow rate of the cells was set at the slow setting). The total cell counts were determined by applying gates based on the dot-plot SSC versus FSC (SSC-A vs. FSC-A). The positioning of the dead cell control population was used to set gates to separate the membrane compromised cells from the membrane intact cells. These gates were applied on the dot-plot SSC vs. FITC (SSC-A vs. FITC-A); both plots were set on log 10 -scale. The SYTOX green fluorescence intensity peak (indicative of membrane compromised cells) was also observed via a histogram plot of FITC vs. count, with FITC-A set on log scale and count set on linear scale. To identify and gate the bacterial population separately from the yeast population present in the multi-species probiotic product, probiotic capsules containing only the lyophilised lactic acid bacterial component of the product mixture were analysed. To account for the possible presence of instrument background noise, filtered PBS-only (blank) controls, which were prepared the same way as for the samples, were analysed on the flow cytometer and events were recorded at the beginning of each flow cytometry assay. To determine the total cell counts, any events detected in PBS controls were subtracted from the total events detected as the proportion of events attributable to background noise. The number of viable cells of the product was expressed as cells/capsule.
Tolerance to Conditions Simulating Gastrointestinal (GI) Tract Passage
The multi-species probiotic capsules were exposed to gastric digestion as per the protocol described by Shehata et al., with slight modifications [38]. Briefly, simulated gastric juice was prepared by suspending 0.3% (w/v) pepsin (Merck, Sydney, Australia) in PBS at pH 2.0. The simulated gastric juice and 0.5% NaCl were added to nine independent 15 mL sterile tubes, each containing a single probiotic capsule. The tubes were incubated at 37 • C with agitation at 100 rpm. At 0, 1.5, and 3 h of incubation, three tubes were removed for processing. Tube contents were vortexed and centrifuged at 7800 rpm at 4 • C for 10 min, followed by PBS washes of the cells until the pH values of the samples were 7.0. The resulting cell suspensions were then processed as described in Section 2.2. Tolerance to gastric digestion was assessed by determining viable counts at each time point using both plate counts and flow cytometric analyses in parallel by splitting the contents of each tube into two for use in each technique. The percent survivability was calculated by comparing the treatment samples (1.5 and 3 h) to the samples collected at 0 h.
For assessing resistance to bile salts, capsule contents were dispensed into tubes containing 0.3% (w/v) of bile salts (Merck, Sydney, Australia) suspended in PBS (pH 7.0). The samples were incubated at 37 • C with agitation at 100 rpm for 3 h. Samples were collected at intervals of 0, 1.5, and 3 h and processed as described above. Three technical replicates of each of the nine samples were diluted and assessed for the viable cell counts and % survivability by plate count method and flow cytometry as described above.
Statistical Analysis
Data were analysed using GraphPad Prism (version 9.4.1,GraphPad Software Inc., San Diego, CA, USA). Data is displayed as mean ± standard deviation (SD). The CFU/capsule and cells/capsule counts were transformed to Log 10 values and for all experiments normality tests were performed using the D'Agostino and Pearson omnibus test to assess the distribution of data. Statistical analysis was performed using Student t-tests to compare the mean differences between two groups, while one-way ANOVA tests were used to compare the mean differences of more than two groups, with p ≤ 0.05 used as the significance level cut-off for both.
Determining Viable Cell Number Variability for Factors Group Australia Probiotic Capsules
The standard plate count technique using MRS culture media was used to count colony forming units and assess the cell count variability between three capsules of the multi-species probiotic product. Viable cell numbers between capsules were determined to be consistent (Table 1), with no significant differences recorded (one way ANOVA, p = 0.59). Serial dilutions of the complete product with mixed bacteria were plated on MRS L-cysteine agar and Streptococcus thermophilus agar plates. While colony forming units were seen on both media, sequencing of these colonies revealed taxonomic assignments to Lactobacillus spp. that are present in the product. Therefore, colony forming units obtained from MRS L-cysteine agar and Streptococcus thermophilus agar plates were not included in any of the plate counting results presented in this study.
Impact of Filtering out Prebiotic Fibres on the Viable Cell Counts of the Multi-Species Probiotic Product
While flow cytometry is a relatively versatile technique, to be suitable for analysis using this technique, any large particles or clumps must first be removed to ensure flow conditions are maintained and prevent clogs in the fluidic path. For this reason, prior to processing probiotic capsule contents, it was necessary to perform a pre-filtering step to remove particles over 100 µm in size. We performed a set of initial experiments to examine how such filtering impacts the flow-cytometry-derived viable cell counts in the multi-species probiotic product. The plate counting method was used to determine viable cell numbers for both the filtrate and the residue retained on the filter to determine the proportion of the viable cell population that was removed by the filtration process. There was a 4.3% reduction in the viable cell numbers in the filtrate compared to the viable cell numbers in capsules not subjected to filtration (Table 2), suggesting that filtration has a minimal impact on the viable cell counts of the product. As the difference in viable cell counts for filtered and unfiltered capsule suspensions was found not to be significant (Student t-test, p = 0.39), and removing the larger particles was necessary to enable reliable flow cytometry analysis of multi-component fibre-containing products, this filtering step was applied in all subsequent work.
Enumeration of Viable Cells in the Multi-Species Probiotic Product
Following preliminary experiments to establish that individual capsules contained relatively consistent numbers of viable cells and that separation of viable cells from fibre was possible, we worked to develop a routine protocol for flow-cytometry-based viable cell enumeration. In addition to multiple strains of bacteria, the tested product also contained yeast. Therefore, to enable the positioning of separate gates for bacterial and yeast populations on the flow cytometric plots, we analysed capsules containing only the lyophilised bacterial mixture alongside standard capsules. The gating for yeast was consistent with the expected trend for larger and more internally complex cells (brown-coloured gate, Figure 1). To discriminate membrane compromised cells from membrane intact viable cells, flow cytometry assays were performed on cells stained with SYTOX TM Green dye. To manually gate the membrane compromised bacterial and yeast cell populations in the product using flow cytometry, the cytograph positioning of heat-killed control populations (green-and brown-coloured gates, respectively, Figure 1A-C) was used. Analyses of the complete product with flow cytometry was then performed to allow both the viable and non-viable (membrane compromised) bacterial populations to be enumerated (representative results shown in Figure 1D-F). Due to the relatively small size of the bacterial cells in the product, it was not possible to set a flow cytometry threshold capable of completely removing all background noise, so PBS-only (blank) control mean cell counts were subtracted from the total bacterial cell counts to more accurately determine viable bacterial cell counts.
Enumeration of Viable Cells in the Multi-Species Probiotic Product
Following preliminary experiments to establish that individual capsules contained relatively consistent numbers of viable cells and that separation of viable cells from fibre was possible, we worked to develop a routine protocol for flow-cytometry-based viable cell enumeration. In addition to multiple strains of bacteria, the tested product also contained yeast. Therefore, to enable the positioning of separate gates for bacterial and yeast populations on the flow cytometric plots, we analysed capsules containing only the lyophilised bacterial mixture alongside standard capsules. The gating for yeast was consistent with the expected trend for larger and more internally complex cells (brown-coloured gate, Figure 1). To discriminate membrane compromised cells from membrane intact viable cells, flow cytometry assays were performed on cells stained with SYTOX TM Green dye. To manually gate the membrane compromised bacterial and yeast cell populations in the product using flow cytometry, the cytograph positioning of heat-killed control populations (green-and brown-coloured gates, respectively, Figure 1A-C) was used. Analyses of the complete product with flow cytometry was then performed to allow both the viable and non-viable (membrane compromised) bacterial populations to be enumerated (representative results shown in Figure 1D-F). Due to the relatively small size of the bacterial cells in the product, it was not possible to set a flow cytometry threshold capable of completely removing all background noise, so PBS-only (blank) control mean cell counts were subtracted from the total bacterial cell counts to more accurately determine viable bacterial cell counts. In order to compare the flow cytometry-derived viable counts for this mixed-species probiotic with that obtained by traditional plate count methods, the standard capsule analyses were performed in parallel. Comparing the culture-based CFU per capsule count to the viable cell counts obtained via flow cytometry, we found the latter showed significantly higher (Student t-test, p < 0.001) counts (Table 3).
Testing the Reproducibility of the Flow Cytometry Protocol
To examine whether results obtained by flow cytometry and plate count methods are reproducible, testing was performed with six additional capsules using both techniques. Similar to our previous observations, higher viable cell counts were obtained using flow cytometry compared with the plate count method (Table 4). Notably, flow cytometry detected consistent numbers of viable cells across the six biological replicates, while for plate count assays there was greater variability, with fold change differences in the CFU/capsule count determined for capsules 1-3 compared to capsules 4-6 (Table 4). The results suggested a higher reproducibility of flow-cytometry-based enumeration of viable cells compared to traditional plate counting using MRS media.
Comparison of Flow Cytometry and Plate Counting Methods for Assessing Viability following Simulated GI Tract Digestion
To ascertain how each enumeration approach compared when used to determine cell survivability following a challenge, we performed assays examining the impact of exposure to gastric juice (0.3% pepsin at pH 2.0) on bacterial viability for the complete probiotic product. Both techniques showed that sizable populations of viable bacterial cells remained after each of the tested exposure periods (Table 5). However, flow cytometry assay-based enumeration indicated considerably higher proportions of bacteria remained viable at both 1.5 and 3 h than what was indicated by plate counting at each time point (Table 5). Following the stomach digestion mimicking assays, we performed standard in vitro bile salt exposure assays, again examining viable cell survivability following different exposure times. Similar trends were observed in this second assay, with flow cytometry indicating substantially higher viable counts after both 1.5 and 3 h bile salt exposure than plate count assays (Table 6).
Discussion
Here, we developed a flow-cytometry-based technique using a commercial viability dye to enumerate viable bacterial cells in a multi-species probiotic product containing bacteria, yeast, and other components. In this work we observed that viable cell counts based on flow enumeration were higher than those derived from plate counting method. This finding is consistent with what has been reported in a number of previous studies on probiotic blends containing Lactobacillus and Bifidobacterium spp. [23,29,39] and on multi-species probiotic products Lacidofil ® , Protecflor ® , and ProbioKid ® (Mirabel, QC, Canada) [40]. It should be noted that the flow cytometry and plate count methods presented in our study are not measuring precisely the same thing, with flow cytometry providing estimates of membrane intact cells per capsule and plate count determining the numbers of colony forming units per capsule under the specific culture conditions applied (MRS medium, incubated aerobically at 37 • C for 72 h). These culture media and conditions used for plate counting typically favour growth of Lactobacillus spp., especially if they are present in a multi-species population. The product tested in this work, whilst comprised predominantly of strains of Lactobacillus spp., did also contain Bifidobacterium longum subsp.infantis and Streptococcus thermophilus. MRS L-cysteine agar and Streptococcus thermophilus agar, which are known to favour the growth of Bifidobacterium and Streptococcus thermophilus, respectively, were used to account for the different genera, but only Lactobacillus spp.were recovered on them. Similarly, work by Sohrabvandi et al., isolating lactic acid bacteria from yogurt, showed Lactobacillus spp.consistently outgrew Bifidobacterium spp., despite providing anaerobic conditions which are considered to be tolerated by both groups [41]. It is possible that competition amongst these multiple species and genera, when co-plated as present in the product, contributed to the lack of viable Bifidobacterium and Streptococcus spp. on the MRS L-cysteine and Streptococcus thermophilus agar plates.
We consistently observed a lower count of colony forming units and higher variation in results when plate counting was used compared to the number of viable cells detected through flow cytometry. This could also be due to the competition between bacterial species and genera induced due to co-plating the mixed population of bacteria present in the tested product. Furthermore, the ability of flow cytometry to detect injured yet membrane intact cells [25], including cells in the VBNC state, may also have contributed to the higher number of cells detected by flow cytometry. The ability of flow cytometry to detect VBNCs and dormant cells, which are generally not detected by plate counting, may have been the reason that the stress tolerance assays resulted in particularly large discrepancies in viable cell counts between flow cytometry and plate counting, as these assays may have resulted in cells within this product entering into a VBNC or dormant state [25], in addition to the competition between the multiple species of bacteria for growth conditions as described above. Furthermore, these limitations associated with plate counting may have exacerbated any minor variation in the proportion of different bacteria present in the product, leading to a greater variation in colony forming units detected. It is also worth noting the technical limitations in enumerating multiple species and genera of bacteria present in products using plate counting due to the different growth conditions required to culture them. This is a significant limitation in assessing the viability of probiotics in complete products containing mixed populations of bacteria, which can be minimised using flow cytometry as we demonstrate here.
The ability of probiotic microorganisms to tolerate GI transit is considered to be an important feature of an effective probiotic product [42] and is therefore assessed at the initial stages of identifying microorganisms with probiotic characteristics. Probiotic survival through the gastrointestinal tract is very much dependent on the bacterial strain and the delivery mechanism of the bacteria (e.g., capsule or powder) [19]. Characteristically, when assessed in a simulated GI tract, probiotic cultures show a decrease in concentration in the stomach followed by an increase in concentration in the intestine, with factors such as the time in the stomach and the composition of foods in the stomach at the time influencing probiotic concentrations throughout the process [43]. Here, we applied flow cytometric enumeration of viable cells to assess the response of a mixed bacterial population in a probiotic product to conditions mimicking independent sections of GI transit. While such testing is routine, it is generally performed on pure cultures during single strain selection [16], instead of on the final probiotic product, and there are limited evaluations of products after freeze drying and formulation [44]. The stage at which such testing is conducted is important, as the survival after exposure to gastric acid and bile can differ significantly, sometimes by several log units depending on formulation, freeze-drying, and storage conditions [45]. One reason for this is that during production, dried lyophilised cells generally undergo rehydration which is challenging to the cell as it moves from a gel-like state to a crystalline liquid state [44]. In this work, lower survivability observed in plate growth assays may have been due to cells entering a transition phase where they could maintain their metabolic activity but were unable to form colonies on the culture medium. Studies that have compared the GI tolerance of the unencapsulated probiotics with the encapsulated probiotics (different encapsulation materials) have shown higher tolerance rates when probiotics are encapsulated [46][47][48]; however, this literature involved capsules that dissolved in bile salt assays, i.e., simulated intestinal conditions rather than in the stomach-mimicking conditions [46][47][48]. In our case, the capsule dissolved in the gastric juice with gentle agitation, suggesting that further work comparing effects of different capsule formulations on viable cell concentrations at different stages of continuous GI transit, transit time variation, and GI transit survivability across multiple complex product formulations would be of interest.
Both of the stressors examined can lead to cell injury, loss in viability, and death. However, the pH buffering capacity, product composition, transit time, and volume of gastric juice are also affected by recent food consumption and composition [49]. Different strains also have differences in bile tolerance due to the individual species' ability to express the bile salt hydrolase enzyme [50]. LAB strains that express more bile salt hydrolase can survive longer in bile salts. Hydrolysed bile salts are a key feature of probiotic bacteria; however, not all probiotics possess this capability to the same degree [5,51]. LAB strains can also possess other abilities to cope with bile exposure, such as active efflux of bile salts and changes in the cell wall and cell membrane composition, which may have downstream effects when assessing cell viability, a question which is worthy of further investigation.
In this work, flow cytometry was used to count the proportion of viable bacterial cells remaining following two different challenges. As the ability to tolerate each of these stressors is likely to differ to some extent between strains with the mixture [38], it would be useful to be able to determine the strain level composition of the surviving population. This could be achieved by using cell sorting to separate viable and non-viable populations following stress assays followed by 16S rRNA amplicon sequencing, which would aid in identification of strains that were more or less tolerant of exposure to such stressors. While the in vitro method used here to simulate stomach digestion and intestinal challenges conform with standard assays used for probiotic research [38,47,48], as discussed above, they have inherent limitations, offering only limited similarity to the complex GI transit of the human body [19,52]. The human digestive system is a dynamic process where there is continuous flow of fluids and enzymes, and varying concentrations of pH and bile salts, throughout the GI tract, which is generally difficult to maintain in an in vitro system [19,52]. Since the in vitro method is not a true representation of the human GI tract system, it is possible that the impact was either milder or more detrimental than what may occur in vivo.
There are many commercial cell viability dyes suitable for flow cytometry applications, with dyes that indicate membrane integrity, such as the SYTOX TM Green dye used in our study, among the most common methods of viability assessment. SYTOX TM Green dye stains nucleic acids in membrane compromised cells, while combinations of other nucleic acid dyes such as SYTO 9/PI allow the detection of total cells (SYTO 9) and enhance the discrimination of membrane compromised cells using PI [53]. Yet, these nucleic acid dyes do not distinguish between metabolically active and dormant cells or damaged cells with intact membranes. While the staining approach used here does not directly identify damaged cells, histogram plotting of the intensity of the fluorescence signal from the SYTOX dye when applied to the probiotic product (Figure 1F) revealed some continuity between the main peaks corresponding to membrane intact and membrane compromised cells, which might indicate the presence of membrane damaged cells. Further flow cytometric examination using additional dyes could be employed to explore the additional information on the physiological status of mixed-species probiotics that can be determined. Additional combinations of fluorescent dyes compatible with flow cytometry [30,54], could be trialled to gain more information about the physiological and metabolic state of cells in probiotic mixtures. Dyes such as CTC, which when reduced by dehydrogenases form fluorescent membrane-impermeant formazan, have been considered useful in providing an indication of actively respiring cells, as absence of formazan production indicates a lack of esterases or dehydrogenases, which could be indicative of dead or damaged cells [32]. It may also be possible to fluorescently tag antibodies specific to particular bacterial strains with viability stains to enable specific enumeration of viable cell numbers of each strain within a mixed population [40]. However, for complex probiotic products, establishing such assays would require considerable set up time and cost, which are likely to make this a less convenient approach for routine testing by probiotic industries.
Probiotic products are increasing in complexity and there are a range of products on the market that combine live probiotic bacteria with other ingredients, including prebiotic fibre and other active but non-living ingredients. For these, flow cytometry enumeration requires an additional processing step to remove large particles to prevent clogging of flow cytometer tubing. Here, we showed that filtration (using 100 µm cell strainers) was successful in preparing such a product for flow cytometry, having a minimal impact on viable cell counts. Such processing is likely to be a necessary step for flow cytometric cell enumeration of similar products containing prebiotic fibres, but independent validation for different products using plate-count-based comparisons is required to ensure there is not a higher proportion of fibre-adherent cells in some products.
Flow cytometry provides various advantages over the standard plate count method. We have shown that flow cytometry can differentiate yeast and bacterial populations and rapidly enumerate viable cell numbers within complex mixtures in a matter of hours, while the traditional plate count method requires 48-72 h to grow visible colonies, providing counts of only culturable cells [40]. Culture-based approaches may also be more variable in assessing stress tolerance depending on recovery times used in assays prior to plating, as stressed cells may transfer to different non-culturable states, possibly resulting in the measured CFU differing to varying degrees from the actual viable cell count [19]. Furthermore, competition among mixed bacterial populations for nutrient sources can lead to outgrowth of certain microorganisms over others, likely leading to undercounting of slower growing or condition-sensitive organisms [25]. In addition to successfully overcoming known plate count limitations, flow cytometry may provide information on various aspects of cell physiology. The ability of flow cytometry to enumerate cells in probiotic mixtures at any stage in the product formulation process has the potential to provide valuable information on the impact of different manufacturing steps or compare alternative techniques for use in product preparation, such as spray drying and freeze drying, on the viability of probiotic microorganisms.
It should be noted that a known issue with flow cytometry is the presence of background noise, which can be especially problematic when analysing relatively small bacterial cells or complex mixtures [27]. While background noise can in some cases be excluded via threshold setting or gating, this is not always possible as the noise may overlap with populations of interest. In such cases, subtracting the average counts for repeated blank measurements from the cell counts of interest, as carried out in this study, has been considered an acceptable approach to overcome this issue [55]. Another option may be to use two fluorescent dyes in combination, which has the potential to reduce the amount of background noise that is inseparable from the signal indicative of stained cells. Additionally, while performing established assays with this technique is much faster than plate counting, it should be acknowledged that the design of reliable flow cytometric assays can take time. These initial steps require use of appropriate controls to optimise instrument settings and enable appropriate gates to be established to count subpopulations of interest.
While successful use of flow cytometry requires one to first establish reproducible protocols and optimise instrument settings for a particular probiotic product, once this has been achieved, the same protocol can subsequently be implemented as a routine, rapid enumeration technique. This has numerous possible applications, such as assessing viable cell counts across different batches of a product or to determine how different stressors impact the product. While some members of the probiotic industry are trialling flow cytometry [30,36], this technique has not yet been adopted by the regulatory bodies in most of countries, including Australia, for their product assessment activities. However, this may change in the future, as this technique shows considerable potential for providing more accurate measurements of viable cells in probiotic products and is already a regulated requirement for other industries such as the dairy industry [27].
Conclusions
Flow cytometry analysis has the potential to provide very detailed information about all cells within both simple and complex samples. However, certain knowledge of the cell population is required to be able to achieve an experimental design that will yield reliable information and reproducible results [56]. Here, we undertook to compare plate counting and flow cytometry approaches for enumeration of viable cells in a probiotic product that contains a complex mixture of bacterial strains, yeast strains, and other active ingredients. For this, we developed a methodology for processing the probiotic capsule to allow flow cytometric analyses, along with instrument protocols and gating strategies to separate out subpopulations and enumerate viable cell fractions. This protocol has the potential to be applied to investigate how a range of different factors affect bacterial viability, from the impact of different manufacturing steps through to exploring batch variability, or shelf life, as product viability and composition can change within a probiotic product over time [57]. This could also be used to examine other factors of high interest to the probiotic industry, such as site-specific activity and lysis along the GI tract.
Flow cytometry was found to have advantages over plate counting in terms of speed, reproducibility, and sensitivity, likely due to the capacity to detect cells in the VBNC state. There are some acknowledged limitations, however, with flow cytometry requiring large particulates in samples, such as probiotic fibres, to be removed prior to analysis. This may, in some instances, affect viable cell counts, either due to high rates of fibre attachment or the potential that removal of additional active ingredients alters viability, possibly due to the protective effects of these additional components of the formulation [57]. Despite such issues, we conclude that flow cytometry has the potential to be used as a routine enumeration technique for rapid and accurate viability assessment for probiotics, especially for mixed-species probiotic products and, as such, is likely to be increasingly used in the coming years.
Figure 1 .
Figure 1. Representative flow cytometric plots of cells populations for heat-killed control populations (A-C) and the complete probiotic product (D-F), all analysed following SYTOX TM Green viability dye staining. In panels (A,D), gating of total bacterial and yeast cells is shown. In plots (B,E), the green gates indicate membrane comprised bacterial cells and the brown gates indicate membrane compromised yeast cells. The coloured histogram peaks in (C,F) show that cells located within the "membrane compromised" gates had the expected high FITC-A intensity and are distinct populations, separate from the membrane intact populations (grey peak in F, absent in the dead control in plot C).
Figure 1 .
Figure 1. Representative flow cytometric plots of cells populations for heat-killed control populations (A-C) and the complete probiotic product (D-F), all analysed following SYTOX TM Green viability dye staining. In panels (A,D), gating of total bacterial and yeast cells is shown. In plots (B,E), the green gates indicate membrane comprised bacterial cells and the brown gates indicate membrane compromised yeast cells. The coloured histogram peaks in (C,F) show that cells located within the "membrane compromised" gates had the expected high FITC-A intensity and are distinct populations, separate from the membrane intact populations (grey peak in (F), absent in the dead control in plot (C)).
Table 1 .
The viable cell counts of three multi-species probiotic capsules determined using the standard plate count method.
Shown as mean ± SD; each capsule analysis was carried out in technical triplicate.
Table 2 .
Plate-count-based assessment of the effect of filtration (100 µm) on viable cells counts for the multi-species probiotic product suspension.
Shown as mean ± SD; analyses were carried out in technical triplicate. CFU-colony forming units.
Table 3 .
The average viable cell counts per capsule of the multi-species probiotic product determined by flow cytometry and plate count methods. Shown as mean ± SD; analyses were carried out on three capsules (as biological replicates), each in technical triplicate. CFU-colony forming unit.
Table 4 .
The viable cell counts of six individual capsules of the multi-species probiotic product obtained by flow cytometry and plate count methods.
Shown as mean ± SD; analyses were carried out in technical triplicate for each of the six capsules. CFU-colony forming unit.
Table 5 .
Impact of simulated gastric juice on the viable cells of multi-species probiotic capsules, determined by flow cytometry and plate count methods. SD; each time point was carried out with three biological replicates, and each in technical triplicate.
Table 6 .
Impact of exposure to bile salts on the viable cells of multi-species probiotic product, determined by flow cytometry and plate count methods.
Shown as mean ± SD; each time point was carried out with three biological replicates, and each in technical triplicate.
==
Domain: Biology Environmental Science
|
[{"added": "2023-09-14T15:21:54.112Z", "created": "2023-09-11T00:00:00.000", "id": "261781062", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2673-8007/3/3/74/pdf?version=1694437135", "pdf_hash": "37574272096d161caace4c581e3e767bf6fe4036", "pdf_src": "ScienceParsePlus", "provenance": "peS2o-0028.json.gz:1525", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "1f2532385725db9806c6e4901f898f0647f946b3", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Integrated Transcriptomic and Metabolomic Analysis Reveal the Underlying Mechanism of Anthocyanin Biosynthesis in Toona sinensis Leaves
Toona sinensis, commonly known as Chinese Toon, is a plant species that possesses noteworthy value as a tree and vegetable. Its tender young buds exhibit a diverse range of colors, primarily determined by the presence and composition of anthocyanins and flavonoids. However, the underlying mechanisms of anthocyanin biosynthesis in Toona sinensis have been rarely reported. To explore the related genes and metabolites associated with composition of leaf color, we conducted an analysis of the transcriptome and metabolome of five distinct Toona clones. The results showed that differentially expressed genes and metabolites involved in anthocyanin biosynthesis pathway were mainly enriched. A conjoint analysis of transcripts and metabolites was carried out in JFC (red) and LFC (green), resulting in the identification of 510 genes and 23 anthocyanin-related metabolites with a positive correlation coefficient greater than 0.8. Among these genes and metabolites, 23 transcription factors and phytohormone-related genes showed strong coefficients with 13 anthocyanin derivates, which mainly belonged to the stable types of delphinidin, cyanidin, peonidin. The core derivative was found to be Cyanidin-3-O-arabinoside, which was present in JFC at 520.93 times the abundance compared to LFC. Additionally, the regulatory network and relative expression levels of genes revealed that the structural genes DFR, ANS, and UFGT1 might be directly or indirectly regulated by the transcription factors SOC1 (MADS-box), CPC (MYB), and bHLH162 (bHLH) to control the accumulation of anthocyanin. The expression of these genes was significantly higher in red clones compared to green clones. Furthermore, RNA-seq results accurately reflected the true expression levels of genes. Overall, this study provides a foundation for future research aimed at manipulating anthocyanin biosynthesis to improve plant coloration or to derive human health benefits.
Introduction
As the national economy grows rapidly, and living standards continue to improve, consumers today demand more than just food and clothing. They desire a healthy diet that also helps prevent diseases. Food that is high in nutrition and has antioxidant properties is particularly appealing [1]. Chinese Toon (Toona sinensis), a tree with both medicinal and edible uses, contains rich chemical substances with functions such as antifungal, antiglycation, or anti-tumor activities. Flavonoid compounds, especially anthocyanin, have been identified as the major bioactive constituents in Toona [2][3][4] which are considered to be of the most beneficial plant-derived antioxidants and have led to increased efforts to analyze the genetic controls of the mechanism of flavonoids synthesis [5,6].
Anthocyanins belong to the flavonoid class of phenolic compounds. Anthocyanins are water-soluble pigments present in plants that contribute to the colors of leaves, flowers, and fruit, which are also crucial for plant growth and beneficial for human health due to their ability to scavenge active oxygens [7]. More than 600 distinct anthocyanin types have been identified so far through previous research [8]. The most prevalent anthocyanin pigments, which generate purple, blue, and red colors, are cyanidin (Cy), delphinidin (Dp), pelargonidin (Pg), peonidin (Pn), petunidin (Pt), and malvidin (Mv) [9]. Anthocyanin biosynthesis and regulation is a complex process influenced by various internal and external factors [10,11].
The investigation of anthocyanin-regulating genes and associated flavonoid metabolites is critical to elucidate the mechanism that governs the biological synthesis of anthocyanins. To study gene expression and metabolic changes in different cultivars, researchers mainly rely on transcriptomic and metabolic profiling, which complement each other [34,35]. In this study, we conducted a dual-level exploration of genes and metabolites that control the pigmentation of Toona sinensis leaves using transcriptome and metabolomics approaches. Specifically, we analyzed the transcriptome and metabolome data of five Toona sinensis clones exhibiting a range of leaf colors, including red, light red, pink, dark, and green, to clarify regulatory networks for biosynthesis-related genes and key metabolomes. Overall, our findings unveil the regulatory mechanisms driving the coloration of Toona sinensis leaves.
Morphological Observation and Anthocyanin Content of Leaves in Toona sinensis
Five clones of Toona sinensis with distinct leaf colors were selected to analyze the regulatory mechanism of anthocyanin biosynthesis. Among the clones, JFC Toona is characterized by brilliant-red leaves and is commonly cultivated in Zhejiang province. The leaf color of other clones was pink buds (FHXN), light-red with a little green buds (BSH), dark-purple buds (HBG), and green buds (LFC) (Figure 1A). Then, the total anthocyanin was extracted and the resulting color were displayed in Figure 1B, which reflect the anthocyanin content of different clones. Total anthocyanin present in the leaves were quantified, with JFC Toona sinensis found to have the highest concentration of anthocyanin, followed by BSH, FHXN and HBG, and LFC (Figure 1C). Obviously, LFC exhibited green leaves due to a lack of anthocyanin accumulation.
The leaf color of other clones was pink buds (FHXN), light-red with a little green buds (BSH), dark-purple buds (HBG), and green buds (LFC) (Figure 1A). Then, the total anthocyanin was extracted and the resulting color were displayed in Figure 1B, which reflect the anthocyanin content of different clones. Total anthocyanin present in the leaves were quantified, with JFC Toona sinensis found to have the highest concentration of anthocyanin, followed by BSH, FHXN and HBG, and LFC (Figure 1C). Obviously, LFC exhibited green leaves due to a lack of anthocyanin accumulation.
Metabolome Analysis of Anthocyanin in Five Toona Clones
To compare the flavonoids and anthocyanin compound composition among five Toona clones, three biological replicates of each clone were sampled to detect anthocyaninrelated metabolites employing the UPLC-MS/MS platform. The resulting metabolite composition data sets underwent PCA analysis (Figure 2A). The PCA plot for the anthocyaninrelated metabolites showed a clear separation between the red, pink, light red, green, dark-purple samples. A total of 43, 40, 44, 40, and 32 anthocyanin-related metabolites were identified in leaves of JFC, BSH, FHXN, HBG, and LFC, respectively (Table S1). A total of 31 anthocyanins were identified in all of the samples. Then, the |log2(fold-change)| ≥ 1 and p value < 0.05 was used to determine significantly metabolites between groups (Figure 2B). In group LFC vs. JFC, 23 upregulated metabolites and 8 downregulated metabolites were found to be significantly different (Figure 2B). In LFC vs. BSH, there were 24 upregulated metabolites and 6 downregulated metabolites (Figure 2B). In BSH vs. JFC, there were 10 upregulated metabolites and 11 downregulated metabolites (Figure 2B). Counting the differential metabolites compared with both LFC and JFC for each individual plant, a Venn diagram was used to show that there were 6 differential metabolites in the comparison group compared with JFC and 24 common differential metabolites compared with LFC (Figure 2C,D). Thirty-one common detected metabolites were used to draw clustering heatmap by TBtools ( [URL] on 13 September 2023). The contents of cyanidin and delphinidin could be mainly responsible for the leaves color of Toona sinensis (Figure 2E).
Metabolome Analysis of Anthocyanin in Five Toona Clones
To compare the flavonoids and anthocyanin compound composition among five Toona clones, three biological replicates of each clone were sampled to detect anthocyanin-related metabolites employing the UPLC-MS/MS platform. The resulting metabolite composition data sets underwent PCA analysis (Figure 2A). The PCA plot for the anthocyanin-related metabolites showed a clear separation between the red, pink, light red, green, dark-purple samples. A total of 43, 40, 44, 40, and 32 anthocyanin-related metabolites were identified in leaves of JFC, BSH, FHXN, HBG, and LFC, respectively (Table S1). A total of 31 anthocyanins were identified in all of the samples. Then, the |log2(fold-change)| ≥ 1 and p value < 0.05 was used to determine significantly metabolites between groups (Figure 2B). In group LFC vs. JFC, 23 upregulated metabolites and 8 downregulated metabolites were found to be significantly different (Figure 2B). In LFC vs. BSH, there were 24 upregulated metabolites and 6 downregulated metabolites (Figure 2B). In BSH vs. JFC, there were 10 upregulated metabolites and 11 downregulated metabolites (Figure 2B). Counting the differential metabolites compared with both LFC and JFC for each individual plant, a Venn diagram was used to show that there were 6 differential metabolites in the comparison group compared with JFC and 24 common differential metabolites compared with LFC (Figure 2C,D). Thirty-one common detected metabolites were used to draw clustering heatmap by TBtools ( [URL] on 13 September 2023). The contents of cyanidin and delphinidin could be mainly responsible for the leaves color of Toona sinensis (Figure 2E).
Transcriptome Analysis of Five Toona sinensis Clones
Fifteen cDNA libraries (each clone with three biological replicates) were sequenced using the Illumina HiSeq platform. After removing low-quality reads, each sample yielded between 42.16 M to 52.03 M clean reads (Supplementary Figure S1). The Q30 base exceeded 92.56% and the guanine-cytosine (GC) content of each sample was greater than 42.27%, indicating acceptable sequencing quality. After filtration, the clean data reached 6.32-7.80Gb for each sample. The genome of Toona sinensis var. Heiyouchun was utilized, which marked a significant was the first chromosome-level genome assembled of Toona, as our reference genome [36]. Furthermore, comparison of all clean reads with this reference genome and subsequently performed gene annotations, the results revealed a high rate of similarity, ranging from 92.26% to 94.03%. These results validate the reliability of our data. Novel genes were defined as unigenes found in the sequencing results but not included in the reference genome (or reference gene set) after reconstructing the transcripts using StringTie v1.3.4dsoftware. Comparison identified 9127 novel genes. The PCA results of transcriptome showed significant differences among five clones (Figure 3A). To investigate the function of unigenes in Toona, we conducted a comparative analy sis with various databases including KEGG, NR, Swiss-Prot, GO, KOG, and Trembl usin BLAST (Diamond version 2.0.9)software. To obtain annotation information on unigene amino acid sequences were derived from the unigene sequences and compared agains the Pfam database using HMMER V3.2 ( [URL] on 10 June 2022) sof ware. Interestingly, when compared to data in the NR database, it was observed that gen sequences from Toona sinensis exhibited a greater degree in similarity to Citrus clementin To investigate the function of unigenes in Toona, we conducted a comparative analysis with various databases including KEGG, NR, Swiss-Prot, GO, KOG, and Trembl using BLAST (Diamond version 2.0.9)software. To obtain annotation information on unigenes, amino acid sequences were derived from the unigene sequences and compared against the Pfam database using HMMER V3.2 ( [URL] on 10 June 2022) software. Interestingly, when compared to data in the NR database, it was observed that gene sequences from Toona sinensis exhibited a greater degree in similarity to Citrus clementina (30.91%), followed by Citrus sinensis (30.47%) as depicted in Supplementary Figure S1B. The function terms of unigenes were annotated in the KOG and GO database, and the function groups were showed in Supplementary Figure S1C,D.
Based on the reference transcriptome, 40,421 genes were assembled from clean reads, which included 31,647 genes annotation in reference genome and 9127 novel transcripts. In addition, 23,909, 16,532, 25,617, 15,566, 32,348, and 25,575 unigenes were annotated by KEGG, NR, SWISSPORT, THEM, GO, and KOG database. There were 14,153 unigenes annotated by all of the databases. A comparison with six public databases (KEGG, NR, SwissProt, Tremble, GO and KOG) annotated a total of 13,295 differentially expressed genes (DEGs) and predicated their potential functions. The DEGs were counted with an absolute |log2fold-change| ≥ 1 and a false discovery rate (FDR) ≤ 0.05 in each compared groups. According to five clones of Toona leaves, 10 pairs of comparisons were performed and the number of up-regulated and down-regulated differential genes was counted in all groups (Figure 3B). Across all compared groups, a total of 310 DEGs were co-detected (Figure 3C). The KEGG-based enrichment analysis revealed that all DEGs compared to LFC were enriched into 140 KEGG pathways. These pathways included flavonoid biosynthesis, phenylpropanoid biosynthesis, and anthocyanin biosynthesis, all of which were significantly enriched (Figure 3D). These terms have previously been associated with anthocyanin production.
Analysis of Anthocyanidin Biosynthetic Pathway Genes and Anthocyanin Derivates in JFC, BSH, and LFC
According to transcriptome and metabolites studies, anthocyanin biosynthesis may be responsible for the distinct red leaf colorations of Toona. Therefore, it is necessary to investigate the mechanisms that underlie this process. Genes encoding anthocyanin biosynthetic enzymes were found to be associated with different compositions of the compounds in the anthocyanin biosynthesis pathways. We constructed a pathway diagram displaying the expression heat map of structural genes and compounds for red, light-red, and green leaves (Figure 4). The PAL gene catalyzed the conversion of L-Phenylalanine to cinnamic acid and was highly expressed in JFC. Among the 4CL genes, two exhibited a gradual decrease in expression from red to green leaves across JFC, BSH, and LFC. CHS, the first key enzyme in the flavonoid pathway, played a pivotal role in catalyzing the synthesis of chalcone. Three CHS genes were differentially expressed in these clones. Two of these genes were significantly up-regulated in BSH and JFC, compared with LFC, whereas one was more highly expressed in LFC than in BSH and JFC. Two CHI genes responsible for converting naringenin chalcone to naringenin exhibited higher expression levels in BSH and lower expression in LFC. The accumulation of anthocyanin was positively correlated with the expression level of these genes. The study found that there was a gradual decrease in the content of naringenin metabolite in JFC, BSH, and LFC. Moreover, the study found that the FLS, DFR, and ANS genes play a crucial role in transforming naringenin into delphinidin, cyanidin, and pelargonidin, respectively, with their expression levels being up-regulated in JFC and down-regulated in LFC (Figure 4). UFGT is a critical enzyme involved in the biosynthesis of stable anthocyanins from unstable anthocyanin glycosides. The study identified two UFGT genes whose expression levels were significantly higher in JFC than in BSH, with the lowest levels observed in LFC, resulting in the highest accumulation of different anthocyanin compounds in JFC compared to LFC. These results revealed that the expression of genes in anthocyanin biosynthesis was consistent with accumulation of anthocyanins.
Correlation Analysis of Differentially Expressed Transcripts and Anthocyanin Compounds to Identify Key TFs and Plant Hormones Genes
To achieve the regulatory network between regulatory genes, including transcription factors and plant hormone-related genes, and anthocyanin metabolites, an association analysis was conducted between transcripts and metabolome (Supplementary Figure S2). The major transcription factors and plant hormone-related genes were then identified using differentially expressed genes and metabolites that showed significant differences between JFC (red) and LFC (green). A conjoint analysis was conducted to establish relationships between 510 genes and 23 metabolites with a positive correlation coefficient greater than 0.8 within JFC and LFC. Among these genes and metabolites, 23 TFs and
Correlation Analysis of Differentially Expressed Transcripts and Anthocyanin Compounds to Identify Key TFs and Plant Hormones Genes
To achieve the regulatory network between regulatory genes, including transcription factors and plant hormone-related genes, and anthocyanin metabolites, an association analysis was conducted between transcripts and metabolome (Supplementary Figure S2). The major transcription factors and plant hormone-related genes were then identified using differentially expressed genes and metabolites that showed significant differences between JFC (red) and LFC (green). A conjoint analysis was conducted to establish relationships between 510 genes and 23 metabolites with a positive correlation coefficient greater than 0.8 within JFC and LFC. Among these genes and metabolites, 23 TFs and phytohormonerelated genes were identified and showed strong correlations with 13 anthocyanin derivates (Table 1). Using Cytoscape 3.10.0,a regulatory network was constructed between these 23 genes and 13 anthocyanin derivatives (Figure 5A). The top 3 TFs associated with the biosynthesis of anthocyanins belonged to AGL9 (MADS-box), SOC1 (MADS-box), MYB (CPC) (Table 1). Furthermore, among the top five compounds, Cyanidin-3-O-arabinoside and Cyanidin-3-O-galactoside emerged as the most significantly different anthocyanin derivatives in Toona sinensis (Table 1). Consequently, a regulatory network was constructed between these top transcription factors and structural regulatory genes involved in anthocyanin biosynthesis (Figure 5B). These findings suggest that MADS-box, MYB, and bHLH transcription factors may regulate DFR, ANS, UFGT, and other structural genes involved in the biosynthesis of anthocyanin in Toona sinensis by either affecting or participating in the regulation of structural gene.
Int. J. Mol. Sci.2023, 24, x FOR PEER REVIEW 9 of 22 phytohormone-related genes were identified and showed strong correlations with 13 anthocyanin derivates (Table 1). Using Cytoscape 3.10.0,a regulatory network was constructed between these 23 genes and 13 anthocyanin derivatives (Figure 5A). The top 3 TFs associated with the biosynthesis of anthocyanins belonged to AGL9 (MADS-box), SOC1 (MADS-box), MYB (CPC) (Table 1). Furthermore, among the top five compounds, Cyanidin-3-O-arabinoside and Cyanidin-3-O-galactoside emerged as the most significantly different anthocyanin derivatives in Toona sinensis (Table 1). Consequently, a regulatory network was constructed between these top transcription factors and structural regulatory genes involved in anthocyanin biosynthesis (Figure 5B). These findings suggest that MADS-box, MYB, and bHLH transcription factors may regulate DFR, ANS, UFGT, and other structural genes involved in the biosynthesis of anthocyanin in Toona sinensis by either affecting or participating in the regulation of structural gene.
Expression Profiles of TFs and Plant Hormone Genes Related to Anthocyanin Biosynthesis
Previous studies have demonstrated that the biosynthesis of anthocyanins requires the involvement of transcription factors, such as MADS-box, MYB, bHLH, and bZIP. In this study, we identified a total of 11 MADS-box genes, 33 MYBs, 42 bHLHs, 36 bZIPs, and 15 WD40 genes in Toona. The RNA expression levels in JFC, BSH, and LFC was compared, which revealed that 12 bHLH, 10 bZIP, 12 MYB, 2 MADS-box, and 1 WD40 were significantly highly expressed in JFC, lowest in LFC (Figure 6A). Furthermore, the heatmap showed that all highly correlated TFs with anthocyanin derivates belonged to these groups.
Expression Profiles of TFs and Plant Hormone Genes Related to Anthocyanin Biosynthesis
Previous studies have demonstrated that the biosynthesis of anthocyanins requires the involvement of transcription factors, such as MADS-box, MYB, bHLH, and bZIP. In this study, we identified a total of 11 MADS-box genes, 33 MYBs, 42 bHLHs, 36 bZIPs, and 15 WD40 genes in Toona. The RNA expression levels in JFC, BSH, and LFC was compared, which revealed that 12 bHLH, 10 bZIP, 12 MYB, 2 MADS-box, and 1 WD40 were significantly highly expressed in JFC, lowest in LFC (Figure 6A). Furthermore, the heatmap showed that all highly correlated TFs with anthocyanin derivates belonged to these groups. Studies suggest that plant hormones have a crucial function in regulating anthocyanin concentrations in leaves and fruits via direct or indirect regulation of transcription factors responsible for anthocyanin biosynthesis. The hormones capable of activating or inhibiting the structural genes involved in the biosynthesis pathway include auxin, gibberellins, cytokinin, brassinosteroids, jasmonic acid, salicylic acid, and abscisic acid. Furthermore, we identified several differentially expressed genes that participate in plant Studies suggest that plant hormones have a crucial function in regulating anthocyanin concentrations in leaves and fruits via direct or indirect regulation of transcription factors responsible for anthocyanin biosynthesis. The hormones capable of activating or inhibiting the structural genes involved in the biosynthesis pathway include auxin, gibberellins, cytokinin, brassinosteroids, jasmonic acid, salicylic acid, and abscisic acid. Furthermore, we identified several differentially expressed genes that participate in plant hormone signal transduction, including 8 Auxin-related genes, 9 CTK-related genes, 10 GA-related genes, 23 BR-related genes, 7 ABA-related genes, 11 ethlye-responsive genes, 6 JA-related genes, and 3SA-related genes. Among these genes, the following exhibited a positive correlation coefficient greater than 0.8 with anthocyanin compounds: 2 Auxinrelated genes (Maker00008503 and Maker00022256), 1 CTK-related gene (Maker00000350), 1 GA-related gene (Maker000117930), 2 BR-related genes (Maker00026791 and novel.6183), 1 Ethylene-responsive genes (Marker00033233), and 2 SA-related genes (Maker00020104 and Maker00032838) (Figure 6B). This indicates that these genes may positively regulate anthocyanin biosynthesis either directly or indirectly.
Expression Analysis of Genes through qRT-PCR
To validate the RNA-seq results, we utilized qRT-PCR to evaluate the expression of 8 structural genes and 4 TFs involved in the anthocyanin biosynthesis pathway. Our observations revealed that bHLH162, CPC, and SOC1 TFs were significantly upregulated in JFC and exhibited the lowest expression levels in LFC (Figure 7A). Additionally, there was a strong correlation between the RNA-seq and the characteristic leaf color of these samples. The qRT-PCR outcomes of structural genes were also significantly correlated with RNA-seq (Figure 7B). Moreover, our data suggested that the expression levels of bHLH162 (Maker00021798, 2.99 times higher in JFC than in LFC), CPC (Maker00014746, 10.48 times), and SOC1 (Maker00029632, 6.67 times) were synchronized with DFR (Marker00011618, 1.84 times), ANS (Marker00000819, 2.95 times), and UFGT-1 (Marker00029333, 3.07 times), indicating that they may be involved in regulating anthocyanin production. These findings demonstrated that transcription data accurately reflect gene expression in the anthocyanin biosynthesis of Toona sinesis.
Expression Analysis of Genes through qRT-PCR
To validate the RNA-seq results, we utilized qRT-PCR to evaluate the expression of 8 structural genes and 4 TFs involved in the anthocyanin biosynthesis pathway. Our observations revealed that bHLH162, CPC, and SOC1 TFs were significantly upregulated in JFC and exhibited the lowest expression levels in LFC (Figure 7A). Additionally, there was a strong correlation between the RNA-seq and the characteristic leaf color of these samples. The qRT-PCR outcomes of structural genes were also significantly correlated with RNA-seq (Figure 7B). Moreover, our data suggested that the expression levels of bHLH162 (Maker00021798, 2.99 times higher in JFC than in LFC), CPC (Maker00014746, 10.48 times), and SOC1 (Maker00029632, 6.67 times) were synchronized with DFR (Marker00011618, 1.84 times), ANS (Marker00000819, 2.95 times), and UFGT-1 (Marker00029333, 3.07 times), indicating that they may be involved in regulating anthocyanin production. These findings demonstrated that transcription data accurately reflect gene expression in the anthocyanin biosynthesis of Toona sinesis.
Discussion
Anthocyanins present in Toona sinensis leaves possess various health benefits, including antioxidant and anti-inflammatory properties, and potential cancer-preventing properties. Understanding the regulatory mechanisms involved in anthocyanin biosynthesis in Toona sinensis leaves has significant implications for human health and agriculture. In this study, we utilized metabolomic and transcriptomic approaches to elucidate the molecular mechanisms and regulatory network involved in anthocyanin biosynthesis in Toona. A total of 32-44 metabolites related to anthocyanins were identified in five different clones of Toona leaves using UPLC-MS/MS. We found that JFC had the highest total anthocyanin content, with Delphinidin, Cyanidin, Peonidin being the most abundant pigments in JFC. Specifically, Cyanidin-3-O-arabinoside was found to be 520.93 times more prevalent in JFC than in LFC, indicating that cyanidin compounds are the dominant anthocyanin in JFC. These findings are consistent with those of previous studies on strawberry petals, which have shown that cyanidins are the primary anthocyanin compound present [37].
Transcriptome sequencing of Toona leaves with different colors can provide valuable insights into the regulating genes involved in anthocyanin biosynthesis. The KEGG pathway enrichment analysis of differentially expressed genes (DEGs) in colored leaves of different clones compared to green leaves revealed significant enrichment of genes involved in flavonoid biosynthesis, phenylpropanoid biosynthesis, and the anthocyanin biosynthesis pathway. This finding supports previous studies demonstrating the involvement of multiple enzymes encoded by early biosynthesis genes and anthocyanin biosynthesis genes in anthocyanin biosynthesis. The study indicates that the biosynthesis of naringenin is a crucial step in the flavonoids metabolic pathway and plays a decisive role in the synthesis of anthocyanin compounds. The PAL, C4H, 4CL, CHS, and CHI are key genes in the flavonoids metabolic pathway and play a decisive role in determining naringenin production. The study found that JFC had a higher naringenin content than BSH and LFC, but only two 4CL genes (Maker00007314, novel.5717)were expressed at high levels in JFC and low levels in LFC. The expression of 4CL genes was highly positively correlated with naringenin content, indicating that the number and function of genes in the naringenin pathway may affect naringenin content. During the biosynthesis of anthocyanins, DFR catalyzes the reduction reaction of flavonoid-3 ,5 -hydroxylase to form anthocyanins, which was confirmed as a key step for regulating anthocyanin types. ANS converts anthocyanin precursors to anthocyanin glycosides. UFGT can catalyze the conversion of unstable anthocyanidins into stable anthocyanins. The absence of these genes directly affects anthocyanins biosynthesis, leading to pigment loss [38,39]. Down-regulation of the apple DFR gene can inhibit both cyanidin and procyanidin accumulation. In Vitis vinifera, the silencing of the DFR gene leads to the absence of anthocyanins [40]. In Duchesnea indica, a plant belonging to the Rosaceae family, decreased expression of the ANS gene results in white-colored fruits [41]. Conversely, overexpressing the SmANS gene in Salvia miltiorrhiza can increase anthocyanin content but decrease the biosynthesis of salvianolic acid [42]. In our study, we found significantly increased expression of the DFR gene (Marker00011618), ANS gene (Marker00000819), UFGT (Marker 00029333 and novel.8170) in JFC and BSH compared to LFC, which may explain the red color leaves in JFC and the green color leaves in LFC. Difference in gene sequence and promoter regions of these structural genes may be the primary factors contributing to difference in gene function and expression, ultimately impacting anthocyanin biosynthesis in different clones.
The MBW protein complex was proven to activate the expression of structural genes involved in anthocyanin production. In addition to the MBW complex, other TFs such as bZIP [32,43], NAC [44], WRKY [45,46], MADS-box [47,48], and zinc finger [49,50] proteins have also shown to regulate anthocyanin biosynthesis in plants. In Toona, we found that MADS-box, MYB, and C2H2 possessed top five TFs related to anthocyanin derivates according to the joint analysis between transcriptome and metabolome. The gene-gene regulatory network also showed that the SOC1, CPC, and bHLH162 were the major TFs regulating the anthocyanin biosynthesis structural genes. SOC1 encodes a MADS box transcription factor and is involved in the regulation of flowering in response to temperature or light in various plants species, such as Arabidopsis [51], Oryza sativa L. [52], Gossypium hissytum [53]. In Toona sinensis, the young leaf color is highly sensitive to temperature, indicating a critical role of SOC1 in temperature or light-induced anthocyanin biosynthesis. These transcription factors might regulate anthocyanin biosynthesis process by directly active or repress gene expression by binding to the promoter regions, or indirectly though protein-protein interactions in model plants and other fruits.
Plant hormones can affect anthocyanin biosynthesis through various mechanisms. Previous studies have reported that overexpression of MdIAA26 in apple calli and Arabidopsis promotes the accumulation of anthocyanin, while auxin inhibit it by degrading the MdIAA26 protein [54]. Exogenous ethylene treatment increased anthocyanin accumulation in grape skins and induced the expression of structural genes (VvPAL, Vv4CH, VvCHS, VvCHI, VvF3H, and VvUFGT) and regulatory genes (VvMYBA1, VvMYBA2, and VvMYBA3) related to anthocyanin biosynthesis [55]. Our study identified nine hormone-related genes with a high correlation to regulate anthocyanins biosynthesis, which are highly expressed in JFC and play a significant role in the controlling leaf color variation in Toona. Leaf color is an important trait that is susceptible to both endogenous and exogenous influences. Variations in hormone signal transduction gene expression reflect the essential mechanisms by which plant hormones affect the anthocyanin biosynthesis pathway. These findings enhance our understanding of how anthocyanin biosynthesis works in Toona and how it can be controlled, with implications for breeding new Toona cultivars. Identifying and understanding the beneficial metabolic constituents and regulatory networks of Toona can facilitate the development and utilization of anthocyanin-enriched varieties.
Plant Materials
The study selected five Chinese toon clones, JinFuChun (JFC), BaShanHong (BSH), LvFuChun (LFC), FenHongXinNiang (FHXN), and HeBeiGu (HBG), which cultivated in Research Institution of Subtropical Forestry, Chinese Academy of Forestry. The colors of these varieties are: brilliant-red buds (JFC), pink buds (FHXN), light-red with a little green buds (BSH), dark-purple buds (HBG), and green buds (LFC). Fresh buds from five different clones were collected, with three replicates taken for each clone, each replicate sourced from different trees. Buds measuring 10-15 cm in length and weighing approximately 5-10 g were collected for each sample and subsequently underwent freeze-drying for RNA-seq and metabolites measures.
Sample Preparation and Extraction
The sample underwent freeze-drying, followed by grinding into a powder (30 Hz, 1.5 min), and subsequently, it was stored at −80 • C until required. Subsequently, 50 mg of the powder was precisely weighed and subjected to extraction using 0.5 mL of methanol/ water/hydrochloric acid (500:500:1, v/v/v). The resulting extract was vigorously vortexed for 5 min, followed by ultrasonication for another 5 min, and then centrifuged at 12,000× g at 4 • C for 3 min. The residue underwent a repeat extraction using the same procedure and conditions. The supernatants obtained were collected and filtered through a 0.22 µm membrane filter (Anpel) prior to LC-MS/MS analysis.
ESI-MS/MS Conditions
Linear ion trap (LIT) and triple quadrupole (QQQ) scans were obtained using a triple quadrupole-linear ion trap mass spectrometer, the AB Scoex Qtrap ® 6500+ LC-MS/MS System (Sciex, Framingham, MA, USA), equipped with an ESI Turbo Ion-Spray interface. The instrument operated in positive ion mode and was controlled by Analyst 1.6.3software (Sciex). The ESI source operated with the following parameters: Ion Source: ESI+, Source Temperature: 550 • C, Ion Spray Voltage (IS): 5500 V, Curtain Gas (CUR): Set at 35 psi. Anthocyanins were subjected to analysis using scheduled multiple reaction monitoring (MRM). Data acquisition was conducted through Analyst 1.6.3software (Sciex). Quantification of all metabolites was performed using Multiquant 3.0.3software (Sciex). Further optimization of declustering potentials (DP) and collision energies (CE) for individual MRM transitions was carried out. A specific set of MRM transitions was monitored for each period, corresponding to the elution of metabolites during that timeframe.
Identification of Differentially Accumulated Metabolites
Significantly regulated metabolites between groups using the identification criterion of the absolute |log 2 (fold-change)| ≥ 1 and p value < 0.05, based on the Student's t-test. Identified metabolites using the Kyoto Encyclopedia of Gene and Genomes (KEGG) compound database available at [URL] on 20 June 2022. Next, we mapped the annotated metabolites to KEGG Pathway database found at [URL] on 20 June 2022. The pathways that significantly regulated metabolites mapped to were integrated into MSEA (metabolite sets enrichment analysis) and evaluated for significance by checking the hypergeometric test's p-values.
RNA Sequencing
Total RNA was extracted from frozen leaves utilizing the RNAprep Pure Plant Kit (Tiangen Biotech, Beijing, China). The Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) was used to evalute the quality of the gathered RNAs by examining their integrity. Subsequently, Poly (A) mRNA was enriched from total RNA using Oligo (dT) magnetic beads. To facilitate sequencing, Poly (A) mRNA was fragmented by an RNA fragmentation kit (Ambion, Austin, TX, USA). First-strand cDNA was produced via transcriptase reaction with random hexamer primers. Next, DNA polymerase I and RNase H enzymes were employed to create the second-strand cDNA (Invitrogen, Carlsbad, CA, USA). Following this, DNA fragments of suitable lengths were obtained, end-repaired, poly(A)-tailed, and connected with sequencing adaptors. Eventually, these fragments underwent Illumina HiSeq™ 2500 platform sequencing.
Transcript Profiles and Annotation
High-quality reads were obtained by processing the raw reads in fastq format using Perl scripts developed in-house. Clean reads were obtained from the raw data by eliminating adaptor sequences, low-quality reads, and reads containing polyN. All downstream analyses were based on clean, highquality data. Gene function was annotated employing several databases: KEGG pathway database, the NCBI non-redundant (Nr) database, the Swiss-Prot protein database, the euKaryotic Clusters of Orthologous Groups (KOG) database, the Gene Ontology (GO) database, and the Pfam database. We analyzed the differentially expressed genes of both groups by utilizing the DESeq R package (version 1.10.1). The DESeq R package utilizes a statistical model based on the negative binomial distribution to identify differentially expressed genes. We corrected the outcomes of all statistical tests by employing the false discovery rate of Benjamini and Hochberg to account for multiple testing. According to DESeq, genes were considered substantially differentially expressed if their adjusted p-value was less than 0.05. We used the top GO R package v.1, which is based on the Kolmogorov-Smirnov test, to carry out GO enrichment analysis of the differentially expressed genes. We performed pathway analysis utilizing the KEGG database ( [URL] on 10 June 2022) to investigate relevant pathways of substantially differentially expressed genes [56][57][58].
Correlation Analysis of Transcriptome and Metabolome
To integrate transcriptome and metabolome datasets, Pearson correlation coefficients were utilized. Gene-metabolite coefficients were obtained by calculating the average expression levels of transcripts and metabolite contents. The fold changes for both transcriptome and metabolome data in each group were also computed. A correlation was considered significant if the Pearson correlation coefficients exceed 0.8 and p-values were less than 0.05 (Table 1). Significant positive correlations between transcription factors (TFs) and anthocyanin derivatives in groups JFC and LFC were detected and visually presented utilizing Cytoscape 3.10.0.
Verification of RNA-Seq Data by qRT-PCR
Total RNA was extracted from Toona sinesis leaves and reverse-transcribed using the Quantscript Reverse Transcriptase Kit. Each clone had three biological replicates, and each sample had three technical replicates. The obtained cDNA served as a template for determining gene expression levels, employing specific primers for genes linked to anthocyanin biosynthesis as well as ACTIN gene (used as an internal control). The primers used in the qRT-PCR analysis were shown in Table 2. The melt curve of ACTIN in the qPCR products was presented in Supplementary Figure S3. The reaction system contained 5 µL 2 × Q3 SYBR qPCR Master Mix-Universal (TOLOBIO), 1 µL cDNA template, 0.5 µL of each forward and reverse primer, and 3 µL of RNase-free water.qRT-PCR was performed using Applied Biosystems 7500 Fast Real-Time PCR System.
Conclusions
In this study, an integrative analysis of the transcriptome and metabolome of five distinct Toona sinensis clones were performed to explore the related genes and metabolites associated with the anthocyanin biosynthesis. Furthermore, our analyses found that the red leaves 'JFC' contained the highest content of stable cyanidin, delphinidin, and peonidin, especially Cyanidin-3-O-arabinoside. The integrated analysis also identified the major transcription factor SOC1, CPC, and bHLH162. Moreover, the regulatory network construction of TFs, metabolic, structural genes reveals the underlying mechanisms of the anthocyanin biosynthesis pathway in Toona sinensis. Overall, this study provides a foundation for future research aimed at manipulating anthocyanin biosynthesis to improve plant coloration or to derive human health benefits.
Figure 1 .
Figure 1. Morphological observation and anthocyanin content of leaves from different Toona clones (JFC, FHXN, BSH, HBG, and LYC).(A) The morphological characteristics of leaves from the five Toona clones.(B) The color of anthocyanin present in the leaves of these clones.(C) The quantification of anthocyanin content in each of the Toona clones. Data are shown as mean ± standard deviation, n = 3. Bars with different letters are significantly different at p < 0.05.
Figure 1 .
Figure 1. Morphological observation and anthocyanin content of leaves from different Toona clones (JFC, FHXN, BSH, HBG, and LYC).(A) The morphological characteristics of leaves from the five Toona clones.(B) The color of anthocyanin present in the leaves of these clones.(C) The quantification of anthocyanin content in each of the Toona clones. Data are shown as mean ± standard deviation, n = 3. Bars with different letters are significantly different at p < 0.05.
Figure 2 .
Figure 2. Targeted metabolome profiles of the anthocyanin biosynthesis pathway.(A) A PCA score plot demonstrates variations among different colored Toona clones.(B) The number of up-and down-regulated metabolites varies across different comparison groups.(C,D) Venn diagram illustrate the number of differential metabolites when compared with JFC and LFC.(E) Heatmaps reveals the differential metabolites present in diverse Toona clones. Three independent replicates of each clone are displayed in the heatmap.
Figure 2 .
Figure 2. Targeted metabolome profiles of the anthocyanin biosynthesis pathway.(A) A PCA score plot demonstrates variations among different colored Toona clones.(B) The number of up-and downregulated metabolites varies across different comparison groups.(C,D) Venn diagram illustrate the number of differential metabolites when compared with JFC and LFC.(E) Heatmaps reveals the differential metabolites present in diverse Toona clones. Three independent replicates of each clone are displayed in the heatmap.
2 Figure 3 .
Figure 3. Transcriptome analysis of leaves among Toona clones.(A) A PCA score plot of the RNA seq results of various colored Toona.(B) The number of up-and down-regulated genes differs acros different comparison groups.(C) A shared number of DEGs across varying comparison groups.(D A KEGG pathway enrichment bubble plot for DEGs between different comparison groups.
Figure 3 .
Figure 3. Transcriptome analysis of leaves among Toona clones.(A) A PCA score plot of the RNA-seq results of various colored Toona.(B) The number of up-and down-regulated genes differs across different comparison groups.(C) A shared number of DEGs across varying comparison groups.(D) A KEGG pathway enrichment bubble plot for DEGs between different comparison groups.
Figure 4 .
Figure 4. Biosynthetic pathway of anthocyanins. The construction of this pathway is based on the KEGG pathway and pertinent literature references. The red box indicated the expression of genes, and the blue box indicated the content of compounds in this pathway.
Figure 4 .
Figure 4. Biosynthetic pathway of anthocyanins. The construction of this pathway is based on the KEGG pathway and pertinent literature references. The red box indicated the expression of genes, and the blue box indicated the content of compounds in this pathway.
Figure 5 .
Figure 5. Regulatory network related to the anthocyanin biosynthesis pathway.(A) Connection network between regulatory genes and metabolites involved in anthocyanin biosynthesis. Purple circles represent transcription factors, light-purple triangle represent plant hormones related genes, green round rectangle represent the metabolites. The size of the cells represented the log2FC (gene or compounds).(B) Regulatory network between transcription factors and structural genes responsible for anthocyanin biosynthesis. Purple circles represent transcription factors, light-purple circles represent structural genes involved in the biosynthesis of anthocyanin pathway.
Figure 5 .
Figure 5. Regulatory network related to the anthocyanin biosynthesis pathway.(A) Connection network between regulatory genes and metabolites involved in anthocyanin biosynthesis. Purple circles represent transcription factors, light-purple triangle represent plant hormones related genes, green round rectangle represent the metabolites. The size of the cells represented the log2FC (gene or compounds).(B) Regulatory network between transcription factors and structural genes responsible for anthocyanin biosynthesis. Purple circles represent transcription factors, light-purple circles represent structural genes involved in the biosynthesis of anthocyanin pathway.
Figure 7 .
Figure 7. Quantitative real-time RT-PCR (qRT-PCR) and RNA-seq analysis of genes involved in anthocyanin biosynthesis pathway and putative transcription regulators.(A) Relative expression level of transcription factors.(B) Relative expression level of structural genes. Data are shown as mean ± standard deviation of three biological replicates.
Figure 7 .
Figure 7. Quantitative real-time RT-PCR (qRT-PCR) and RNA-seq analysis of genes involved in anthocyanin biosynthesis pathway and putative transcription regulators.(A) Relative expression level of transcription factors.(B) Relative expression level of structural genes. Data are shown as mean ± standard deviation of three biological replicates.
Table 1 .
Correlation analysis of transcription factors and plant hormone-related genes with anthocyanin-related metabolites.
Table 2 .
The primer sequences used in this study.
==
Domain: Biology Environmental Science
|
[{"added": "2023-10-26T15:30:51.328Z", "created": "2023-10-01T00:00:00.000", "id": "264473149", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1422-0067/24/20/15459/pdf?version=1698055459", "pdf_hash": "dc1be98babf39c85d549ff479ade954ed05a047f", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:441", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "0ec7c9be07c752664798620ec00cc25dc34ea26e", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Caste-biased patterns of brain investment in the subterranean termite Reticulitermes flavipes
Summary Investment into neural tissue is expected to reflect the specific sensory and behavioral capabilities of a particular organism. Termites are eusocial insects that exhibit a caste system in which individuals can develop into one of several morphologically and behaviorally distinct castes. However, it is unclear to what extent these differences between castes are reflected in the anatomy of the brain. To address this question, we used deformation-based morphometry to conduct pairwise comparisons between the brains of different castes in the eastern subterranean termite, Reticulitermes flavipes. Workers exhibited enlargement in the antennal lobes and mushroom bodies, while reproductives showed increased investment into the optic lobes and central body. In addition, caste-specific enlargement was observed in regions that could not be mapped to distinct neuropils, most notably in soldiers. These findings demonstrate a significant influence of caste development on brain anatomy in termites alongside convergence with eusocial hymenopteran systems.
INTRODUCTION
The brain is the center of all behavioral and sensory processes and has remained a popular subject of research for centuries.2][3] Due in part to these costs, patterns of brain investment differ significantly across the animal kingdom, coinciding with marked differences in sensory and cognitive abilities. A relatively simple example of this phenomenon can be observed in the development of visual versus olfactory sensory systems in day-and night-active insect species. In insects, the primary centers of visual and olfactory processing are the optic lobes and antennal lobes, respectively. 4,57][8][9] Similar examples of differential investment in these two regions can be observed across Drosophila species, 10 between the paper wasp Polistes dominula and its obligate social parasite Polistes sulcifer, 11 and between the ant Formica fusca and its obligate social parasite Polyergus mexicanus. 12These patterns of investment suggest a tradeoff between development of particular regions of the brain, consistent with the idea of neural tissue as energetically costly.
Even within the same species, it is possible to observe significant differences in brain anatomy between individuals. This is especially true in the social insects, which exhibit a division of labor manifested as a caste system. Castes can be broadly categorized as either reproductive or non-reproductive and generally show clear differences in morphology and behavior among one another. 13These differences extend to the structure of the brain, although to date the majority of research in this area has focused on the social Hymenoptera.5][16][17][18][19] This trend is reversed in sweat bees and paper wasps, potentially due to the demands associated with maintaining reproductive dominance in these groups. 20,21Comparisons of workers to soldiers across several ant species have shown expansion of the mushroom bodies in workers, along with expansion in the antennal lobes, which process chemosensory cues. 17,22,23Significant changes in brain anatomy are also well-characterized within castes, most notably in relation to the nurse-forager transition exhibited by honeybee workers. 24Foragers, even those that develop precociously, always possess larger mushroom bodies than nurse bees of any age. 25Complex differences in brain allometry have also been described in the leafcutter ant Atta cephalotes, which exhibits several worker subcastes: the smallest workers remain in the nest and tend to fungus gardens, intermediate-sized workers harvest leaves outside the nest, and the largest workers fill a defensive role. 26,27Examined together, these and other results indicate that the processes of caste and subcaste development are associated with significant changes in brain structure. Given the role of the brain, these changes are expected to accommodate the specialized task repertoires and sensory demands of different castes and subcastes. 28ermites represent another major group of social insects and exhibit complex caste development pathways. In addition to workers and reproductives, termites possess a morphologically distinct soldier caste that is responsible for colony defense. In the lower termites, soldiers differentiate directly from workers rather than following a separate developmental trajectory, as in ants. 29The process of worker-to-soldier differentiation requires only two molts but results in a significant change in the appearance and behavior of the individual, with soldiers
OPEN ACCESS
showing high aggression but an inability to perform any of the tasks that workers are responsible for. 30While rare under natural conditions, it is also possible for workers to develop into ergatoids, which function as supplementary reproductives. The sense of vision is of particular note in termites, as workers, soldiers, and ergatoids in all but the most basal termite lineages lack external eyes. 31However, alates, which are responsible for dispersing and founding new colonies in which they act as the primary reproductives following insemination, possess functional eyes and wings to aid in their dispersal. Eye and wing development is progressively observed beginning from the nymph stage, which follows a separate trajectory from worker development. 29uch extreme differences in sensory abilities and behavior are likely to be reflected in the brain, although to date only a handful of studies have examined this aspect of termite biology. In the dampwood termite Hodotermopsis sjostedti, soldiers show expansion in the mandibular motor neurons relative to workers, as well as in distinct clusters of octopaminergic and tyraminergic neurons. 32In dampwood termites of genus Zootermopsis, reproductives show expansion in the optic lobes relative to non-reproductive castes, while workers show expansion in the antennal lobes and mushroom bodies relative to soldiers and reproductives. 19,33Among non-dampwood termite species, Reticulitermes speratus reproductives show expansion in the optic lobes relative to non-reproductive castes, 34 while Procornitermes araujoi workers show expansion in the mushroom bodies relative to soldiers. 35These studies are informative, but there is still much that remains a mystery regarding the influence of termite caste development on the brain, particularly as patterns of investment may differ significantly among species with different lifestyles.
The goal of this study was to characterize differences in brain anatomy across castes in the eastern subterranean termite, Reticulitermes flavipes. Reticulitermes flavipes is a widespread, subterranean termite species that nests within the soil and exploits multiple wood sources at a time using an interconnecting system of tunnels. 36This lifestyle contrasts heavily with that displayed by previously studied Hodotermopsis and Zootermopsis species, which nest within a single piece of wood and notably do not forage. This difference in lifestyle should impose different cognitive demands on individuals, thus influencing the evolution of brain allometry both between and within species. A study of differences in neural tissue investment among castes in termite species exhibiting a variety of lifestyles can therefore lead to a greater understanding of the links between brain allometry and the traits that it influences. With this in mind, we hypothesized that, given the vast differences in behavioral repertoires and sensory requirements between R. flavipes castes, we would find differences in neural tissue investment between these castes. To test this hypothesis, we dissected and imaged brains from five different caste phenotypes (workers, soldiers, and three reproductive caste phenotypes: ergatoids, nymphs, and alates), then compared brain anatomy among caste phenotypes using deformation-based morphometry before validating our results manually by measuring and comparing the volumes of distinct brain neuropils.
Deformation-based morphometry
Significant differences in brain anatomy were observed between different pairs of termite castes, both generally throughout the brain and in four distinct neuropils: the antennal lobes (ALs), optic lobes (OLs), mushroom bodies (MBs), and central body (CB) (Figure 1; Table 1). First, a series of pairwise comparisons between the three worker-derived castes (workers, soldiers, and reproductives, represented as ergatoids) was carried out (Figure 2). The ALs and MBs showed enlargement in workers relative to both soldiers and ergatoids. The OLs showed enlargement in ergatoids relative to both workers and soldiers. Finally, enlargement at or close to the central complex (CX) was identified in soldiers and ergatoids relative to workers. Although other regions showing caste-specific enlargement were identified, these regions could not be mapped to distinct neuropils. Next, a series of pairwise comparisons between workers, nymphs, and alates was carried out (Figure 3). The ALs showed enlargement in both workers and nymphs relative to alates, while the MBs showed enlargement in workers relative to both nymphs and alates. The OLs showed enlargement in nymphs and alates relative to workers and in alates relative to nymphs. Enlargement at or close to the CX was identified in nymphs and alates relative to workers. As in the comparisons of the worker-derived castes, other regions showing phenotype-specific enlargement were identified but could not be mapped to distinct neuropils.
Finally, the presence of sexual dimorphism within castes was tested in all five caste phenotypes by comparing male and female brains (Figure 4). Sex had no effect on brain anatomy within caste phenotypes.
Workers showed expansion in the antennal lobes and mushroom bodies
Our analysis revealed diverse patterns of caste-specific brain enlargement between R. flavipes termite caste phenotypes. Workers comprise the majority of a termite colony and consistently showed enlargement in two key brain regions, the ALs and the MBs. The ALs are the primary olfactory centers of the insect brain and receive input directly from olfactory sensory neurons located along the antennae, 5 while the MBs act as centers for multisensory integration, learning, and memory. 37Both of these regions commonly appear as showing caste-specific enlargement in eusocial insects, most often in the worker caste relative to reproductives and/or defensive castes or subcastes. Species in which this is the case include the honeybee Apis mellifera, [14][15][16] the turtle ant Cephalotes varians, 17 the carpenter ant Camponotus ocreatus, 18 the leafcutter ant Atta cephalotes, 26 those of the army ant genus Eciton, 22 those of the ant genus Pheidole, 23 and those of the termite genera Zootermopsis and Procornitermes. 19,35In species that show this pattern of investment, workers possess complex behavioral repertoires that may require stronger cognitive processing capabilities relative to other castes. When the inverse is true, as in the sweat bee Augochlorella aurata, 20 reproductives may face unique cognitive challenges not encountered by the worker caste.
In the case of R. flavipes, workers are responsible for an array of tasks, which includes foraging, nest construction and repair, and brood care. By contrast, the repertoires of the remaining castes are highly restricted. Soldiers are responsible for defending the colony from attack, while alates and ergatoids are responsible for reproduction. 29Considering the reduced number and complexity of tasks that non-worker castes are responsible for, a low demand for investment into neural tissue might be expected. The role of nymphs in the colony is less clear. In Coptotermes formosanus, nymphs are capable of feeding themselves and caring for soldiers when workers are absent, 38 although it is unknown if nymphs contribute labor under field conditions. Nymphs are also occasionally found among foraging populations, although it has been suggested that they are feeding themselves in this case rather than aiding in foraging. 39Nevertheless, nymphs at least appear to be able to perform some tasks associated with the worker repertoire. That AL volume was not found to significantly differ between workers and nymphs may reflect an increased dependency on chemical communication in these two phenotypes relative to soldiers and sexually mature reproductives. Correlations between task complexity and MB volume have been documented within the eusocial Hymenoptera, particularly in the case of the age polyethism exhibited by honeybee workers. Young workers remain within the hive and act as nurses, while older workers regularly exit the hive and act as foragers. 40In the case of honeybees, foraging is a complex task that requires strong navigational ability and the ability to recognize and remember the locations of nectar resources. Foragers show expansion in MB volume relative to nurses at all ages, even when foraging is induced precociously. 25,41A similar trend is observed in the leafcutter ant Atta cephalotes, in which medium-sized workers that exhibit a wide behavioral repertoire possess larger MBs than those specialized toward certain tasks, such as fungus garden tending or nest defense. 26Likewise, expansion of the ALs, as is observed in Eciton army ant and Zootermopsis termite workers, 19,22 suggests increased sensitivity to chemosensory cues, including the pheromones that constitute a major portion of termite communication. Interestingly, only the MBs were found to be enlarged in workers relative to soldiers in the higher termite Procornitermes araujoi, 35 although the reason why is unclear. Overall, expansion of these regions in eusocial insects appears to correlate with an increase in task complexity or task repertoire size, or both.
Based on the results of our validation, the ALs of workers were approximately 8% and 16% larger than those of nymphs and alates, respectively, when whole brain volume was taken into account (Figure 5), while their MBs were approximately 78% and 67% larger (Figure 6). Worker ALs were approximately 13% and 15% larger than those of soldiers and ergatoids, respectively (Figure 5), while their MBs were approximately 10% and 3% larger (Figure 6). Given that soldiers and ergatoids differentiate directly from workers in R. flavipes, the possible magnitude of change in a particular brain region may be physiologically constrained, despite any theoretical differences in cognitive function between castes. Worker and nymph/alate development separate into different trajectories at the third instar, 29 providing a much earlier origin point and greater length of time for specializations in regional brain volume to develop.
Figure 7. Comparison of optic lobe (OL) volume among R flavipes caste phenotypes
Volume of each OL is expressed as a percent of whole brain volume to account for individual variation in brain size. OL volume was measured in 6 individuals per phenotype; left and right OLs were measured separately to generate 12 measurements per phenotype. Individuals of all phenotypes except alates were pooled from a total of five colonies and randomly selected for validation, while all alates were collected from one colony. Bars represent mean +SD, while points represent individual OLs. Letters denote significance groups as determined using a one-way ANOVA followed by Tukey's post-hoc test, where p < 0.05 was used as the threshold for significance. Individuals of all phenotypes except alates were pooled from a total of five colonies and randomly selected for validation, while all alates were collected from one colony. Bars represent mean +SD, while points represent individual MBs. Letters denote significance groups as determined using a one-way ANOVA followed by Tukey's post-hoc test, where p < 0.05 was used as the threshold for significance.
Reproductives showed expansion in the optic lobes and central body
We examined three phenotypes of the reproductive caste, two of which-nymphs and alates-follow a separate developmental trajectory from workers, while one-ergatoids-differentiates directly from workers. Nymphs and alates showed significant expansion in the OLs, which are involved in visual processing, 4 relative to all worker-derived castes. The largest difference was observed between alates and workers, in which an approximately 22-fold difference in OL volume was recorded (Figure 7). Our DBM analysis identified expansion in ergatoid OLs relative to those of workers and soldiers, although the results of our validation did not confirm this significance. However, the OLs of ergatoids were approximately 67% and 20% larger than those of workers and soldiers, respectively (Figure 7).
Alates are the only termite caste phenotype that possesses functional eyes and wings. In R. flavipes, mature colonies produce alates throughout winter, which disperse in spring to search for mates and find new colonies. 42Shortly after pairing, the alates lose their wings and retreat underground into total darkness. Despite the transient usage of eyes in termites, they are integral to the process of colony foundation. Therefore, it is expected that alates would invest significantly into optic sensory systems, while the remaining castes would not. The OLs of the worker-derived castes show close to zero development relative to those of nymphs and alates (Figure 9). Interestingly, the OLs of ergatoids were, on average, larger than those of workers and soldiers. It is possible that the process of sexual maturation in R. flavipes is, to an extent, coupled with development of visual sensory systems. Although R. flavipes ergatoids do not exhibit any external eye development, ergatoids with vestigial eyes are observed in termites from the genera Nasutitermes and Mastotermes [43][44][45][46] and in the family Termitidae. 47aekawa et al. noted a lack of external eye development in ergatoids of the congeneric Reticulitermes speratus. 48It is possible that in ergatoids of this species, as well, only minor enlargement of the OLs occurs without compound eye development. Our DBM analysis also identified regions of caste-specific enlargement close to the OLs in nymphs, alates, and ergatoids. These regions may include portions of the lateral protocerebrum to which OL neurons project, as is observed in other insects, 49,50 although further investigation is necessary to validate and determine the source of this enlargement.
We also observed expansion in the CBs of nymphs and alates relative to those of the worker-derived castes, particularly workers. The CBs of nymphs and alates were approximately 31% and 44% larger than those of workers, respectively (Figure 8). The CB, along with the protocerebral bridge and paired noduli, comprises the CX. 51The CX, as a whole, is involved in a number of behavioral processes but notably functions as the center of sensory-motor integration in the insect brain. 52It is possible that expansion of the CB in alates and, by extension, nymphs is in some way related to the complex task of flight that only they are capable of. Investment into the CX in general has also been suggested to improve navigation ability in dark, subterranean environments in ants. 26Given that many extant termite genera, such as Zootermopsis, nest within a single piece of wood and thus exhibit a simple nest structure in comparison to R. flavipes and other subterranean termites, it would be of future interest to investigate whether relative CX investment differs in termite species exhibiting different levels of nest complexity.
Regional expansion in soldiers
Soldiers are significantly more aggressive than workers, and in R. flavipes the transition from worker to soldier is accomplished in just two molts. 29In addition to the loss of AL and MB volume discussed previously, expansion in other regions of the brain is likely necessary to complete the transition to the soldier behavioral syndrome. Our DBM analysis identified large regions of soldier-specific expansion close to the ALs relative to both workers and ergatoids. Although this region could not be mapped to a distinct neuropil, it overlaps with a population of neurons previously identified as showing soldier-specific expansion in the dampwood termite Hodotermopsis sjostedti. 32It was suggested by the authors of this study that this neuron population may be involved in modulating the aggressive responses of soldiers, citing research in locusts in which an analogous population of neurons was demonstrated to influence responses to stressful stimuli. 53A second possibility is that this region contains neurons that project to the subesophageal zone (SEZ). In termites, the SEZ is separate from the central brain and is enlarged in soldiers relative to workers. 54Specifically, the SEZ contains mandibular motor neurons, which control movement of the mouthparts and are likewise enlarged in soldiers. Further study is necessary to confirm the source of soldier-specific expansion, as well as its potential function.
Sexual dimorphism was not observed in any caste
In addition to comparing brain anatomy between termite castes, we tested for the possibility of sexual dimorphism in brain anatomy within castes. Unlike the eusocial Hymenoptera, which follow a haplodiploid sex determination system and exhibit colonies composed exclusively of females, termites are diploid and each caste is generally made up of both male and female individuals. 55Our DBM analysis did not identify sexual dimorphism within any of the five caste phenotypes observed in this study. This result is not unexpected, as neither R. flavipes nor termites in general are known to show sexual dimorphism in relation to behavior. There are, however, other forms of sexual dimorphism exhibited among termites that could potentially influence brain anatomy. Sexual size dimorphism is observed in the workers of many termite species, which in turn leads to sex-specific biases in the soldier caste, or to soldier castes consisting of a single sex. 56Alates also tend to show sexual size dimorphism, with females being larger than males. 57,58Overall, these differences within castes may be too minor to produce significant changes in brain anatomy, although investigation into a larger variety of termite species would be necessary to confirm this.
Limitations of the study
Here, we have provided a broad overview of the major differences in brain anatomy observed between R. flavipes castes. Structural changes observed in workers relative to the other castes show convergence with the eusocial Hymenoptera, while reproductives show clear development of the visual sensory system. The functions of other identified regions of enlargement, such as those observed in soldiers, are more speculative. However, these results provide a foundation for future work intending to establish a causal link between caste-specific enlargement of the brain and caste-specific behaviors. Comparative study between termite species is also of interest as more information in this area becomes available, as termites exhibit a wide array of lifestyles that may be associated with different cognitive demands. Relative to R. flavipes, single-piece nesting genera such as Zootermopsis do not need to forage because they live within their food source. Likewise, open-air foraging is observed within some termite species, [59][60][61] which may introduce additional navigational complexity to the task of foraging.
Notably, we were not able to include primary reproductives in our study due to a lack of availability. After colony foundation, the founding alates-now the colony's primary reproductives-undergo a number of behavioral and physiological changes to suit their new role, which may be accompanied with changes in the brain. In addition to enlargement of the gonads, 62 primary reproductives exhibit a negative phototaxis and gradual degeneration of the compound eyes, 63,64 which has shown to be associated with OL degradation and loss of visual acuity in the congeneric Reticulitermes speratus. 34verall, the differences that we have observed contribute to a larger picture of how behavior may influence neural tissue investment. Termites represent a case in which the tasks that an individual organism is generally responsible for, including foraging, reproduction, and self-defense, are decoupled, and distributed among behaviorally specialized castes. How these different patterns of investment may have evolved and whether they remain consistent across the termite clade represent questions that can be elucidated through future work.
STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following: that workers and nymphs may belong to one of several different instars, which may influence brain development, we attempted to control for the influence of age by only using individuals of approximately the same body size. 65,66rgatoids were induced separately using a model orphaning assay. 67Groups of 100 workers of approximately the same size were collected from a single colony and placed inside a 5.5-cm diameter Petri dish containing a moistened paper towel disc. Petri dishes were sealed with Parafilm, then placed in a 27 G 1 C incubator in total darkness. Beginning from 1 month following setup, Petri dishes were observed every other day for ergatoid differentiation. Ergatoids were identified by morphology and sexed, then returned to the Petri dish until an ergatoid of the opposite sex differentiated. Once an ergatoid of each sex had differentiated, one of the two ergatoids was removed from the Petri dish and placed in a smaller 2.5-cm Petri dish containing 10 nestmate workers. Once approximately 20 ergatoids of each sex had been collected, dissection of all ergatoids was carried out in a single day. Approximately 3 months passed from setup of the orphaning assay to the date of ergatoid dissection.
METHOD DETAILS Brain dissection, fixation, and imaging
Reticulitermes flavipes brains were collected and fixed using a previously described antibody staining protocol, 68 with adjustments. Termites were anesthetized on ice, then dissected in PBS. Brains were fixed in 2% paraformaldehyde (PFA) for 55 min, then washed in PBS mixed with 10% Triton X-100 (PBT) three times for 15 min each. Next, brains were incubated in PBT containing 5% normal goat serum (NGS; Thermo Fisher Scientific; Waltham, MA, USA) for 1 h, then incubated overnight at 4 C in PBT containing 5% NGS and the primary antibody, mouse anti-nc82 (Developmental Studies Hybridoma Bank; Iowa City, IA, USA), at 1:50. The following day, brains were washed in PBT three times for 15 min each, then incubated overnight in PBT containing 5% NGS and the secondary antibody, Alexa Fluor 546 goat anti-mouse IgG (Thermo Fisher Scientific), at 1:500.
The following day, brains were fixed in 4% PFA for 4 hours, washed in PBT three times for 15 min each, then mounted onto a coverslip coated in a poly-L-lysine solution. The poly-L-lysine solution was generated by mixing 25 mg of poly-L-lysine hydrobromide (Sigma-Aldrich; St. Louis, MO, USA) with 2 mL of Invitrogen Ultrapure distilled water (Thermo Fisher Scientific) until the powder had dissolved, then transferring the mixture to a 50 mL conical vial, adding another 30 mL of distilled water, and then adding 64 mL of Kodak Photo-Flo 200 Solution (Kodak; Rochester, NY, USA). After mounting them onto the coverslip, brains were dehydrated in graded ethanol (30%, 50%, 70%, 95%, 100%, 100%, 100%) for 5 min per step, then cleared in xylene (100%, 100%, 100%) for 5 min per step. Finally, the coverslip was placed on a slide treated with several drops of DPX (Thermo Fisher Scientific) and ventilated in darkness within a fume hood for at least 2 days prior to imaging.
Images of R. flavipes brains were acquired using a Leica SP8 DLS laser scanning confocal microscope at the Arts & Sciences Imaging Center at the University of Kentucky. Whole brains were imaged at 1024 3 1024 pixel resolution using a 103 dry objective (HC PL APO 103/0.40). Image stacks were generated by capturing images of brains at 1 mm intervals, which were then saved as. TIF files.
Template brain generation
Template brain generation was carried out using the registration software Computational Morphometry Toolkit (CMTK; [URL] Fiji ( [URL] stacks were rotated so that all were oriented in the same direction, and the number of images per stack was normalized to 150. Processed image stacks were then exported as NRRD files. Shape-averaged template brains were generated using CMTK's iterative_shape_averaging function. For the worker, nymph, and alate templates, individual brains of the corresponding caste were rated on the bases of symmetry and overall shape, and the 5 highest-rated brains of each sex were selected for use in template generation. A worker-alate intercaste template brain was also generated by averaging the worker and alate templates, as the difference in optic lobe size between these two castes was too large to perform a suitable comparison using the worker template.
Deformation-based morphometry
Deformation-based morphometry (DBM) was carried out using CMTK. The following pairwise comparisons were performed: worker-soldier, worker-ergatoid, soldier-ergatoid, worker-nymph, worker-alate, and nymph-alate. In addition, pairwise comparisons were performed between males and females of the worker, soldier, ergatoid, nymph, and alate caste phenotypes. A summary of the number of brains and template used for each comparison is presented in Table 1. For each pairwise comparison, nonrigid registration of each individual brain to the corresponding template was performed using CMTK's registration, affine, and reformatx functions, resulting in a Jacobian file as output. Jacobian files indicated the degree of per-voxel expansion or shrinkage exhibited in an individual brain relative to the template, expressed as a numerical value. Specifically, voxels with values < 1.0 indicated regions of lesser volume in the individual brain than in the template, while those with values > 1.0 indicated regions of greater volume in the individual brain than in the template.
Next, Jacobian files were downsampled using a custom Fiji script provided by S. C. (personal communication). Jacobian files from each caste or sex group were then compared to one another by performing a per-voxel t-test using CMTK's ttest function. Significant threshold t-values were determined in R (v4.1.1; [URL]/) by performing permutation tests. Each test was repeated 10,000 times and quantiles of 2.5% and 97.5% were selected as threshold t-values to use in data visualization. Data were visualized in Amira (v2020.1,Thermo Fisher Scientific).
Figure 1 .
Figure 1. Brain regions measured in validation study Mushroom bodies are not visualized in lower panel so that central body is visible.
Figure 2 .
Figure 2. Pairwise DBM-based analysis of worker-derived R flavipes castes Colored regions indicate brain regions identified as significantly enlarged in one caste relative to the other (A) Worker-soldier comparison.(B) Worker-ergatoid comparison.(C) Soldier-ergatoid comparison.
Figure 3 .
Figure 3. Pairwise DBM-based analysis of R flavipes workers, nymphs, and alates Colored regions indicate brain regions identified as significantly enlarged in one caste phenotype relative to the other (A) Worker-nymph comparison.(B) Worker-alate comparison.(C) Nymph-alate comparison.
Figure 5 .
Figure 5. Comparison of antennal lobe (AL) volume among R flavipes caste phenotypes Volume of each AL is expressed as a percent of whole brain volume to account for individual variation in brain size. AL volume was measured in 6 individuals per phenotype; left and right ALs were measured separately to generate 12 measurements per phenotype. Individuals of all phenotypes except alates were pooled from a total of five colonies and randomly selected for validation, while all alates were collected from one colony. Bars represent mean +SD, while points represent individual ALs. Letters denote significance groups as determined using a one-way ANOVA followed by Tukey's post-hoc test, where p < 0.05 was used as the threshold for significance.
Figure 4 .
Figure 4. Pairwise DBM-based analysis of sexual dimorphism among R flavipes caste phenotypes Colored regions indicate brain regions identified as significantly enlarged in one sex relative to the other (A) Worker comparison.(B) Soldier comparison.(C) Ergatoid comparison.(D) Nymph comparison.(E) Alate comparison.
Figure 6 .
Figure 6. Comparison of mushroom body (MB) volume among R flavipes caste phenotypes Volume of each MB is expressed as a percent of whole brain volume to account for individual variation in brain size. MB volume was measured in 6 individuals per phenotype; left and right MBs were measured separately to generate 12 measurements per phenotype. Individuals of all phenotypes except alates were pooled from a total of five colonies and randomly selected for validation, while all alates were collected from one colony. Bars represent mean +SD, while points represent individual MBs. Letters denote significance groups as determined using a one-way ANOVA followed by Tukey's post-hoc test, where p < 0.05 was used as the threshold for significance.
Figure 8 .
Figure 8. Comparison of central body (CB) volume among R flavipes caste phenotypes Volume of each CB is expressed as a percent of whole brain volume to account for individual variation in brain size. CB volume was measured in 6 individuals per phenotype. Individuals of all phenotypes except alates were pooled from a total of five colonies and randomly selected for validation, while all alates were collected from one colony. Bars represent mean +SD, while points represent individual CBs. Letters denote significance groups as determined using a oneway ANOVA followed by Tukey's post-hoc test, where p < 0.05 was used as the threshold for significance.(A) CB volume expressed as a percent of whole brain volume without adjustment.(B) CB volume expressed as a percent of whole brain volume minus the volume of the optic lobes.
Figure 9 .
Figure 9. Visual comparison of optic lobes Caste phenotypes used in this study included worker (A), nymph (B), and alate (C). In each case, the optic lobe was highlighted by red circle.
Table 1 .
Summary of number of brains and template used for each pairwise comparison
==
Domain: Biology Environmental Science
|
[{"added": "2024-05-25T15:16:47.067Z", "created": "2024-05-01T00:00:00.000", "id": "269998350", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": null, "oa_url": null, "pdf_hash": "e69f955c94353a2fefb1528ce00e18e00ecd98f2", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:3873", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "5324fd3eba2c397adfd5aa02cf4293f192e7af62", "year": 2024}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
High-Temperature Stress Effect on the Red Cusk-Eel (Geypterus chilensis) Liver: Transcriptional Modulation and Oxidative Stress Damage
Simple Summary The red cusk-eel (Genypterus chilensis) is a native Chilean species important for aquaculture diversification in Chile. The effect of high-temperature stress on the liver, a key organ for fish metabolism, is unknown. In this study we determined for the first time the effects of high-temperature stress on the liver of red cusk-eel. The results showed that high-temperature stress increased hepatic enzyme activity in the plasma of stressed fish. Additionally, this stressor generated oxidative damage in liver, and generated a transcriptional response with 1239 down-regulated and 1339 up-regulated transcripts associated with several processes, including unfolded protein response, heat shock response and oxidative stress, among others. Together, these results indicate that high-temperature stress generates a relevant impact on liver, with should be considered for the aquaculture and fisheries industry of this species under a climate change scenario. Abstract Environmental stressors, such as temperature, are relevant factors that could generate a negative effect on several tissues in fish. A key fish species for Chilean aquaculture diversification is the red cusk-eel (Genypterus chilensis), a native fish for which knowledge on environmental stressors effects is limited. This study evaluated the effects of high-temperature stress on the liver of red cusk-eel in control (14 °C) and high-temperature (19 °C) groups using multiple approaches: determination of plasmatic hepatic enzymes (ALT, AST, and AP), oxidative damage evaluation (AP sites, lipid peroxidation, and carbonylated proteins), and RNA-seq analysis. High-temperature stress generated a significant increase in hepatic enzyme activity in plasma. In the liver, a transcriptional regulation was observed, with 1239 down-regulated and 1339 up-regulated transcripts. Additionally, high-temperature stress generated oxidative stress in the liver, with oxidative damage and transcriptional modulation of the antioxidant response. Furthermore, an unfolded protein response was observed, with several pathways enriched, as well as a heat shock response, with several heat shock proteins up regulated, suggesting candidate biomarkers (i.e., serpinh1) for thermal stress evaluation in this species. The present study shows that high-temperature stress generated a major effect on the liver of red cusk-eel, knowledge to consider for the aquaculture and fisheries of this species.
Introduction
Environmental factors are important for the physiology of fish, particularly those associated with water conditions. Among these, temperature, pH, and dissolved oxygen (DO) are crucial for the homeostasis of the fish in marine environments, and changes in these factors could lead to generating a stressful status for the animal [1]. It has been reported that an increase in water temperature could lead to stress and negative effects on marine fish, including salmonids [2,3], catfish [4], Atlantic cod [5], and gilthead seabream [6]. Understanding the effects of water temperature on fish is crucial in the actual scenario of climate change, considering the average rising sea temperature per decade [7]. Additionally, the effect of climate change on sea temperatures is expected to influence relevant phenomena of the Pacific coast, such as El Niño-Southern Oscillation (ENSO), increasing their intensity and frequency [8], which is relevant for marine species of the South Pacific coast, considering the increase in water temperatures associated with this phenomena. Therefore, it is important to understand the effect of an increase in temperature and how it affects the stress response in marine fish with aquaculture potential.
Chile is a relevant country in seafood production, with an important fishery and aquaculture industry. Additionally, the Chilean aquaculture sector is recognized for its important salmon and mussel industries [9]. However, in the last decade, an important effort has been made by the public private association to diversify Chilean aquaculture with native fish of commercial value. One of these species is the red cusk-eel (Genypterus chilensis), part of the Genypterus genus, endemic to the South Pacific coast, an economically relevant fish for fisheries and, recently, part of the Chilean aquaculture diversification program [10,11], with recent elucidation of the complete production cycle [12]. The red cusk-eel is a demersal fish with a carnivorous diet; it lives on rocky bottoms [13] and is characterized reproductively as a multiple spawner [11]. The high value of its flesh makes this species an attractive product; however, tons of fisheries have presented variable levels, with a decreasing tendency in the last decade [14]. In this sense, it is important to understand how environmental factors could affect this species and the mechanisms involved.
Stress in fish can be characterized as an adaptative response to danger, which generates physiological changes to prepare the fish to respond and survive threats. This stress response in fish is mediated by the hypothalamic-pituitary-interrenal (HPI) axis though several key chemical mediators, including corticotropin releasing factor, adrenocorticotropic hormone, α-melanocyte-stimulating hormone, adrenaline, and cortisol, a key hormone that increases in plasma under stress [15]. If the stressor is maintained for prolongated periods, a chronic stress status is generated in the fish, leading to several negative physiological effects, including decreased growth, reproductive problems, behavior modifications, and immune response [16,17]. Moreover, at a cellular level, the stress can lead to an increase in reactive oxygen species (ROS), which could lead to an oxidative stress status, as previously observed for several teleost fish species [18][19][20][21][22], an effect also observed for red cusk-eel in response to several stressors [23][24][25]. Red cusk-eel has shown low tolerance to intensive farming stressors, with limited information related to the stress response capacity in this species [13]. However, the specific stress response varies according to species, as well as the effect on each tissue. Our previous studies on G. chilensis showed a variable tissue response under handling stress, with altered metabolic status in the liver, a modulation of the immune response in the head kidney, and an induction of atrophy in skeletal muscle through coding and noncoding regulation [26]. Additionally, it has been observed that thermal stress could induce muscle atrophy in this species [24], as well as oxidative damage in eggs, with a minor effect on the ovary [23]. One of the most important organs for fish metabolism is the liver, which is directly involved in stress response by metabolizing and liberating stored energy to respond to stress [17]. Nevertheless, the liver response to thermal stress in Genypterus species has not been previously studied, nor has the impact of this stressor on the oxidative status of this tissue. It is important to consider that sea temperatures will increase through sea heat waves due to the effect of ENSO under a climate change scenario, which will affect the Chilean coast associated with the geographic range of G. chilensis. Therefore, the objective of this study was to evaluate the effect of high-temperature stress on the liver of G. chilensis in terms of the transcriptomic and oxidative stress status to determine the negative impact of this type of stressor on liver.
Ethics Statement
All procedures with the red cusk-eel individuals and all scientific activities adhered to animal welfare procedures and were approved by the bioethical committees of the Universidad Andres Bello (007/2018) and the National Commission for Scientific and Technological Research (CONICYT) of the Chilean government.
Fish Sampling and Experimental Design
In this study, we used reproductively immature red cusk-eel juveniles (G. chilensis) of 12 months of age (average weight of 665 ± 52.7 g; average length of 60 ± 4.8 cm), collected from the Centro de Investigación Marina de Quintay (CIMARQ), maintained under natural photoperiod conditions (L:D 12:12), and controlled temperature (14 ± 1 • C), and fed daily with commercial pellet food. Fish were separated into control and stress groups, with the stress group subjected to a standardized thermal stress protocol proven to generate stress in red cusk-eel [24]. Briefly, this protocol consists of increasing the temperature over 24 h at a rate of 1 • C in 5 h. This protocol maintains the thermal stress temperature (19 ± 1 • C) for 5 days. This high temperature protocol was selected considering heat waves observed in the summer season on the Chilean coast in recent years [27]. The control group was maintained at the control temperature (14 ± 1 • C) during the assay. At the end of the experiment, six individuals per group (two tanks per group, with three animals sampled per tank, total of N = 12 sampled fish) were netted and sampled. Blood samples were collected via caudal puncture using heparinized tubes, immediately centrifuged at 5000× g for 10 min for serum obtention and stored at −80 • C until analysis. After blood sampling, fish were euthanized (overdose of anesthetic 3-aminobenzoic acid ethyl ester, 300 mg/L). Fish livers were collected and stored for RNA extraction in RNAsave solution (Biological Industries, Cromwell, CT, USA) or immediately frozen in liquid nitrogen and stored at −80 • C until analysis for oxidative damage evaluation.
AST, ALT and AP Evaluation
The plasmatic activities of aspartate aminotransferase (AST), alanine aminotransferase (ALT), and alkaline phosphatase (AP) were determined using commercially available kits from Valtek (Santiago, Chile) following the manufacturer's instruction. Briefly, these kits determined the enzymatic activity (IU/L) via the generation of colorimetric products from glutamate (colorimetric product: 450 nm), pyruvate (535 nm), and p-nitrophenol (405 nm) for AST, ALT and AP activity, respectively.
Oxidative Stress Assays in Liver
To determine the oxidative damage in the liver of red cusk-eel in response to hightemperature stress, DNA oxidative damage, protein carbonylation, and lipid peroxidation were evaluated using commercially available kits. The DNA oxidative damage assay was performed with genomic DNA (gDNA) purified from 25 mg of the liver using the DNAzol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer's protocol and quantified with the Epoch Spectrophotometer System (BioTek, Winooski, VT, USA). Then, the apurinic/apyrimidinic sites (AP sites) were determined in the gDNA using the OxiSelect Oxidative DNA Damage Quantification Kit (Cell Biolabs, CA, USA) following the manufacturer's protocol. The protein carbonylation assay was performed using total protein extracted from 100 mg of the liver in 1 mL of lysis buffer containing 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, and 1% NP-40, solubilized at 4 • C after 12,000× g centrifugation. Proteins were quantified using the Pierce BCA Protein Assay Kit (Thermo Scientific, Batavia, IL, USA). Then, carbonylated protein content was quantified using the OxiSelect Protein Carbonyl Spectrophotometric Assay (Cell Biolabs, San Diego, CA, USA) following the manufacturer's protocol. The lipid peroxidation determination in the liver was performed using the OxiSelect HNE Adduct Competitive ELISA Kit (Cell Biolabs, San Diego, CA, USA), following the manufacturer's protocol. This kit determines lipid peroxidation through the quantification of hydroxynonenal (HNE) protein adducts in the extracted proteins of the liver.
Liver RNA Extraction, Library Preparation and Illumina Sequencing
Total RNA was extracted from the liver stored in RNAsave solution (Biological Industries, Cromwell, CT, USA) using the TRIzol ® reagent (Invitrogen, Carlsbad, CA, USA) protocol. Total RNA was quantified by fluorometry with the Qubit ® RNA quantitation assay (Invitrogen, Carlsbad, CA, USA) and purity was determined according to 260/280 ratio using the Epoch Spectrophotometer System (BioTek, Winooski, VT, USA). The RNA integrity was measured according to RNA Quality Measurement Number (RQN) through a Fragment Analyzer with the Standard Sensitivity RNA Analysis kit (Advanced Analytical Technologies, Fiorenzuola, Italy), selecting samples with RQN ≥ 8. The cDNA libraries construction were performed with 1 µg of total RNA per sample using the Illumina ® TruSeq RNA Sample Prep Kit v2 (Illumina ® , San Diego, CA, USA), following the manufacturer's protocol. The sizes of the mRNA libraries were determined through a Fragment Analyzer using the NGS Fragment Analysis kit (Advanced Analytical Technologies) and quantified by qPCR using the Kapa Library Quantification kit (Roche, Little Falls, NJ, USA). Paired-end sequencing (2 × 100 bp) was performed on a Hiseq 2500 (Illumina ® ) platform in Macrogen Inc. (Seul, South Korea).
Reads Filtering, Differential Expression, and GO Enrichment Analysis
The raw reads obtained from Illumina sequencing were trimmed to remove the remaining Illumina adapter, low-quality sequences, and short sequences (<50 bp), using the CLC Genomics Workbench v.7.0.3 software. The filtered reads were mapped to a G. chilensis reference transcriptome previously annotated by our group ( [28], NCBI accession number SRS614525) using the CLC Genomics Workbench v.7.0.3 software with the following parameters: mismatch cost = 2, insertion cost = 3, deletion cost = 3, length fraction = 0.8, and a similarity fraction = 0.8. The expression values were used for clustering and heatmap chart generation with R. Differential expression analysis was performed with the R package DESeq2 (version 1.2.10) [29] to determine differentially expressed transcripts in the liver between the control and stress groups. Transcripts presenting an adjusted p-value of <0.05 and an absolute log2 fold change of >1 were considered as differentially expressed between the groups. Enrichment analysis was performed on the list of differentially expressed transcripts and GO terms to determine the overrepresented processes in response to high-temperature stress in the liver, considering up and down-expressed transcripts. This analysis was performed using the enrichment analysis tool implemented in the Blast2GO software [30]. Additionally, the Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathway database was used to build the represented pathways through the KEGG Automatic Annotation Server (KAAS) [31] using the KEGG Orthology (KO) identifiers of the differentially expressed list.
RNA-seq Validation by qPCR
Total RNA previously extracted was DNAse-treated to remove residual gDNA, and 1 µg of RNA was reverse transcribed into cDNA using the QuantiTect ® Reverse Transcription kit (Qiagen, Germantown, MD, USA), following the manufacturer's protocol. A total of thirteen differentially expressed transcripts were selected for qPCR validation, corresponding to: hsp60, hsp70, gpx7, ddit4, leptin, msh2, msh3, c1ql1, ccl20, atg12, atg4b, casp3 and c3. Primers were designed with Primer 3 software v0.4.0 ( [URL]3/, accessed on 6 May 2021) using the reference transcriptome previously described. The primer sequences, amplicon size, Tm, and efficiency are presented in Table 1. The qPCR was performed in a Stratagene MX3000P qPCR system (Stratagene, La Jolla, CA, USA). All qPCR assays were performed in triplicates, using no-template and no-RT controls, in compliance with the MIQE guidelines [32]. The qPCR reaction mixture contained 7.5 µL of 2× Brilliant ® II SYBR ® Green master mix (Agilent Technologies, CA, USA), 100 ng of cDNA per reaction and 200 nM of each primer in a 15 µL final volume. Thermal cycling conditions were an initial activation of 2 min at 95 • C, followed by 40 cycles of 30 s at 95 • C, 30 s at 62 • C, and 30 s at 72 • C. A melt curve analysis was performed to confirm a single qPCR product and a standard curve using two-fold series dilutions was used to estimate the efficiency of each primer set. The expression of target genes was normalized using the geometric means of two reference genes (actb and taf12) previously validated for red cusk-eel in the liver [26] and following the methodology described by [33].
Statistical Analysis
Significant differences between means of the control and stress groups for DNA damage (AP sites), enzymatic activity (AST, ALT, and AP) and differential expression of the qPCR validated genes were determined using a t-test with a significance threshold of p < 0.05. All statistical analyses were performed using GraphPad Prism, v.5.00 (GraphPad Software, San Diego, CA, USA).
Hepatic Enzyme Activity and Oxidative Stress Response to High-Temperature Stress
We previously reported that thermal stress for 5 days significantly increased plasmatic levels of cortisol and glucose in G. chilensis [24]. To understand how thermal stress affects the metabolism in this species, we studied the effect of high temperature on the liver. To evaluate the effect on the hepatic function of this type of stressor, we measured plasmatic markers of liver damage, i.e., ALT, AST, and AP enzymatic activity. High-temperature stress significantly increased the plasmatic activity of ALT ( Figure 1A), AST ( Figure 1B) and AP ( Figure 1C) in the stress group, evidencing altered hepatic function in the liver of stressed fish. To determine the oxidative damage in the liver generated by high-temperature stress, the DNA oxidative damage, protein carbonylation and lipid peroxidation were determined in the liver. High temperatures generated DNA damage, evidenced by the significant increase in apurinic/apyrimidinic sites in the stressed group (Figure 2A). Oxidative damage was also observed in lipid and proteins, determined by a significant increase in protein carbonylation ( Figure 2B) and lipid peroxidation ( Figure 2C) in response to high temperature in the stress group.
Differentially Expressed Transcripts in Hepatic Response to High-Temperature Stress
To understand the complexity of the stress response in the liver of G. chilensis, we performed RNA-seq analysis on the liver of each experimental group. The sequencing generated a total of 754,678,455 paired-end reads, with an average of 58,052,189 raw paired-end reads per library. The raw data are available from NCBI under BioProject , and lipid peroxidation in terms of HNE adducts (C) in control and stress groups. Bars represent the mean ± SEM. Significant differences between the control and stress groups are indicated by asterisks; * (p < 0.05), ** (p < 0.01) and **** (p < 0.0001).
Differentially Expressed Transcripts in Hepatic Response to High-Temperature Stress
To understand the complexity of the stress response in the liver of G. chilensis, we performed RNA-seq analysis on the liver of each experimental group. The sequencing The raw data are available from NCBI under BioProject PRJNA835467 with BioSamples accession number SAMN28102858, SAMN28102859, SAMN28102860, SAMN28102861, SAMN28102862, SAMN28102863, SAMN28102864, SAMN28102865, SAMN28102866, and SAMN28102867. After trimming by quality, adapters, and size, we obtained an average of 58,018,122 high-quality filtered paired-end reads per library (Table 2). These reads were mapped to the G. chilensis reference transcriptome (NCBI accession number SRS614525), obtaining an average of 85.2% mapped reads (Table 2). Expression values were used for normalization and differential expression analysis with the R package DESeq2 (version 1.2.10) [29], obtaining a total of 2578 differentially expressed transcripts between control and stressed groups. Of these transcripts, 1239 were down-regulated (Table S1) and 1339 were up-regulated (Table S2) in the stressed group (Figure 3 and Figure S1), evidencing five clusters of transcripts with different patterns of expression for the control and thermal stress groups ( Figure S2). SAMN28102865, SAMN28102866, and SAMN28102867. After trimming by quality, adapters, and size, we obtained an average of 58,018,122 high-quality filtered paired-end reads per library (Table 2). These reads were mapped to the G. chilensis reference transcriptome (NCBI accession number SRS614525), obtaining an average of 85.2% mapped reads (Table 2). Expression values were used for normalization and differential expression analysis with the R package DESeq2 (version 1.2.10) [29], obtaining a total of 2578 differentially expressed transcripts between control and stressed groups. Of these transcripts, 1239 were down-regulated (Table S1) and 1339 were up-regulated (Table S2) in the stressed group (Figures 3 and S1), evidencing five clusters of transcripts with different patterns of expression for the control and thermal stress groups ( Figure S2).
GO Enrichment and Pathway Analysis in the Liver
The differentially expressed transcripts were used for an enrichment analysis using GO terms to determine the overrepresented processes in response to high-temperature stress in the up-and down-regulated transcripts of the liver. The enrichment analysis showed 38 enriched processes for up-regulated transcripts (Table 3), including 18 biological processes (BP), 16 cellular components (CC), and 4 molecular functions (MF). Several of the enriched terms were related to protein metabolism, including protein folding (GO:0006457), protein transport (GO:0015031), protein localization (GO:0008104), protein retention in ER lumen (GO:0006621), and unfolded protein binding (GO:0051082). Additionally, the BP term response to heat (GO:0009408) was also enriched in the up-regulated transcripts in the high-temperature stress group. No enriched GO terms were identified in the downregulated transcripts (FDR < 0.05). In the KEGG pathway analysis, we identified several pathways represented in the down-regulated and up-regulated transcripts, including apoptosis, cell cycle, MAPK signaling pathway, autophagy, protein processing in endoplasmic reticulum, and ubiquitinmediated proteolysis. Among these pathways, protein processing in endoplasmic reticulum was one of the most represented on the differentially expressed transcripts (Figure 4), with 31 up-regulated and three down-regulated transcripts represented in this pathway, including several heat shock proteins; heat shock protein 40 (hsp40, also known as dnaJ homolog subfamily B member 1), heat shock protein 70 (hsp70) and heat shock protein 90 (hsp90).
(casp3) and complement C3-like (c3) ( Table 1). All of the selected differentially expressed transcripts on RNA-seq were validated by qPCR, confirming the results of the RNA-seq analysis ( Figure 5), with a Pearson's correlation coefficient of r = 0.847, confirming the significant differential expression of all tested transcripts by qPCR between the control and high-temperature group ( Figure S3). Figure 5. qPCR validation of selected differentially expressed transcripts in the liver of G. chilensis in response to high-temperature stress. The transcript expression levels were normalized with the geometric means of actb and taf12. The differential expression levels according to qPCR (black bars) and RNA-seq (gray bars) for these selected genes are expressed as log2 fold changes. The log2FC represents the expression change in the stress group compared with the control group. Results are expressed as the mean ± standard error. Significant differences in the validated qPCR data between control and stress groups are indicated by asterisks in the log2 Fold qPCR bars; (* pvalue < 0.05).
Discussion
Stress in fish is a relevant issue for aquaculture species and native populations of fish [15]. The environmental stress factor associated with water parameters represents a key issue in fish, especially considering the variations in the environment in the shortand long-term associated with global warming, climate change, ENSO, and pollution of the oceans, which could lead to modification of sea temperature, pH, and DO level, as well as increased levels of pollutants, including microplastics and toxic compounds, which could generate stress in teleost fish [7,8,[34][35][36]. These environmental stressors could also affect the red cusk-eel, a relevant species for Chilean fisheries and aquaculture diversification. However, studies aimed at understanding the effect of stressors in Genypterus species are limited [25,26,37,38], with no information about the impact of thermal stress on the liver for Genypterus species. In this sense, thermal stress has been previously studied in other tissues of Genypterus species, including the skeletal muscle, ovary, and post-ovulatory eggs [23,24]. Therefore, to understand the effect of high- Figure 5. qPCR validation of selected differentially expressed transcripts in the liver of G. chilensis in response to high-temperature stress. The transcript expression levels were normalized with the geometric means of actb and taf12. The differential expression levels according to qPCR (black bars) and RNA-seq (gray bars) for these selected genes are expressed as log 2 fold changes. The log 2 FC represents the expression change in the stress group compared with the control group. Results are expressed as the mean ± standard error. Significant differences in the validated qPCR data between control and stress groups are indicated by asterisks in the log2 Fold qPCR bars; (* p-value < 0.05).
Discussion
Stress in fish is a relevant issue for aquaculture species and native populations of fish [15]. The environmental stress factor associated with water parameters represents a key issue in fish, especially considering the variations in the environment in the short-and long-term associated with global warming, climate change, ENSO, and pollution of the oceans, which could lead to modification of sea temperature, pH, and DO level, as well as increased levels of pollutants, including microplastics and toxic compounds, which could generate stress in teleost fish [7,8,[34][35][36]. These environmental stressors could also affect the red cusk-eel, a relevant species for Chilean fisheries and aquaculture diversification. However, studies aimed at understanding the effect of stressors in Genypterus species are limited [25,26,37,38], with no information about the impact of thermal stress on the liver for Genypterus species. In this sense, thermal stress has been previously studied in other tissues of Genypterus species, including the skeletal muscle, ovary, and post-ovulatory eggs [23,24]. Therefore, to understand the effect of high-temperature stress at the hepatic level, we evaluated the hepatic enzymes, oxidative stress response, and transcriptional regulation in red cusk-eel.
High-Temperature Effect on Hepatic Enzymes
In a previous study, we determined that high temperature (19 • C) could generate a stress response in red cusk-eel with an increase in the plasmatic level of cortisol and glucose [24]. In the present study, we determined that a high temperature increased the plasmatic activity of ALT, AST, and AP enzymes, evidencing the effect at the hepatic level. This effect was previously observed in other teleost fish under stress conditions, including the plasmatic activity of ALT and AST under high densities and metal pollution in rohu (Labeo rohita) [39,40], metal toxicity and pesticides in Nile tilapia (Oreochromis niloticus) [41,42] and spotted snakehead (Channa punctatus) [43], and elevated ALT, AST and AP under handling stress in red cusk-eel [25]. In terms of high-temperature stress, different results have been observed, with an increase in the plasmatic activity of AST and ALT in pufferfish (Takifugu obscurus) [44], which is consistent with our results for red cusk-eel, while no effect for AST and ALT was observed in Turbot (Scophthalmus maximus) [45] under high-temperature stress, evidencing that the hepatic response to thermal stress could vary according to fish species. The increased levels on ALT, AST an AP enzymes could be indicative of liver dysfunction, reflecting hepatocyte damage due to thermal stress, which is concordant with the results observed in mammals [46].
Oxidative Stress under High-Temperature Stress
Oxidative stress corresponds to a disturbance between the production of ROS, which can accumulate in cells, and the antioxidant defenses generated by the cellular systems to detoxify these ROS [47]. OS participate in different normal cellular functions, acting as a second messenger in signal transduction [48], but they can also generate cellular damage, including oxidative damage generating lipid peroxidation, as well as damage to proteins and DNA [49]. Environmental stress, including thermal stress, can lead to oxidative stress status in marine animals [3]. It was previously observed that thermal stress could modulate the oxidative stress status in several tissues of teleost fish [50,51]. One of the relevant organs in which oxidative status could be affected by thermal stress is the liver, as it was previously observed that thermal stress by low temperature could increase antioxidant enzymes in milkfish (Chanos Chanos) [52], as well as generate oxidative damage, leading to lipid peroxidation in Pacu (Piaractus mesopotamicus) under low-temperature stress [53,54]. High-temperature stress could also generate a relevant impact on the oxidative status of fish, with increased antioxidant enzyme activities, as observed for Senegalese sole (Solea senegalensis) [55]. Additionally, oxidative damage was observed in the liver of rohu (Labeo rohita), with lipid peroxidation and DNA fragmentation [56]. This is consistent with the findings of our study, evidencing that thermal stress generates an important oxidative effect on the liver of red cusk-eel. This response was also observed at the transcriptional level, with the up-regulation of GPx genes in several tissues of teleost fish under thermal stress, including black porgy (Acanthopagrus schlegeli) [57] and pufferfish [58], as well as in red cusk-eel eggs under thermal stress [23], where gpx1 was increased, an effect not observed for this species in skeletal muscle [24]. Additionally, the effect of high-temperature stress on the liver related to DNA damage was also observed at the transcriptional level, with the up-regulation of genes involved in DNA mismatch repair (msh2 and msh3), concordant with the previously reported effect of thermal stress on zebrafish (Danio rerio) [59] and American lobster [60]. However, the thermal stress response associated with oxidative stress in teleost fish could vary according to species and tissues, as observed in sheepshead minnow (Cyprinodon variegatus), where a limited effect on antioxidant enzymes and no lipid peroxidation were present [61], in contrast to the variable lipid peroxidation observed in Senegalese sole [55]. We previously observed in red cusk-eel that the impact of high-temperature stress on oxidative damage could vary according to tissue, with lipid peroxidation and DNA damage observed for skeletal muscle [24], but no oxidative damage was observed in the ovaries under high-temperature stress [23], showing that the liver is a sensitive organ under thermal stress in red cusk-eel under an ENSO temperature increase scenario.
High-Temperature Stress in Hepatic Protein Processing and Folding
Temperature can modulate several cellular processes through gene expression regulation [2]. Here, we used RNA-seq analysis to evaluate, for the first time, the effect of thermal stress on the liver of red cusk-eel, considering that no studies have previously evaluated the temperature effect on the liver in any species of the Genypterus genus. We observed a higher hepatic transcriptional response to high-temperature stress compared to other types of stressors, such as handling stress, previously observed for red cusk-eel (4.6 times the differentially expressed transcripts) [25], evidencing that a high temperature could have a higher transcriptional impact on the liver than others stressors in this species. At the cellular level, the function of endoplasmic reticulum is key to protein synthesis, folding, and exporting [62]. However, external processes such as thermal stress could generate alterations in homeostasis, affecting normal protein folding, which leads to endoplasmic reticulum (ER) stress. To alleviate this stress, the unfolded protein response molecular mechanism is activated in cells [63]. This ER stress and subsequent unfolded protein response is concordant with our results in the liver, where we observed enriched processes associated with this type of response in the up-regulated genes in the stress group, including protein folding, protein transport, protein localization, protein retention in ER lumen, and unfolded protein binding. In this sense, we found several genes associated with protein processing and folding pathways, including transcripts involved in: protein export, such as signal recognition particle 14 and 19 kDa (srp14 and srp19) and signal peptidase complex subunit 2-like (spcs2) [64]; transcripts associated with protein processing in endoplasmic reticulum (Figure 4), with part of this pathway down-regulated, including TNF receptorassociated factor 2-like (traf2), and mainly an up-regulation of this pathway, including ubiquilin-4 (ubqln4), cytoskeleton-associated protein 4 (ckap4); ribosome-binding protein 1 (rrbp1, also known as p180); and glucosidase 2 subunit beta (prkcsh). Others include transcripts associated with the unfolded protein response, including eukaryotic translation initiation factor 2 alpha kinase 1 (eik2ak1, also known as hri), 3 (eik2ak3, also known as perk) and 5 (eik2ak5), cyclic AMP-dependent transcription factor ATF-4-like (atf4) and DNA damage-inducible transcript 3 -like (ddit4, also known as chop). The unfolded protein response could be initiated by EIK2AK3 kinase activation of eIF2α, leading to ribosome inhibition and attenuating protein synthesis [65]. Additionally, the ATF4 gene was activated, which would lead to DDIT4 gene regulation and an antioxidant response in cells [66], evidencing that thermal stress in the liver of red cusk-eel modulates the protein processing and generates an unfolded protein response that is not able to control the oxidative stress and damage in this tissue. This is concordant with the results observed in previous studies where unfolded protein response genes were activated under thermal stress in mammals [67,68] and fish, e.g., in the liver of Tambaqui (Colossoma macropomum) [69] and gilthead sea bream (Sparus aurata) [70].
Heat Shock Protein as Thermal Stress Biomarkers
The cellular response to thermal stress is a key process to preserve protein integrity; this is known as the heat shock response, which includes heat shock proteins (Hsps) to re-fold the proteins damaged by temperature [71]. This response was observed in the liver of red cusk-eel with the enrichment of the term response to heat (GO:0009408), highlighting several heat shock proteins differentially expressed in response to high-temperature stress, including hsp40, hsp60, hsp70, hsp90, and serpinh1. The up-regulation of these Hsps was previously observed in other fish under thermal stress, with an increase in the liver of three-spined stickleback (Gasterosteus aculeatus) (hsp60, hsp70, and hsp90) [72], Atlantic salmon (Salmo salar) (hsp70) [73,74], Atlantic cod (Gadus morhua) (hsp70 and hsp90) [74], and Wuchang bream (Megalobrama amblycephala) (hsp60, hsp70, and hsp90) [75]. The hsp60 and hsp70 expression levels were also regulated under thermal stress in the postovulatory eggs and skeletal muscle of red cusk-eel [23,24], showing that these genes could be valuable biomarkers of thermal stress in this species. However, the most up-regulated gene in the liver under high-temperature stress was serpinh1 (5.7log 2 Fold Change, Table S2). SerpinH1 (also known as Hsp47) is a chaperone involved in the biosynthesis of collagen at the ER [76], with a key role in the restoration of homeostasis during high-temperature stress and oxidative stress [22]. It was shown to be a good biomarker for thermal stress in several salmonid species, including rainbow trout (Oncorhynchus mykiss), sockeye salmon (Oncorhynchus nerka), and Chinook salmon (Oncorhynchus tshawytscha) [3,77], as well as zebrafish [78], evidencing that sherpinh1 is one of the most suitable transcriptional biomarkers of high-temperature stress in fish. This was also the case for red cusk-eel in this study, representing a useful tool to evaluate thermal stress status in this species under a climate change scenario.
Conclusions
The present study evaluated for the first time the effects of high-temperature stress in the liver of G. chilensis, using a multiple approach of plasmatic hepatic enzymes, oxidative damage evaluation and RNA-seq analysis. We showed that high-temperature stress under heatwaves ENSO-associated scenario generated a major effect in the liver, affecting hepatic enzymes, generating oxidative damage in this tissue, as well as generating an unfolded protein response at the molecular level in several associated pathways, including a heat shock response, evidencing the affection of red cusk-eel under this type of stressor. This study contributes to knowledge about thermal stress under a climate change scenario, generating candidate biomarkers for thermal stress evaluation in this species, information that should be relevant for the aquaculture and fisheries industry of red cusk-eel.
Supplementary Materials: The following supporting information can be downloaded at: [URL]:// www.mdpi.com/article/10.3390/biology11070990/s1, Figure S1: Differentially expressed transcripts in response to high-temperature stress in G. chilensis expressed as average transcription per group; Figure S2: Differentially expressed transcripts and clustering groups in response to high-temperature stress in G. chilensis; Figure S3: Validation of selected differentially expressed transcripts by qPCR on liver of G. chilensis on response to high-temperature stress; Table S1: Down-regulated differentially expressed transcripts between control and stress groups of G. chilensis; Table S2: Up-regulated differentially expressed transcripts between control and stress groups of G. chilensis.
Informed Consent Statement: Not applicable.
Data Availability Statement: The raw read sequences obtained from liver sequencing of red cuskeel were deposited in NCBI under BioProject PRJNA835467, with BioSample accession number SAMN28102858, SAMN28102859, SAMN28102860, SAMN28102861, SAMN28102862, SAMN28102863, SAMN28102864, SAMN28102865, SAMN28102866, and SAMN28102867. The datasets generated and/or analyzed in the present study are available from the corresponding author upon reasonable request.
==
Domain: Biology Environmental Science
|
[{"added": "2022-07-02T15:02:59.734Z", "created": "2022-06-29T00:00:00.000", "id": "250200078", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2079-7737/11/7/990/pdf?version=1657092129", "pdf_hash": "3b367c281a99d5b50e58f8c75f9a01fb458a4836", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:490", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "61c88da6aa20225d2f120647b80437d996938b60", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Unraveling the Adaptive Significance of Mitochondrial Genome Variability of Drosophila obscura †
: Drosophila obscura is a very common fruit fly inhabiting European forests. This species has a large number of mitochondrial haplotypes of Cyt b gene. We used experimental lines of D. obscura to test the adaptive significance of intra-population variability of the mitochondrial genome (mtDNA) and selective forces that maintain it. We chose three isofemale lines with distinct mitochondrial haplotypes of Cyt b gene from each of the two populations sampled in Serbia. Using back-crossing, we created nine experimental lines for each population with all combinations of mtDNA haplotypes and nuclear genetic backgrounds (nuDNA). Individuals of both sexes were tested separately for desiccation resistance at two temperatures. Cox proportional hazards model, with four factors: mtDNA, nuDNA, sex and temperature was used to analyze the survival data. In some comparisons we noticed significant effect of mtDNA on desiccation resistance, while all of them showed significant effect of interaction between mitochondrial and nuclear genome. Temperature in interaction with mtDNA or mito-nuclear genotype more frequently showed significant effect on desic-cation resistance compared to sex in interaction with mtDNA or mito-nuclear genotype. Our result show adaptive significance of intra-population variation of mtDNA and importance of interactions between mitochondrial and nuclear genome on fitness. Temperature specific mito-nuclear interaction rather than sex-specific selection on mito-nuclear genotypes maintains mtDNA variability in this model species.
Introduction
Variation in the mtDNA has long been considered selectively neutral [1][2]. This view was based solely on the properties of the mitochondrial genome (mtDNA), it is haploid, there is no dominance and its inheritance is mostly uniparental. Although it generally codes a small number of genes, their products are enrolled in very important processes which enable eukaryotic cells their high energy efficiency. Hence, variation in mtDNA that affects an organism's fitness would swiftly be eliminated or fixed by natural selection.
However, the modern view of selective processes that act on mtDNA takes into consideration functional interrelations between mitochondrial and nuclear genomes (nuDNA) [3,4]. Interactions between the two genomes are complex and encompass several important biological processes such as cellular respiration, mtDNA replication, transcription, and translation. All of these processes require sequences coded from both genomes [5][6][7][8]. The adaptive inter-population variability in mtDNA is easily explained with the different selective regimes in different populations. However, the existence of stable intrapopulation variability is quite problematic to explain. While theoretical models predict the action of either negative frequency-dependent selection [8][9] or sex-specific selection on mito-nuclear genotypes [10], empirical research in this field lacks. Some data show that spatial or temporal variation in the environment, through the action of selection on mitonuclear genotypes, could maintain intra-population variation in mtDNA [3,11]. We used a set of experimental lines of D. obscura to test differences in desiccation resistance, which is an important fitness component. We tested the influence of nuDNA, sex, and temperature on desiccation resistance for bearers of different mtDNA haplotypes.
Materials and Methods
We chose three isofemale lines with distinct mitochondrial haplotypes of the Cyt b gene from each of the two populations sampled in Serbia (Figures 1 and 2) Those lines were maintained in the laboratory on a standard corn-meal medium for multiple generations. Using 14 generations of backcrossing, we created nine experimental lines for each population with all combinations of mtDNA haplotypes and nuclear genetic backgrounds (nuDNA). Backcrossing procedure included mating of 10 virgin females of specific haplotype with twice as many virgin males with the desired nuclear genetic background. For each specific genotype, more than 30 individuals of both sexes were tested separately for desiccation resistance at two temperatures (16 °C and 19 °C). Flies were inspected in small plastic tubes hourly after the experiment was set. Cox proportional hazards model, with four factors: mtDNA, nuDNA, sex, and temperature was used to analyze the survival data. Starting from the full model with all the interactions of the four factors, we subtracted term by term, in order to identify the minimal adequate model, having the lowest Akaike information criterion (AIC) score, in both populations and all genotype interactions.
Results
Due to the proportional hazards assumption being violated, in the first population, we had to stratify the data by both temperature and nuDNA factors, while in the second population all the variables satisfied the proportional hazards assumption, and stratification was not necessary. In both populations, females generally survived longer than their counterpart males. Flies of both sexes survived longer on the lower temperature in both populations, as was expected.
In the first population, we noticed a significant effect of mtDNA genotype on desiccation resistance in two out of three pairwise comparisons, not being significant in comparison II vs III which differ by six mutations although all synonymous. The mtDNA x nuDNA interaction significantly influenced the survival time in all three comparisons. While sex was a significant factor for survival under desiccation stress in all three comparisons, mtDNA x sex was significant in two comparisons, and for the combination of mtDNA x nuDNA x sex significance was not observed. Both Genotype by temperature interactions (mtDNA x temp and mtDNA x nuDNA x temp) were statistically significant in two out of three comparisons in the first population. Interaction between mtDNA, sex, and temperature was highly significant in all comparisons. Results of the Cox proportional hazards model factors and their significance for the three pairwise comparisons in the first population are given in Table 1.
Table 1. The effect of mitochondrial haplotype (mt), nuclear genetic background (nu), sex, temperature and their interaction on desiccation resistance for the first population of D. obscura. In the second population, there was no need for stratification in our models. Both nuDNA and temperature were highly significant in all three comparisons. Mitochondrial haplotype significantly influenced survival in only one pairwise comparison, while the genotype interaction (mtDNA x nuDNA) significantly influenced resistance to desiccation stress in all comparisons. Interestingly, in the second population both sex and mtDNA x sex had a significant effect on desiccation resistance only in one pairwise comparison, while the combination of mtDNA x nuDNA x sex, was conversely significant in two of the three pairs. The second population showed a significant influence of genotype by temperature interaction (mtDNA x Temp as well as mtDNA x nuDNA x Temp) in all three comparisons. Sex x temp x mtDNA interaction term was highly significant in two comparisons. Full results of the Cox models' factors and their significance for the three pairwise comparisons in the second population are given in Table 2.
Compari
Table 2. The effect of mitochondrial haplotype (mt), nuclear genetic background (nu), sex, temperature and their interaction on desiccation resistance for the second population of D. obscura.
Discussion
Our results have several important findings. They show that mtDNA variants (on their own) have a significant effect on fitness, which was measured by desiccation resistance. The desiccation resistance is an important fitness component determining which fruit flies cope with harsh environmental conditions of water deprivation in dry months in natural habitats. This way our results support a body of growing evidence of non-neutrality of mtDNA variability [4,12,13], and more importantly they give evidence for the adaptive significance of intra-population variation in mtDNA [14][15][16]. All pairwise comparisons showed a significant effect of interaction between mitochondrial and nuclear genome. This result is not surprising considering that the Cyt b gene is part of respiratory complex III which also includes subunits coded by nuDNA. Therefore, subunits coded from different genomes have to be co-adapted for the proper energy production in mitochondria. If one haplotype is combined with a non-matching nuclear genetic background, a decrease in fitness is expected.
With females of D. obscura being generally bigger than males, it is expected that sex is a significant factor in all comparisons, with larger females surviving longer in dry conditions. In some of the isofemale lines from second population, sex ratio skewed towards females was observed, consequently fewer males were available for the experiment. This deviation from equal sex ratio could be the result of a mutation on the X chromosome, that is lethal to all the male carriers, and in females it decreases their fitness, so in the experiment we had only healthy males and some females with the mutation.
The sex to genotype interaction on fitness is observed in several pairwise comparisons. The significant interaction between mtDNA and sex is in line with the Mother's Course Hypothesis [17,18]. Due to maternal inheritance of mtDNA, mutations that are harmful only in males, having no effect or being beneficial to females, cannot be eliminated by natural selection. Interaction between sex and mito-nuclear genotype is observed only in two cases, indicating a difference in direction or strength of selection on mitonuclear genotypes depending on the sex of the individual. This type of selection is marked as one of the balancing mechanisms that maintain stable intra-population variation in mtDNA [10], but there is limited evidence that empirically supports this theoretical ground [19,20].
Figure 1 .
Figure 1. Haplotypes I, II and III from the first population. Small black circles are mutations, Nnonsynonymous; S-synonymous.
Figure 2 .
Figure 2. Haplotypes IV, V and VI from the second population. Small black circles are mutations, N-nonsynonymous; S-synonymous.
==
Domain: Biology Environmental Science
|
[{"added": "2021-12-24T16:20:38.930Z", "created": "2021-07-01T00:00:00.000", "id": "245437079", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://sciforum.net/paper/download/10522/manuscript", "pdf_hash": "40055db6f8c1f639ad17ec53402a32ca0a502f9f", "pdf_src": "ScienceParsePlus", "provenance": "peS2o-0028.json.gz:5501", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "3a5bf73abe231d9b7fde328013fa1efa95af2f37", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
A New Cell Line from the Brain of Red Hybrid Tilapia (Oreochromis spp.) for Tilapia Lake Virus Propagation
Simple Summary Simple Summary: In this study, a new cell line derived from red hybrid tilapia brain tissue, RHTiB, was established. This fibroblast-like cell line was maintained for over 50 passages with optimal growth at 25 °C in Leibovitz-15 medium with 10% fetal bovine serum at pH 7.4. Genetic and chromosomal analyses confirmed its origin from red hybrid tilapia. Additionally, immunofluorescence and RT-qPCR tests revealed successful TiLV propagation in RHTiB cell lines. The development of this novel cell line offers valuable prospects in enhancing diagnostic techniques in red hybrid tilapia. Abstract Tilapia lake virus (TiLV) presents a substantial threat to global tilapia production. Despite the development of numerous cell lines for TiLV isolation and propagation, none have been specifically derived from red hybrid tilapia (Oreochromis spp.). In this study, we successfully established a new cell line, RHTiB, from the red hybrid tilapia brain. RHTiB cells were cultured for 1.5 years through over 50 passages and demonstrated optimal growth at 25 °C in Leibovitz-15 medium supplemented with 10% fetal bovine serum at pH 7.4. Morphologically, RHTiB cells displayed a fibroblast-like appearance, and cytochrome oxidase I gene sequencing confirmed their origin from Oreochromis spp. Mycoplasma contamination testing yielded negative results. The revival rate of the cells post-cryopreservation was observed to be between 75 and 80% after 30 days. Chromosomal analysis at the 25th passage revealed a diploid count of 22 pairs (2n = 44). While no visible cytopathic effects were observed, both immunofluorescence microscopy and RT-qPCR analysis demonstrated successful TiLV propagation in the RHTiB cell line, with a maximum TiLV concentration of 107.82 ± 0.22 viral copies/400 ng cDNA after 9 days of incubation. The establishment of this species-specific cell line represents a valuable advancement in the diagnostic and isolation tools for viral diseases potentially impacting red hybrid tilapia.
Introduction
Tilapia (Oreochromis spp.) is known for its fast growth, adaptability to different environments, and resistance to various microbial diseases [1,2] and is thus a commercially important fish species in aquaculture. The global production of tilapia has expanded in Asia, Africa, and the Middle East, with the Food and Agriculture Organization (FAO) reporting record production exceeding 6 million tons in 2020, which makes tilapia the second most farmed fish species after carp [3]. However, the emergence and rapid spread of the highly virulent tilapia lake virus (TiLV), officially named by the International Committee on Taxonomy of Viruses as Tilapinevirus tilapiae [4], poses a significant threat to the worldwide tilapia industry and has resulted in considerable economic losses [5][6][7][8].
Since its initial identification in Israel in 2014 [5], TiLV has been detected in more than 17 countries [9,10], with recent outbreaks in Vietnam [11] and China [12]. TiLV infections have been associated with high mortality rates of up to 90% in red tilapia (Oreochromis spp.) [9]. Such infections are characterized by severe clinical signs, such as lethargy, ocular abnormalities, and skin erosion [5,9]. While a variety of diagnostic methods have been developed to detect TiLV, including molecular assays [13], cell culture remains the gold standard for virus isolation [14,15].
Despite the high susceptibility of various cell lines derived from different piscine species to TiLV infection, such as those from largemouth bass (Micropterus salmoides), hybrid snakehead (Channa argus × Channa maculata), and perch (Siniperca chuatsi) [16], the host specificity of viruses requires the use of appropriate cell lines for viral isolation and propagation [17]. While TiLV may infect other fish cells, such as E-11 cells, which are commonly used nowadays for TiLV propagation and isolation [5], previous studies suggest that TiLV is highly host-specific, mainly infecting tilapia cell lines [16,[18][19][20][21][22]. However, no primary cell line derived from red hybrid tilapia has been established or tested for TiLV propagation. In this study, we therefore aimed to establish a new cell line from red hybrid tilapia, to determine its optimal growth conditions, and to evaluate its susceptibility to TiLV propagation.
Tissue Preparation for Primary Cell Cultures
Red hybrid tilapia fingerlings aged 2 months and weighing 10.58 ± 3.71 g were obtained from a hatchery farm in Phetchaburi Province, Thailand. The fish were euthanized using eugenol at a concentration of 2 mL per 1000 mL of water, in line with the guidelines approved by the Institutional Animal Care and Use Committee of Kasetsart University (protocol number ACKU64-VET-071) for animal handling and care.
The primary cell culture was attempted by following the tissue explant method [23]. In brief, the organs (brain, liver, spleen, heart, gills, fins, and ocular muscles) were aseptically dissected, washed three times with sterile phosphate-buffered saline (PBS) supplemented with an antibiotic-antimycotic solution (200 U/mL of penicillin, 0.2 mg/mL of streptomycin, and 0.5 µg/mL of Amphotericin B; Sigma-Aldrich, St Louis, MO, USA), and cut into 1 mm 3 pieces using sterile scissors. The tissues were then seeded in triplicate into 25 cm 2 cell culture flasks (Corning Glass Work, Corning, NY, USA) containing 5 mL of Leibovitz's L-15 medium at pH 7.4 supplemented with 20% fetal bovine serum (FBS; Gibco, Buffalo, NY, USA) and the antibiotic-antimycotic solution in a low-temperature incubator without CO 2 at 25 • C. The medium was replaced every 7 days, and the cultures were monitored daily using an inverted microscope (CKX53, Olympus, Tokyo, Japan) at 40× magnification.
Subculture and Maintenance
Upon reaching 80-90% confluence, the cells were washed twice with sterile PBS (pH 7.4) and trypsinized with 0.125% trypsin-ethylenediaminetetraacetic acid (Thermo Fisher Scientific, Grand Island, NY, USA). The detached cells were suspended in fresh L-15 medium supplemented with 20% FBS, 200 U/mL of penicillin, 0.2 mg/mL of streptomycin, and 0.5 µg/mL of Amphotericin B and centrifuged at 500 RCF at 4 • C for 5 min. The supernatant was then discarded. The cell pellet was resuspended in L-15 medium containing 20% FBS and seeded into new 25 cm 2 cell culture flasks with L-15 medium and 20% FBS. Subsequent passages were split at a 1:2 ratio from the second passage onward. After the 10th passage, the cells were maintained in L-15 medium containing 10% FBS, 100 U/mL of penicillin, 0.1 mg/mL of streptomycin, and 0.25 µg/mL of Amphotericin B. The medium was replaced every 7 days, with daily monitoring under an inverted microscope (CKX53, Olympus, Tokyo, Japan) at 40× magnification.
Cryopreservation and Recovery
The cells at 75-90% confluence were collected via trypsinization for cryopreservation. Following centrifugation at 500 RCF at 4 • C for 5 min, the supernatant was discarded, and the cell pellet was resuspended in a freezing medium consisting of 10% dimethyl sulfoxide in 90% FBS. The suspensions were aliquoted into sterile cryovials and stored overnight at −80 • C before being transferred to liquid nitrogen for long-term storage. After 30 days, the frozen cells were thawed in a water bath at 37 • C. The frozen medium was removed, and the cells were suspended in L-15 medium with 10% FBS, 100 U/mL of penicillin, 0.1 mg/mL of streptomycin, and 0.25 µg/mL of Amphotericin B. The cell viability was assessed using trypan blue staining, and the cell count was obtained with a hemocytometer.
Cell Growth Characteristics
To determine the optimal growth conditions, including the FBS concentrations and pH levels, the cells at passage 20 were seeded in duplicate at a density of 9.0 × 10 3 cells/well in 96-well cell culture plates. For FBS concentration optimization, the cells were seeded separately in L-15 media containing 5%, 10%, 15%, or 20% FBS. For pH optimization, the cells were seeded in 100 µL of L-15 medium with 20% FBS and pH levels ranging from pH 7 to 8. The pH of the L-15 medium was adjusted by adding 1 N hydrochloric acid or 1 N sodium hydroxide. The cell viability was quantified on days 1, 3, 5, and 7 using a Cell Counting Assay Kit-8 (Sigma-Aldrich, St. Louis, MO, USA), in line with a previously described procedure [24]. The optimal temperature for cell growth was assessed by incubating the cells in duplicate in 24-well cell culture plates (6.0 × 10 4 cells/well) at 25 • C, 28 • C, and 30 • C. The cell viability was determined on days 1, 3, 5, and 7 by trypan blue staining and quantified using the Countess™ II Automated Cell Counter (Invitrogen, ThermoFisher Scientific, Waltham, MA, USA).
Identification of Origin Cell Line Using Mitochondrial Markers
To verify the specificity of the new cell line, which we named red hybrid tilapia brain (RHTiB) cells, the amplification of the mitochondrial cytochrome oxidase subunit 1 (cox1) gene was conducted. The genomic DNA was isolated from the RHTiB cells using a Quick-DNA Miniprep Plus Kit (Zymo Research, Orange, CA, USA), in accordance with the manufacturer's instructions. The fragments of the cox1 gene were amplified with the Fish F1 and Fish R1 universal primers under polymerase chain reaction (PCR) conditions, as described by Ward et al. [25]. The amplified product was visualized in 1.5% agarose gel containing ethidium bromide and subjected to Sanger sequencing (1st Base, Singapore). The forward and reverse nucleotide sequences of the cox1 gene were assembled using CLC Genomic Workbench version 11.0.1 (Qiagen, Hilden, Germany), and the cox1 nucleotide sequence was deposited in the National Center for Biotechnology Information GenBank (www.ncbi.nlm.nih.gov/genbank,accessed on 15 January 2023). A maximum likelihood phylogenetic tree and a genetic distances table were subsequently generated for different fish species from the cox1 nucleotide sequence using the Kimura 2-parameter model in the Molecular Evolutionary Genetics Analysis X (MEGAX) tool [26].
Detection of Mycoplasma and Snakehead Retrovirus Contamination in RHTiB Cell Line
We screened for Mycoplasma contamination in the RHTiB cells after passages 10 and 20 using the LookOut Mycoplasma PCR Detection Kit (Sigma-Aldrich, St Louis, MO, USA), in line with the manufacturer's instructions. The E-11 cell line (catalogue no.01110916) derived from snakehead fish fry (Channa striata) was used as a reference for the Mycoplasma contamination screening. Positive and negative controls were included in the analysis. The PCR amplicons were separated via 1.5% agarose gel electrophoresis and visualized using a ChemiDoc Imaging System (Bio-Rad, Hercules, CA, USA).
To detect the potential contamination of the snakehead retrovirus in the E-11 and RHTiB cell lines, the total RNA was extracted using GENEzol ® Reagent (Geneaid, Taipei, Taiwan), in accordance with the manufacturer's protocols. The RNA concentration was adjusted to 200 ng/µL with nuclease-free water, and the first-strand cDNA was synthesized using a ReverTra Ace ® kit (Toyobo, Osaka, Japan), according to the manufacturer's recommended protocol. PCR was carried out according to Nishizawa et al. [27] with forward (5 ′ -TGGTACCCATGGATACAGGTACCTCA-3 ′ ) and reverse primers (5 ′ -TGTCAGACATGGCCTGTACT) in a T100 thermocycler (Bio-Rad, Hercules, CA, USA). The PCR products were visualized via 1.5% agarose gel electrophoresis and analyzed using the ChemiDoc Imaging System (Bio-Rad, Hercules, CA, USA).
Chromosome Analysis of RHTiB Cells
The chromosome analysis was prepared using the RHTiB cells at passage 25. The cell cultures at 60-70% confluence were treated with 10 µg/mL of KaryoMAX™ Colcemid™ Solution (Invitrogen, ThermoFisher Scientific, Waltham, MA, USA) for 40 min. The cells were subsequently digested, centrifuged, and resuspended in a 0.075 M potassium chloride hypotonic solution for 20 min and fixed in a cold methanol and acetic acid mixture at a 3:1 ratio. The fixed cells were dropped vertically onto glass slides, stained with 5% Giemsa solution for 5 min, and the metaphase chromosomes were observed under a light microscope at 100× magnification using an Olympus VS120 Slide Scanner (Olympus, Tokyo, Japan).
Immunofluorescent Assay to Detect TiLV in RHTiB Cell Line
We used an immunofluorescent assay (IFA) to confirm the susceptibility of the RHTiB cell lines to TiLV. Briefly, 1 × 10 5 of RHTiB cells were plated on a cell culture chamber slide (SPL Life Sciences, Pocheon-si, Republic of Korea) and allowed to propagate in L-15 medium containing 5% FBS at 25 • C until 80-90% confluency was reached. After washing, the cells were inoculated with the TiLV strain VETKU-TV08 at 0.1 MOI or a control medium for 1 h at 25 • C. The cells were then rinsed and incubated with 2% FBS L-15 medium at 25 • C for an additional 24 h. After fixation with ice-cold 100% methanol for 10 min and permeabilization with 0.3% triton-X 100 for 10 min, a blocking solution (2% bovine serum albumin in PBS) was applied at 25 • C for 30 min. The cells were then probed with rabbit polyclonal antibodies against TiLV [28] in blocking solution at a dilution of 1:100 overnight at 4 • C. The cells were subsequently rinsed three times with PBS and incubated with a secondary antibody (Goat Anti-Rabbit IgG-Alexa Fluor™ 488, Abcam, Carlsbad, CA, USA) in blocking solution at a dilution of 1:500 for 1 h at room temperature. The cellular nuclei were stained with diaminophenylindole (DAPI) at a final concentration of 1 µg/mL (Sigma-Aldrich, St Louis, MO, USA), before visualization under a confocal microscope (Fluoview 3000, Olympus, Tokyo, Japan).
Propagation of TiLV in RHTiB Cell Line under Varying pH and Temperature Conditions
A volume of 500 µL TiLV was inoculated into each well of a 24-well cell culture plate containing confluent RHTiB cells. After incubation for 1 h at 25 • C, the viral suspensions were replaced with L-15 medium containing 2% FBS per well. To investigate the impact of varying pH levels (7, 7.2, 7.4, 7.6, and 7.8) and temperatures (25 • C, 28 • C, and 30 • C), the L-15 medium was added to the 24-well cell culture plates under different pH levels and temperatures. The propagation of TiLV under the different cell culture conditions was confirmed using RT-qPCR on days 1, 5, 7, and 9. The total RNA was extracted from the cell culture supernatant using GENEzol ® Reagent (Geneaid, Taipei, Taiwan) and processed for RT-qPCR analysis in line with the published protocol of Yamkasem et al. [29]. The cytopathic effects (CPE) were monitored regularly through an inverted microscope (CKX53, Olympus, Tokyo, Japan) at 40× magnification, with the E-11 cell line as a positive control.
Statistical Analysis
The data on cell growth and virus propagation were analyzed via two-way ANOVA followed by Tukey's multiple comparisons post hoc tests using the GraphPad Prism soft-ware, version 8.3.0 (GraphPad, San Diego, CA, USA). Statistical significance was considered at p < 0.05.
Primary Cell Isolation and Culture
We attempted to isolate primary cell lines from various organs of red hybrid tilapia, including the brain, spleen, heart, fins, and ocular muscles. Interestingly, only the cells derived from the brain exhibited continuous growth without degenerative or morphological changes (Figure 1). The cells from most of the other tissues showed limited proliferation and viability during the 14 days of the primary culture (Figure S1). Consequently, for further experimentation, the brain-derived cells were selected and designated as the RHTiB cell line. When establishing the primary cell culture, the brain tissue explants showed rapid cell proliferation and formed a monolayer of cells (1.2 × 10 6 cells/mL) within 2-3 weeks of tissue seeding (Figure 1A). The cells subsequently reached 80-90% confluence in successive passages within 7-9 days of subculturing (Figure 1B). The RHTiB cells exhibited a fibroblast-like morphology during their early growth phase, and this remained stable through the subsequent passages. This cell line has been maintained for 1.5 years and has undergone over 50 subcultures.
Primary Cell Isolation and Culture
We attempted to isolate primary cell lines from various organs of red hybrid tilapia, including the brain, spleen, heart, fins, and ocular muscles. Interestingly, only the cells derived from the brain exhibited continuous growth without degenerative or morphological changes (Figure 1). The cells from most of the other tissues showed limited proliferation and viability during the 14 days of the primary culture (Figure S1). Consequently, for further experimentation, the brain-derived cells were selected and designated as the RHTiB cell line. When establishing the primary cell culture, the brain tissue explants showed rapid cell proliferation and formed a monolayer of cells (1.2 × 10 6 cells/mL) within 2-3 weeks of tissue seeding (Figure 1A). The cells subsequently reached 80-90% confluence in successive passages within 7-9 days of subculturing (Figure 1B). The RHTiB cells exhibited a fibroblast-like morphology during their early growth phase, and this remained stable through the subsequent passages. This cell line has been maintained for 1.5 years and has undergone over 50 subcultures.
The viability of the cryopreserved RHTiB cells was assessed after 30 days in storage. Remarkably, approximately 75-80% of the initial RHTiB cell population was successfully revived and demonstrated optimal growth when cultured in L-15 medium supplemented with 20% FBS. The revived cells achieved confluency within 7 days of seeding in a 25 cm 2 cell culture flask and displayed no observable changes in cell morphology or detachment following cryopreservation and thawing.
Optimization of RHTiB Cell Growth Conditions
The optimal growth conditions for the RHTiB cells were investigated by examining the effects of the FBS concentration, pH, and temperature. We found that the RHTiB cells grew well at 25-30 °C in the L-15 medium supplemented with 20% FBS at pH 7.4. Specifically, we observed a direct link between the FBS concentration and the growth rate. In 20% FBS, the RHTiB cells demonstrated the highest growth rate, with an increase from 3.3 × 10 5 cells/mL to 1.35 × 10 6 cells/mL within 7 days, while the cultures with 5-15% FBS exhibited slower growth at 1.8 × 10 5 cells/mL to 1.05 × 10 6 cells/mL at 7 days of primary The viability of the cryopreserved RHTiB cells was assessed after 30 days in storage. Remarkably, approximately 75-80% of the initial RHTiB cell population was successfully revived and demonstrated optimal growth when cultured in L-15 medium supplemented with 20% FBS. The revived cells achieved confluency within 7 days of seeding in a 25 cm 2 cell culture flask and displayed no observable changes in cell morphology or detachment following cryopreservation and thawing.
Optimization of RHTiB Cell Growth Conditions
The optimal growth conditions for the RHTiB cells were investigated by examining the effects of the FBS concentration, pH, and temperature. We found that the RHTiB cells grew well at 25-30 • C in the L-15 medium supplemented with 20% FBS at pH 7.4. Specifically, we observed a direct link between the FBS concentration and the growth rate. In 20% FBS, the RHTiB cells demonstrated the highest growth rate, with an increase from 3.3 × 10 5 cells/mL to 1.35 × 10 6 cells/mL within 7 days, while the cultures with 5-15% FBS exhibited slower growth at 1.8 × 10 5 cells/mL to 1.05 × 10 6 cells/mL at 7 days of primary culture (dpc; Figure 2A). The optimal growth of the RHTiB cells occurred at pH 7.4, where the cell numbers increased from 4.1 × 10 5 cells/mL to 1.33 × 10 6 cells/mL at 7 dpc with no observed morphological changes or detachment (Figure 2B). Remarkably, despite showing higher live cell counts of 1.58 × 10 6 cells/mL to 1.96 × 10 6 cells/mL at 7 dpc, increasing the pH from 7.6 to 8 led to noticeable cell shrinkage and detachment. Finally, on the same day, the number of RHTiB cells cultured at 25 • C, 28 • C, and 30 • C did not differ significantly (p > 0.05; Figure 2C). Specifically, the number of RHTiB cells ranged from 5.5 × 10 4 to 9.15 × 10 4 cells/mL at day 1, 1.75 × 10 5 to 1.8 × 10 5 cells/mL at day 3, 2.7 × 10 5 to 2.87 × 10 5 cells/mL at day 5, and 3.01 × 10 5 to 3.42 × 10 5 cells/mL at day 7, within these temperatures.culture (dpc; Figure 2A). The optimal growth of the RHTiB cells occurred at pH 7.4, where the cell numbers increased from 4.1 × 10 5 cells/mL to 1.33 × 10 6 cells/mL at 7 dpc with no observed morphological changes or detachment (Figure 2B). Remarkably, despite showing higher live cell counts of 1.58 × 10 6 cells/mL to 1.96 × 10 6 cells/mL at 7 dpc, increasing the pH from 7.6 to 8 led to noticeable cell shrinkage and detachment. Finally, on the same day, the number of RHTiB cells cultured at 25 °C, 28 °C, and 30 °C did not differ significantly (p > 0.05; Figure 2C). Specifically, the number of RHTiB cells ranged from 5.5 × 10 4 to 9.15 × 10 4 cells/mL at day 1, 1.75 × 10 5 to 1.8 × 10 5 cells/mL at day 3, 2.7 × 10 5 to 2.87 × 10 5 cells/mL at day 5, and 3.01 × 10 5 to 3.42 × 10 5 cells/mL at day 7, within these temperatures.
Species Identification of RHTiB Cell Line Using cox1 Gene
We evaluated the origin of the RHTiB cell line through the amplification and sequencing of partial fragments of the cox1 gene. Our analysis of the cox1 gene fragments from the RHTiB cells revealed the expected PCR products at 650 bp (Figure 3A). Subsequent BLAST analysis showed significant sequence similarity of 99.22% between the cox1 nucleotide sequence of the RHTiB and Nile tilapia (GenBank accession number MK355381.1). Furthermore, the genetic distances of the cox1 DNA sequence between the RHTiB and Nile tilapia were 0.01 (Table S1). The phylogenetic analysis constructed using the cox1 nucleotide sequence placed the RHTiB cell line within the Oreochromis genus, with marked divergence from both Nile tilapia and other tilapia species (Figure 3B). The nucleotide sequence of the cox1 gene of RHTiB was deposited at the NCBI GenBank under the accession number OQ351327. Further chromosome analysis revealed that more than 75% of the RHTiB cells had a diploid karyotype during the metaphase, with normal chromosomes at 2n = 44 (Figures 4 and S2).
Species Identification of RHTiB Cell Line Using cox1 Gene
We evaluated the origin of the RHTiB cell line through the amplification and sequencing of partial fragments of the cox1 gene. Our analysis of the cox1 gene fragments from the RHTiB cells revealed the expected PCR products at 650 bp (Figure 3A). Subsequent BLAST analysis showed significant sequence similarity of 99.22% between the cox1 nucleotide sequence of the RHTiB and Nile tilapia (GenBank accession number MK355381.1). Furthermore, the genetic distances of the cox1 DNA sequence between the RHTiB and Nile tilapia were 0.01 (Table S1). The phylogenetic analysis constructed using the cox1 nucleotide sequence placed the RHTiB cell line within the Oreochromis genus, with marked divergence from both Nile tilapia and other tilapia species (Figure 3B). The nucleotide sequence of the cox1 gene of RHTiB was deposited at the NCBI GenBank under the accession number OQ351327. Further chromosome analysis revealed that more than 75% of the RHTiB cells had a diploid karyotype during the metaphase, with normal chromosomes at 2n = 44 (
Detection of Mycoplasma and Snakehead Retrovirus Contamination
Mycoplasma and snakehead retrovirus contamination in the RHTiB cell line were con firmed through PCR amplification with the E-11 cell line as a reference. The supernatant collected from both the RHTiB and E-11 cell lines cultured in L-15 media did not exhibi any amplification corresponding to the Mycoplasma-specific 259 bp amplicon, as observed in the positive control (Figure 5A). Additionally, for the snakehead retrovirus contamina tion screening, no amplification was detected in the RHTiB cells. However, a specific am plicon of approximately 730 bp was shown in the E-11 cell line (Figure 5B).
Detection of TiLV Infection in RHTiB Using Immunofluorescence
We evaluated the susceptibility of the RHTiB cell line to TiLV infection using an im munofluorescence technique. The specific rabbit polyclonal antibody against TiLV, la beled with Alexa Fluor™ 488, was incubated to detect the localization of TiLV in infected cells. Although no clear CPE was observed in the infected RHTiB cells (Figure 6A), which was in contrast to the clear CPE in the E-11 cells (Figure 6B), a strong fluorescence signa was observed in the cytoplasm of the RHTiB cells at 24 h post-TiLV inoculation (Figure
Detection of Mycoplasma and Snakehead Retrovirus Contamination
Mycoplasma and snakehead retrovirus contamination in the RHTiB cell line were confirmed through PCR amplification with the E-11 cell line as a reference. The supernatants collected from both the RHTiB and E-11 cell lines cultured in L-15 media did not exhibit any amplification corresponding to the Mycoplasma-specific 259 bp amplicon, as observed in the positive control (Figure 5A). Additionally, for the snakehead retrovirus contamination screening, no amplification was detected in the RHTiB cells. However, a specific amplicon of approximately 730 bp was shown in the E-11 cell line (Figure 5B).
x FOR PEER REVIEW 8 of 14
Detection of Mycoplasma and Snakehead Retrovirus Contamination
Mycoplasma and snakehead retrovirus contamination in the RHTiB cell line were confirmed through PCR amplification with the E-11 cell line as a reference. The supernatants collected from both the RHTiB and E-11 cell lines cultured in L-15 media did not exhibit any amplification corresponding to the Mycoplasma-specific 259 bp amplicon, as observed in the positive control (Figure 5A). Additionally, for the snakehead retrovirus contamination screening, no amplification was detected in the RHTiB cells. However, a specific amplicon of approximately 730 bp was shown in the E-11 cell line (Figure 5B).
Detection of TiLV Infection in RHTiB Using Immunofluorescence
We evaluated the susceptibility of the RHTiB cell line to TiLV infection using an immunofluorescence technique. The specific rabbit polyclonal antibody against TiLV, labeled with Alexa Fluor™ 488, was incubated to detect the localization of TiLV in infected cells. Although no clear CPE was observed in the infected RHTiB cells (Figure 6A), which was in contrast to the clear CPE in the E-11 cells (Figure 6B), a strong fluorescence signal was observed in the cytoplasm of the RHTiB cells at 24 h post-TiLV inoculation (Figures
Detection of TiLV Infection in RHTiB Using Immunofluorescence
We evaluated the susceptibility of the RHTiB cell line to TiLV infection using an immunofluorescence technique. The specific rabbit polyclonal antibody against TiLV, labeled with Alexa Fluor™ 488, was incubated to detect the localization of TiLV in infected cells. Although no clear CPE was observed in the infected RHTiB cells (Figure 6A), which was in contrast to the clear CPE in the E-11 cells (Figure 6B), a strong fluorescence signal was observed in the cytoplasm of the RHTiB cells at 24 h post-TiLV inoculation (Figures 6D and S2). No TiLV fluorescence signal was observed in the uninfected RHTiB cells (Figure 6C).
Propagation of TiLV in RHTiB Cells under Different pH and Temperature Conditions
The RHTiB cells were exposed to TiLV under various culture conditions to assess the influence of the pH and temperature on the propagation of the virus. The cultures were maintained at various pH levels (7, 7.2, 7.4, 7.6, and 7.8) and temperatures (25 • C, 28 • C, and 30 • C) for 9 days. The viral copy numbers were quantified using RT-qPCR analysis. On day 7 following TiLV inoculation, a significant (p < 0.05) increase in the viral copy numbers was observed at pH 7.6 (10 7.32 ± 0.01 viral copies/400 ng cDNA) and 7.8 (10 7.30 ± 0.27 viral copies/400 ng cDNA) compared to pH 7 (10 5.63 ± 0.27 viral copies/400 ng cDNA), 7.2 (10 5.55 ± 0.02 viral copies/400 ng cDNA), and 7.4 (10 5.71 ± 0.13 viral copies/400 ng cDNA). However, on day 9, the highest TiLV copy numbers (10 7.82 ± 0.22 viral copies/400 ng cDNA) were detected at pH 7, while the lowest levels (10 7.00 ± 0.13 viral copies/400 ng cDNA) were recorded at pH 7.6 (Figure 7A).differences (p > 0.05) were observed in the TiLV copy numbers in the RHTiB cell line when exposed to temperatures that varied between 25 °C and 30 °C (Figure 7B). Although the highest TiLV copy number was detected on day 9 at 30 °C (10 6.51 ± 0.49 viral copies/400 ng cDNA), there was no statistically significant difference among the temperatures on this day.
Discussion
The recent outbreaks of TiLV in several countries and the inclusion of TiLV among the World Organisation for Animal Health's listed diseases has highlighted the need for TiLV disease surveillance and the development of diagnostic tools. Cell culture is considered the gold standard technique for virus isolation but requires cell lines specific to each virus. In this study, we isolated, characterized, and tested the susceptibility of new cell lines from red hybrid tilapia for TiLV propagation. Red hybrid tilapia, which is commonly cultured in Southeast Asia, is highly susceptible to TiLV. TiLV infection often leads to high morbidity and mortality rates [9]. It is detectable in various organs, including the liver, With respect to the influence of the temperature on TiLV propagation, no significant differences (p > 0.05) were observed in the TiLV copy numbers in the RHTiB cell line when exposed to temperatures that varied between 25 • C and 30 • C (Figure 7B). Although the highest TiLV copy number was detected on day 9 at 30 • C (10 6.51 ± 0.49 viral copies/400 ng cDNA), there was no statistically significant difference among the temperatures on this day.
Discussion
The recent outbreaks of TiLV in several countries and the inclusion of TiLV among the World Organisation for Animal Health's listed diseases has highlighted the need for TiLV disease surveillance and the development of diagnostic tools. Cell culture is considered the gold standard technique for virus isolation but requires cell lines specific to each virus. In this study, we isolated, characterized, and tested the susceptibility of new cell lines from red hybrid tilapia for TiLV propagation. Red hybrid tilapia, which is commonly cultured in Southeast Asia, is highly susceptible to TiLV. TiLV infection often leads to high morbidity and mortality rates [9]. It is detectable in various organs, including the liver, spleen, head kidney, gills, heart, intestines, and gonads of infected tilapia, with the liver and brain as the major target tissues of this virus [9,30,31]. Establishing a tilapia-susceptible cell line from this species would therefore aid in future studies of the pathogenesis of TiLV [16,19]. Several tilapia cell lines already exist, including TiB [19], On1B [18], and TA-02 [22], from O. niloticus and OmB from O. mossambicus [32]. In the present article, we report the first brain cell line isolated from red hybrid tilapia, namely the RHTiB cell line, using the tissue explant method. The brain tissue explants demonstrated rapid proliferation during the early growth phase. A monolayer of cells formed within 2-3 weeks of seeding and exhibited a homogeneous population of fibroblast-like cells. Similar findings have been reported in previous studies, which have described a fibroblast dominance in the primary cell cultures derived from tilapia brain tissue [18][19][20][21][22]. Although we successfully isolated primary cells from the brain of red tilapia, our attempts to isolate primary cells from other organs, including the spleen, heart, fins, muscles, liver, and gills, have not yet been successful. The failure to propagate and sustain these primary cells may be influenced by various factors, such as unsuitable culture conditions, the nature of each cell type, the initial cell seeding density, and the choice of growth factors and enzymes used during the subculturing processes [33,34]. Optimizing these factors is important for future efforts aimed at isolating primary cells, which could serve as a tool to investigate the pathobiology of TiLV.
To verify the origin of the RHTiB cells, we amplified and sequenced the mitochondrial cox1 gene, which is a well-established DNA marker [35]. Additionally, chromosomal analysis confirmed a normal diploid karyotype (2n = 44) consistent with Oreochromis spp. The RHTiB cells were also free of Mycoplasma contamination and showed promising cryopreservation capabilities, with revival efficiencies reaching 80%, which is comparable to those of other established fish cell lines [18,19,36]. Interestingly, the growth of the RHTiB cells was significantly influenced by the FBS concentration and pH but not the temperature. Similar to other tilapia cell lines [18,22] and cell lines derived from other fish species, including striped catfish (Pangasianodon hypophthalmus) [37] and catfish (Clarias dussumieri) [38], the RHTiB cells showed maximal growth in the L-15 medium supplemented with 20% FBS. FBS is a complex mixture of nutrients essential for cell growth and thus likely provides an optimal environment for faster proliferation at higher concentrations [39,40]. However, in the interest of long-term cost-effectiveness, we found that 10% FBS was sufficient for RHTiB cell propagation, which is consistent with the findings of a previous study [19]. Similar to primary Nile tilapia cells [19], the optimal pH for RHTiB cell growth was 7.4, which is the physiological pH that supports the optimal activity of most cellular protein and enzyme activities. While tilapia can tolerate a wider pH range between 6 and 8.5 [41], we observed cell shrinkage and detachment at higher pH levels, possibly due to alterations in the cell membrane properties [42]. Nevertheless, additional conditions for cell and virus propagation including different culture media, CO 2 concentrations, and humidity should be explored for the new cell line.
Similar to the optimal temperatures reported for other tilapia brain cells, the RHTiB cells demonstrated robust growth across a temperature range of 25 • C to 30 • C [19,22]. As a poikilothermic species, tilapia naturally adapts to water temperatures within this range, which is relevant to the observed optimal growth of the RHTiB cells. Similar observations of optimal growth within this temperature range have been reported for various tropical fish species, including tilapia [22,43]. Unlike the E-11 and other tilapia brain-derived cell lines, the RHTiB cells did not exhibit CPE following TiLV inoculation [16,19]. However, both IFA and RT-qPCR revealed RHTiB susceptibility and permissiveness to TiLV propagation. Similarly, in previous studies, IFA was applied to detect TiLV in Nile tilapia-derived TiB cells [16,19] and Mozambique tilapia-derived bulbus arteriosus cells [44], which suggests the potential of these cell lines in further TiLV entry mechanism research.
Further investigations revealed that the pH could impact TiLV propagation in RHTiB cells. We noted a significant increase in TiLV replication within the L-15 medium at a pH ranging from 7 to 7.8, with the highest TiLV concentration observed on day 9 at pH 7. Similarly, despite high variation, the highest TiLV copy number observed in this study was detected on day 9 at 30 • C. While the different tested temperatures did not exhibit significant differences, these findings are consistent with previous reports of optimal TiLV propagation at temperatures between 25 • C and 30 • C [45]. This highlights the potential of the RHTiB cell line as a valuable tool in studying the dynamics of TiLV infection in tilapia. The ability of RHTiB cells to support TiLV replication without displaying CPE makes them a unique resource for the investigation of host cell-virus interactions during prolonged infections and opens up opportunities for future TiLV research.
Conclusions
We successfully established and characterized the RHTiB cell line, which is the first reported brain cell line derived from red hybrid tilapia. This new cell line holds great potential for various applications, including TiLV detection and vaccine development, and as a tool to study host-virus interactions.
Supplementary Materials:
The following supporting information can be downloaded at: [URL]1, Figure S1. Primary cells derived from the fins, ocular muscles, heart, and spleen on days 8 and 14, respectively. The scale bar represents 200 µm. S1. Genetic distance estimates among fish species using cox1 nucleotide sequences.
Figure 1 .
Figure 1. Establishment of the RHTiB cell line derived from the brain tissue of red hybrid tilapia, Oreochromis spp.(A) The outgrowth of RHTiB cells from a tissue explant on day 3 (40× magnification).(B) A confluent monolayer of RHTiB cells (40× magnification).
Figure 1 .
Figure 1. Establishment of the RHTiB cell line derived from the brain tissue of red hybrid tilapia, Oreochromis spp.(A) The outgrowth of RHTiB cells from a tissue explant on day 3 (40× magnification).(B) A confluent monolayer of RHTiB cells (40× magnification).
Figure 2 .
Figure 2. Growth characteristics of RHTiB cells under various culture conditions over a 7-day period.(A) Growth of RHTiB cells under different concentrations of fetal bovine serum (FBS; %) in L-15 medium.(B) Growth of RHTiB cells under different pH levels in L-15 medium.(C) Growth of RHTiB cells at different incubation temperatures (°C) in L-15 medium. The cells were incubated without CO2 in an incubator. The mean values with the standard error of the mean (SE) are presented (n = 3), and significant differences are denoted as * p < 0.05, ** p < 0.01, and *** p < 0.001.
Figure 2 .
Figure 2. Growth characteristics of RHTiB cells under various culture conditions over a 7-day period.(A) Growth of RHTiB cells under different concentrations of fetal bovine serum (FBS; %) in L-15 medium.(B) Growth of RHTiB cells under different pH levels in L-15 medium.(C) Growth of RHTiB cells at different incubation temperatures ( • C) in L-15 medium. The cells were incubated without CO 2 in an incubator. The mean values with the standard error of the mean (SE) are presented (n = 3), and significant differences are denoted as * p < 0.05, ** p < 0.01, and *** p < 0.001.
Figure 3 .
Figure 3. Molecular characterization of RHTiB cell line through amplification of mitochondrial gene.(A) Polymerase chain reaction (PCR) amplification of a partial nucleotide sequence of the cox1 gene from the RHTiB cell line. MW = 100 bp molecular weight marker, lane 1 = 650 bp PCR amplicon from RHTiB cells.(B) Phylogenetic analysis based on the cox1 nucleotide sequence of the RHTiB cell line and a sequence from other fish species using the Molecular Evolutionary Genetics Analysis X tool. The analysis included 18 nucleotide sequences in the final dataset. The GenBank accession number of the reference sequence is provided before each fish name in the phylogenetic tree. The numbers at the nodes show the bootstrap values obtained from 1000 replicates.
Figure 3 .
Figure 3. Molecular characterization of RHTiB cell line through amplification of mitochondrial gene.(A) Polymerase chain reaction (PCR) amplification of a partial nucleotide sequence of the cox1 gene from the RHTiB cell line. MW = 100 bp molecular weight marker, lane 1 = 650 bp PCR amplicon from RHTiB cells.(B) Phylogenetic analysis based on the cox1 nucleotide sequence of the RHTiB cell line and a sequence from other fish species using the Molecular Evolutionary Genetics Analysis X tool. The analysis included 18 nucleotide sequences in the final dataset. The GenBank accession number of the reference sequence is provided before each fish name in the phylogenetic tree. The numbers at the nodes show the bootstrap values obtained from 1000 replicates.
Figure 5 .
Figure 5. Screening of Mycoplasma and the snakehead retrovirus in the E-11 and RHTiB cell lines (A) Mycoplasma screening of the E-11 and RHTiB cell lines. A distinct 481 bp band on the agaros gel during Mycoplasma screening indicated the presence of internal control DNA, demonstrating th expected performance from the test kit.(B) Snakehead retrovirus screening of the E-11 and RHTiB cell lines.
Figure 5 .
Figure 5. Screening of Mycoplasma and the snakehead retrovirus in the E-11 and RHTiB cell lines.(A) Mycoplasma screening of the E-11 and RHTiB cell lines. A distinct 481 bp band on the agarose gel during Mycoplasma screening indicated the presence of internal control DNA, demonstrating the expected performance from the test kit.(B) Snakehead retrovirus screening of the E-11 and RHTiB cell lines.
Figure 5 .
Figure 5. Screening of Mycoplasma and the snakehead retrovirus in the E-11 and RHTiB cell lines.(A) Mycoplasma screening of the E-11 and RHTiB cell lines. A distinct 481 bp band on the agarose gel during Mycoplasma screening indicated the presence of internal control DNA, demonstrating the expected performance from the test kit.(B) Snakehead retrovirus screening of the E-11 and RHTiB cell lines.
Figure 6 .
Figure 6. Photomicrographs illustrating the cytopathic effects (CPE) in the cell lines infected with tilapia lake virus (TiLV) and the immunofluorescence staining of the TiLV-infected RHTiB cells.(A) The RHTiB cell lines infected with TiLV at 4 days post-infection (dpi), showing no observable CPE formation (40× magnification).(B) The E-11 cell line infected with TiLV at 4 dpi, displaying CPE in the form of cell lysis (black arrow), cell rounding, clumping, and the formation of syncytial cells (40× magnification).(C) Uninfected RHTiB cells showing diaminophenylindole (DAPI) staining of the nuclei of the normal cells.(D) Discrete green fluorescence signals (white arrows) in the cytoplasm of infected cells after 24 h of virus inoculation. The blue color indicates the DAPI staining of the nucleus. The scale bar represents 20 µm.
Figure 6 .
Figure 6. Photomicrographs illustrating the cytopathic effects (CPE) in the cell lines infected with tilapia lake virus (TiLV) and the immunofluorescence staining of the TiLV-infected RHTiB cells.(A) The RHTiB cell lines infected with TiLV at 4 days post-infection (dpi), showing no observable CPE formation (40× magnification).(B) The E-11 cell line infected with TiLV at 4 dpi, displaying CPE in the form of cell lysis (black arrow), cell rounding, clumping, and the formation of syncytial cells (40× magnification).(C) Uninfected RHTiB cells showing diaminophenylindole (DAPI) staining of the nuclei of the normal cells.(D) Discrete green fluorescence signals (white arrows) in the cytoplasm of infected cells after 24 h of virus inoculation. The blue color indicates the DAPI staining of the nucleus. The scale bar represents 20 µm.
Figure 7 .
Figure 7. Quantification of the tilapia lake virus (TiLV) copy number (log10 viral copies/400 ng of cDNA) in RHTiB cell lines cultured under different conditions.(A) Viral copies in the RHTiB cell line at five pH levels.(B) Viral copies in the RHTiB cell line grown at three different incubation temperatures. The values indicate significant differences (*** p < 0.001 and **** p < 0.0001) and are presented as the means ± standard error of the mean (SE; n = 3).
Figure 7 .
Figure 7. Quantification of the tilapia lake virus (TiLV) copy number (log 10 viral copies/400 ng of cDNA) in RHTiB cell lines cultured under different conditions.(A) Viral copies in the RHTiB cell line at five pH levels.(B) Viral copies in the RHTiB cell line grown at three different incubation temperatures. The values indicate significant differences (*** p < 0.001 and **** p < 0.0001) and are presented as the means ± standard error of the mean (SE; n = 3).
Figure S2. Additional representative images of RHTiB cells (A) Chromosome analysis of a RHTiB cell with Giemsa staining, observed at a magnification of 100×.(B) Fluorescence images of RHTiB cells infected with TiLV at 24 h post infection.40× magnification. Table
==
Domain: Agricultural and Food Sciences Biology Environmental Science
|
[{"added": "2024-05-24T15:04:44.399Z", "created": "2024-05-22T00:00:00.000", "id": "269979930", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2076-2615/14/11/1522/pdf?version=1716356211", "pdf_hash": "f57527c21bab7a5f7cb226bfcd69fc0048402e69", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:9723", "s2fieldsofstudy": ["Biology", "Environmental Science", "Agricultural and Food Sciences"], "sha1": "34d0406a07b182255041eee1699d1be7155ff52f", "year": 2024}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
CoIN: co-inducible nitrate expression system for secondary metabolites in Aspergillus nidulans
Sequencing of fungal species has demonstrated the existence of thousands of putative secondary metabolite gene clusters, the majority of them harboring a unique set of genes thought to participate in production of distinct small molecules. Despite the ready identification of key enzymes and potential cluster genes by bioinformatics techniques in sequenced genomes, the expression and identification of fungal secondary metabolites in the native host is often hampered as the genes might not be expressed under laboratory conditions and the species might not be amenable to genetic manipulation. To overcome these restrictions, we developed an inducible expression system in the genetic model Aspergillus nidulans. We genetically engineered a strain of A. nidulans devoid of producing eight of the most abundant endogenous secondary metabolites to express the sterigmatocystin Zn(II)2Cys6 transcription factor-encoding gene aflR and its cofactor aflS under control of the nitrate inducible niiA/niaD promoter. Furthermore, we identified a subset of promoters from the sterigmatocystin gene cluster that are under nitrate-inducible AflR/S control in our production strain in order to yield coordinated expression without the risks from reusing a single inducible promoter. As proof of concept, we used this system to produce β-carotene from the carotenoid gene cluster of Fusarium fujikuroi. Utilizing one-step yeast recombinational cloning, we developed an inducible expression system in the genetic model A. nidulans and show that it can be successfully used to produce commercially valuable metabolites.
Background
Natural products or secondary metabolites (SMs) have been invaluable as platforms for developing front-line drugs. Between 1981 and 2010, 5% of the 1031 new chemical entities approved as drugs by the Food and Drug Administration (FDA) were natural products or derivatives, including 48.6% of cancer medications [36]. In addition, SMs are major sources of innovative therapeutic agents for both bacterial and fungal infectious diseases, lipid disorders, and immunomodulation [16]. Fungal SMs have proven to be a particularly important source of new leads with useful pharmaceutical activities. A literature survey of fungal metabolites, covering 1500 fungal SMs that were isolated and characterized between 1993 and 2001, showed that more than half of the molecules had antibacterial, antifungal or antitumor activity [41]. However, the full metabolic potential of the majority of existing fungal species has not been investigated. Major roadblocks in this endeavor are that some species are not cultivable under laboratory conditions and/or their SM gene clusters are silent. Previous strategies on activating fungal SMs have focused mainly on (1) activating endogenous gene clusters by over-expressing the pathway-specific transcription factor [13,43,54]), (2) manipulating global regulators [11,28], and (3) expressing the entire gene cluster in a heterologous host [8]. Although successful in some cases, these strategies have significant disadvantages. As not all fungal species are easily amenable to genetic manipulation, strategies that focus on endogenous activation are impossible in these species. This prevents the option of over-expressing a cluster-specific transcription factor, which has been the most successful approach to activating cryptic clusters thus far (reviewed in [51]). In addition, not all SM clusters contain transcription factors and although some clusters have been activated by overexpressing every gene in the cluster [14,53], this adds labor and time to the process and may be hard to achieve with clusters containing many genes.
Previous approaches expressing fungal gene clusters in heterologous hosts (mainly Saccharomyces cerevisiae or Aspergillus spp.) focused on amplification of the entire gene cluster including native promoters. Although these approaches lead to expression of the targeted gene clusters in some cases [55], the use of native promoters cannot guarantee controlled activation of the genes. Therefore, the identification and use of defined promoters presents an alternative means to activate clusters. Cloning of entire gene clusters can be achieved by PCR-based amplification of the desired DNA region and subsequent yeast recombination-based cloning [55]. Promoter exchanges using this technique rely on the identification of different promoter regions as the use of identical promoter sequences is impossible due to the homologous recombination among promoters [14].
Our goal was to identify a series of distinct promoters that could be activated in one step and, furthermore, activated under an inducible system as many SMs exhibit antifungal properties that could be toxic to the heterologous host [12,46]. Thus, we designed a strain of genetic model organism Aspergillus nidulans that contains an inducible genetic construct which allows for expression of the positive acting transcriptional elements of the sterigmatocystin (ST) gene cluster, aflR and aflS (formerly aflJ, [20]). The ST gene cluster contains 25 distinct genes and it is known that the transcription factor AflR and its cofactor AflS are responsible for ST production [9]. We constructed a strain of A. nidulans with its endogenous ST cluster removed but with aflR/aflS placed back into the strain under the control of a nitrate inducible divergent promoter (niiA(p)/niaD(p)) [44,45], thereby allowing controlled aflR/aflS expression based on culture conditions. We tested the expression of all 25 ST promoters by AflR/AflS in this strain and identified eight ST promoters specifically regulated by nitrate induction of aflR/S. To test the system for expression of a fungal secondary metabolite, we cloned the carotenoid gene cluster from Fusarium fujikuroi and placed it under control of these inducible ST promoters. We show that the derived A. nidulans transformants produce β-carotene in competitive levels to existing systems using our technology.
Fungal strains and culture conditions
Aspergillus nidulans strains used in this study are listed in Additional file 1: Table S1. Fusarium fujikuroi IMI58289 [51] was used for carRA, carB, and ggs1 amplification as a reference for carotenoid production. Strains were maintained as glycerol stocks and activated on solid glucose minimal medium (GMM) at 37 °C with appropriate supplements [48]. For experiments in Fig. 1, nitrate was replaced with equimolar ammonium tartrate and Fig. 1 Nitrate-inducible aflR/S expression in A. nidulans. a Schematic overview of the aflR/S expression strain TPMW2.3 that harbors the nitrateinducible niaD/niiA promoter-driven aflR/S genes at the native sterigmatocystin cluster locus. b Northern blot analysis of nitrate-dependent aflR/S expression in A. nidulans TPMW2.3. Strains were grown in 50 mL of GMM with 35 mM glutamine as nitrogen source supplemented with 5 mM uracil/uridine and riboflavin for 24 h at 250 rpm at 37 °C. The mycelia were washed and shifted into new media containing either NH 4 + or NO 3 − as nitrogen sources plus supplements and grown at 250 rpm and 37 °C for 1 h before RNA extraction. Strains were grown in duplicate, indicated genes were probed and rRNA was visualized as loading control supplemented with 5 mM uracil and uridine, respectively. For solidified media, Noble Agar (Difco ™ , BD, USA) was added at 16 g/L. For pyrG auxotrophs, the growth medium was supplemented with 5 mM uridine and uracil. For riboB auxotrophs, the growth medium was supplemented with 5 mM riboflavin. For pyroA auxotrophs, the growth medium was supplemented with 5 mM pyridoxine. Conidia were harvested in 0.01% Tween 80 and enumerated using a hemocytometer. For RNA analysis, indicated strains were inoculated into 50 mL of liquid GMM with 35 mM glutamine as nitrogen source at 5 × 10 6 conidia/mL in duplicate and grown at 37 °C and 250 rpm for 24 h in ambient light conditions. The cultures were shifted into new GMM media either containing 70 mM nitrate or 35 mM glutamine as nitrogen source for 1 h. The mycelium was harvested and lyophilized before RNA extraction. For carotenoid production 5 × 10 6 conidia of indicated strains were inoculated on 20 mL liquid stationary GMM with either containing 70 mM nitrate or 35 mM glutamine as nitrogen source for 3 days at 37 °C (A. nidulans) or 29 °C (F. fujikuroi) in the dark. To distinguish between carotenoid production in the mycelia and spores, the strains were grown on liquid stationary GMM media (described above) for 3 days at 37 °C in the light to induce sporulation. Mycelia and spores were resuspended in 0.01% (v/v) Tween80, vortexed to separate spores from mycelia. Spores were separated from mycelia by filtration.
Yeast recombinational cloning
Yeast strain BJ5464 (MATalpha, ura3-52, trp1, leu2-∆1, his3-∆200, pep4::HIS3, prb1-∆1.6R, can1, GAL) was inoculated into 25-50 mL of appropriate media (2× YPDA) and incubated at 30 °C at 200 rpm overnight. The concentration of overnight culture was determined using OD 600 with a 1 × 10 7 cells/mL set to an OD reading of 1.0. Then, 1.25 × 10 9 cells were centrifuged at 3000×g for 5 min. Fresh media was added to the pelleted cells and added to a baffled flask containing 250 mL of 2× YPAD to a final concentration of 5 × 10 6 cells/mL and incubated at 30 °C at 200 rpm until the cell titer reached 2 × 10 7 cells/mL. Cells were harvested by centrifugation at 3000×g for 5 min. Supernatant was removed and the cells were washed with double distilled H 2 O (ddH 2 O). The cells were transferred to one 50 mL falcon tube and washed an additional time with ddH 2 O before they were centrifuged at 3000×g for 5 min. Cells were resuspended in 5% glycerol and 10% DMSO to a final concentration of 2 × 10 9 cells/mL aliquoted in 100 µL. These cells can be frozen at − 80 °C for several weeks. Before transformation, cells were pelleted and the supernatant removed. For transformation 250 ng of the digested backbone vector and 500 ng of each DNA PCR product (see below) were added and adjusted to a final volume of 14 µL with ddH 2 O. The DNA mixture was added to the yeast along with 260 µL of a 50% (w/v) polyethyleneglycol 3600, 36 µL 1 M lithium acetate and 50 µL of denatured sheared salmon sperm DNA (2 mg/mL). The mixture was vortexed and incubated at 42 °C for 45 min. Cells were centrifuged at 13,000×g for 30 s and the supernatant was removed. The cells were carefully resuspended in 1 mL ddH 2 O and 200-500 µL were spread on solidified synthetic drop out media containing all necessary supplements without uracil for selection. Plates were incubated at 30 °C for 3-5 days.
Plasmid isolation from yeast
All colonies from a transformation plate were scraped and incubated overnight in liquid synthetic yeast drop out solution containing the appropriate supplements without uracil at 200 rpm at 30 °C. One mL was pelleted and the supernatant removed. 200 µL of STC buffer (50 mM Tris-HCl pH 7.5, 1.2 M sorbitol, 50 mM CaCl 2 ) including 3 µL Zymolase was added and incubated at 37 °C for 1 h. To the mixture, 200 µL of 1% (w/v) sodium dodecyl sulfate (SDS) in 200 mM NaOH were added and inverted. The solution was neutralized by adding 240 µL of 3 M potassium acetate, pH 5.5 and inverted. The mixture was centrifuged and the supernatant mixed with 600 µL isopropanol, inverted and centrifuged at maximum speed for 10 min. The supernatant was removed and the pellet was washed with 70% (v/v) ethanol. The pellet was air dried and resuspended in 30 µL ddH 2 O.
Transformation of Escherichia coli and plasmid conformation
Following standard techniques [23], 10 µL of the yeast plasmid extract were transformed into E. coli and positive colonies were selected on media containing ampicillin. Plasmids from colonies were isolated using standard procedures [23]. Plasmids were restriction digested with appropriate enzymes to confirm correct insertion. Additional confirmation was achieved using PCR amplification of fused DNA products. To ensure correct DNA sequences for expression plasmids, Sanger sequencing was performed. The correct plasmids were then grown in a 50 mL culture and plasmids were isolated using the Quantum Prep ® Plasmid Midiprep Kit (Biorad) according to the manufacturers' instructions. Before fungal transformation, the plasmids were linearized using AscI.
Plasmid construction and fungal transformation
Expression fragments were created by yeast recombinational cloning as described above. All primers used are listed in Additional file 1: Table S2 and all plasmids are listed in Additional file 1: Table S3. For assembling the nitrate inducible aflR/S construct, six fragments total were amplified and eventually cloned into the AscI digested plasmid backbone of pYHC-yA-riboB [57]. The 3′ flanks of stcA and stcW were amplified from A. nidulans LO8030 DNA using primer pairs stcA3′-F/-R and stcW3′-F/-R with the -R primers containing 5′ overlaps to the respective site of AscI digested plasmid backbone and the -F primers having overlaps to the aflS terminator and pyroA cassette, respectively. The bidirectional niaD/niiA promoter region was amplified from A. nidulans LO8030 [15,38] with overlaps to the open reading frames of aflR and aflS using primer pairs nitrate-F/-R. The open reading frame of aflR including 500 bp of terminator was amplified from A. nidulans FGSC 4A DNA using primer pair aflR-F/-R where the -R primer had a 5′ overhang to the terminator region of the A. fumigatus pyroA gene. The open reading frame of aflS was amplified from A. nidulans FGSC 4A DNA using primer pairs aflS-F/-R with the -R primer having a 5′ overlap to the -F primer used to amplify the stcW flank. The pyroA cassette was retrieved through PstI restriction digest of pJMP61 [7]. After yeast recombinational cloning, the plasmid pPMW1 was created. For sterigmatocystin promoter studies the entire bidirectional promoter region between two open reading frames was cloned or, in the case of monodirectional promoters 500 bp upstream of the open reading frame was amplified using primer pairs stc "gene name"-pF/stc "gene name"-pR including 5′ overlaps to the wA 5′ flank and the open reading frame of pyrG gene from A. fumigatus CEA10. Plasmid pYHC-wA-pyrG [55] was linearized using NheI. After yeast recombineering, plasmid pAN "stcGene" were yielded. For constructing the carotenoid expression plasmid pJSF1, the bidirectional promoter region between stcA and stcB was amplified using primer pair stcAB-cF/-cR with overlaps to the carotenoid cluster genes carRA and carB from F. fujikuroi IMI58289. The open reading frame of carRA including 500 bp terminator region was amplified from F. fujikuroi IMI58289 DNA using primer pairs carRA-F/-R with the -R primer including an overlap to the wA 3′ flank of plasmid pYHC-wA-pyrG. The open reading frame of carB including 500 bp terminator region was amplified from F. fujikuroi IMI58289 DNA using primer pairs carB-cF/-cR with the -cR primer including an overlap to the wA 5′ flank of plasmid pYHC-wA-pyrG. All fragments were assembled using yeast recombinational cloning into EcoRI/XhoI linearized pYHC-wA-pyrG resulting in pJSF1. pJSF2 was assembled in a similar process using EcoRI/XhoI linearized pYHC-yA-riboB and PCR amplicons of the stcM promoter (amplified with stcM-cF/-cR) and ggs1 (amplified with ggs1-cF/-cR).
Transformation of A. nidulans was performed as previously described [40]. For selection of nitrate inducible aflR/S strains, A. nidulans LO8030 was used as the recipient strain. pPMW1 was linearized using AscI, transformed into LO8030, and transformants were selected on media where pyridoxine was omitted and uracil/uridine and riboflavin were supplemented yielding strain TPMW2.3. For selection of stc promoter test strains, TPMW2.3 was used as the recipient strain. pAN "stc-Gene" plasmids were linearized using SbfI and transformants were selected on media were riboflavin was omitted and uracil/uridine was supplemented yielding strains TANx and TAASx (see Additional file 1: Table S1). To create a riboflavin prototrophic strain, TPMW2.3 was transformed with SbfI linearized pYHC-yA-riboB and selected on media where riboflavin was omitted and uracil/uridine was supplemented yielding strain TPMW7.2. For selection of car expression strains, TPMW7.2 was used as the recipient strain and SbfI linearized pTJSF1 was transformed into TPMW7.2 and transformants selected on media omitting uracil/uridine yielding strain TJSF1.1. An auxotrophic control strain was generated by using TPMW7.2 as recipient strain and SbfI linearized pYHC-wA-pyrG was transformed and selected on media without supplements yielding strain TPMW8.2. For DNA isolation, all fungal strains were grown for 24 h at 37 °C (Aspergillus) or 29 °C (Fusarium) in steady state liquid GMM, supplemented appropriately as described by Shimizu and Keller [48]. Single integration was confirmed by Southern analysis as described by [23] using P 32 -labelled probes created by amplification of the indicated DNA fragment in Additional file 2: Figures S1-S4.
Carotenoid analysis
Carotenoids were extracted and analyzed as previously described [18,19]. Briefly, carotenoids were extracted with acetone from freeze dried mycelia and were separated by thin layer chromatography developed in light petroleum/diethyl ether/acetone (4:1:1; v/v/v). The bands were scraped out and dissolved in acetone. Highperformance liquid chromatography (HPLC) was used to analyze the β-carotene content by comparison to an authentic standard. HPLC separation was performed on using a ZORBAX Eclipse XDB-C18 column (Agilent, 4.6 mm by 150 mm with a 5 μm particle size) by using a binary gradient of methanol/t-butylmethyl ether (1:1) (v/v) as solvent A and methanol/t-butylmethyl ether/ water (5:1:1) (v/v/v) as solvent B using a Flexar Binary Liquid Chromatography (LC) Pump (PerkinElmer) coupled to a Flexar LC Autosampler (Perkin Elmer) and a Flexar PDA Plus Detector (PerkinElmer). The binary gradient started with a linear step from 0 A to 57% A in 45 min and an additional linear gradient to 100% A in 0.5 min and hold for 25 min at a flow rate of 2 mL/min. Identification and relative quantification of secondary metabolites was performed using Chromera Manager (PerkinElmer) by comparison to an authentic standard (Sigma Aldrich).
Results
The sterigmatocystin (ST) gene cluster of A. nidulans is known to harbor 25 genes involved in biosynthesis of sterigmatocystin [9]. While environmental regulation of the ST gene cluster is complex and not well understood, it was the first cluster that identified a gene product encoded within the cluster itself to function as a clusterspecific Zn(II) 2 Cys 6 transcription factor, called AflR [21]. The gene encoding AflR shares a bidirectional promoter with aflS encoding a transcriptional cofactor of AflR [20]. We replaced the native promoter of aflR/S with the well characterized niaD/niiA promoter which is induced by the presence of nitrate in the absence of other nitrogen sources [10]. We confirmed nitrate-dependent expression of aflR/S by northern blot analysis (Fig. 1b).
Next, we set out to test which of the 25 stc gene promoters would be nitrate-inducible in our production strain (TPMW2.3). Since TPMW2.3 is a uracil/uridine and riboflavin auxotroph, we designed plasmids that contain each of the 25 stc gene promoters, respectively, driving expression of the A. fumigatus pyrG gene along with a riboB selectable marker flanked by bordering regions of the yA locus ( Fig. 2a; Additional file 1: Table S3; Additional file 2: Fig. S2). Using a minimalized promoter selection strategy, we chose promoter regions as follows: For unidirectional stc genes, the promoter region was amplified from the first base after the stop codon of the first gene to the start codon of the second gene, but not exceeding 1 kb. In cases of bidirectional promoters, the entire region between the two start codons was chosen, not exceeding 1 kb. We selected 25 strains for each stc promoter for riboflavin prototrophy, exhibiting yellow spore color, and a control strain that did not include a stc promoter. To test for the ability of AflR/S to induce AfpyrG expression driven by each of the stc promoters, we grew them on media containing either nitrate (induces aflR/S expression; Fig. 1b) or ammonium, and supplemented with or without uracil/uridine (Fig. 2b, c). The growth assay showed that eight of the tested promoters (stcA, stcB, stcI, stcM, stcN, stcQ, stcV, and stcW) exhibited the desired ability to grow on plates without uracil/uridine supplementation on nitrate containing media only (Fig. 2b, c), demonstrating specific expression under induction conditions. Three of the promoters tested exhibited leaky expression (stcC, stcD, and stcE) as we observed colony growth of strains on media containing ammonium (Fig. 2b, c) where aflR/S should not be induced (Fig. 1b). In order to confirm control of the identified promoters by AflR, we investigated expression of all stc cluster genes in an A. nidulans WT and an isogenic ∆aflR knock-out strain [56] under sterigmatocystin production conditions. We found that in addition to the eight promoters identified in our plate assay, most of the remaining cluster genes were also expressed in an AflR-dependent manner (Fig. 2d). We speculate that either the length of the chosen promoters, the difference in culture conditions, or the insufficient expression level could be responsible for the observed discrepancies of stc activation of the AfpyrG reporter gene.
To test the functionality of our expression system genes responsible for carotenoid production from Fusarium fujikuroi [2][3][4] were expressed in our A. nidulans nitrate-inducible aflR/S strain TPMW2.3. In F. fujikuroi, the geranylgeranyl diphosphate (GGDP) synthase gene ggs1 is responsible for production of GGDP [34], which is a substrate for CarRA and CarB, encoded by two of the clustered carotenoid biosynthetic genes needed for β-carotene production [31]. We inserted the ggs1 gene driven by the stcM promoter and 0.5 kb of the native terminator, a riboflavin selectable marker flanked by the yA border regions (Additional file 1: Table S3; Additional file 2: Fig. S3), and the two carotenoid cluster genes carRA and carB including 0.5 kb of the native terminator regions, responsible for β-carotene production [31] under control of the bidirectional stcA/B promoter flanked by the wA border regions (Additional file 1: Table S3; Additional file 2: Fig. S3). Both plasmids were linearized and transformed into TPMW2.3 consecutively, yielding strain TJSF3.1 (Fig. 3a; Additional file 1: Table S1; Additional file 2: Fig. S4). Nitrate-inducible expression of ggs1, carRA, and carB was confirmed by northern blot analysis compared to a prototroph control strain that produced white spores (TPMW8.2) (Fig. 3b). When the two strains were grown on nitrate containing media, TJSF3.1 exhibited a characteristic orange color that was absent in the control (Fig. 3c). Characterization of carotene production by HPLC showed that the strain TJSF3.1 produced 125 µg β-carotene per gram mycelial dry weight in our experimental setting (Fig. 3d, e). The production of β-carotene was significantly higher on nitrate induction media than on non-induction media containing glutamine (Fig. 3d, e). The control strain TPMW8.2 did not show any carotene production (Fig. 3d).
As carotenoid production in N. crassa and Fusarium spp. occurs in both mycelia and spores [5], we asked whether a similar distribution would occur in our production strain. Therefore, mycelia and spores were assessed individually for β-carotene content. Carotenoids were only produced in the mycelia and not in the spores (Fig. 4a). Since the GGDP produce by Ggs1 is also utilized for ergosterol production in F. fujikuroi [34] we set out to investigate if the homolog of ggs1 in A. nidulans (AN0654, 54% identity, e-value: 4.0 −118 ) would be sufficient for β-carotene production. A strain was constructed that only expressed carRA and carB called TJSF1.1 (Additional file 2: Fig. S4). When carotenoid production between TJSF1.1 (carRA and carB) and TJSF3.1 (ggs1, carRA, and carB) was compared no significant difference under inducing conditions was observed (Fig. 4b), suggesting that AN0654 is sufficient to provide the maximum amount of GGDP that can be funneled into carotenoid production. As it is known that one of the bottlenecks during carotenogenesis is the production of mevalonate (a GGDP precursor) by the 3-hydroxyl-3-methyl-glutaryl-conenzyme A reductase (HMG CoA reductase) [1], we grew the two production strains on nitrate media supplemented with mevalonate before carotenoid quantification. However, we did not find any difference in production levels between the strains grown with or without mevalonate (Fig. 4c).
Discussion
Many efforts have been made to increase expression of fungal natural products [6,35]. Apart from increasing production in the native host, a major focus has been on developing heterologous expression systems. Heterologous systems have the advantage that they can be carried out in a safe host system without toxic byproducts, that is easily amenable to genetic manipulation and preferably inducible [50]. Traditional approaches are laborious as they are mainly based on over-expressing each natural product cluster individually or sequentially, thereby relying on multiple selection markers that limit the number of genes expressed and subsequently reduce the chemical complexity of the natural product produced [55]. There have been successful reports on marker recycling to overcome this issue [14,38], but these approaches involve multiple time-consuming transformation steps. Additionally, construction of the expression plasmids or cassettes has been achieved by labor intensive restriction enzyme-or fusion PCR-mediated methods [22,25]. The ease of yeast recombinational cloning [39] has been exploited for a wide range of molecular methods, including gene knock-out libraries [17] and expression systems [47] in filamentous fungi as well as yeast itself [8].
Technically, yeast recombinational cloning allows for the assembly of multiple PCR fragments up to a vector size of several ten thousand kilo bases [39]. One of the major hurdles to overcome during yeast recombineering, is undesired recombination among multiple identical DNA regions. Recently, a study in A. terreus demonstrated the requirement of an AflR-like transcription factor, TerR, for expression of all twelve terrein cluster genes [24] similar to our expression data for AflR-dependency of all 25 stc genes. The system was subsequently used to control one of the terrein promoters in a heterologous expression system in A. niger to demonstrate activation of orsA from A. nidulans [24]. Here, we have demonstrated the specific induction of eight of the 25 stc promoters to control the expression of a reporter gene (pyrG). Subsequently, we have utilized three of the eight promoters to successfully express three Fusarium spp. derived genes responsible for carotenoid production in A. nidulans and confirm functionality of their gene products by detection of β-carotene. As a precursor of vitamin A, β-carotene has long been in the focus of biotechnology. The most prominent example of heterologous gene expression leading to the production of β-carotene is the development of Golden Rice by Monsanto [37]. However, β-carotenoid production was also achieved in baker's yeast and bacteria. The amounts of carotenoids produced by the production strain constructed in this study are equivalent to the first production strain engineered in yeast [52]. Recent advances in manipulating metabolic pathways and genetic elements have increased the production in baker's yeast [8,30], and similar approaches could be undertaken to increase production in A. nidulans.
Notably, production of β-carotene is restricted to the mycelia in the A. nidulans production strains investigated in this study, whereas in other fungal species like Neurospora crassa and F. fujikuroi, carotenes are predominantly produced in asexual spores [26,31]. One explanation might be the developmentally controlled expression pattern of the niaD promoter, as it was shown to be only transiently expressed at early stages of conidiation, but not at later time points [33]. Sterigmatocystin and aflatoxin in Aspergillus species are predominantly found in the mycelial fraction and a complex fusion network of vesicles containing different precursors and biosynthetic enzymes that ensure correct cellular localization of these secondary metabolites is being unveiled by several studies [32,42,49]. Additionally, pioneering work in A. fumigatus has demonstrated that certain natural products are predominantly produced in the asexual spores (conidia) [27,29]. These findings suggest that conidial directed cellular pathways in the native host (Fusarium) may differ significantly from Aspergillus as location of β-carotene is not the same in the native and heterologous host. Determining which factors control the direction of fungal natural products to certain developmental structures in different fungal species will be a fascinating future task.
Conclusions
This study presents a new heterologous expression system for fungal natural products in the genetic model organism A. nidulans. The system described here makes use of the ability to co-express, minimally, eight promoters by a fungal-specific Zn(II) 2 Cys 6 transcription factor, AflR, and its cofactor AflS. By replacing the intrinsic bidirectional aflR/S promoter with a nitrate bidirectional Fig. 4 Tissue-and media-specific β-carotene production. a Comparison of β-carotene production in spores and mycelia of TJSF3.1 in triplicates. The strains were grown under nitrate inducing conditions and carotenoid production was normalized to the amount produced in mycelia. b Comparison of β-carotenoid production between TJSF1.1 (carRA and carB) and TJSF3.1 (ggs1, carRA and carB). Strains were grown under nitrate inducible conditions and β-carotene was quantified based on normalized dry weight in triplicates. No significant difference be could be detected. c Comparison of β-carotene production between TJSf1.1 (carRA and carB) and TJSF3.1 (ggs1, carRA and carB) grown on nitrate inducible conditions either supplemented with (+) or without (−) 10 mM mevalonate in triplicates. No significant difference be could be detected inducible promoter, all eight identified genes can be simultaneously activated and repressed. As all eight identified promoters differ in their DNA sequence, the system has the potential to utilize one-step yeast recombinational cloning for assembly of entire secondary metabolite gene clusters. Here, we demonstrate the production of β-carotene by heterologous expression of three genes from F. fujikuroi. The inducibility of the system also is useful for production of toxic metabolites at a stage when the host strain has accumulated a significant biomass.
==
Domain: Biology Environmental Science Chemistry
|
[{"added": "2023-02-01T14:24:44.950Z", "created": "2018-03-13T00:00:00.000", "id": "256420040", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1186/s40694-018-0049-2", "pdf_hash": "a101a5574948396119847de6b68773b314635546", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:4418", "s2fieldsofstudy": ["Biology", "Environmental Science", "Chemistry"], "sha1": "a101a5574948396119847de6b68773b314635546", "year": 2018}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
The anti-inflammatory role of zDHHC23 through the promotion of macrophage M2 polarization and macrophage necroptosis in large yellow croaker (Larimichthys crocea)
Zinc finger Asp-His-His-Cys motif-containing (zDHHC) proteins, known for their palmitoyltransferase (PAT) activity, play crucial roles in diverse cellular processes, including immune regulation. However, their non-palmitoyltransferase immunomodulatory functions and involvement in teleost immune responses remain underexplored. In this study, we systematically characterized the zDHHC family in the large yellow croaker (Larimichthys crocea), identifying 22 members. Phylogenetic analysis unveiled that each of the 22 LczDHHCs formed distinct clusters with their orthologues from other teleost species. Furthermore, all LczDHHCs exhibited a highly conserved DHHC domain, as confirmed by tertiary structure prediction. Notably, LczDHHC23 exhibited the most pronounced upregulation following Pseudomonas plecoglossicida (P. plecoglossicida) infection of macrophage/monocyte cells (MO/MΦ). Silencing LczDHHC23 led to heightened pro-inflammatory cytokine expression and diminished anti-inflammatory cytokine levels in MO/MΦ during infection, indicating its anti-inflammatory role. Functionally, LczDHHC23 facilitated M2-type macrophage polarization, as evidenced by a significant skewing of MO/MΦ towards the pro-inflammatory M1 phenotype upon LczDHHC23 knockdown, along with the inhibition of MO/MΦ necroptosis induced by P. plecoglossicida infection. These findings highlight the non-PAT immunomodulatory function of LczDHHC23 in teleost immune regulation, broadening our understanding of zDHHC proteins in host-pathogen interactions, suggesting LczDHHC23 as a potential therapeutic target for immune modulation in aquatic species.
Introduction
The zinc finger Asp-His-His-Cys motif-containing (zDHHC) proteins constitute a family of palmitoyltransferases (PATs) primarily tasked with catalyzing the addition of palmitate moieties onto protein substrates, a process known as Spalmitoylation (1). To date, 24 zDHHC members have been identified in mammalian species, with 23 in humans (zDHHC1-9, 11-24) and 24 in mice (zDHHC1-9, 11-25) (1). All of these members are characterized by the presence of a 50-residue long DHHC (Asp-His-His-Cys) domain, which serves as the active center for PATs (2). S-palmitoylation, the reversible attachment of palmitate to cysteine residues via thioester bonds, plays crucial roles in regulating protein membrane localization, trafficking, stability, and function (3). As integral players in protein Spalmitoylation, zDHHCs hence contribute significantly to the modulation of diverse cellular processes, including signal transduction, membrane dynamics, and protein-protein interactions (4). Recently, research has focused on the role of zDHHCs in modulating immune responses. Molecules involved in innate immunity, such as the pattern recognition receptor TLR2 (Toll-like receptor 2), the cytoplasmic DNA receptor cGAS (cyclic GMP-AMP synthase), the cytoplasmic bacterial peptidoglycan receptor NOD1 and NOD2 (nucleotide-binding oligomerization domain-containing protein 1/2), the type I interferon receptor IFNAR (interferon alpha/beta receptor), the transcription factor STAT3 (signal transducer and activator of transcription 3), and the cytokine TNF-a, have all been reported to undergo palmitoylation by various zDHHCs (5)(6)(7)(8)(9). Moreover, zDHHCs participate in the palmitoylation of adaptive immune molecules, such as CD4, CD8, and LAT (the inker for activation of T cells), which are associated with T cell function (10)(11)(12). Additionally, molecules involved in B cell function, like CD81, LAB (the linker for activation of B cells), were also palmitoylated by zDHHCs (13,14). Despite primarily functioning as PAT for immune-related molecules, the non-PAT roles of zDHHCs in immunity are largely unexplored. Only zDHHC1 and zDHHC11 has been identified as a positive regulator of DNA virus-triggered signaling via STING interaction, independent of their PAT activity (15,16). Therefore, investigating the non-PAT functions of the zDHHC family is a highly valuable area for further research.
To date, our understanding of the composition and function of the zDHHC family in lower species, including teleosts, remains limited. Systematic studies are needed to determine the number of zDHHCs present in teleosts and their respective functions, as well as to investigate the evolutionary clues of this family. Only five zDHHCs (zDHHC1, 13, 15, 16, and 17) have been reported and characterized for their functions in teleosts, with studies focusing mainly on zebrafish (zDHHC13, 15, 16, and 17) (17)(18)(19)(20). Independently of their PAT activity, zebrafish zDHHC13 regulates embryonic differentiation through interaction with Smad6, zDHHC15b is implicated in the differentiation of dopamine (DA) neurons, and the anchor protein domain of zDHHC17 influences neuronal axon growth by regulating the formation of the TrkA tubulin complex (17,19,20). Only zDHHC16 promotes telencephalic neural stem cell proliferation by modulating the FGF/ERK signaling pathway through its PAT activity (18). Research on the immune regulatory roles of zDHHCs in teleosts is further limited. Only zDHHC1, identified in grass carp (Ctenopharyngodon idella) and Chinese perch (Siniperca chuatsi), has been reported to possess antiviral functions (21,22).
In the current study, we systematically screened the large yellow croaker genome for members of the zDHHC family, identifying 22 zDHHC proteins (LczDHHC1-9, [11][12][13][14][15][16][17][18][20][21][22][23][24]. Subsequently, we cloned all 22 zDHHCs and conducted analyses on their evolutionary relationships, protein structures, and importantly, their potential association with Pseudomonas plecoglossicida (P.plecoglossicida) infection, a significant pathogenic bacterium affecting large yellow croaker (23). Among them, LczDHHC23 exhibited the highest upregulation in P. plecoglossicida infected large yellow croaker macrophage/monocyte cells (MO/MF). However, the immune regulatory functions of zDHHC23 have not been previously explored, even in mammals. In mammals, zDHHC23 primarily palmitoylates molecules implicated in mTOR signaling and tumorigenesis (24,25). Therefore, investigating the mechanisms underlying the notable upregulation of LczDHHC23 expression post-infection holds significance. By silencing LczDHHC23 expression, we observed an upregulation of proinflammatory cytokines and a downregulation of antiinflammatory cytokines during P. plecoglossicida infection, indicating the anti-inflammatory role of LczDHHC23. Further investigation revealed that LczDHHC23 predominantly promotes M2-type macrophage polarization while inhibiting M1-type polarization, as also evidenced by enhanced phagocytic activity in LczDHHC23-knockdown MO/MF. Additionally, LczDHHC23 was found to facilitate the necrosis of P. plecoglossicida-infected MO/ MF, as evidenced by delayed and reduced phosphorylation of necrosis markers receptor−interacting serine/threonine kinase (RIP)1, RIP3, and mixed lineage kinase domain-like (MLKL) upon compromised LczDHHC23 expression. To our knowledge, our investigation represents the first study to elucidate the role of zDHHC23 in immune cells during infection, potentially laying the groundwork for a deeper understanding of zDHHC23's function in modulating anti-pathogen immune responses.
Sampling and challenging
Healthy large yellow croakers (100-120 g) were purchased from Xiangshan county farm (NingBo, China). All fish were temporarily kept in a recirculating seawater system maintained at 18-20°C for at least two weeks. All animal experiments were approved by the Institutional Animal Care and Use Committee of Ningbo University and conducted in accordance with the Guide for the Care and Use of Laboratory Animals issued by the National Institutes of Health. P. plecoglossicida strain (NZBD9) was cultivated at 18°C in Luria-Bertani (LB) broth with shaking and collected until reaching the logarithmic growth stage. The bacteria were washed with sterile phosphate-buffered solution (PBS), and then diluted to a final 5 ×10 4 colony forming units (CFU)/100g fish in 100 mL PBS for the in vivo challenge test. The large yellow croakers were intraperitoneally injected with P. plecoglossicida, while the control group received an equivalent volume of PBS. The tissue samples in each group (gill, head kidney, intestine, liver, and spleen) were collected at 0, 6, 12, 24, 48 and 72 hours after infection (hpi), and promptly snapfrozen in liquid nitrogen and stored at -80°C for subsequent Real-time quantitative polymerase chain reaction (RT-qPCR) analysis. The primers used are listed in Supplementary Table 1.
Cloning of the LczDHHC genes and bioinformatic analysis
The 22 LczDHHC genes were cloned via reverse transcription PCR (RT-PCR), using sequences obtained from the NCBI database. Subsequently, the PCR products were sequenced for validation. The primers used are listed in Supplementary Table 1. Phylogenetic analysis of the whole family was conducted using the neighbor-joining method, supported by 1000 bootstrap repetitions in MEGA 7.0 software. The phylogenetic tree was modified using iTOL programs. Transmembrane (TM) domain analysis of LczDHHC23 was predicted by the TMHMM Server v. 2.0 program ( [URL] sequence alignment was generated using ClustalW ( [URL]:// clustalw.ddbj.nig.ac.jp/). The domain information of LczDHHC23 was predicted using SMART ( [URL] the three-dimensional structures of all LczDHHCs were predicted using SWISS-MODEL ( [URL] by visualization of the PDB files using PyMOL software version 3.1. Sequences used in this study are listed in Supplementary Table 2.
Large yellow croaker head-kidneyderived MO/MF isolation and P.plecoglossicida stimulation
Large yellow croaker head-kidney-derived MO/MF were isolated as previously described (26,27). Leukocyte-enriched fractions were obtained by applying dissociated head-kidney to a Ficoll density gradient (1.077 g/mL; #17144002 GE Healthcare, Chicago, IL, USA). The cells were then seeded in 6 well plates at a density of 2×10 7 /mL and cultured overnight at 26°C under 5% CO2. After washing away the nonadherent cells, attached cells were incubated with complete DMEM medium (10% fetal bovine serum, 100 U/mL penicillin, and 100mg/mL streptomycin) and cultured under the same conditions. Live P. plecoglossicida diluted in PBS at a multiplicity of infection (MOI) of 2 or PBS alone were added into cell culture medium. Cells were collected at 4, 8, 12, and 24 hpi for RNA extraction, and RT-qPCR was conducted as described. The primers used are listed in Supplementary Table 1.
RNA extraction and RT-qPCR
Total RNA of both large yellow croaker tissues and MO/MF were isolated using RNAiso (#9108/9109, TaKaRa, Dalian, China), treated with DNase I (#2270A, TaKaRa), and reversed transcribed into first-strand cDNA using AMV reverse transcriptase (#2621, TaKaRa) according to manufacturer protocol. RT-qPCR was conducted on an ABI StepOne real-time PCR system (Applied Biosystems, Foster City, CA, USA) using SYBR premix Ex Taq II (#RR82WR, TaKaRa). The thermal cycling conditions were as follows: initial denaturation at 95°C for 10 s, followed by 40 cycles of amplification (95°C for 50 s and 60°C for 20 s), and final melting curve analysis (95°C for 60 s, 55°C for 30 s, and 95°C for 30 s). Relative gene expression was calculated using the 2 −DDCT method and the data were normalized against Lc18S rRNA. The primers used are listed in Supplementary Table 1. Each PCR trial was performed in triplicate and repeated at least three times.
Constructing of LczDHHC23 eukaryotic expression plasmids
The full-length open reading frame (ORF) of LczDHHC23 was amplified by PCR using PrimeSTAR GXL DNA polymerase (#R050A, Takara). The PCR product was ligated with pcDNA-HA vector to generate the HA-tagged LczDHHC23 plasmid using the pEASY ® -Basic Seamless Cloning and Assembly Kit (#CU201-02, TransGen Biotech). The plasmid was subsequently transformed into competent Escherichia coli (E.coli) cells. The bacterial solution was spread onto LB nutrient agar plates, and single colonies were selected from overnight cultures for sequencing.
Subcellular localization of LczDHHC23
Hela cells were seeded at a density of 1×10 5 /mL and transfected with HA-tagged LczDHHC23 plasmids. After 36 h of transfection, the cells were fixed with 4% paraformaldehyde in PBS (pH=7.4) and permeabilized with 0.5% saponin. After blocking, the cells were incubated with appropriate primary and secondary antibodies. Confocal images were obtained using Zeiss LSM 880 confocal microscope (Carl Zeiss AG, Oberkochen, Germany) and analyzed using the ZEN Blue software.
RNA interference
LczDHHC23-specific small interfering RNA (RiboBio, Guangzhou, China) was transfected into large yellow croaker MO/MF using Lipofectamine ™ RNAiMAX Transfection Reagent (#13778150, Invitrogen/Life Technologies, Carlsbad, CA, USA) according to the previous studies for 24 and 48 h to evaluate the knock down efficiency (21,(28)(29)(30). The scrambled siRNA was used as the control. To evaluate the role of LczDHHC23 on MO/MF function following P. plecoglossicida stimulation, the isolated MO/ MФ were transfected with 30 pmol LczDHHC23 siRNA, or scrambled siRNA for 24h before infected with P.plecoglossicida.
In vitro bacterial-killing assay
Isolated MO/MF were transfected with LczDHHC23 siRNA, or corresponding controls, for 24 h before being infected with live P. plecoglossicida at an MOI of 4 (30,31). The phagocytosis of bacteria was proceeded for 30 minutes at 26°C under 5% CO 2 . The remaining bacteria attached to cell surface were killed using gentamicin (50 mg/mL), and then washed with sterile PBS. Each set of interfered MO/MФ were divided into two groups. One group (the uptake group) was lysed immediately with 1% Triton X-100 solution and plated onto solid LB agar medium to assess bacterial uptake. The other group (the kill group) was incubated for an additional 1.5 h before being lysed and plated on LB agar medium. After incubation at 28°C for 24 h, the CFUs of the plates were calculated. Bacterial survival was determined by dividing the CFUs in the kill group by those in the uptake group. Three independent experiments were performed.
MO/MФ polarization assay
To investigate the impact of LczDHHC23 on MO/MФ polarization, LPS-induced M1-type and cAMP-induced M2-type MO/MФ were prepared according to previous reported (32). MO/ MФ cells were transfected with LczDHHC23 siRNA, or corresponding controls for 24 h before treatment with LPS (50 mg/mL; #L4391, Sigma Aldrich) or a cAMP analog (dibutyryl cAMP; 0.5 mg/mL, #28745-M, Sigma-Aldrich) for 18 h. Expression levels of the pro-inflammatory cytokines (IL-1b and IL-6) and the anti-inflammatory cytokines (IL-10 and TGF-b) were determined. Moreover, the markers of M1-type MO/MФ (C-X-C motif chemokine ligand 9 (CXCL9) and induced nitric oxide synthase (iNOS)) and the markers of M2-type MO/MФ (secreted phosphoprotein 1 (SPP1) and arginase) were also evaluated (33). In addition, the iNOS (M1-type) and arginase activities (M2-type) were assessed.iNOS activity was measured using a nitric oxide synthase assay kit (fluorescence probe method; #S0024, Beyotime, Shanghai, China) following the manufacturer's instructions. The relative iNOS activity of each group was expressed as fold change relative to the value of the control group. Arginase activity was measured using an arginase activity assay kit (#MAK112, Sigma-Aldrich) following the manufacturer's protocol. Absorbance was read at 430 nm, and arginase activity (U/L) was calculated according to comparison with urea-standard data.
Necroptosis assay
Large yellow croaker MO/MF were transfected with LczDHHC23 siRNA, or corresponding controls, for 24 h prior to infection with P. plecoglossicida at an MOI of 10 (30, 31). Cell samples were collected at 1, 2, 4 and 8 hpi. The collected cells were washed and then labeled with Annexin V-FITC and propidium iodide (PI) for 15 minutes using a FITC Annexin V apoptosis detection kit I (#556547, BD Pharmingen, San Diego, CA, USA). Apoptosis was evaluated by flow cytometry using the MACSQuant Analyzer 10 (Miltenyi Biotec) within 15 min of staining, and data were analyzed using MACSQuant analysis software (Miltenyi Biotec). Three independent experiments were performed.
Western blotting
Large yellow croaker MO/MF cells were seeded in 6 well plates at a density of 2×10 7 /mL the day before the experiment. Cells were washed twice with PBS and infected with P.plecoglossicida at an MOI of 10. Subsequently, cells were collected at 0, 1, 2, 4 and 8 hpi and lyzed by lysis buffer (20 mM Tris-HCl [pH=8.0], 2 mM EDTA, 120 mM NaCl, 1% NP-40) containing phosphatase inhibitor (#A32963, Thermo fisher). The soluble protein concentration was measured using the Bradford method. Proteins were separated on 10% or 15% SDS-PAGE, transferred to polyvinylidene difluoride (PVDF) membranes (#IPVH00010, EMD Millipore). The membranes were blocked with 5% non-fat milk for 1 h at room temperature, then incubated overnight at 4°C with apoptosis/ necroptosis antibodies (#92570, Cell Signaling Technology), followed by incubation with appropriate HRP-conjugated secondary antibodies for 1 h at room temperature. The blots were subsequently visualized using a chemiluminescent detection system with ECL western blotting detection reagents (#32106, Thermo Fisher Scientific).
Statistical analysis
All data are presented as the means ± SEM. Statistical analysis was performed using one-way analysis of variance (ANOVA) with the Prism 8.02 software (GraphPad Software, San Diego, CA, USA). The p values *p < 0.05 were considered statistically significant.
Cloning and bioinformatic analysis of the LczDHHC members
Twenty-two LczDHHCs have been cloned from the large yellow croaker (LczDHHC1-9, 11-18, 20-24). Unlike observed in mammals, the zDHHC family in the large yellow croaker does not include zDHHC19 and zDHHC25 (1). Phylogenetic analysis revealed three major clades grouping the 22 subfamilies, with each of these 22 LczDHHCs forming clusters with their respective orthologues from other teleost species (Figure 1A). The tertiary structures of the LczDHHCs were predicted using SWISS-MODEL, revealing a highly conserved DHHC domain and 4 to 7 transmembrane (TM) domains in all members (Figures 1B-D; Supplementary Figure 1). Additionally, LczDHHC6 contains an SH3 domain, while LczDHHC13 and LczDHHC17 feature an ankyrin repeat domain. The TM helices of these LczDHHCs form pocket-like structures, and the DHHC domain contains 2 or more b-hairpin structures, except for LczDHHC13. This arrangement is reminiscent of the crystal structure of zDHHC proteins reported in humans and zebrafish (34). Typically, the DHHC domain of most LczDHHC proteins (excluding LczDHHC13, 17, and 23) is situated between the second and third TM domains (Figures 1B-D; Supplementary Figure 1).
LczDHHC23 exhibits the most significant response to P. plecoglossicida infection among all zDHHCs
We screened the 22 LczDHHCs for changes in expression during P. plecoglossicida infection in large yellow croaker head kidney-derived MO/MF. Among them, LczDHHC8, LczDHHC11, and LczDHHC23 showed significant induction, with LczDHHC23 exhibiting the highest induction, reaching over 300-fold at 24 hpi (Figures 1E-G). The expression changes of other LczDHHCs were minor, with induction or suppression within 20-fold (Supplementary Figures 2, 3). Consequently, we focused on studying the role of LczDHHC23 in regulating the immune response induced by P. plecoglossicida infection in the subsequent investigation. The LczDHHC23 sequence spans 2355 nucleotides (nt), with an open reading frame (ORF) of 1178 base pairs (bp), encoding a protein of 392 amino acids (aa). A phylogenetic tree was constructed based on the amino acid sequences of zDHHC23 proteins from teleosts and other species. The analysis revealed that teleost zDHHC23s formed a distinct cluster, with LczDHHC23 showing the closest relationship to that of Nibea albiflora (Supplementary Figure 4A). Multiple sequence alignment demonstrated a high degree of conservation among zDHHC23 orthologs, comprising six transmembrane (TM) domains and a highly conserved DHHC domain (Supplementary Figure 4B). We further compared the tertiary structure of LczDHHC23 with that of human orthologue (HszDHHC23), and found that they both possess six TM helices, forming a pocket-like structure similar to that of most members of the zDHHC family. Additionally, they feature a highly conserved DHHC domain positioned between the fourth and fifth TM domains, contrasting with most zDHHC family proteins where it situated between the second and third TM domains. As depicted in Figure 2A, the DHHC domain of LczDHHC23 contains two b-hairpin structures, whereas the DHHC domain of HszDHHC23 has more. This suggests that the DHHC domain of HszDHHC23 has a more complex structure compared to that of the LczDHHC23.
Subcellular localization and tissue distribution pattern of LczDHHC23
We further evaluated the molecular characteristics of LczDHHC23. Previous studies have indicated that members of the zDHHC family primarily localize to the endoplasmic reticulum, the Golgi apparatus, or both (35). Our subcellular localization analysis via immunofluorescence revealed that LczDHHC23 predominantly co-localizes with the Golgi apparatus marker TGN38, rather than the endoplasmic reticulum marker Calnexin (Figure 2B), indicating a difference from observations in mammals (36). The tissue expression pattern of LczDHHC23 in major immune tissues (intestine, head kidney, liver, gill, and spleen) of healthy large yellow croaker revealed that the spleen exhibited the highest levels of LczDHHC23 transcripts, followed by the liver and head kidney (Supplementary Figure 5A). Following P. plecoglossicida injection, the expression of LczDHHC23 significantly increased in all examined tissues except for the liver, which showed a decrease at 12 hpi. The highest expression levels of LczDHHC23 in the head kidney and gill were observed at 12 hpi, followed by a decrease. In the intestine, LczDHHC23 expression initially decreased significantly and then sharply increased. LczDHHC23 levels in the spleen increased at 12 hours and remained stable until 72 hpi (Supplementary Figures 5B-F).
LczDHHC23 plays an anti-inflammatory role in MO/MF during P. plecoglossicida infection
To assess the impact of LczDHHC23 on immune responses following P. plecoglossicida infection, we initially synthesized LczDHHC23 siRNA (5'-CACCACTGCATCTGGATAA-3') to downregulate its expression in large yellow croaker MO/MF. The expression of LczDHHC23 was notably reduced after 24 and 48 hours of interference compared to the control group, reaching 9.8% and 43.8%, respectively (Figure 2C). After confirming the efficacy of LczDHHC23 siRNA knockdown, we investigated the impact of LczDHHC23 expression interference on the expression of pro-inflammatory (IL-1b and IL-6) and anti-inflammatory (IL-10 and TGF-b) cytokines following P. plecoglossicida infection. The results revealed that inhibiting LczDHHC23 significantly upregulated the expression of proinflammatory cytokines (IL-1b and IL-6; Figures 2D, E) and downregulated the expression of anti-inflammatory cytokines (IL-10 and TGF-b; Figures 2F, G), compared to the control group. These findings suggest that LczDHHC23 may indeed play an antiinflammatory role in MO/MF during P. plecoglossicida infection in large yellow croaker.
LczDHHC23 promotes M2-type polarization of large yellow croaker MO/MF
We further evaluated whether LczDHHC23 exert its antiinflammatory roles by affecting the polarization of large yellow croaker MO/MF. Fish MO/MФ can differentiate into two main types: the M1-type, characterized by the expression of proinflammatory cytokines and the production of reactive oxygen species (ROS) and NO, and the M2-type, characterized by the expression of anti-inflammatory cytokines and increased arginase activity (33,37). We induced M1-type and M2-type polarization of isolated large yellow croaker MO/MF using lipopolysaccharide (LPS) and cyclic adenosine monophosphate (cAMP), respectively, as previously reported (32). As shown in Figure 3, knockdown of LczDHHC23 further promoted the induced expression of proinflammatory cytokines (IL-1b and IL-6) and M1-type markers (CXCL9 and iNOS), as well as the iNOS activity in M1-type MO/ MF (Figures 3A, B, I, J, M). Conversely, LczDHHC23 silencing further decreased the already low expression of anti-inflammatory cytokines (IL-10 and TGF-b) in M1-type MO/MF, indicating an inhibiting role of LczDHHC23 on the M1-type function (Figures 3E, F). Moreover, LczDHHC23 silencing significantly upregulated the expression of IL-1b and IL-6 in cAMP-induced M2-type MO/MF, while downregulating the expression of IL-10, TGF-b, as well as the M2 marker SSP1, and the arginase activity (Figures 3C, D, G, H, K, L, N). These findings suggest that LczDHHC23 may promote M2-type MO/ MF polarization to exert its anti-inflammatory function. In support of this, we observed a significant augmentation in bacterial killing activity in LczDHHC23-silenced MO/MF. Specifically, the bacterial killing activity decreased from 22.84% to 6.07% in LczDHHC23-deficient cells compared to control cells (Supplementary Figure 6).4A, B). Upon silencing of LczDHHC23, the heightened phosphorylation of RIP1, RIP3, and MLKL induced by P. plecoglossicida infection was mitigated or delayed (Figure 4B), indicating the involvement of LczDHHC23 in promoting immune cell necroptosis during pathogen infection.
Discussion
Previous studies have underestimated the non-PAT immunomodulatory functions of zDHHC family proteins, but primarily focused on the palmitoylation modification of immunerelated molecules mediated by the zDHHC members (11,15,16). Also, research on the immune regulatory roles of zDHHCs in teleosts remains limited, with only zDHHC1 molecules investigated for their antiviral functions (21,22). Herein, we identified 22 zDHHC proteins in the large yellow croaker genome, excluding zDHHC10 and 19, and analyzed their evolutionary relationships. Interestingly, we observed the absence of the zDHHC19 gene in teleosts, contrasting with its presence in mammals (1). Among the identified zDHHC proteins, LczDHHC23 stood out due to its significant upregulation following P. plecoglloscida infection in MO/MF, prompting further investigation into its potential immune regulatory role. Our choice to focus on LczDHHC23 was also influenced by previous studies implicating zDHHC23 in immune-related processes, particularly its involvement in mTOR signaling and tumorigenesis (24, 25). Thus, we aimed to elucidate the specific immune functions of LczDHHC23 in the context of pathogen infection induced immune responses in teleosts.
Our bioinformatics analyses provided insights into the functional domains and evolutionary patterns of LczDHHC23. Phylogenetic analyses revealed its clustering with zDHHC23 from teleosts, particularly Nibea albiflora, suggesting evolutionary conservation. Furthermore, structural predictions indicated similarities between LczDHHC23 and HszDHHC23 proteins, particularly in their DHHC domains and transmembrane regions. Despite these similarities, we noted differences in the subcellular localization of LczDHHC23, predominantly within the Golgi apparatus in contrast to mammalian studies (36).
In examining the immune function of LczDHHC23, we observed its widespread expression in immune-related tissues of healthy large yellow croaker, with significant upregulation following P. plecoglloscida infection, except in the liver. The expression pattern of LczDHHC23 initially increased, followed by a decrease, in both the liver and gills after P. plecoglossicida infection. This suggests that the immune system might boost immune cell activity and inflammatory signaling pathways by upregulating the expression of the zDHHC23 gene to combat pathogen invasion, while its gradual decrease in expression aims to alleviate excessive inflammation and immune-related damage (40). These findings underscore the involvement of LczDHHC23 in the immune response against bacterial infection (41, 42). Silencing LczDHHC23 expression in MO/MF resulted in heightened proinflammatory cytokine expression and reduced anti-inflammatory cytokine levels during P. plecoglossicida infection, indicating its anti-inflammatory role. Meanwhile, we observed a significant increase in the expression of TGF-b at 4 h, whereas the expression of IL-1b and IL-6 significantly decreased at 4 and 24 h, respectively. Typically, MO/MF release inflammatory cytokines rapidly upon pathogen infection. However, to mitigate excessive inflammation and immune damage, the expression of inhibitory inflammatory cytokines may gradually rise (43). This result could be attributed to the regulatory role of zDHHC23, a non-PAT, in the cytokine signaling pathway. Currently, there is insufficient literature on the non-PAT functions of the zDHHC family and their involvement in cytokine signaling pathways. Further research is warranted to elucidate specific targets and underlying mechanisms.
Moreover, our study revealed the impact of LczDHHC23 deficiency on MO/MF polarization, shifting the balance towards the pro-inflammatory M1 phenotype. These findings further support the anti-inflammatory role of LczDHHC23 in MO/MF and its role in promoting M2-typ MO/MF polarization (32,44). Further, silencing LczDHHC23 reduced P. plecoglloscida infection induced necroptosis of MO/MF, which was also supported by decreased phosphorylation levels of necroptosis markers, suggesting a protective role against cell death with LczDHHC23 deficiency (45,46). Meanwhile, we also observed a slight decrease in the expression of cleaved Caspase-3 and 8 in the LczDHHC23 knockout group 2 hpi (Figure 4B). However, no further significant differences were observed with prolonged infection. Furthermore, upon calculating the relative grey intensity of the Cleaved Caspase-3 band to that of GAPDH (data not shown), we found that the minor decrease of the 2 hpi band were not statistically significant. We thought that the weak cleavage of Caspase-3 may not be significant enough to indicate a notable effect. Additionally, our statistical analysis of flow cytometry data representing apoptosis revealed no significant difference between the LczDHHC23 knockdown and control groups (Figure 4A). Therefore, we thought the influence of LczDHHC23 on cell death induced by infection primarily pertains to necroptosis. Given the emerging role of necroptosis in host defense against pathogens, elucidating the precise mechanisms underlying LczDHHC23-mediated regulation of necroptosis could provide valuable insights into host-pathogen interactions in teleosts.
In summary, our study elucidates the immune regulatory functions of LczDHHC23 in large yellow croaker, highlighting its anti-inflammatory properties in MO/MF. By modulating cytokine expression, polarization, and necroptosis, LczDHHC23 plays a crucial anti-inflammatory role in the immune response against bacterial infection. However, further investigations are warranted to unravel the precise mechanisms underlying LczDHHC23mediated immune regulation and its potential as a therapeutic target for immune modulation in teleosts.
1
FIGURE 1 Bioinformatic analysis and expression of LczDHHC family in response to P. plecoglossicida infection in head kidney-derived MO/MF.(A) Phylogenetic tree showing the divergence of zDHHC proteins among teleost species, generated using the neighbor-joining method in MEGA 7.0 with 1000 bootstrap replications and modified using iTOL.(B-D) Predicted tertiary structures of LczDHHC family proteins with five (B), six (C), and seven (D) transmembrane (TM) domains. Structures were modeled using SWISS-MODEL and visualized using PyMOL software version 3.1. LczDHHC 1, 4, 11, 22, and 24 have five TM domains (B), LczDHHC17 and LczDHHC 23 have six TM domains (C), and LczDHHC13 has seven TM domains (D). Additionally, LczDHHC13 and 17 feature an ankyrin repeat domain. TM domains are depicted in yellow, DHHC domains in red, and ankyrin repeat domains in deep blue.(E-G)Quantitative analysis of significantly upregulated expression of LczDHHCs in response to P. plecoglossicida infection, including LczDHHC8 (E), LczDHHC11 (F), and LczDHHC23 (G). MO/MF from large yellow croakers were infected with P. plecoglossicida at an MOI of 2, with PBS-treated cells as controls. Samples were collected at 0, 4, 8, 12, and 24 hpi. Expression levels of each LczDHHC mRNA were normalized to Lc18S rRNA and then to the 0 h PBS control using the 2 -DDCT method. Data represent the means ± SEM of three replicates.*p < 0.05 and **p < 0.01.
3. 6 2 3
FIGURE 2 Molecular characterization of LczDHHC23 and its impact on cytokine expression of P. plecoglossicida-infected MO/MF.(A) Comparison of the tertiary structures of LczDHHC23 and Homo sapiens zDHHC23 (HszDHHC23). The tertiary structures were modeled using SWISS-MODEL and visualized using PyMOL software version 3.1. TM domains are represented in yellow, and DHHC domain is depicted in red.(B) Immunofluorescence images showing the subcellular localization of LczDHHC23 with the endoplasmic reticulum marker Calnexin and Golgi apparatus marker TGN38. HeLa cells were seeded at a density of 1 × 10 5 /mL and transfected with HA-tagged LczDHHC23 plasmid. After 36 h of transfection, the cells were fixed and examined using a confocal microscope. The white line indicates the color pickup line, with the fluorescence intensity curve shown in green and red on the right side, analyzed with ZEN software. Scale bars, 5 mm. Results are representative of three independent experiments.(C-G) Impact of LczDHHC23 on the cytokine expression in P. plecoglossicida infected MO/MF.(C) RT-qPCR assesses the interference efficiency of LczDHHC23 siRNA. Transcript levels of LczDHHC23 were normalized against 18S rRNA and then to the 24 h scrambled siRNA control group using the 2 -DDCT method.*p < 0.05 and **p < 0.01.(D-G)RT-qPCR analysis evaluates the effects of LczDHHC23 interference on cytokine mRNA expression in P. plecoglossicida-infected MO/MF. Large yellow croaker MO/MF were transfected with LczDHHC23 siRNA for 24 h before P. plecoglossicida infection at an MOI of 2. Samples were collected at 0, 4, 8, 12, and 24 hpi. Expressions levels of IL-1b, IL-6, IL-10, and TGF-b were detected using RT-qPCR.mRNA expression was normalized to that of 18S rRNA and then to the respective 0 h control using the 2 -DDCT method. Data represent the means ± SEM of three replicates.*p < 0.05 and **p < 0.01.
4
FIGURE 4Impact of LczDHHC23 deficiency on P. plecoglossicida-induced necroptosis of large yellow croaker MO/MФ.(A) MO/MF were transfected with LczDHHC23 siRNA for 24 h prior to P. plecoglossicida infection at an MOI of 10. After 1, 2, 4 and 8 h of infection, cells were harvested and stained with Annexin V-FITC and PI for flow cytometry analysis. Live cells, are negative for both PI and Annexin V-FITC (lower left quadrant); dead cells are positive for PI but negative for Annexin V-FITC (lower right quadrant); necroptotic cells are positive for both PI and Annexin V-FITC (upper right quadrant), while early apoptotic cells are positive for Annexin V-FITC but negative for PI (upper left quadrant). Histograms represent the percentage of apoptosis and necroptosis. Data represent the means ± SEM of three replicates.**p < 0.01, ns represent not significant.(B) Cells collected from the above experiment were also collected and lysed for Western blot analysis to evaluate the expression of necroptosis-related proteins (RIP1, RIP3, MLKL, p-RIP1, p-RIP3, and p-MLKL) and apoptosis-related proteins (cleaved Caspase-3 and cleaved Caspase-8). GAPDH was used as an internal control.
and considering the impact of inflammatory responses on cell states during pathogen infection (38), we further assessed the involvement of LczDHHC23 in the apoptosis/necroptosis of P. plecoglossicidainfected large yellow croaker MO/MF. The findings indicate that P.
==
Domain: Biology Environmental Science Medicine
|
[{"added": "2024-05-31T15:11:53.094Z", "created": "2024-05-29T00:00:00.000", "id": "270131139", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2024.1401626/pdf", "pdf_hash": "9e74dc99711e35e148b082ac2fb9867ea65f06d1", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:2380", "s2fieldsofstudy": ["Biology", "Environmental Science", "Medicine"], "sha1": "68dfedf2109575bff00291bd1d3e38911756b7b3", "year": 2024}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Prioritizing Land and Sea Conservation Investments to Protect Coral Reefs
Background Coral reefs have exceptional biodiversity, support the livelihoods of millions of people, and are threatened by multiple human activities on land (e.g. farming) and in the sea (e.g. overfishing). Most conservation efforts occur at local scales and, when effective, can increase the resilience of coral reefs to global threats such as climate change (e.g. warming water and ocean acidification). Limited resources for conservation require that we efficiently prioritize where and how to best sustain coral reef ecosystems. Methodology/Principal Findings Here we develop the first prioritization approach that can guide regional-scale conservation investments in land- and sea-based conservation actions that cost-effectively mitigate threats to coral reefs, and apply it to the Coral Triangle, an area of significant global attention and funding. Using information on threats to marine ecosystems, effectiveness of management actions at abating threats, and the management and opportunity costs of actions, we calculate the rate of return on investment in two conservation actions in sixteen ecoregions. We discover that marine conservation almost always trumps terrestrial conservation within any ecoregion, but terrestrial conservation in one ecoregion can be a better investment than marine conservation in another. We show how these results could be used to allocate a limited budget for conservation and compare them to priorities based on individual criteria. Conclusions/Significance Previous prioritization approaches do not consider both land and sea-based threats or the socioeconomic costs of conserving coral reefs. A simple and transparent approach like ours is essential to support effective coral reef conservation decisions in a large and diverse region like the Coral Triangle, but can be applied at any scale and to other marine ecosystems.
Introduction
Coral reefs are the world's most diverse marine ecosystem and are vital to hundreds of millions of people as a source of nutrition, economic opportunity, and storm protection [1]. Due to climate change and local impacts, the state of coral reefs is grim and their protection is urgent [2,3,4]. As with all conservation, limited resources for coral reef protection require that we prioritize where and how to act to efficiently sustain coral reef ecosystems [5].
Local-scale threats to coral reefs originate from both land-and sea-based human activities (e.g. over-fishing, nutrient runoff from farming) [6]. Where both exist, conservation strategies should consider each of them [7,8]. The allocation of conservation resources to coral reefs should depend on which strategies most efficiently reduce their threats [9]. Sophisticated approaches for identifying marine conservation priorities exist [7,10,11], but fail to explicitly address threats originating on land and the associated costs of mitigating these threats through conservation action. Effective conservation prioritization should provide guidance on how to distribute funds between land-and sea-based conservation actions to protect coral reefs.
We address this deficiency by developing the first explicit method for prioritizing conservation actions and locations to costeffectively mitigate land-and sea-based threats to marine ecosystems and apply it to the Coral Triangle, one of the world's highest conservation priorities [6,12]. The multi-lateral Coral Triangle Initiative on Coral Reefs, Fisheries and Food Security (formalized in May 2009) is the focus of significant global conservation attention with financial commitments of at least US $400 million ( [URL]). This amount is likely to be insufficient to achieve the Initiative's goals, thus investments must be prioritized.
Standard advice from business and economics is to invest in projects where the rates of return on investment are the highest [13]. This approach has been applied to the conservation of terrestrial biodiversity [13,14,15], but it has yet to be applied to marine conservation. Any application of the return on investment approach requires an explicit statement of overall objective. The objective in previous studies has been species focused (e.g. maximize the number of species conserved). Here, our objective is to maximize threat reduction to coral reefs across the Coral Triangle's ecoregions through investment in land-and sea-based conservation actions.
Achieving the objective relies upon a rigorous problem formulation and information on threats to marine ecosystems, effectiveness of management actions at abating threats, and the economic costs of actions. We considered 8 threats to coral reefs (Table 1), each associated either with agricultural run-off or fishing, and two actions that reduce their impact on coral reefs: effective management of coastal watersheds and coral reefs. We refer to the places where these actions are implemented as protected areas, but acknowledge that effective management is rare and involves more than just dedicating protected areas, especially in the Coral Triangle [16,17,18]. As a result, protected areas require funding for management and the restriction of profitable activities. Thus, we estimated the management costs and value of foregone usage to farmers and fishers (i.e. opportunity cost) of protected areas.
Using this information, we calculated the rate of return on investment of each action in each ecoregion (denoted ''ecoactions'') [13], where the rate is calculated as the reduction of threats (return) per dollar spent on their reduction (investment). We ranked each ecoaction (e.g. effective management of coral reefs in the Bird's Head ecoregion) in terms of how cost-effective it is at mitigating threats to coral reefs for two scenarios. Scenario 1 reflects investment of management costs alone, whereas scenario 2 also considers opportunity costs. We demonstrate how these rankings can be used to allocate a limited budget for conservation and we compare our results to those based on individual criteria (e.g. cost, species richness).
Materials and Methods
Our method for prioritizing land and sea conservation investments to protect marine ecosystems involves five general steps (Figure 1), described below.
Step 1: Define conservation objective The first step in formulating any conservation resource allocation problem is to define a quantifiable objective. Our objective was to maximize threat reduction to coral reefs across the Coral Triangle's ecoregions through investment in land and seabased conservation actions. We used 16 marine ecoregions that were defined on the basis of coral diversity and endemism, each of which contains 503-553 zooxanthellate coral species [19].
Step 2: Identify threats to ecosystem The second step is to determine the threats to, and their relative impact on, the marine ecosystem. We considered the threats that could be mitigated with local-scale conservation action and their relative impact on coral reefs (Table 1). We used data from Halpern et al. [6] that depicts the impact of anthropogenic drivers of change (henceforth referred to as threats), to each 1 km 2 section of coral reefs [6,20].
Step 3: Identify conservation actions to abate threats The third step is to identify conservation actions and their effectiveness at abating the threats identified in step two. We determined the area available (i.e. not cleared or effectively managed in each ecoregion) for implementing two actions (Table S1): 1) effective management of coastal watersheds; 2) effective management of coral reefs [18]. We assume that each threat reduces linearly with protection of the ecoregion.
A surrogate must be used to represent where and how much of the land and sea is effectively managed at present, as this information does not exist across the Coral Triangle. In theory, protected areas are effectively managed; however, in practice, only a subset of protected areas is effectively managed for biodiversity conservation [16,18]. Therefore, we estimated which coral reefs and terrestrial protected areas are effectively managed based on a few simple guidelines, described below.
Land-based conservation. We only considered sub-catchments that reach the ocean and consider them as part of an ecoregion if their coastal pour-point emptied into the marine portion of that region. We used sub-catchment boundaries and coastal pour-point data from Halpern et al. [6]. We assume that the protected areas are effectively managed in areas containing native vegetation. In this analysis, we used terrestrial protected areas with an IUCN designation from the World Database on Protected Areas from the World Commission on Protected Areas from December 2007. Using SPOT vegetation satellite data, we determined the amount of protected areas containing native vegetation from 2000 [21]. For each ecoregion, we calculated the proportion of land protected under two scenarios: 1) Pessimistic scenario, where vegetated areas in only the more stringently protected areas (i.e. IUCN 1-4) are effective and 2) Optimistic scenario, where vegetated areas in all types of protected areas recognized by the IUCN (IUCN 1-6) are effective.
Marine conservation. We used the global coral reef atlas [22], compiled by the World Conservation Monitoring Centre at the United Nations Environment Programme, to determine the location of coral reefs with each ecoregion. Data indicate the presence/absence of coral reefs for each 1 km 2 cell. Mora et al. [18] provided an assessment on the extent and effectiveness of coral reef protected areas. Each protected area was classified by its regulations on extraction (no-take, take, or multi-purpose) and risk of poaching (low, medium, high). For each ecoregion, we calculated the proportion of coral reef protected under two scenarios: 1) Pessimistic scenario, where only areas with no extraction (no-take, low poaching) are effective at protecting the reefs and 2) Optimistic scenario, where areas with limited extraction (no-take or multipurpose for any level of poaching) are effective at protecting the reefs. We show results that use the amount protected under the pessimistic scenario for terrestrial and marine conservation. However, the ranking results were insensitive to the information used as there is little difference between the amounts protected under each scenario.
Step 4: Calculate costs of implementing actions
We predicted the annual management and opportunity costs associated with land and marine protected areas (Table S1). When applying the method with opportunity costs, we assume that the ecoaction excludes extractive activities and causes economic losses that cannot be recovered in another place or industry. However, in reality, conservation can deliver benefits (e.g. improved fishing yields) that may compensate for some economic losses [23].
Management costs. We used a model developed by Moore et al. [24] to predict the management costs of terrestrial protected areas in each ecoregion, as done in Kark et al [25] and Bode et al [26]. The model states that the cost of managing a protected area is a nonlinear function of the size of the proposed protected area, the Purchasing Power Parity (PPP) of the nation, and the Gross National Income (GNI) of the nation: where all logarithms are of base ten. We modeled the area of protected areas in each ecoregion by the median size of the existing vegetated protected areas (IUCN 1-6) in that ecoregion.
We used a model developed by Balmford et al. [27] to predict the management costs of marine protected areas in each ecoregion. The model states that the cost of managing a marine protected area is a nonlinear function of the size of the proposed protected area, distance of area from land, and the PPP of the nation: where all logarithms are of base ten. We modeled the area of protected areas in each ecoregion by the median size of the existing no-take or multipurpose coral reef protected areas in that ecoregion. We modeled the distance of coral reef protected areas in each ecoregion by the median distance of coral reefs from land. The economic data we used to inform the models described below were obtained from the 2006 International Monetary Fund's Financial Statistics [URL]/. To calculate the PPP, we divided the PPP conversion rate (local $/ international $) reported by the Monetary Fund by the exchange rate (local $/US $). We substituted missing GNI information with the Gross Domestic Product.
As some ecoregions span multiple countries, the management costs are therefore likely to vary substantially. Our analyses treat each region as a homogeneous entity, where the cost is calculated using the Balmford-Moore models [24,27], with parameter values that are the area-weighted average of the constituent nations' exclusive economic zone. The area-weighting method is applied to the other predictor variables as done in Bode et al [26]. Opportunity costs, land. We estimated the opportunity costs of agricultural production from implementation of a protected area that excludes cultivation. The agriculture opportunity cost represents the potential foregone economic returns from agricultural production (cropping and grazing) on areas containing native vegetation [28]. The potential economic returns from agricultural production are estimated at a 59 resolution by the maximum of the potential crop and livestock yields based on land capability, multiplied by the producer price [28]. For each ecoregion, we calculated the maximum potential agricultural profits per unit area of native vegetation.
Opportunity costs, marine. We estimated the lost opportunity costs to fishermen (C) from implementation of a coral reef protected area that excludes fishing: C annual cost, US$ km -2 À Á annual catch, tonneÞÃ catch value, US$ t -1 À Á reef area, km 2 À Á Spatially explicit information on catch rates for small-scale fisheries was determined for each 1 km 2 of coral reef by Halpern et al. [29] from the FAO and Sea Around Us Project (SAUP). Although this is the best available data for artisanal fishing, it is modeled and based on many crude assumptions. Development of a new artisanal fishing model for the Coral Triangle that considers the spatial distribution of catch, population size of species across the region, historical fishing, and fishing method is an area of further research. We summed the catch rates across all coral reefs within each ecoregion. The value (US$, year 2000) of reef fish in each country are provided by the SAUP for reported landings from 1950-2004 [30]. We used the maximum value reported per country to prevent underestimating the value over time. Like management costs, opportunity costs in some ecoregions vary substantially because they span multiple countries. Our analyses treat each region as one entity using the area-weighted average of the constituent nation's exclusive economic zone.
Step 5: Invest where the rate of return on investment is highest The final step is to mathematically formulate the resource allocation problem and determine the rate of return (i.e. reduction of threats) on investment (i.e. cost of reducing threats) of each ecoaction. The overall impact, I i , that a set of threats (k = 1,…,8) have on a 1 km 2 section of coral reef (i = 1,…,N) was defined by Halpern et al. [29] as a weighted sum of land-and sea-based threats where L ik and C ik are threat values originating from the land and sea, respectively, and a k is a weighting reflecting the relative impact of threat k on coral reefs (Table 1).
In step 3, we made the assumption that threat, k, in any 1 km 2 section of reef, i, is reduced linearly with protection of the ecoregion, j: L ik~1 {l j and C ik~1 {c j , where l j and c j are the proportion of terrestrial and coral reef protected area, respectively. Therefore, the average threat impacting coral reefs in each ecoregion (j = 1,…,16) can be written as a function of how much of the land and sea that we protect in an ecoregion, where N j is the number of reef pixels (i) in ecoregion j and S j is the set of indices that determine if pixel i is in region j. In doing this, we assume that the benefit of protection is evenly spread across the ecoregion. This relationship could be modified if more discrete regions were targeted. The proportion of the ecoregion protected is the sum of the portion currently protected (l oj and c oj ) and the portion protected by additional investment. To account for the cost of additional protection, the proportion protected after additional investment made can be expressed as the proportion of additional investment made (x j and y j ) relative to the total cost of land and ocean available for protection (a j and b j ), respectively: l j~loj z x j a j and c j~coj z y j b j : The rate of return (threat reduction) on investment of each ecoaction can then be calculated for each land and sea-based conservation action, respectively: The greater the rate of return on investment per ecoaction, the higher priority it is for investment. In order to achieve the conservation objective, investments should be made in high priority ecoactions unless there are ecoregional or action-specific constraints (e.g. budget or area targets). We show how a budget and area constraint influences the distribution of an arbitrary budget of US $1 B, $400 M, and $100 M. The area constraint ensures that a priority ecoaction receives funding for no more than a designated percentage of its available area, which we arbitrarily selected to be 30%.
Ranking
We applied our prioritization approach to rank ecoactions using different costs and found a high concordance in the rankings (Spearman's rank correlation of 0.88, p,0.001). We present our ranking results for both scenarios at two scales ( Fig. 2): across the entire Coral Triangle and within each ecoregion. At the Coral Triangle scale, we found that terrestrial conservation in one ecoregion is sometimes a higher priority than marine conservation in another ecoregion, especially in scenario 1 (management costs only). For example, the highest ranking terrestrial action (E, North Arafura ecoregion) has a larger return on investment than marine conservation in half of the ecoregions.
Within any particular ecoregion, marine conservation is almost always a higher priority than terrestrial conservation. The one exception is in the North Philippines (Scenario 1), where the marine management cost is substantially larger than on the land (Table S1).
Budget Allocation
We demonstrate two ways these rankings can be used to allocate limited conservation resources under scenario 1 (management costs) (Fig. 3). First, for three different budgets (US $ 1 B, 400 M, and 100 M), we allocate money to the highest ranking ecoactions until it is spent (Fig. 3a). This assumes that within an ecoregion, all available (i.e. not currently protected or developed) coral reefs and land can be effectively managed, which is likely to be unrealistic. Thus, we show how a budget would be distributed to the highest ranking ecoactions if we cap the allocation at protection of thirty percent of the available reef or land affecting the reef (Fig. 3b). To explore the sensitivity of our results to the threat weighting values (Table 1), we performed the analysis with the range of weighting values provided by experts (n = 24) and found that the rankings were robust to these variations (Spearman's rank correlations .0.99, p,0.001). Regardless of weighting values used, rankings for the top seven ecoactions were always the same. The remainder of ecoactions typically did not change rank and never changed by more than four places (Table S2). Depending on the budget, how these subtle discrepancies could impact the distribution of funding are important considerations.
Comparison
We compare our ecoregional rankings to those based on individual criteria (Table 2). Since other approaches do not consider marine and terrestrial conservation actions simultaneously, we compare our ecoregional rankings for marine actions only. We found a lack of concordance between approaches (Spearman's rank correlations from -0.22 to 0.3), indicating that they would recommend different investment priorities. Using estimated management costs, we show how a budget of US $400 M for management of land-and sea-based threats would be distributed We compare rankings on the basis of 1) Return on investment (ROI) analysis for marine conservation for both scenarios; 2) Coral reef species richness; 3) Annual opportunity and management cost (lower cost, higher rank); 4) Average cumulative impact on coral reefs from all human activities [6]. Higher return, richness, and impact values were given a higher rank and equivalent values were assigned the same rank. The spatial location of the ecoregions is indicated by letter in Fig. 2. *Thirty percent of the available reef or land affecting the reef would be effectively managed with a fixed budget of US $400 M. doi:10.1371/journal.pone.0012431.t002 following each ranking scheme ( Table 2). For example, we found that prioritization on cumulative threats alone would only provide enough funding for effective management of 6% of one ecoregion, whereas our approach would ensure that 30% of seven ecoregions were managed.
Discussion
The purpose of this paper is to illustrate a novel approach for delivering cost-effective outcomes for marine conservation that can explicitly trade-off resource allocation decisions among land-and sea-based conservation actions to protect marine ecosystems. Our approach is useful in guiding managers and policy makers in making decisions on where and in which actions to invest. Although it is useful for supporting broad scale resource allocation decisions, the results are not necessarily applicable to all places within an ecoregion as the conservation context may vary between communities [31]. However, the method can also be applied at a local-scale (e.g. provincial or catchment level), using more conservation actions (e.g., run-off management, improved agricultural practices, fishing gear-based management). In addition, more specific data on social and economic costs would need to be estimated for a local-scale application as the data we used may be too coarse, especially for management costs. The effectiveness of any local conservation plan in this region is reliant upon community involvement and the consideration of indigenous knowledge, management practices, and property rights [31,32,33].
One of the key results -terrestrial conservation in one ecoregion is sometimes a higher priority than marine conservation in another ecoregion -is contrary to current conservation strategies, which typically do not trade-off marine and terrestrial conservation actions to protect marine ecosystems, and suggests that more costeffective conservation outcomes could be achieved using our method. Although another key result -within any particular ecoregion, marine conservation is almost always a higher priority than terrestrial conservation within an ecoregion -generally supports current management practice in any given place, greater conservation outcomes could be achieved when the entire region is considered.
Incorporating different socioeconomic costs did not significantly affect outcomes. However, decisions following each scenario are likely to have different social and economic implications. For example, investments including opportunity costs (Scenario 2) are more likely to minimize impact on fishers and farmers as they were explicitly considered in the analysis [34,35,36]. However, scenario 2 assumes that people would be compensated for displacement due to conservation and that conservation actions preclude subsistence farming and fishing, both of which are unlikely.
We assume that each threat reduces linearly with protection of the ecoregion (Step 3, materials and methods). This represents the most parsimonious relationship between threat and protection but could easily be modified if more detailed information were available for each ecoaction. This type of information is difficult to obtain as effective monitoring and good quality data relevant to this is lacking [14]. However, in a region with little protection and a limited budget for conservation, the use of a non-linear function that demonstrates diminishing returns may not substantially impact the results. Testing this on a specific region where this type of information could be obtained would be informative. Assessing the benefits of conservation actions, including the relationship between reducing threats and biodiversity, is a significant challenge and research priority in conservation.
Other applications of the return on investment framework to inform the allocation of resources to protect terrestrial biodiversity use a non-linear benefit function based on the species-area relationship where the total number of species (S) present in area (A) is a power-law function of that area [13,14]: S~aA z . This relationship is an appropriate estimation of the benefits of protection when the objective is to conserve species, as in these studies; however, it is not applicable to our objective (i.e. threat reduction to coral reefs). Although we aimed to solve one objective, application of the return on investment thinking can be used to solve a range of conservation objectives to conserve marine ecosystems [13].
Priorities and investment plans following our approach versus that based on individual criteria (Table 2) would be substantially different. In addition, prioritization on species information alone, for example, will not be able to inform how funding should be divided between management actions on the land and in the sea. Similar confusion can arise if we prioritize only on cost or threat.
Our method could be adapted to provide more specific guidelines on how much and when (i.e. timing of investments) to invest in ecoactions [13]. Such analyses may require information on budget (size and constraints), benefits of conservation (e.g. payments for ecosystem services), more specific conservation actions, social adaptive capacity indicating the likelihood of a project succeeding (e.g. willingness of people to forego resources) [31], distribution of species, more opportunity costs (e.g. aquaculture and forestry), a better understanding of the effectiveness of management actions, coral reef resilience [37], and other relevant threats (e.g. sedimentation from deforestation). At any scale, neglecting to properly address social costs to resource users will most likely lead to unsuccessful conservation plans [38,39]. These are areas of further research.
Although we apply our prioritization approach to the Coral Triangle Initiative, we acknowledge that our analysis is focused on a small aspect of the conservation problem in the Coral Triangle. In addition to identifying priority areas for effective management (Goal 1 in the Regional Plan of Action), the Coral Triangle Initiative aims to achieve outcomes relevant to fisheries management, climate change adaptation, and threatened species [40]. However, it is important to note that effective management of coral reefs at a local scale can increase their resilience to global threats such as climate change [9].
A simple, transparent, and economically grounded approach like ours is essential to making any conservation decisions in a large and diverse region like the Coral Triangle, where the budget is primarily financed from international aid. Effective conservation of marine resources must consider land-and sea-based human activities and their management costs [41]. The lack of a defensible resource allocation plan could lead to costly and contentious conservation strategies that do not protect biodiversity, impeding additional global funding to one of the world's most biodiverse and threatened regions.
Supporting Information
Table S1 Cost and protected area data for coastal catchments and coral reefs in each ecoregion. Found at: doi:10.1371/journal.pone.0012431.s001 (0.04 MB DOC)
Table S2
Ranking results from scenario 1 compared to results from the impact weighting value sensitivity analysis. We perform the analysis with the maximum and minimum impact weighting values provided by experts for the land-and sea-based threats. Found at: doi:10.1371/journal.pone.0012431.s002 (0.05 MB DOC)
==
Domain: Biology Environmental Science Medicine
|
[{"added": "2014-10-01T00:00:00.000Z", "created": "2010-08-30T00:00:00.000", "id": "9378759", "metadata": {"extfieldsofstudy": ["Biology", "Environmental Science", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0012431&type=printable", "pdf_hash": "9b71d8cb9a1d7910256fae083ae4d2fb348fa7d4", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:1027", "s2fieldsofstudy": ["Environmental Science"], "sha1": "38c0c5f3bd859a824d31a6d1a5a89165ea0b51c9", "year": 2010}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Exposure to airborne bacteria depends upon vertical stratification and vegetation complexity
Exposure to biodiverse aerobiomes supports human health, but it is unclear which ecological factors influence exposure. Few studies have investigated near-surface green space aerobiome dynamics, and no studies have reported aerobiome vertical stratification in different urban green spaces. We used columnar sampling and next generation sequencing of the bacterial 16S rRNA gene, combined with geospatial and network analyses to investigate urban green space aerobiome spatio-compositional dynamics. We show a strong effect of habitat on bacterial diversity and network complexity. We observed aerobiome vertical stratification and network complexity that was contingent on habitat type. Tree density, closer proximity, and canopy coverage associated with greater aerobiome alpha diversity. Grassland aerobiomes exhibited greater proportions of putative pathogens compared to scrub, and also stratified vertically. We provide novel insights into the urban ecosystem with potential importance for public health, whereby the possibility of differential aerobiome exposures appears to depend on habitat type and height in the airspace. This has important implications for managing urban landscapes for the regulation of aerobiome exposure.
near-surface aerobiomes in urban green spaces. Mhuireach et al. 22 showed that aerobiomes in urban green and grey spaces had distinct compositions. Subsequent studies have shown vegetation type has a potential modulating effect on aerobiome diversity and composition 23,24 . Stewart et al. 25 found that aerobiomes varied in composition and function between urban and suburban sites. Mhuireach et al. (2019) identified localised influences on aerobiomes, including weather and land management 22,26 . Our recent work has also demonstrated aerobiome vertical stratification between ground level and 2 m heights in an urban green space 27 . Together, these studies suggest that individuals may be exposed to different aerobiomes depending on the type of habitat visited and human-scale height-based variation in environmental aerobiomes. Consequently, understanding the effects of habitat and height--and their interactions-on aerobiomes could have important implications for public health.
There is growing recognition that urban green spaces are important for human health and wellbeing through provision of psychosocial and biological benefits [28][29][30][31][32] . Gaining a deeper understanding of urban green space aerobiome exposure potential could inform public health and environmental management strategies in the future. In this study, we used an innovative columnar sampling method to sample aerobiome bacterial communities in three urban green space habitat types in the Adelaide Parklands, South Australia. These habitats included amenity grasslands, woodland/scrub (dominated by native Eucalyptus spp. trees and shrubs; henceforth referred to as 'scrub'), and bare ground habitat; each is a typical urban green space habitat. We conducted next generation sequencing of the bacterial 16S rRNA gene to characterise the diversity, composition and network complexity of aerobiomes. We also applied geospatial analytical methods to explore the potential influence of trees on the micro-biodiversity of aerobiomes. Our primary objectives were to: (a) assess aerobiome composition and microbiodiversity differences between the three habitats; (b) compare aerobiome vertical stratification between the different habitats; (c) assess whether tree density, distance to trees, and tree canopy coverage influenced bacterial alpha diversity; and, (d) to assess any differences in known pathogenic bacterial taxa between habitats and sampling heights.
Results
Bacterial communities were dominated by three key phyla in all three habitats: Proteobacteria, Bacteroidetes, and Actinobacteria, however, abundance differed depending on height (Fig. S1) (full description of sequencing reads in Supplementary Materials, Appendix B). We now present the results in order of the objectives (a-d) set out in the Introduction.
Comparison of bacterial alpha diversity between habitats. We found that bacterial alpha diversity of the soil differed significantly between habitats (ANOVA F = 3.95, df = 1, p = 0.03) (Fig. 1). The soil microbiome from the scrub habitat was significantly more biodiverse than the grassland habitat (Tukey multiple comparison Table S1). We also tested for mean alpha diversity differences between dates and sites, showing that sampling dates and individual sites were generally not a factor in alpha diversity variation with nearly 90% of comparisons showing non-significant results (Appendix C, Table S2). We also found that bacterial alpha diversity of the air differed significantly between bare ground and scrub habitats (Chi-squared = 11.3, df = 1, p ≤ 0.01), with the scrub aerobiome being more biodiverse than the bare ground. Aerobiome alpha diversity of scrub and grassland were also significantly different (Chi-squared = 24.8, df = 1, p ≤ 0.01), and the scrub aerobiome was the most biodiverse. No significant difference was observed in alpha diversity between bare ground and grassland habitats (Chi-squared = 0.46, df = 1, p ≤ 0.49).
Comparison of bacterial beta diversity between habitats. We observed clear differences in aerobiome compositions (beta diversity) (PERMANOVA, df = 2, F = 3.7, R 2 = 0.07, p ≤ 0.01, permutations = 999) and soil samples (PERMANOVA, df = 2, F = 6.8, R 2 = 0.36, p ≤ 0.01, permutations = 999 among habitats (Fig. 2). For air samples, all habitats displayed significantly distinct bacterial communities, where habitat type explained 7% of the variation in bacterial community composition. However, there was significant heterogeneity in dispersion (PERMDISP, F = 13, p ≤ 0.01). For soil only, habitat type explained 36% of the variation in bacterial community composition, however, this increased significantly to 75% and 74% when comparing scrub to grassland and scrub to bare ground, respectively (PERMANOVA, df = 5, F = 7, R 2 = 0.75 and 0.74, p ≤ 0.01). There was no significant heterogeneity in dispersion (PERMDISP, F = 2, p = 0.07). We carried out PERMANOVA analysis for the beta diversity/community composition (using the centre log ratio transformation) for each site and date prior to running the broader compositional analysis. This step showed that there was no statistically significant variation in bacterial community composition between either dates or sites (Supplementary Materials, Appendix C, Table S2).
Vertical stratification: alpha diversity. For the bare ground habitat, we observed a strong negative correlation between alpha diversity (air and soil for all sites/dates) and sampling height from ground level to 2 m (Pearson's r = -0.75, df = 39, p ≤ 0.01) (Fig. 3a). Alpha diversity (Shannon scores) ranged from 1.2 to 5.93 and was highest at soil level, followed by lower air sampling levels (0.0-0.5 m) and upper sampling levels (1.0-2.0 m), respectively. Analysis of air-only samples also showed a significant negative correlation between height and bacterial alpha diversity, demonstrating vertical stratification in this bare ground habitat (Pearson's r = − 0.60, df = 30 p ≤ 0.01). For the grassland aerobiome, the correlation was weak and not significant, indicating that vertical stratification was not detected in this habitat (Pearson's r = 0.03, df = 34, p = 0.86). When analysing air and soil (for all sites/dates) together, we observed a significant negative correlation between alpha diversity and sampling height (Fig. 3b). Alpha diversity ranged from 1.2 to 5.9 and was highest at soil level.
In the scrub aerobiome, we observed a significant negative correlation between alpha diversity (air and soil for all sites/dates) and sampling height from ground level to 2 m (Pearson's r = − 0.59, df = 39, p ≤ 0.01) (Fig. 3c). Bacterial alpha diversity in the scrub habitat ranged from 1 to 6 (Shannon score) and was highest at soil level, followed by lower air sampling levels (0.0-0.5 m) and upper sampling levels (1.0-2.0 m), respectively. Analysis of air-only samples showed a significant negative correlation between height and bacterial alpha diversity, demonstrating vertical stratification in this scrub habitat (Pearson's r = − 0.38, df = 30, p = 0.03). See Supplementary Materials, Appendix B, Table S3 for a statistical comparison of data between individual dates and sites (90% were non-significant relationships, indicating low inter-date and site variation).
Vertical stratification: beta diversity. Sampling heights in the bare ground habitat displayed disparate bacterial compositions. Sampling height explained 29% of the variation in bacterial community composition when all air sampling heights were included (PERMANOVA df = 4, F = 3.67, R 2 = 0.29, p ≤ 0.01, permutations = 999). Analysis of air samples for the bare ground habitat in isolation showed that sampling height still explained 25% of the variation in bacterial community composition (df = 3, F = 3.06, R 2 = 0.25, p ≤ 0.01, permutations = 999).
Air sampling heights in the grassland habitat displayed disparate bacterial communities to the soil. Sampling height explained 24% of the variation in bacterial community composition when all air sampling heights were included (PERMANOVA df = 4, F = 3.17, R 2 = 0.24, p ≤ 0.01, permutations = 999). However, analysis of grassland air samples in isolation showed that sampling height only explained 9% of the variation in bacterial community composition and was not statistically significant (df = 3, F = 1.06, R 2 = 0.09, p = 0.24, permutations = 999).
Sampling heights in the scrub habitat displayed disparate bacterial communities. Sampling height explained 22% of the variation in bacterial community composition when all air sampling heights and soil were included (PERMANOVA df = 4, F = 2.9, R 2 = 0.22, p ≤ 0.01, permutations = 999). Analysis of air samples in isolation showed that sampling height still explained 11% of the variation in bacterial community composition (df = 3, F = 1.30, Vertical stratification: aerobiome network analysis. In spite of differences in bacterial community composition and alpha diversity among the three study sites, network analyses showed an increase in the community complexity and interactions, defined by node degree and network size, at lower heights as compared to higher heights (Fig. 4). Bacterial operational taxonomic units (OTUs) in the scrub habitat at 0-0.5 m heights had the highest node degree, while the OTUs in the grassland habitat 1-2 m had the lowest node degree. At lower heights, the average association of any OTU in the grassland was less (node degree = 2.7) than the average association of OTUs for scrub (node degree = 4.9) and bare ground (node degree = 4.7) habitats. At upper heights, node degree for OTUs was highest for bare ground (2.7) followed by scrub (1.8) and grassland (1.7). Evaluation of link type, either positive or negative links, suggested a positive association among most OTUs, except for scrub 1-2 m which only had a small number of negative associations (Fig. 4). Comparisons of modularity between heights across the study sites suggested an increase in the network modularity at higher heights, despite the decrease in network connectance and node degree. Percentage of change in the modularity between heights was highest in the grassland (~ 50%), although there were fewer nodes per module.
The relationship between tree metrics and bacterial alpha diversity. In the air samples, we found Differentially abundant and notable taxa. There were 53 differentially abundant genera across habitat types (based on log-twofold-change with adjusted p ≤ 0.05). The top three, for example, in the scrub habitat were: Gillisia, Sphingobium, and Kutzneria; in grassland: Parvibaculum, BSV43, and Pseudomonas; and in bare ground: Rudanella, Bacteroides, and Actinomyces. We also observed vertical stratification of differentially abundant taxa and at the putative species level. After unclassified taxa were removed, we confirmed identity (100% match) via Basic Local Alignment Search Tool (BLAST) against the NCBI database 33 . In the bare ground habitat, we found 30 differentially abundant taxa assigned at the putative species level. Sixteen of these significantly decreased in relative abundance with sampling height and 14 significantly increased (p ≤ 0.01). In the grassland habitat, we found 40 differentially abundant taxa assigned at the putative species level. Thirty-two of these significantly decreased with sampling height and 8 significantly increased (p ≤ 0.01). In the scrub habitat, we found 16 differentially abundant taxa assigned at the putative species level. Ten of these significantly increased with sampling height and 6 significantly decreased (p ≤ 0.01). Using BLAST and a literature search, we found putative differentially abundant human pathogens in each habitat (Fig. 7). A 2-sample test for equality of proportions with continuity correction showed a significant difference in proportions of identifiable pathogenic species between grassland and scrub (Chi-squared = 5.57, df = 1, p ≤ 0.02) but not between other habitats, where grassland samples exhibited significantly greater proportions of identifiable pathogenic species compared to scrub. Moreover, 87% of these significantly decreased with sampling height based on log-twofold-change differential analysis (p ≤ 0.01). These taxa contain bacteria that have been associated with a number of diseases, including infective endocarditis (Rothia mucilaginosa) 34 and gut mucosal damage (Prevotella copri) 35 . More information on these diseases can be found in Supplementary Materials, Appendix D, Fig. S3.
Environmental metadata. In terms of the environmental metadata, there was only one significant association with bacterial alpha diversity; aerobiome alpha diversity decreased significantly in scrub habitat as windspeed increased (Spearman's r s = − 0.88, β = − 0.88 (− 0.98 to − 0.5), p ≤ 0.01) (full details in Supplementary Materials, Appendix C, Table S7).
Discussion
Here we show that aerobiome alpha and beta diversity (community composition) differed significantly between urban green space habitat type, and that aerobiome diversity, composition and network complexity also stratified vertically. The level to which this occurred was dependent on habitat type. Therefore, potential bacterial exposure levels and transfer loads to humans will likely differ depending on habitat type as well a person's height and behaviour. We also confirmed low variation in bacterial alpha and beta diversity between sampling dates and sites (Appendix C). Our results confirmed that more trees, closer proximity to trees, and greater canopy coverage associate with higher aerobiome diversity, which could have important implications for landscape management and public health as growing emphasis is placed on designing and managing green spaces for 36 . We also found that grassland samples exhibited significantly greater proportions of identifiable pathogenic bacteria compared to scrub, and their abundance decreased significantly with sampling height. Our study was conducted only in the Adelaide Parklands, South Australia and therefore may not be representative of urban green spaces and climatic zones in other areas. Future work should explore these trends in additional geographical, socioeconomic, cultural areas to understand both generalisability and opportunities to optimise green space exposure for health benefits. 5,14,39 . For example, environmental microbiomes are essential in the development and regulation of immunity 1,6 , and soil-derived butyrate-producing bacteria may supplement gut bacteria and have anxiety-reducing effects 5 . Importantly, urban green space exposure can result in transmission of environmentally-derived bacteria to the skin and airways 21 . Furthermore, a recent study showed that transfer of bacteria from biodiverse environments enhanced immunoregulatory pathways in children 40 . Consequently, environments with different levels of bacterial diversity may affect the potentiality of bacterial exposure levels and transfer loads, warranting further research. We found differentially abundant putative pathogenic taxa and showed significant differences in proportions between grassland and scrub habitat samples. In other words, amenity grassland seemed to exhibit a significantly greater proportion of (identifiable) pathogenic species compared to scrub samples. However, considerably more research is needed to fully explore the validity and generalisability of these results. As with many microbial ecology studies, only identifiable bacterial taxa were used in the differential abundance and analyses that identified the pathogenic taxa (i.e., unclassified taxa were removed). This could result in recording bias with implications for validity. Our results suggest that tree density, distance to nearest trees, and tree canopy cover could have a considerable influence on aerobiome alpha diversity. This corroborates reports of trees acting as stationary vectors, spreading bacterial cells in the air 41 . Complex plant detritus (leaf litter) and organic matter at the base of trees, and corresponding soil-microbe systems, may also contribute to tree-associated aerobiomes. The number of trees and amount of canopy coverage within a given radius correlated strongly with alpha diversity. Furthermore, negative correlations were shown between distance to nearest trees and bacterial alpha diversity for air and soil. This supports the results of the tree density associations and suggests that closeness to trees could be important. These results could have important implications for landscape management and public health. Indeed, there have been widespread calls to improve urban ecosystem services by augmenting tree coverage (e.g., to help reduce urban heat island effects 42 , support wildlife 43,44 , improve sleep 45,46 , and capture precipitation to reduce flood risk 47 ). There is also a need to restore complex vegetation communities and host-microbiota interactions that provide multifunctional roles in urban ecosystems 37,48,49 . An important limitation in our study was that tree species and structural diversity metrics were not used. These additional measures could have enriched the quality of analysis and implications of our results and further research that takes these factors into account is needed. However, our findings suggest additional co-benefits from increasing urban tree coverage due to its potential to mediate aerobiome alpha diversity. Our results also corroborate other studies showing microbial alpha diversity increasing along densely-urban to semi-natural environmental gradients 50,51 .
Our results suggest that aerobiome beta diversity (compositional differences) differs between habitats. The results imply that microbial communities in the soil of the scrub habitat are significantly different to bare ground www.nature.com/scientificreports/ and grassland, which are more compositionally aligned. It is possible that bacterial homogeneity between grassland and bare ground is attributed to homogeneity of vegetation complexity 52 . In other words, phyllosphere (total above-ground portion of plants) and rhizosphere (soil root zone) presence and complexity create conditions for different microbial relationships and thus compositional disparity with less botanically-complex or depauperate habitats 38,39 . Taken together with the alpha diversity results, significantly more bacterial species and unique communities exists in scrub habitat samples compared to grassland and bare ground samples. This could mean that humans are exposed to a greater diversity of bacteria in the scrub habitat. Future studies should focus on the functional relevance of these findings. Moreover, combining samples from across consecutive days has limitations in that uncertainty increases. Although our PERMANOVA tests showed significantly low variation across dates and sites, future work should increase inter-date sample sizes and apply uncertainty-reducing methods such as mixed effect models.
Aerobiome vertical stratification. In our study, vertical stratification in bacterial alpha and beta diversity occurred in the bare ground and scrub habitat. However, for the grassland aerobiome, both alpha and beta diversity were relatively stable as height increased. This is the first study to demonstrate that aerobiome vertical stratification is contingent on habitat type, which is important for potential human exposure. As mentioned, urban green space exposure can result in transfer of environmental bacteria to the skin and respiratory tract 21 , and our study shows that the composition and diversity of aerobiome bacteria may differ between heights (from ground level to 2 m). Consequently, there could be different bacterial exposure levels and transfer loads depending on a person's height and activity 27 , however, further confirmatory research is needed. Our results suggest that this may not be the case in amenity grassland where bacterial alpha and beta diversity exhibited high levels of homogeneity among heights. Further research is required to determine the reasons for the lack of vertical stratification in grassland. However, we hypothesise that lower baseline diversity, bacterial resources, openness and airflow in this habitat may be contributing factors. Our study also provides some evidence that different urban green space habitats and heights may not only affect exposure levels and transfer loads of bacterial diversity, but also the presence of notable and potentially pathogenic species for humans. The relative abundance of pathogens identified in the grassland habitat decreased significantly with sampling height. It is possible that a number of these potential pathogens may originate from larger air-sheds (consistent with increasing relative abundance with height), however grasslands may have lesser capacity, compared to scrub or bare ground, to present barriers to this broader airflow or contribute to a more locally distinctive aerobiome. These findings highlight the need for further empirical studies focusing on functional interactions in the environment-aerobiome-health axis. Future studies should also try to control for factors such as human presence/movement through the study sites. Our network analyses also provided evidence to support aerobiome vertical stratification. We saw a decrease in bacterial interactions and network complexities with increased network modularity at higher heights compared to lower heights across habitats, which might be attributed to reduced bacterial diversity with sampling height. This pattern might be due to increasing influence, with increasing height, of diluted and somewhat homogenised aerobiomes from larger airsheds, representing the physical mixing of air (and therefore aerobiomes) from multiple different and distant ecological sources. Increased modularity with reduced network size and interactions may also indicate the existence of relatively simplified, yet modular bacterial communities at higher heights. This could be the function of sparse food resources, especially if associations in the networks reflect niche-based interactions. Increased modularity indicates the presence of dense connections between bacteria within modules but sparse connections between bacteria in different modules, whereas reduced connectance means reduced probability of interactions between any pair of bacteria. Increased modularity with reduced connectance often indicates ecological stability 53 . Moreover, presence of mostly positive associations might also suggest cooperation for resources or lack of competition among the interacting OTUs in the community. While associationbased networks allow a depiction of potential interactions among OTUs and portray community structure, they do not separate niche-based and biological interactions. Experiments with cultures are recommended if future researchers wish to dissociate interaction types and understand the biological mechanisms behind such interactions and network complexity. This action could be important to gain a greater ecological understanding of aerobiome assembly (including vertical stratification), dynamics, and the potentiality of bacterial exposure. Our results provide strong evidence that vertical stratification is a key factor not only in aerobiome diversity and composition, but also in aerobiome interactions, community structure and complexity.
In conclusion, our study provides evidence that bacterial alpha and beta diversity differed significantly between habitats, with scrub habitat providing the most biodiverse aerobiomes. We provide evidence supporting the presence of aerobiome vertical stratification in bacterial community diversity, composition and complexity, which also differed in a habitat-dependent manner. Our results confirmed that more trees, closer proximity to trees, and greater canopy coverage associated with higher alpha diversity of the aerobiome. Finally, we found that grassland samples exhibited significantly greater proportions of identifiable putative pathogenic bacteria compared to scrub, and their richness decreased significantly with sampling height. As discussed, there is growing evidence to suggest that exposure to biodiverse aerobiomes may contribute towards the development and regulation of immunity and support mental health 1,4-6 . Gaining a greater understanding of bacterial transmission routes, exposure levels, transfer loads, and downstream health implications is required. This aerobiome characterisation study provides novel insights into the urban ecosystem to help encourage further empirical investigations. Future research should focus on the functional interactions between humans and the aerobiome. Although additional research is required, our findings also support calls to increase urban tree cover. Exploring the mediatory roles of trees in aerobiome compositional and functional diversity could have important implications for landscape management and public health.
Materials and methods
Site selection. Our study was undertaken in the southern Adelaide Parklands (Kaurna Warra Pintyanthi), South Australia, which comprised nine vegetated plots that spanned approx. 18 ha (central geographic coordinates: latitude − 34.937866, longitude 138.60747). The nine plots included three amenity grasslands, three scrub, and three bare ground (exposed soil) habitats. There were several justifications for selecting this site: (1) the southern Parklands occur within the Upper Outwash Plain soil boundary (coalescing alluvial soil, draining the Eden Fault Block), which provided broad consistency in soil geochemistry; (2) a single section of the Parklands provided control over potential microgeographic variation effects on the aerobiome (e.g., distance to coast, elevation, orientation, aspect, and dominant vegetation communities); (3) the Parkland habitats are representative of the types of green spaces that urban residents are regularly exposed to when commuting or recreating; and, (4) the City of Adelaide provided guidance in the selection process, identifying accessible (and inaccessible) plots.
We defined the boundaries of the nine plots (as polygons) in QGIS 3 (v3.0.2) in conjunction with the City of Adelaide. Using spatial shapefiles for the plot boundaries, we generated random point algorithms to provide random sampling points within each of the nine study plots (Fig. 8a). We recorded geographic coordinates for each sampling point and programmed them into a handheld global positioning system (GPS) receiver. We operated the GPS receiver in the field, allowing us to pinpoint the locations for the sampling stations.
Sampling equipment. Sampling stations (Fig. 8b-e) were constructed using timber (42 mm × 28 mm × 2.7 m), steel brackets, hooks and guy-lines 27 . We secured lab-grade clear plastic petri dishes (bases and lids) to the sampling stations, which were used to sample the aerobiome via passive sampling 22 .
On-site setup. We installed the sampling stations on site between 0600 and 0800 h on the 4th, 5th and 6th November 2019. At 0800 h, sampling stations were decontaminated using a 5% Decon 90 solution. We then installed the petri dishes for passive sampling. The data loggers were also decontaminated. In the scrub habitat (defined as vegetation dominated by locally native shrubs, usually ≤ 5 m tall, with scattered trees) 54 , the nearest trees and shrubs were between 2 and 5 m from the sampling stations, and all trees were < 10 m height and 20-50 cm in diameter at breast height 25 . Sampling protocol. We installed temperature and relative humidity data loggers at each sampling station 27 .
We programmed each logger to record data at 8-s intervals for the entire sampling period. At the start of each sampling day, we calibrated the dataloggers using a mercury thermometer (Gerotherm) and a sling psychrometer (Sper Scientific 736700). We collected other metadata including windspeed and soil pH (Alotpower digital meter). We inserted the pH meter into the soil for a period of 1-min before taking a reading (manufacturer's instructions). We obtained data for windspeed and direction from Adelaide's meteorological weather station at Ngayirdapira (West Terrace): Lat: − 34.93, Lon: 138.58, Height: 29.32 m. We also used a handheld anemometer (Digitech QM-1644) to record these parameters hourly at each sampling site 22 .
Soil samples.
We used a small, decontaminated shovel to collect topsoil samples and stored these in sterile 50 mL falcon tubes. We collected five topsoil samples (approx. 0-5 cm depth) at equidistant sampling points, 20-30 cm from the stem of each sampling station 55 . We pooled and homogenised the soil samples, passed them through a decontaminated 1 mm pore sieve, and placed them in new sterile and labelled 50 mL Falcon tubes. We included field controls for the soil by opening sterile falcon tubes for 60 s in the equipment box at each site 56 . We placed all soil and field control samples immediately into an ice box, and stored the samples in the lab at -80 °C prior to further processing 57 . In total, we collected 45 soil subsamples per sampling day across nine sampling stations with three temporal replicates (on three consecutive days). We pooled and homogenised subsamples by sampling station and day, which gave a total of 27 homogenised samples (nine per sampling station) plus 9 field controls.
Aerobiome samples. To collect aerobiome samples, we used a passive sampling technique, following established protocols 22,26 . We installed petri dishes (100 mm × 15 mm) with Velcro tabs on the sampling stations at four sampling heights: ground level (i.e., 0.0 m), 0.5 m, 1 m, and 2 m. As with soil samples, aerobiome samples were collected on three consecutive days. The total height of the sampling station was 2 m from ground level (95% of typical adult male heights lie within 2 SD at 1.93 m, and 1.78 m for females based on a study across Europe, North America, Australia and East-Asia) 58 . Various human characteristics informed the height selection (e.g., representation of adults vs. children, and different activities such as sitting, crawling, walking) 27,59,60 . We decontaminated the steel plates supporting the petri dishes with 5% Decon 90 solution.
We secured the petri dishes to the sampling stations (Fig. 8), leaving them open for 6-8 h 22 , and closing them at the end of the sampling period. To reduce contamination, new disposable laboratory gloves were worn for each vertical sampling point. Once sampling was complete, we sealed the petri dishes using Parafilm, labelled and transported them to the laboratory (on ice) for storage at -80 °C 26 . We collected field control samples by leaving unused petri dishes for 60 s in the equipment box and sealing them at each site. DNA extraction, amplification and sequencing. We extracted DNA from soil and air samples at the facilities of the Evolutionary Biology Unit, South Australian Museum. Using a digital number randomiser, we processed samples on a randomised basis. We processed the low biomass air samples prior to the higher biomass soil samples to minimise cross-contamination. 5,26,27,61 . All swabbing was carried out in a laminar flow cabinet type 1 (License no. 926207) and each sample was swabbed for 30 s. Samples from the base and lids of each petri dish for each height were pooled (intra-height pooling only). We cut the swabs directly into Eppendorf tubes (2 mL). We used Qiagen QIAamp DNA Blood Mini Kits to extract DNA from the swabs and extraction blank controls. For extraction blank controls, we used sterile water and reagents instead of a sample and all DNA extraction 62 in R. Prior to analysis of compositional data, we used centre log-ratio (clr) transformations 63 . Information acquired from this approach is directly relatable to the environment 64 . We generated violin plots with ggplot2 65 to visualise the distribution of the alpha diversity scores for each habitat and height. Bacterial beta diversity was visualised using ordination plots of Aitchison distances based on clr-transformations of OTU abundances. Ordination plots show low-dimensional ordination space in which similar samples are plotted close together, and dissimilar samples are plotted far apart. We used permutational multivariate analysis of variance (PERMANOVA) to test for compositional differences between different sites, habitats and sampling heights, and permutation tests for homogeneity of multivariate dispersions using vegan 66 in R. Pearson's product-moment and Spearman's rank correlation tests were used to examine correlations between habitat, sampling height and alpha diversity scores. Using phyloseq, we calculated OTU relative abundances to examine the distribution of taxa that may have potential implications for public health. We used DESeq2 67 in R to conduct differential abundance analysis for all air samples based on log-twofold-change. To compare presence and proportions of taxa we used 2-sample tests for equality of proportions with continuity corrections. We also applied bootstrap resampling to assign a measure of accuracy to sample estimates for the Spearman's correlations, using a minimum of 1000 iterations. This was carried out with the psych 68 and boot 69 packages in R.
In order to understand the effect of vertical stratification on bacterial interactions and community structures, we evaluated association-networks of bacterial OTUs. We combined the OTU database from 0 to 0.5 m and 1 to 2 m for each site, and constructed two networks per site (i.e., lower and upper height), such that in total six networks were evaluated across the three habitats. In the evaluated network, nodes represent OTUs and links exists between a pair of OTUs if their frequencies are significantly associated (absolute abundance > 0.7, p ≤ 0.01). The type of association, whether positive or negative, was represented with blue and red links, respectively. To account for compositional bias associated with OTU data, we used SparCC 70 to define associations, and only OTUs with sequence counts > 10 were included. Randomly permuted (n = 100) data were used to estimate significance of associations, and igraph 71 was used to visualize and evaluate the plots. We also ran Spearman's correlation tests with bootstrap resampling to determine whether environmental metadata (pH, temperature, windspeed) associated with bacterial alpha diversity. Outliers were considered as data points more than 1.5× above the third quartile or below the first quartile. Geospatial analyses. We investigated possible relationships between aerobiome samples and surrounding vegetation properties using spatial buffer zones. For the buffer analysis, we used vector geoprocessing tools in QGIS 3. Buffer sizes of 10 m, 25 m, 50 m, and 100 m were considered appropriate for the study scale. Similar distances have been used in previous green space and epidemiology studies 31,[72][73][74] . A 100 m maximum buffer radius was chosen; at greater distances, effects would no longer be local to the sampling points (i.e., they would overlap with other sampling points). To determine tree canopy cover within each buffer radii, ESRI shapefiles were imported into i-Tree Canopy 75 . This enabled random sampling points (between 50 and 250 points per buffer) and selection of land cover classification and associated metrics overlaid with Landsat 8 satellite imagery [75][76][77] . Tree count and distance measures were acquired using geometry tools in QGIS 3. We explored the potential relationship between tree metrics and mean bacterial alpha diversity values of combined air-only sampling heights (n = 27 samples).
Data availability
All data and code used in this study are available on the UK Data Service ReShare. Data collection ID: 854551.
==
Domain: Biology Environmental Science
|
[{"added": "2020-11-19T09:16:38.137Z", "created": "2020-11-11T00:00:00.000", "id": "227062633", "metadata": {"extfieldsofstudy": ["Biology", "Environmental Science"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41598-021-89065-y.pdf", "pdf_hash": "5d465be5bcc305a3b1db7c29e89c69272968c6e3", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:7233", "s2fieldsofstudy": ["Environmental Science"], "sha1": "8b3b72e6a8a0cc9628e3b15cdd0901ce03f701e8", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Standardising Visual Control Devices for Tsetse Flies: Central and West African Species Glossina palpalis palpalis
Background Glossina palpalis palpalis (G. p. palpalis) is one of the principal vectors of sleeping sickness and nagana in Africa with a geographical range stretching from Liberia in West Africa to Angola in Central Africa. It inhabits tropical rain forest but has also adapted to urban settlements. We set out to standardize a long-lasting, practical and cost-effective visually attractive device that would induce the strongest landing response by G. p. palpalis for future use as an insecticide-impregnated tool in area-wide population suppression of this fly across its range. Methodology/Principal Findings Trials were conducted in wet and dry seasons in the Ivory Coast, Cameroon, the Democratic Republic of Congo and Angola to measure the performance of traps (biconical, monoconical and pyramidal) and targets of different sizes and colours, with and without chemical baits, at different population densities and under different environmental conditions. Adhesive film was used as a practical enumerator at these remote locations to compare landing efficiencies of devices. Independent of season and country, both phthalogen blue-black and blue-black-blue 1 m2 targets covered with adhesive film proved to be as good as traps in phthalogen blue or turquoise blue for capturing G. p. palpalis. Trap efficiency varied (8–51%). There was no difference between the performance of blue-black and blue-black-blue 1 m2 targets. Baiting with chemicals augmented the overall performance of targets relative to traps. Landings on smaller phthalogen blue-black 0.25 m2 square targets were not significantly different from either 1 m2 blue-black-blue or blue-black square targets. Three times more flies were captured per unit area on the smaller device. Conclusions/Significance Blue-black 0.25 m2 cloth targets show promise as simple cost effective devices for management of G. p. palpalis as they can be used for both control when impregnated with insecticide and for population sampling when covered with adhesive film.
Introduction
Human and Animal African Trypanosomiasis (sleeping sickness and nagana) are still a major constraint on the social and economic development of sub-Saharan Africa, [1]. The diseases affect the health of people and their livestock, resulting in reduced food production and increased poverty [2][3][4]. Tsetse flies (Diptera: Glossinidae) transmit the trypanosomes that cause these illnesses for which a vaccine has still to be discovered. The antigenic variation of the pathogen is a major constraint on the development of a vaccine [5,6]. Although new treatments based on Nifurtimox and Eflornithine are promising [7], sleeping sickness is still difficult to treat, particularly in the second phase of the disease [8][9][10]. For the treatment of nagana in livestock, the initial success of trypanocides is increasingly compromised as trypanosomes continue to develop resistance across Africa [11].
G. p. palpalis is one of the principal vectors of sleeping sickness and nagana across large areas of central and West Africa. Its geographical range corresponds to the coastal belt of tropical rain forest stretching from Liberia in West Africa to Angola in Central Africa [12,13]. However, it can also adapt to man-modified environments, including large urban settlements [14][15][16][17]. Studies on microsatellite populations have shown that there is some genetic variability in this subspecies, probably related to geographical distance at a macro-geographical scale [18] and that at a micro-geographical scale the degree of variation is closely related to the extent of habitat fragmentation [19], as is the case with G. palpalis gambiensis in Burkina Faso [20].
In the face of the continuing difficulties to treat human and animal trypanosomiasis, the reduction and eradication of the tsetse fly vector remains one of the most effective methods to control both diseases. Amongst the different control methods that have been employed, the deployment of visually attractive traps and targets impregnated with insecticide is the most widely used as it is one of the most accessible and efficient methods of control. Historically, the first trapping devices for controlling tsetse were black overalls worn by workers, coated in glue and hung up in the plantations of Sao Tome and Principe in 1910 [21]. Later, in the 1930s, Harris [22][23][24] developed a trap that was employed with great success in Zululand. A further series of trap types followed but was rarely used for controlling tsetse. After the Second World War, trapping was abandoned as a control method in favour of widespread spraying with DDT. It was only in the 1970s that trapping was seriously considered again, thanks to the development of the standard biconical trap by Challier and Laveissière [25] for trapping palpalis and fusca group tsetse. Based on this model, simpler traps, the pyramidal [26][27][28] and monoconical ''Vavoua'' [29], were developed in the1980s to increase trapping efficiency and reduce manufacturing costs. Both traps are still regularly used for controlling G. p. palpalis [30,31], with over 60,000 insecticide-impregnated pyramidal traps deployed in Angola alone since 2008. In order to reduce control costs further, simpler two-dimensional targets were developed [32]. Green established that highest catches of G. p. palpalis are obtained on targets made of phthalogen blue cloth with its exceptionally high reflectivity in the blue part of the light spectrum [33]. The same author went on to show that two-colour targets incorporating phthalogen blue with either black or white are better at catching G. p. palpalis than single-colour ones [34]. Recent research has focused on the cost-effectiveness of using smaller targets [35,36], and chemical attractants [37].
Within the Africa-wide WHO-TDR initiative to develop innovative control strategies for tsetse, we set out to standardize long-lasting, visually-attractive devices for G. p. palpalis, and to see if their efficiency and cost-effectiveness could be improved. The trials were based on existing trap/target/bait technology used at each location following a similar experimental approach throughout Africa [38,39]. Trials were conducted in wet and dry seasons in the Ivory Coast, Cameroon, the Democratic Republic of Congo and Angola to measure the performance of pyramidal, monoconical and biconical traps and targets in phthalogen blue cloth and various alternatives at different population densities and seasons under different environmental conditions across its continental range. A simple enumeration method (adhesive film) was used at these sometimes remote locations to compare trapping efficiencies of devices made of well-characterized colour-fast fabrics. The relative performance of devices was also compared with and without baits. The goal was to determine the most practical and cost effective device/material that would induce the strongest landing response in G. p. palpalis for future use in areawide population suppression of this fly with insecticide-impregnated devices.
Study sites
Studies were conducted in four countries: three in central Africa (Angola, Cameroon and the Democratic Republic of the Congo) and one in West Africa (Ivory Coast; Figure 1). Any study made on private land had the owner's consent. A brief description of each site is given below.
Angola. Three sets of studies were undertaken at the same location along the Onzo River near Tabi in northwest Angola (S 08u 099 240, E 13u 269 410). The site supports intact gallery woodland, surrounded by savannah grassland and bush; there are no domestic animals and the human population density is low but wild animals are still relatively abundant. A first set of field trials took place in 2010 in the wet season (January) and was repeated at the same site in the dry season (June). A second series of trials was conducted in 2010 in the wet season (November) and a third series in 2012 in the dry season (May).
Cameroon. One set of field trials was conducted around Bechati near Fontem, in the South-West Cameroon (N 05u 409 3.60, E 09u 549 550), a hilly region with numerous streams with fragmented indigenous forest and plantations (bananas, palm oil). The local human population density is high and there are many domestic animals. The trials took place in 2009 in the wet season (May) and were repeated at the same location in the dry season (December), but catches in the dry season were too low to be analysed.
Democratic Republic of the Congo (DRC). Two sets of field trials were conducted along the Ndongwa and Kamba watercourses near Malanga about 200 km south-west of Kinshasa (S 05u 329 220, E 14u 219 070). The site is in an area of wooded savannah of Hyparrhenia spp. and Panicum maximum grasses with riverside gallery forest, palm oil and coconut plantations. It is an area of intense human activity with numerous free-roaming goats and pigs and is an endemic focus for sleeping sickness. The trials were carried out in 2010 in the wet season; the first set in February and the second set in November.
Ivory Coast. Two sets of field trials were conducted near Markouguié, Azaguié, 65 km north west of Abidjan (W 04u 089 490, N 05u 379 310) in a hilly region with numerous wet hollows and streams. The area is vegetated by a mosaic of relict indigenous forest and agricultural plantations of bananas, papaya and commercial flowers with livestock rearing (cattle, pigs and chickens) and fish-farming. The first set of trials took place in 2009 in the dry season (December) and was repeated again in 2010, in the wet season (April). A second set of trials was conducted in 2010 in the wet season (November).
Author Summary
G. p. palpalis is one of the principal tsetse fly vectors of African Trypanosomiasis. Its range stretches from Liberia in West Africa to Angola in Central Africa. G. p. palpalis inhabits tropical rain forest but has also adapted to urban settlements. Reduction of tsetse populations remains one of the most effective methods to control disease transmission to man and animals, and development of visuallyattractive insecticide-impregnated traps and targets for palpalis group tsetse dates from half a century ago. Here we describe field experiments made in wet and dry seasons in the Ivory Coast, Cameroon, Democratic Republic of Congo and Angola to establish the most efficient, long-lasting and practical object that induces the strongest landing response in G. p. palpalis. Independent of season and country, both phthalogen blue-black 1 m 2 cloth targets covered with adhesive film proved as good as traps in phthalogen blue or turquoise blue cloth when employed as capturing and landing devices for G. p. palpalis. Pyramidal trap efficiency was inconsistent. As landings on 0.25 m 2 square phthalogen blue-black targets were not significantly different from landings on the 1 m 2 targets, these smaller targets show promise as simple cost effective devices for the management of G. p. palpalis populations.
Catching devices, materials and baits
Five catching devices were tested: standard biconical [25], monoconical (Vavoua type) [29] and pyramidal [27] traps ( Figure 2), and two target designs: a 1 m 2 regular square cloth target (equal vertical rectangles of blue and black, Figure 2) and a 0.91 m 2 Ivory Coast target, 85 cm wide by 107 cm high made of two vertical strips of black cloth (17.5 cm wide) on either side of a single blue panel [32]. In Angola, two additional target designs were evaluated in one set of trials: a square 1 m 2 target of equal vertical rectangles of black-blue-black cloth and a reduced regular square target of 0.25 m 2 with vertical rectangles of blue and black cloth.
To monitor the numbers of tsetse landing on targets, one-sided sticky adhesive film (Rentokil FE45, UK) was attached to both sides of the targets. This film was also attached to the cloth component of traps in some experiments to enumerate flies that land on traps but may not be captured. To assess the influence of adhesive film, particularly its shininess, on landing responses, the number of flies attracted to non-sticky targets was compared to targets covered with adhesive film by using an electric grid of fine electrocuting copper wires (spaced 8 mm apart) mounted in front and behind the targets [40]. A potential difference of 40 KV was applied between adjacent wires and tsetse that landed on the Etarget were electrocuted and fell into a tray (3 cm deep) of soapy water. E-targets are assumed to be invisible to savannah tsetse [40,41], but this assumption has hardly ever been tested on riverine species. Recently, Tirados et al. (2011) [36] showed for the first time that many G. p. palpalis are caught with traditional etargets set up on their own. A 1:4:8 mixture of 3-n-propylphenol (P), 1-octen-3-ol (O), and p-cresol (C) was used as an attractant for experiments comparing baited devices based on general efficacy for several tsetse. The mixture was prepared at origin by the supplier (Ubichem Research LTD, Budapest/Hungary) with a global purity of 98%. Sachets made of 500 gauge/0.125 mm polyethylene containing 3 g of the mixture were placed below the catching devices, 10 cm above the ground, alongside a 250 ml bottle buried up to the shoulders containing acetone (A) with a 2 mm aperture in the stopper. This combination, termed the POCA bait, was made up according to the method described by Torr et al. [42].
Experimental design
Best trapping device and blue material. To assess which was the best catching device and the most attractive blue material, experiments were carried out to compare between four to six devices in a Latin square design of days6sites6treatments, with three simultaneous replicates. Trap positions were always .100 m apart and flies from each device were counted after 24 hours at each position. The various devices and blue materials tested were: biconical traps (in standard blue cotton or S250 phthalogen blue cotton/polyester); monoconical traps (in standard blue cotton or S250 phthalogen blue cotton/polyester), pyramidal traps (in standard blue cotton or turquoise blue polyester/cellulose or Top Notch blue polyester) and a regular target in standard blue cotton and an Ivorian target in standard blue cotton or S250 blue cotton/polyester. The four to six device experiment (depending on location) was repeated using the POCA bait after the unbaited trial was completed in the same general area, with trapping positions .200 m apart. The objective was to determine whether baiting changed the performance ranking of the devices/fabrics (Table 1).
Comparing traps versus targets as landing devices. To assess the efficiency of 3-d traps versus 2-d targets as landing devices, catches on either pyramidal (Angola and the DRC) or monoconical (Ivory Coast) traps with sticky adhesive film on the cloth component were compared to targets covered with adhesive film. This gave a surface area of 2 m 2 of adhesive film for the pyramidal trap and regular target and 0.9 m 2 for the monoconical trap. All devices used to measure landing responses were made of standard phthalogen blue cotton. Flies caught in the cage of the traps with adhesive film on the cloth component were not included in the total for this comparison. Pyramidal and monoconical traps not treated with the adhesive film were included as controls to estimate trap efficiency (percentage flies caught in the control compared to those caught in the cage and on the cloth by the trap with adhesive film). In the DRC and Angola, a three-day experiment was conducted to compare three devices in four replicates. In the Ivory Coast, three devices were compared in four replicates in a six-day experiment (three days per set of two replicates; Figure 3). The trapping positions were always .100 m apart and flies of each sex from each device were counted after 24 hours at each position.
There was an additional five-day experiment in Angola to compare the performance of pyramidal traps to three different target types: a regular square 1 m 2 target (equal vertical rectangles of blue and black); a square 1 m 2 Ivorian type target of equal vertical rectangles of black-blue-black cloth and a reduced regular square target (equal vertical rectangles of blue and black) of 0.25 m 2 ( Figure 4).
Testing adhesive film. To assess whether the addition of the adhesive film could affect the attraction of tsetse to a catching device, a comparison was made in the Ivory Coast between catches of tsetse attracted to a 1 m 2 regular square cloth target (equal vertical rectangles of blue and black), with no film applied and targets covered on both sides by the adhesive film with the sticky side inwards. The two types of targets were placed within electric grids (above), orientated E-W, and the experiments were conducted following a 262 Latin square design of days6sites6 treatments, with two replicates, over eight days. The experiments were carried out simultaneously from 10:00 am to 02:00pm each day and trapping positions were always .100 m from one another.
Statistical analysis. In all trials randomization was set up using design.lsd in the package agricolae [43], R version 2.15.1 [44]. Data were analysed using a linear model in R version 2.15.1 [44], including the following additional packages: MASS [45] and multcomp [46]. Analysis was performed on log (x+1) transformed data including day and position as additional explanatory parameters, and Tukey contrasts were calculated to compare treatments. The Wilcoxon paired test was used to compare fly landings on the blue and black portions of targets. Sex ratios of fly captures by treatments within an experiment were compared using a generalized linear model with a binomial response. Unless otherwise specified, results are presented as detransformed means.
Best trapping device and blue material
In the Ivory Coast the target with adhesive film consistently captured significantly more flies than the traps. The better performance of the target was less consistent in the other three countries, where on at least one occasion, the traps performed equally as well as the targets and actually outperformed the target in Cameroon (Table 1).
There was no difference between the performance of the same trapping device made from different blue cloths (P.0.05; Table 1) with the exception of the dry season experiments in Angola where the pyramidal trap in standard blue proved significantly better than equivalents in either turquoise or Top Notch blue (P,0.05; Table 1). Sex ratios varied between the field experiments but were not significantly different (P.0.05) on the various devices and blue cloths in a given experiment and season. For example, in Angola (wet season) the male to female ratio only varied between 0.55 and 0.63.
Performance of POCA-baited trapping devices
The relative rankings of POCA-baited devices were very similar to those in the unbaited trials, but the capture rate on the target covered in adhesive film increased relative to the number of flies caught in the cages of the traps in all countries, most noticeably in the Ivory Coast and Angola ( Table 1). The POCA bait did not affect the relative performance of the biconical compared to the monoconical trap in the Ivory Coast, but in Cameroon the performance of the biconical trap was improved to equal that of the pyramidal traps. As in the unbaited trials, there was no difference between the performance of the same trapping device made from different blue cloths (P.0.05). Sex ratios varied between the field experiments but were not significantly different (P.0.05) on the various devices and blue cloths in a given experiment and season.
Best landing device
Very similar numbers of flies landed on the traps and targets in Angola and the Ivory Coast and the slight differences recorded are not significant (P.0.05; Figure 3). In contrast, twice as many flies landed on the target compared to the pyramidal trap in the DRC (P,0.01; Figure 3), although in this experiment almost twice as many flies were caught in the cage of the pyramidal trap as on the cloth component of the trap covered with adhesive film (Figure 3) and the pyramidal control caught twice as many flies as the pyramidal trap with adhesive film. In all three countries, a relatively large proportion of flies did not land on the cloth part of the trap but was caught in the cage of the traps with film (18% Angola, 33% Ivory Coast, 62% DR Congo). The proportion of females caught was slightly higher in the cage of the traps covered in adhesive film, compared to the cages of the controls in DR Congo and the Ivory Coast (12% more), but this difference was not significant. In Angola twice as many males were attracted to the pyramidal control, but this is based on only two replicates due to weather damage to the third replicate.
Optimal target colour configuration and size
In the experiment conducted in Angola, the 1 m 2 targets in blue-black (regular) and black-blue-black (Ivory Coast style) equal sized vertical stripes covered with adhesive film caught very similar numbers of flies (14 and 11 flies/day, respectively; P.0.05 Figure 4). There was a significant preference for landing on the black portion on both targets (60% and 71% on the black, respectively; P,0.05), although actual fly numbers on the black were very similar on both target types. This experiment also served to confirm an earlier finding at the same location, namely that similar numbers of flies landed on targets as on the cloth panels of the pyramidal traps (P.0.05, Figure 4). Contrary to this, the pyramidal control (without adhesive film) caught few flies on this occasion (compare Figures 3 and 4). The daily landing rate of flies on the smaller 0.25 m 2 blue-black square target was 70% of the total recorded on the 1 m 2 square target, despite being only a quarter of the size (10 and 14 flies per day, respectively; Figure 4) and this difference was not significant (P.0.05). When the landing rates are corrected to an equal target size of 1 m 2 , the landing rate on the smaller target is nearly triple that on the standard target (40 flies/day/m 2 and 14 flies/day/m 2 , respectively).
Efficiency of pyramidal and monoconical traps
Trap efficiency, defined here as the proportion of flies caught in the cage of the unaltered trap relative to those caught in the cage and on the cloth by the same trap with adhesive film, has been estimated by dividing the mean daily catch of the unaltered pyramidal and monoconical traps by the mean daily catch of the matching traps with adhesive film on the cloth (flies caught on the adhesive film and in the cage; Figure 3 and Table 2). From these results, trap efficiency is estimated at 51% for the monoconical trap in the Ivory Coast, and at 34% for the pyramidal trap in Angola, although the pyramidal estimate is based on a reduced sample size, due to weather damage during the Angolan trials ( Table 2). It was not possible to estimate trap efficiency for the pyramidal traps in the DRC as fly catches were higher in the control (Figure 3 and Table 2).
Effects of adhesive film
Experiments with electric grids to kill flies indicate that the application of adhesive film to a 1 m 2 regular square cloth target (equal vertical rectangles of blue and black), reduced by over half the total number of G. p. palpalis that apparently attempted to land on the device. The detransformed catch index compared to the unmodified target is 0.45 (P#0.01; Table 3), affecting both sexes equally. The effect of the adhesive film on fly behaviour nevertheless differed for the blue and black sections of the target. The adhesive film had little effect on numbers landing on the blue section, but in contrast, on the black section, addition of the adhesive film reduced catches by about two-thirds (P,0.001; Table 3). This response was recorded for both sexes.
Discussion
This study shows that independent of season and country, both phthalogen blue-black and blue-black-blue 1 m 2 targets covered with adhesive film proved to be as good as monoconical Figure 3. Daily catches of G. palpalis palpalis by devices with and without adhesive film. Pyramidal pyramidal trap; monoconical monoconical trap; target blue-black 1 m 2 target. dtr. mean detransformed mean. The target and the cloth portions of traps were covered with adhesive film to compare the propensity of flies to land on the different devices. Catch rates of traps are divided into fly catches on the cloth part and those trapped in the cage of the trap. The limits of the boxes indicate the twenty-fifth and seventy-fifth percentiles, the solid line in the box is the median, the capped bars indicate the tenth and the ninetieth percentiles, and data points outside these limits are plotted as circles. doi:10.1371/journal.pntd.0002601.g003 and pyramidal traps in phthalogen blue or turquoise blue for capturing G. p. palpalis. There was no difference in the performance of blue-black and blue-black-blue targets types. Trap efficiency varied between countries and seasons. Baiting with chemicals augmented the overall performance of targets relative to traps. When 1 m 2 targets and the panels of monoconical and pyramidal traps were covered with adhesive film, fly landings were as high on the traps as on the targets. However, the performance of the pyramidal trap as a landing device was not the same between countries. Fly landings on smaller phthalogen blueblack 0.25 m 2 square targets were not significantly lower than on either 1 m 2 blue-black-blue or blue-black square targets. In fact three times more flies were captured per unit area on the smaller device.
Comparison of unbaited trapping devices
Taken overall, the combined results from the four countries suggest that the addition of adhesive film to targets in blue and black makes them equal to or more efficient than traps at capturing G. p. palpalis, in most situations but not always. Indeed, earlier studies in the Ivory Coast by Laveissière and Penchenier (2000) [47] suggested that the monoconical (Vavoua) is more efficient for attracting G. p. palpalis than black-blue-black and blueblack targets. However, our results imply that G. p. palpalis attraction to targets is underestimated in the presence of adhesive film by up to 50% which would mean that the targets systematically surpass traps as landing devices. It is the landing response that underlies the principle of using insecticide-impregnated targets as control devices for tsetse. To determine whether traps impregnated with insecticide (which has been a practice in West and Central Africa to control G. p. palpalis [26,47] and is still common practice in Angola) are more or less efficient than targets at inducing a landing response, a second series of trials was conducted with both the targets and the cloth panels of the traps covered with adhesive film to enumerate the flies which land (see below under performance of targets versus traps as landing devices below).
Effect of the POCA bait on trap and target performance
As the baited and unbaited trials were sequential at each location they cannot be compared directly. Baits were used to see whether they increased trap efficiency as has been shown for other tsetse species [48], but they appear to have had little impact on trap entry for G. p. palpalis, with the exception of an improved entry rate for the biconical trap in Cameroon. In comparison to the unbaited trials, the POCA bait improved catches on the targets relative to the traps in all countries, but most noticeably in Angola, and in the DRC (by a factor of three and two respectively). This confirms observations made by Rayaisse et al. (2010) [37] who found that odours could increase visual responses to a black target in G. p. palpalis in the Ivory Coast. However, considering the efficacy of smaller targets for G. p. palpalis (see below), one could ask how much effort should one invest in deploying and maintaining chemical baits in control campaigns (some of which are toxic, e.g. phenols) when it may be possible to compensate adequately by simply deploying more targets.
Effect of fabric types
The blue fabrics chosen for these experiments (phthalogen blue cotton, polyester or cotton/polyester and turquoise blue polyester/ viscose) were manufactured with differences in fabric texture and with clear differences in blue-green colour, yet with only one exception (Angola, dry season) all performed equally well in capturing G. p. palpalis. These results agree with findings for the same fabrics tested in similar devices for several riverine and savannah tsetse species in East and West Africa [38,39]. Phthalogen blue cotton cloth has been used for about 30 years in tsetse sampling and control, and is the standard against which all other blues should be compared for attractive properties [49]. The fact that phthalogen blue cotton only remains in limited production has resulted in the ad hoc use of several alternative blue fabrics in tsetse control, some of which are less than optimal for attracting tsetse [50]. The turquoise blue fabric produced in Kenya by Sunflag for these experiments using generic dyes performed well in our studies, confirming that a deep turquoise can be used as a practical alternative to phthalogen blue [51]. Generic dyes are less colour-fast than phthalogen blue cloths, but fading was not a problem in the central African climate after twelve months exposure. However, in humid hot conditions, the cloth must be treated with an anti-mould additive to prevent discolouring due to fungal developments. In contrast, although the 100% polyester blue from Top Notch has excellent colour-fastness it is prohibitively expensive. There is clearly a need to develop a biodegradable and inexpensive replacement for phthalogen blue cotton.
Performance of targets versus traps as landing devices
The adhesive film used to count flies for this comparison (as in Rayaisse et al. (2012) and Mramba et al. (2013) [38,39]) was found to reduce landings by G p. palpalis by half on the 1 m 2 blue-black target, accounted for in the main by reduced landings on the black portion of the target. We assume that landings on panels of monoconical and pyramidal traps are affected to the same extent by the presence of the adhesive film. In any case, the surface area of blue and black parts of pyramidal traps and targets covered with adhesive film were the same in these field trials. The two trap types performed equally as well as the target as a landing device in both Angola and the Ivory Coast. In contrast to this, over twice as many flies landed on the target as on the cloth portion of the pyramidal trap in the DRC. This may be partially explained by the behavioural responses of G. p. palpalis as a relatively high proportion of flies were captured in the cage of the adhesive traps in the DRC (62%) as well as in the similarly treated monoconical and pyramidal traps in the Ivory Coast and Angola (33% and 18%, respectively). This is in contrast to the results of identical experiments conducted on other tsetse species where very few flies flew directly into the cage (Glossina swynnertoni: 7% in the cage of a pyramidal trap, [39], G. tachinoides: 5% and G. morsitans submorsitans 2% in the cage of a monoconical trap [38]). The only exception was the closely related G. palpalis gambiensis with 20% of flies counted in the cage of a monoconical trap [38]. This indicates an apparent propensity of these two palpalis group tsetse to enter the cone of pyramidal and monoconical traps without first landing on the cloth panels. If this is the case, then the efficacy of an insecticide-impregnated pyramidal trap as a fly killing device would rely on the ability of the less physically robust trap netting as well as the cloth panels to retain insecticide over time, factors which argue against its use as control a device for G. p. palpalis.
Optimal target colour configuration and size
The 2012 field trial in Angola shows that alighting by G. p. palpalis was the same on the standard blue-black and Ivory Coast type black-blue-black 1 m 2 targets covered with adhesive film, with a noticeable preference for landing on the black portion on both targets (60% and 71%, respectively). These results would suggest that there is little difference between the two target designs to induce landing by G. p. palpalis. In contrast, landing was equally divided between the blue and black panels on the pyramidal trap. However, the trials using electric grids in the Ivory Coast show that numbers of G. p. palpalis landing on the black portion of the targets would be three times higher on unmodified targets and similar results were recorded using the same experimental approach for the closely related G. p. gambiensis in Burkina Faso [38]. Capture rates using e-nets must be interpreted with a certain amount of caution as recent findings by Tirados et al. [36] have shown that e-nets on their own have a certain attraction for G. p. palpalis.
The 2012 Angolan trial also included a 0.560.5 m blue-black target to test if smaller devices could prove effective for G. p.
palpalis as has recently been demonstrated for this species in West Africa [35] and a range of riverine and a savannah tsetse spp. [35,36,39,52,53]. Landings by G. p. palpalis on the 0.25 m 2 blueblack target in Angola were not significantly different to those on either the blue-black or blue-black-blue 1 m 2 targets covered with adhesive film. In fact, fly catches normalised by unit area were three times higher on the smaller device. This confirms the threefold higher attraction per unit area recorded for G. p. palpalis to 0.25 m 2 black cloth targets over 1 m 2 targets of the same colour by Tirados et al. in the Ivory Coast [36]). The same field study revealed that square and vertical oblong targets are equally attractive to G. p. palpalis and that 0.25 m 2 is near the optimum target size. Such devices are also less prone to wind damage and theft because of their smaller size.
Efficiency of pyramidal and monoconical traps
It is a well-established fact that traps used for tsetse capture only a proportion of the flies that are attracted to their vicinity or that may even land on them [38,39]. For example, the efficacy of the biconical trap has been estimated at between 8 to 27% for G. p. palpalis [37]. The efficacy of the monoconical and pyramidal traps used in this study was also found to vary widely. In the Ivory Coast, the efficiency of the monoconical trap was up to 51% (November 2010 experiment), whereas in Angola the efficiency of the pyramidal trap was estimated at 34% in the 2010 field trial, but at just 8% in the second trial at the same location in 2012. From our results, the differences in the performance of a trap type for G. p. palpalis cannot be ascribed to known population structuring in this species across its West and Central African range [18,19,54] as inconsistencies in the performance of the same pyramidal trap were recorded in successive years at two sites in this study. The much higher catches recorded in Angola and the Ivory Coast on sticky targets indicate that the use of traps alone for monitoring can result in the underestimation of fly population densities.
Concluding remarks
There is a need for reliable and inexpensive devices for population suppression and monitoring of G. p. palpalis across the diverse range of natural and man-made habitats this species occupies from West Africa to Central Africa. Targets that attract flies to land on insecticide-impregnated surfaces are most suitable for population suppression of this vector. We have found no significant difference between the performance of regular blueblack and traditional blue-black-blue 1 m 2 targets in experiments performed in West and Central Africa. Furthermore, our results show that landings by G. p. palpalis on 0.25 m 2 blue-black targets are not significantly different from those on either blue-black or blue-black-blue 1 m 2 targets, with three times more flies per unit area on the smaller device. It is thus possible that a number of smaller insecticide-impregnated targets in blue and black could achieve the same result as larger targets in G. p. palpalis control campaigns across its geographical range. However, the most effective size of devices for controlling G. p. palpalis in terms of the costs of fabrication, deployment and maintenance of large targets versus a higher number of smaller targets needs to be determined through field trials. Either phthalogen or turquoise blue cloth would be suitable for these visual control devices.
Effective control requires adaptive management [55] whereby tsetse populations are monitored and disease-transmission hot spots are identified for additional intervention. [56]. Pyramidal/ monoconical traps could be used for initial monitoring, but our findings indicate that fly numbers caught in the cage of a pyramidal trap should be multiplied three to ten-fold to provide a more realistic estimate of the G. p. palpalis population visiting the device. However, for long-term eradication goals, the detection of very low-density residual pockets is also critical and 0.25 m 2 targets covered with adhesive film would be a more effective tool, as already been proven in the eradication programme against G. p. gambiensis in the Loos islands (Guinea) (J-B Rayaisse, pers comm.).
==
Domain: Biology Environmental Science Medicine Geography
|
[{"added": "2016-05-12T22:15:10.714Z", "created": "2014-01-01T00:00:00.000", "id": "18568905", "metadata": {"extfieldsofstudy": ["Biology", "Geography", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosntds/article/file?id=10.1371/journal.pntd.0002601&type=printable", "pdf_hash": "864b750707081cbe205d8f6301ede695e9d5e2d5", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:9547", "s2fieldsofstudy": ["Environmental Science", "Medicine"], "sha1": "ea3a2f38974a251a46ad8ea34af68525383eec07", "year": 2014}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
A new method for quantifying heterochrony in evolutionary lineages
Abstract. The occupation of new environments by evolutionary lineages is frequently associated with morphological changes. This covariation of ecotype and phenotype is expected due to the process of natural selection, whereby environmental pressures lead to the proliferation of morphological variants that are a better fit for the prevailing abiotic conditions. One primary mechanism by which phenotypic variants are known to arise is through changes in the timing or duration of organismal development resulting in alterations to adult morphology, a process known as heterochrony. While numerous studies have demonstrated heterochronic trends in association with environmental gradients, few have done so within a phylogenetic context. Understanding species interrelationships is necessary to determine whether morphological change is due to heterochronic processes; however, research is hampered by the lack of a quantitative metric with which to assess the degree of heterochronic traits expressed within and among species. Here I present a new metric for quantifying heterochronic change, expressed as a heterochronic weighting, and apply it to xiphosuran chelicerates within a phylogenetic context to reveal concerted independent heterochronic trends. These trends correlate with shifts in environmental occupation from marine to nonmarine habitats, resulting in a macroevolutionary ratchet. Critically, the distribution of heterochronic weightings among species shows evidence of being influenced by both historical, phylogenetic processes and external ecological pressures. Heterochronic weighting proves to be an effective method to quantify heterochronic trends within a phylogenetic framework and is readily applicable to any group of organisms that have well-defined morphological characteristics, ontogenetic information, and resolved internal relationships.
Introduction
Understanding the evolutionary patterns and processes that drive the generation of new phenotypes and occupation of previously unexploited environments (the occurrence of novelty and innovation, sensu Erwin 2017) is a fundamental goal of evolutionary study. The topic of novelty and innovation has most frequently been explored through the lens of adaptive radiation (Losos 2010;Yoder et al. 2010), which posits that ecological opportunities combined with chance phenotypic evolution permits exploration of new ecospace (Stroud and Losos 2016). This model of adaptive radiation has been invoked as the dominant causal factor of major evolutionary events in Earth's history, including the Cambrian explosion (Erwin and Valentine 2013), invasion of freshwater ecosystems (Davis et al. 2018), colonization of land (Benton 2010), and recovery from mass extinction events (Toljagić and Butler 2013). In turn, the innovation of phenotypes is recognized to occur through heritable alterations in the timing of an organism's development, a process known as heterochrony (McNamara and McKinney 2005;McNamara 2012;Colangelo et al. 2019). Therefore, heterochronic processes result in new morphological characteristics that can allow organisms to exploit new environments and subsequently diversify.
Natural biological systems are complex entities, in which both environmental interaction and historical components are potentially equally important (Eldredge and Salthe 1984;Vrba and Eldredge 1984;Erwin 2015b;Lamsdell et al. 2017;Congreve et al. 2018). Evolutionary trajectories are defined by the interaction of two semi-independent genealogical and ecological hierarchies (Eldredge and Salthe 1984;Congreve et al. 2018); as such, ecological changes have the potential to influence heterochronic processes to the same extent as heterochronic changes might alter the potential for ecological occupation. Therefore, pluralistic approaches are required to determine the relative importance of ecological or evolutionary mechanisms behind biotic responses (Lamsdell et al. 2017;Tucker et al. 2017;Congreve et al. 2018), representing a synthesis of macroevolutionary and macroecological topics and methods.
Heterochrony, however, has rarely been studied within an explicitly phylogenetic context (Bardin et al. 2017). Cranial evolution of anguimorphan lizards (Bhullar 2012) and theropod dinosaurs ) has been examined by placing heterochronic changes within a phylogenetic hypothesis, but no attempt to incorporate ecological occupation into such investigations has been made. These studies have also focused on only a subset of morphological characters that may show heterochronic trends, which may obscure overall patterns of heterochrony within lineages, as traits display patterns of mosaic evolution (Hopkins and Lidgard 2012;Hunt et al. 2015), whereby individual traits may exhibit different modes of evolution within a lineage. Recognizing overall heterochronic trends within a lineage beyond the trajectories of individual traits is key to understanding the role of heterochrony in evolution. Furthermore, appropriate methods to quantitatively compare heterochronic trends between lineages have been lacking until now. Here, I use phylogenetic paleoecology to analyze patterns in ecological affinity and heterochronic trends across xiphosuran chelicerates and present a new metric for quantitatively representing heterochronic changes within an evolutionary lineage.
Quantifying
Heterochrony.-A novel approach for quantifying heterochrony is proposed herein, whereby a character matrix comprising a series of multistate characters coded for each species within the analysis is developed. Each character represents an aspect of morphology that may exhibit a paedomorphic, peramorphic, or neutral heterochronic expression. The peramorphic and paedomorphic conditions for each character are determined based on a ranked series of criteria: 1. direct observations of ontogenetic changes within the target species; 2. ontogenetic changes observed in closely related species; 3. ontogenetic changes observed in extant relatives; 4. comparison with outgroup juvenile morphology or ontogeny for character polarity; and 5. comparison with outgroup adult morphology for character polarity.
Once the polarity of the morphology of heterochronic expression for each character is established, characters are coded for each species and assigned a score. A paedomorphic condition is assigned a negative (−1) score, a peramorphic condition is assigned a positive (+1) score, and a neutral condition is assigned a score of 0. If a character cannot be coded for a given species, either due to it not being preserved or its condition being otherwise unclear, it is not given a score and does not contribute to the analysis. The total number of characters within the matrix depends on the number of identifiable heterochronically influenced traits within a lineage; as such, organisms with more complex morphologies and betterunderstood ontogenies will result in larger heterochronic matrices. A larger matrix allows for increased granularity in the heterochronic weighting, and ideally a matrix will comprise 10 or more characters; however, it is possible to generate a heterochronic weighting for a taxon that has as few as three characters coded. Each of the characters coded for a species in the matrix contributes to the heterochronic weighting of the species, calculated as: where Hw is the heterochronic weighting of species j, derived from the mean of the combined heterochronic scores (η) of n characters, resulting in a value between 1.00 (more peramorphic) and −1.00 (more paedomorphic). In turn, the heterochronic weighting of a given clade is calculated as: where [Hw] is the heterochronic weighting of clade k, derived from the mean of the combined heterochronic weightings (Hw) in N species, again resulting in a value between 1.00 and −1.00. The heterochronic weighting of a clade therefore represents the average of the heterochronic weightings of its constituent species; it is explicitly not an ancestral state reconstruction. What heterochronic weighting does ensure is that all taxa within an analysis have a directly comparable metric, permitting analysis of heterochronic trends and correlation with other evolutionary phenomena. While the scale of the heterochrony metric raw values within an analysis is relative and cannot be directly compared between analyses, patterns and timing of heterochronic shifts can be directly compared between datasets in an identical manner to quantitative analyses of morphospace (Korn et al. 2013;Hopkins and Gerber 2017).
To determine whether the heterochronic weighting of a clade represents a concerted trend shift distinct from what could be expected from random, nondirectional evolution, the observed clade scores are randomized across the tree topology. In total, 100,000 randomizations are performed, with the distribution of the randomized heterochronic weightings then collated into a histogram for each clade to which the observed clade heterochronic weighting can be directly compared. If the actual weighting score falls within either tail of the distribution, the weighting is significantly different from what would be expected under random (nondirectional) character change. Randomizing the heterochronic weightings across the observed phylogeny accounts for vagaries such as clade size and topological relationships that may impact the distribution of heterochronic weighting scores. The heterochronic weights retrieved for a given clade are therefore compared with a distribution of weight scores over an identical topology; as such, randomized distributions should be generated for each topology used in an analysis. In this way, disparate analyses can determine whether clades fall within the tails of the expected heterochronic weighting distribution given their phylogenetic topology.
Node-versus Tip-based Analyses.-Scores for heterochronic weighting can be applied two ways, through either a node-based or a tipbased analysis (Fig. 1). Applying heterochronic weighting through node-based calculations accurately encompasses the fact that whether a condition for a given taxon is peramorphic, paedomorphic, or neutral is dependent on the condition in its ancestor as inferred by ancestral state reconstruction, and so it is possible for the polarity of a character to change over the evolutionary history of a clade. As such, when applying node-based calculations, it is critical to document the polarity of each character for each node in the phylogeny. Heterochronic weights are calculated for nodes and are reset at the base of each branch of the phylogeny, with individual character weights representing the inferred shift or transition from the ancestral condition. Node-based application of heterochronic weighting more accurately reflects the actual process of heterochrony; however, it is sensitive to sampling, as failing to sample a species can result in an incorrect assumption of character polarity for a given node, while large gaps in sampling can result in unnatural stacking of transitions at sampled nodes. The analysis also becomes incredibly sensitive to phylogenetic topology, as the relative order in which nodes occur will have a major impact on character polarity. Therefore, node-based heterochronic weighting ideally requires a wellconstrained and dated phylogeny for an evenly sampled group with good ontogenetic data. In practice, only a minority of clades may combine all of these attributes, limiting the extent to which heterochronic weighting can be applied from a node-based perspective-although the promise of phylogenetic advances in a variety of molluscan groups such as ammonoids and gastropods, which regularly preserve details of their ontogeny and have a good fossil record, marks them as potential candidates for nodebased heterochronic weighting.
Alternatively, the tip-based application of heterochronic weighting provides a grand average of heterochronic traits within observed taxa compared with the root character polarity. Rather than tracing the transition of the heterochronic event, tip-based heterochronic weighting quantifies the overall outcome of heterochronic events. This makes the method less precise than its node-based articulation, as it may fail to recognize relative polarity changes in characteristics, but is more broadly applicable to groups with uneven sampling or uncertain phylogenetic topology. Tip-based heterochronic weighting gives an overall FIGURE 1. Example of heterochronic weightings calculated from three traits evolving across a lineage comprising taxa A-G. In the top tree, evolution of the three traits is shown with their condition (peramorphic + 1, paedomorphic −1, or neutral 0) for each species and internal node of the phylogeny shown in boxes. Transitions between character states are shown beneath each branch. The polarity of a transition is dependent on the condition of the character at the preceding node; therefore, a transition to 0 from −1 would be positive (a peramorphic transition), while a transition to 0 from + 1 would be negative (a paedomorphic transition). Node-based calculations are shown on the bottom left, where heterochronic weights are derived from the transitions leading to each node or tip species, while tip-based calculations of heterochronic weights derived from the terminal character conditions of tip species are shown on the bottom right. Both analytical variations accurately capture the overall peramorphic trend among species A and B and the paedomorphic trend from species E to G. Notably, the tip-based application of the method fails to recognize the peramorphic reversal in species F; however, tipbased heterochronic weights would recognize the peramorphic influence if this were to develop into a long-term trend. Node-and tip-based calculations of heterochronic weights are therefore both equally accurate with regard to recognizing overall trends, but node-based calculations are more precise.
indication of peramorphic or paedomorphic trends over the evolutionary history of a clade; in this way, it is similar to the method recently used by Martynov et al. (2020) to study paedomorphosis in nudibranchs, with the distinction that heterochronic weighting provides a quantitative rather than qualitative assessment of the degree of peramorphic or paedomorphic traits within a species. For the present study, heterochronic weighting was applied using the tip-based procedure, as sampling of Xiphosura is uneven across their evolutionary history, as demonstrated by the lack of Silurian exemplars. This is considered an appropriate field test of the method, as I suspect that tip-based analyses will form the bulk of any subsequent studies that adopt heterochronic weighting. Heterochronic Character Matrix.-A heterochronic character matrix for Xiphosura was assembled comprising 20 characters. Characters and their peramorphic and paedomorphic conditions were developed following the criteria detailed in "Quantifying Heterochrony". The characters encompass a wide variety of xiphosuran morphology, equally divided between traits of the prosoma (anterior tagma) and traits of the thoracetron (posterior tagma) (Figs. 2, 3, Table 1). The resulting matrix, coded for 54 xiphosurid taxa, is available in the Supplementary Material. While a number of aspects of these characters also appear in the phylogenetic character matrix, several (such as body size) are not considered appropriate phylogenetic characters and are not incorporated into the phylogenetic analysis. Those heterochronic characters that are included comprise only a minority (5%) of the total phylogenetic characters, ensuring a degree of independence between the two character sets.
Determination of polarity of character morphs for assigning peramorphic and paedomorphic conditions was performed with reference to the observed ontogenetic trajectories of extinct and extant species. The most extensive work on xiphosurid development and ontogeny has focused on the extant American species Limulus polyphemus, with studies of embryonic (Scholl 1977;Sekiguchi et al. 1982;Sekiguchi 1988;Shuster and Sekiguchi 2003; Haug and Rötzer 2018a) and postembryonic (Sekiguchi et al. 1988a,b;Shuster and Sekiguchi 2003) ontogeny supplemented by detailed studies of the early development of the thoracic opercula and book gills (Farley 2010) and compound lateral eyes (Harzsch et al. 2006). Comparatively few studies of the extant Asian species' ontogenies exist; those few that do focus on Tachypleus tridentatus (Sekiguchi et al. 1982(Sekiguchi et al. , 1988aSekiguchi 1988), often via direct comparison with the development of Limulus.
Ontogenetic data also exist for a number of extinct taxa from across the xiphosurid phylogeny. The bellinurines Euproops danae and Euproops sp., from the Carboniferous of the United States and Germany, respectively, have been subject to detailed study of their postlarval development (Schultka 2000;Haug et al. 2012;Haug and Rötzer 2018b;Tashman et al. 2019). Ontogenetic data have also been reported from the bellinurine Alanops magnificus, known from the Carboniferous Montceau-les-Mines Konservat-Lagerstätte in France, a species considered to demonstrate a paedomorphically derived adult morphology (Racheboeuf et al. 2002). Bellinurine embryonic larvae preserved within egg clutches have also been reported from the Carboniferous of Russia and assigned to the newly erected species Xiphosuroides khakassicus (Shpinev and Vasilenko 2018), which probably represent the larvae of an existing species of Bellinurus. Outside of the bellinurines, juvenile and adult individuals of the paleolimulid Paleolimulus kunguricus, from the Permian of Russia have been described (Naugolnykh 2017). Subadult and adult stages have also been differentiated among the available material of the Cretaceous tachypleine Tachypleus syriacus from the Hâqel and Hjoûla Konservat-Lagerstätten of Lebanon (Lamsdell and McKenzie 2015), while the Cretaceous mesolimulid Mesolimulus tafraoutensis from the Gara Sbaa Konservat-Lagerstätte of Morocco is known from juvenile material (Lamsdell et al. 2020).
This diversity in phylogenetic sampling of xiphosurid ontogeny reveals a number of consistent trends in horseshoe crab development across their evolutionary history that can still be observed in the ontogeny of modern species (Fig. 4). These evolutionarily conserved FIGURE 2. Heterochronic characters coded for Xiphosura encompassing aspects of overall body size and prosomal morphology, showing paedomorphic (−1), neutral (0), and peramorphic (+1) conditions. A character unavailable for coding in a species is considered missing data (?) and does not contribute to the species score. ontogenetic trends include a progressive reduction in expression of the prosomal ophthalmic ridge from the larval form through to the adult, a relative decrease in the length of the FIGURE 3. Heterochronic characters coded for Xiphosura encompassing aspects of thoracetron and telson morphology, showing paedomorphic (−1), neutral (0), and peramorphic (+1) conditions. A character unavailable for coding in a species is considered missing data (?) and does not contribute to the species score. prosomal appendages, a reduction in expression of the free lobe segment, an increase in the number and complexity of opisthosomal opercula, and an increase in the relative length of the telson compared with the body. It is these traits, alongside other observed consistent trends in xiphosuran ontogeny, that were incorporated into the heterochronic character matrix.
Phylogenetic Analysis.-A phylogenetic framework for Xiphosura was constructed using an expanded version of the chelicerate character matrix from Lamsdell (2016) . Three taxa were removed from the matrix: Anacontium brevis, which is a synonym of the co-occurring Anacontium carpenteri (Anderson 1997); Liomesaspis leonardensis, which is poorly preserved and lacks diagnostic characters separating it from Liomesaspis laevis (see Tasch 1961); and Willwerathia laticeps, which may not be a chelicerate (Lamsdell 2019). In total, 54 Xiphosura (sensu Lamsdell 2013Lamsdell , 2016 were included, and it is these taxa that were the focus of subsequent analyses of heterochrony and ecological affinity, with the remaining 99 taxa within the matrix serving to adequately root and demonstrate the monophyly of the xiphosuran clade. Additionally, the character coding for Yunnanolimulus luopingensis was updated to incorporate information from recently described specimens exhibiting exceptional preservation (Hu et al. 2017). Four new characters were also added to the matrix, encompassing cardiac lobe width (character 32), the cardiac lobe bearing a median cardiac Selden and Siveter (1987) and Lamsdell (2013). Tree inference was performed using Bayesian statistical analysis, which has been shown to outperform maximum parsimony analyses of simulated data (Wright and Hillis 2014;O'Reilly et al. 2016;Puttick et al. 2017Puttick et al. , 2019, although tree topologies derived from empirical data analyzed under parsimony methods exhibit higher stratigraphic congruence than those retrieved from Bayesian analysis (Sansom et al. 2018). Bayesian inference was performed using Markov chain Monte Carlo analyses as implemented in MrBayes 3.2.6 (Huelsenbeck and Ronquist 2001), with four independent runs of 100,000,000 generations and four chains each under the maximum-likelihood model for discrete morphological character data (Mkv + Γ; Lewis 2001), with gamma-distributed rate variation among sites. All characters were treated as unordered and equally weighted (Congreve and Lamsdell 2016). Trees were sampled with a frequency of every 100 generations, resulting in 1,000,000 trees per run. The first 25,000,000 generations (250,000 sampled trees) of each run were discarded as burn-in, and the 50% majority rule consensus tree was calculated from the remaining 750,000 sampled trees across all four runs; this represents the optimal summary of phylogenetic relationships given the available data (Holder et al. 2008). Posterior probabilities were calculated from the frequency at which a clade occurred among the sampled trees included in the consensus tree.
Additionally, maximum parsimony analysis was performed using TNT (Goloboff et al. 2008) (made available with the sponsorship of the Willi Hennig Society) to test for topological congruence between Bayesian and parsimony methods. The search strategy employed 100,000 random addition sequences with all characters unordered and of equal weight (Congreve and Lamsdell 2016), each followed by tree bisection-reconnection branch swapping (the mult command in TNT). Jackknife (Farris et al. 1996), bootstrap (Felsenstein 1985), and Bremer (Bremer 1994) support values were also calculated in TNT and the ensemble Consistency, Retention and Rescaled Consistency Indices were calculated in Mesquite 3.02 (Maddison and Maddison 2018). Bootstrapping was performed with 50% resampling for 1,000 repetitions, while jackknifing was performed using simple addition sequence and tree bisectionreconnection branch swapping for 1,000 repetitions with 33% character deletion.
Ecological Affinity.-Previous work by Kiessling and Aberhan (2007) and Hopkins (2014) has set out a series of standard environmental categories for organisms found in marine environments; latitudinal occupation, substrate preference, bathymetry, and reef association. For the present study, however, a novel characteristic is considered: salinity. Previous work has shown that horseshoe crabs invaded nonmarine environments multiple times over their evolutionary history (Lamsdell 2016), a transition associated with a variety of biomechanical and physiological challenges (Lamsdell in press), and it is these macroevolutionary events that comprise the focus of the work herein. As with the characteristics of Kiessling and Aberhan (2007) and Hopkins (2014), salinity is reduced to a binary variable, and species are assigned either a marine or nonmarine (including freshwater and deltaic/transitional marginal-marine environments with a large amount of continental flora and fauna) affinity. The ecological affinity for each taxon was estimated through the modified method of Miller and Connolly (2001). This is a departure from Hopkins (2014), who implemented a refined version of the Bayesian estimation of affinity developed by Simpson and Harnik (2009), which has traditionally been applied to higher taxa (e.g., genera). The drawback of applying the Bayesian estimation method to species data is that it requires multiple samples; Simpson and Harnik (2009) only considered genera with at least four occurrences. This is impossible for many species in the fossil record, particularly unmineralized marine organisms (which include horseshoe crabs), the majority of which are known solely from single localities-representing only a single occurrence. In contrast, the method of Miller and Connolly (2001) simply calculates the affinity of a taxon for a given environment as the number of occurrences within the environment divided by the total number of occurrences, resulting in a metric ranging from 0 to 1; when combined, the value of the metric for each of the two possible affinities and no clear affinity will always sum to 1. The advantage of this method is that it can operate with only a single observation, with the drawback that lower sample sizes increase the chances of a false positive of ecological affinity. However, species tend to have a narrow range of habitat preferences (Saupe et al. 2014(Saupe et al. , 2015, thereby reducing the chance of occurrences outside their preferred environments; when species do have broader habitat preferences, they also tend to have broader geographic ranges, increasing the probability of fossilization and thereby increasing the available sample size, reducing the chance of false positives. The resulting metrics for ecological affinity were interpreted as any single value >0.5 indicating a preference for that environment. When no single environment scored >0.5, the taxon was considered to exhibit no affinity. Ties (an instance where the highest-scoring environment achieved a 0.5) were also considered an indication of no affinity. In practice, as the majority of horseshoe crab species are known from only one or a handful of localities, most species showed a clear ecological affinity (see Supplementary Material).
Phylogenetic Paleoecology.-The emergent field of phylogenetic paleoecology leverages phylogenetic information to assess the longterm evolutionary significance of ecological trends within lineages through the synthesis of phylogenetic theory and quantitative paleoecology (Lamsdell et al. 2017). Using the historical phylogenetic and ecological framework, it is possible to test for phylogenetic or ecological signal in the distribution of heterochronic weightings, thereby accounting for the possible influence of both the genealogical and ecological biological hierarchies (Congreve et al. 2018).
The phylogenetic framework also permits for the estimation of the ancestral conditions of both morphological and ecological characteristics. To determine the number of transitions between marine and nonmarine environments, the ecological affinity of each species as determined above was mapped onto the resulting phylogenetic tree and parsimony-based ancestral state reconstruction performed in Mesquite 3.02 (Maddison and Maddison 2018). The inferred ancestral ecological affinities were used to polarize the internal nodes of the tree, thereby establishing whether the ecological affinities of tip species represent inherited habitat preferences or are the result of a shift in the occupied niche.
Multivariate statistical tests (permutational multivariate analysis of variance [PERMA-NOVA] using the Euclidian distance measure) were performed in the statistical software package PAST (Hammer et al. 2001) and using the adonis function in the R (R Core Team 2018) package vegan (Oksanen et al. 2019) to ascertain the statistical significance of differences in heterochronic weightings across ecological affinities and phylogenetic clades. Several analyses were performed: comparison between marine and nonmarine habitats, comparison between clades, and a test of the interaction between clade and habitat categories. For the comparison between marine and nonmarine habitats, all species were included in the analysis. However, for analyses that required the assignment of species to clades, only species assignable to Bellinurina, Paleolimulidae, Austrolimulidae, or Limulidae were included. This resulted in the exclusion of Lunataspis aurora and Kasibelinurus amoricum (stem Xiphosurida), Bellinuroopsis rossicus and Rolfeia fouldenensis (stem Limulina), and Valloisella lievinensis (stem Limulidae), as their inclusion would necessitate the formulation of unnatural, paraphyletic "groups" that possess no evolutionary cohesion and could serve to bias analyses. Significance was estimated by permutation across groups with 10,000 replicates, including Bonferroni correction to correct for multiple comparisons in the case of comparison between clades.
To test whether the distributions of species' heterochronic weightings within clades exhibit directional trends, Spearman's (1904) rank correlation coefficient ρ, calculated using the cor.test function in R (R Core Team 2018), was used to compare the heterochronic weighting of observed taxa within a clade to their distance from the root of the clade. A statistically significant ρ, in combination with a statistically significant clade heterochronic weighting [Hw], was considered evidence of a concerted heterochronic trend within the clade. The distance of a species from the root of the clade was determined using the cladistic rank method (Gauthier et al. 1988;Norell and Novacek 1992a,b;Norell 1993;Benton and Storrs 1994;Benton and Hitchin 1997;Carrano 2000Carrano , 2006Wagner and Sidor 2000;Poulin 2005;Prado and Alberdi 2008;Huttenlocker 2014). Cladistic rank is determined by counting the sequence of primary nodes in a cladogram from the basal node to the ultimate node (Fig. 5). The main axis of the cladogram is defined as the internal branch supporting the most inclusive clade after each bifurcation. Smaller offshoot clades are reduced to polytomies for the purpose of assigning a clade rank, with each constituent species within a polytomy receiving the same rank. The assignment of equal rank to all constituents of an offshoot clade is a conservative approach that ensures any revealed trend is the majority pattern demonstrated by the most inclusive clades within the cladogram, without biasing the analysis through exclusion of taxa located in offshoot clades. This was done for each of the four main xiphosurid clades: Bellinurina, Paleolimulidae, Austrolimulidae, and Limulidae.
Results
Analysis of the phylogenetic character matrix resulted in Bayesian inference and parsimony optimality criteria retrieving a concordant tree topology, with the strict parsimony consensus of six most parsimonious trees having an identical hypothesis of xiphosuran relationships to the Bayesian majority rule consensus ( Fig. 6; for details of the parsimony tree and the full Bayesian tree, see Supplementary Material). The phylogenetic topology, while broadly consistent with those of previous analyses (Lamsdell 2013(Lamsdell , 2016Lamsdell and McKenzie 2015), is more stratigraphically congruent due to movement in the placement of K. amoricum, 'Kasibelinurus' randalli, and Casterolimulus kletti. Previously resolving as a paraphyletic grade at the base of Xiphosura, K. amoricum now resolves as the sister taxon to the clade comprising the two suborders Limulina and Bellinurina, while 'Kasibelinurus' randalli resolves at the base of Bellinurina. This leaves L. aurora, the sole described Ordovician species, as the sister taxon to all other Xiphosura. Casterolimulus, previously resolved as the sole post-Triassic representative of Austrolimulidae, now forms a clade with Victalimulus within Limulidae. Victalimulus and Casterolimulus are both Cretaceous taxa known from nonmarine environments, and the notion that they share common ancestry has been suggested previously (Lamsdell in press). The internal relationships of the bellinurines and limulids are also better resolved, and a sister clade to the xiphosuran crown group (composed of Tachypleinae and Limulinae) comprising Mesolimulus, 'Limulitella' vicensis, 'Limulus' woodwardi, Victalimulus, and Casterolimulus is recognized for the first time.
The distribution of environmental affinity across the phylogeny of Xiphosura supports previous assertions that the group has invaded nonmarine environments from the marine realm multiple times over the course of its evolutionary history (Lamsdell 2016). However, whereas previous estimates suggested that the transition from a marine to nonmarine environmental affinity had occurred at least five times during xiphosuran evolution, the present results indicate such a transition has occurred only four, or potentially as few as three, times (Fig. 6). Ancestral state reconstruction demonstrates that a marine affinity is plesiomorphic for Xiphosura and that shifts to a nonmarine affinity occurred for the clades Bellinurina and Austrolimulidae, as well as Valloisella and the sister taxa Victalimulus and Casterolimulus. While Bellinurina and the Victalimulus/ Casterolimulus clade are clearly derived from marine-dwelling lineages that subsequently transitioned to nonmarine environments, the situation for Valloisella and the austrolimulids FIGURE 6. Bayesian phylogeny of xiphosurids showing environmental affinity of salinity and heterochronic weighting mapped onto the tree. Environmental affinity is indicated on the branches (blue, marine; brown, nonmarine), heterochronic weighting is shown at the tips alongside the taxon names through heat-map shading (green, more paedomorphic; orange, more peramorphic). Bayesian posterior probabilities are shown below each node. The clades shown in Figs. 7 and 8 are labeled alongside the tree. is less clear. Due to the tree topology, with Valloisella-a nonmarine xiphosurid-resolving as sister taxon to a large clade divided between the predominantly marine Limulidae and the nonmarine Austrolimulidae, ancestral state reconstruction is unable to determine a clear environmental affinity for the internal nodes between Limulidae, Austrolimulidae, and Valloisella. These results infer a number of potential evolutionary scenarios. One possibility is that Valloisella and Austrolimulidae independently transitioned to nonmarine environments and a marine affinity is plesiomorphic for Limulidae. Alternatively, Valloisella, Austrolimulidae, and Limulidae may all be derived from fully euryhaline ancestors that genuinely have no environmental affinity and occupied both marine and nonmarine environments; in this case, the transition to a nonmarine affinity in Valloisella and Austrolimulidae, and a marine affinity in Limulidae, would be the result of a reduction in their ancestral occupied niche. Finally, it is possible that the transition to nonmarine environments occurred in the ancestors of Valloisella, Austrolimulidae, and Limulidae and that a nonmarine affinity is therefore plesiomorphic for all three taxa, with Limulidae having undergone a subsequent reversal to reoccupy marine environments.
Heterochronic weightings are unevenly distributed across the phylogeny, with more extreme (i.e., higher positive or negative) values occurring among taxa with a nonmarine environmental affinity. This distinction is borne out statistically, with PERMANOVA tests showing that the variance of heterochronic weightings of all species occupying marine environments is significantly distinct from that of those occupying nonmarine environments (Table 2). Statistically significant differences also separate the heterochronic weightings of the xiphosuran clades Bellinurina, Paleolimulidae, Austrolimulidae, and Limulidae (Table 3). Environmental affinity and phylogenetic relatedness therefore both influence heterochronic weightings; however, there appears to be no interaction between the two factors, indicating that they exhibit conflicting signals and do not covary (Table 4).
Comparing the heterochronic weightings of Bellinurina, Paleolimulidae, Limulidae, and Austrolimulidae to the distribution of randomized heterochronic weightings reveals the heterochronic weightings of Bellinurina, Austrolimulidae, and Limulidae to be significantly different from what would be expected from random, while that of Paleolimulidae falls within what could be explained from a random distribution (Fig. 7). Interestingly, Spearman's rank correlation (Fig. 8), indicates directional trends in heterochronic weighting within Austrolimulidae and Bellinurina, but no significant trend within Limulidae or Paleolimulidae. The heterochronic weightings of Bellinurina and Austrolimulidae show a consistent directional trend as demonstrated by locally estimated scatterplot smoothing (LOESS) regression (although Bellinurina do undergo a slight shift in trajectory among their highest-ranked clades), while Paleolimulidae and Limulidae exhibit random directional shifts across the phylogeny. This distinction is potentially TABLE 2. One-way permutational multivariate analysis of variance (F (1,53) = 4.197, η 2 = 0.075, p = 0.0424), 10,000 permutations, Euclidean distance measure. Value in regular font is the p-value, value in italics is the raw F-value. Total sum of squares = 5.120, within-group sum of squares = 4.737, between-group sum of squares = 0.383.
Marine Nonmarine
Marine -4.197 Nonmarine 0.045 - borne out in the position of the observed heterochronic weightings relative to the randomized distribution of heterochronic weightings: the observed weightings for Bellinurina and Austrolimulidae sit far outside the randomized distribution, expressing a value that does not occur within the randomized weights.
In comparison, the observed heterochronic weighting for Limulidae falls within the tail of the randomized distribution, with a value equal to that of a set of those retrieved from the randomizations.
Discussion
Shifts in ecological affinity correlate with changes in evolutionary regime in Xiphosura. Clades that invade nonmarine environments exhibit distinct differences in the prevalence of heterochronic traits in comparison to those that inhabit the marine realm (Fig. 6), with Austrolimulidae demonstrating increased prevalence of peramorphy, while paedomorphy is prevalent among Bellinurina (Fig. 7). Paedomorphic traits have long been recognized in bellinurines (Haug et al. 2012;Lamsdell in press), including their retention of long, gracile prosomal appendages into adulthood; visible opisthosomal segmentation; and elongated dorsal prosomal shield spines. Austrolimulids, meanwhile, develop elongate and splayed prosomal genal spines; reduce the size of their opisthosomal tergopleura; and exhibit enlarged, posteriorly positioned lateral eyes-all of which are recognized as peramorphic conditions herein. Interestingly, lineages that make the transition to nonmarine environments demonstrate concerted and enduring heterochronic trends (Fig. 8) that persist for millions of years, with species progressively exhibiting an increasingly greater number of paedomorphic (in Bellinurina) or peramorphic (in Austrolimulidae) traits. It seems that shifts in environmental occupation set these lineages along a heterochronic trajectory resulting in a directional bias (Gould 1982), whereby changes in the timing or rate of development produce innovative morphologies, the selection of which-mediated by the environment-result in increasingly specialized phenotypes. Heterochronic processes are known to be one of the primary ways by which morphological innovation ( It is notable that only rarely do either of the heterochronic trends observed here show any indication of heterochronic reversals and that in both cases the trend is associated with a marked shift in diversity dynamics, with Bellinurina increasing in diversity before rapidly going extinct and Austrolimulidae exhibiting decreased rates of speciation and a lower diversity even in relation to other Xiphosura (Lamsdell 2016, in press). The combination of limited reversal and shift in diversity dynamics suggests that these directional trends may represent macroevolutionary ratchets (trends where reversals are rare and ultimately result in increased extinction risk or a decrease in rates of origination: Van Valkenburgh 1991Van Valkenburgh et al. 2004), although whether this is due to consistent environmental pressure or some inherent property of the developmental processes operating is unclear. Numerous consistent peramorphic and paedomorphic evolutionary trends associated with environmental gradients-termed peramorphoclines and paedomorphoclines (McNamara 1982), or more generally heteroclines (McKinney 1999), although these terms conflate process (heterochrony) and evolutionary outcome (directional bias or macroevolutionary ratcheting)-have been recognized in the fossil record (McNamara 1982(McNamara , 1986(McNamara , 1988Simms 1988;Korn 1995;Crônier et al. 1998;McKinney 1999;Poty 2010;Fernandez-Lopez and Pavia 2015). Most described heterochronic trends do not exhibit heterochronic reversals, with only a few notable exceptions (Gerber 2011), although it should be noted that none of these studies were performed within a phylogenetic framework. It has previously been suggested that concerted heterochronic trends are always controlled by environmental factors (McKinney 1986(McKinney , 1988; however, subsequent studies have documented heterochronic trends occurring apparently independently of any environmental gradient (Breton 1997). In the present study, xiphosurans possibly undergo one reversal in environmental affinity, with the ancestors of the predominantly marine Limulidae potentially having a nonmarine or mixed environmental affinity (Fig. 6). The ramifications of this are twofold. First, it would suggest that xiphosurans survived the end-Permian mass extinction by occupying nonmarine environments and returned to the marine realm during the subsequent recovery, a possibility first suggested by Błażejowski et al. (2017). Second, Limulidae display on average more positive heterochronic weightings than other marine taxa (Fig. 7), although these peramorphic traits do not manifest as part of a concerted trend (Fig. 8) as they do in the limulid nonmarine sister group, Austrolimulidae, which is characterized by extreme peramorphism. This opens up the possibility that, if limulids have reoccupied marine environments from an ancestral nonmarine habitat, the elevated occurrence of peramorphic character conditions in the clade may be a relict of the peramorphic trajectory that continued in austrolimulids. If this were to be the case, it would suggest that returning to marine environments stopped the heterochronic bias in limulids, and perhaps most interestingly, that the changes the lineage had undergone while occupying nonmarine environments were not subsequently reversed.
The exact mechanism by which heterochrony operated in the cases observed here is uncertain. It is notoriously difficult (arguably impossible) to discern between changes in timing and changes in rate of development without detailed ontogenetic sequences of consecutive species (Gould 1988;Jones 1988;McKinney 1988;Allmon 1994;McNamara and McKinney 2005;Bardin et al. 2017). Nevertheless, it is possible to recognize broad peramorphic and paedomorphic trends. Why these processes unfolded in a ratchet-like fashion is also unclear. Classical natural selection (Darwin 1859) may explain the evolution of increasingly unusual morphologies; however, it is unclear whether these morphologies are truly specialized or simply bizarre. Another possibility among the Bellinurina is that physiological changes required in order for xiphosurans to tolerate low-salinity environments for extended periods of time may have been accomplished through the retention of larval physiology into adulthood (Lamsdell in press). The larvae of modern horseshoe crabs are extremely tolerant of salinities lower than 35‰ (Shuster 1982;Ehlinger and Tankersley 2007;Botton et al. 2010), and paedomorphic processes permitting the maintenance of this tolerance in adults may have also resulted in the retention of larval morphological characteristics. The broad, shallow prosomal carapaces of austrolimulids, meanwhile, may have developed in response to unidirectional hydrodynamic environments that the group encountered as it radiated in lacustrine environments. It is worth considering that macroevolutionary ratchets in even closely related groups may occur through distinct mechanisms.
Ultimately, one of the more interesting outcomes of the study is the support for the quasi-independence of the signal imparted by history and ecology in evolution. History (as represented by phylogeny) and ecology are both significant sources of variation among heterochronic weights but do not interact (Table 4), indicating that although they both exert influence on the distribution of heterochronic weights, they do so with conflicting signals. This conflict is due to the impact of the quasi-independent genealogical and ecological biological hierarchies (Congreve et al. 2018), whereby historical contingency limits the morphological framework for subsequent adaptation to ecological pressures (Gould and Lewontin 1979;Eldredge and Salthe 1984;Anderson and Allmon 2018). Logically, and as hinted at by the data discussed here, this tension between competing hierarchies can also extend to developmental frameworks as the mediating factors by which morphologic phenotypes are expressed.
Conclusions
Applying this new method for quantifying heterochrony, expressed as a heterochronic weighting, within a phylogenetic context reveals concerted independent heterochronic trends in xiphosurans. These trends correlate with, and may be driven by, shifts in environmental occupation from marine to nonmarine habitats, resulting in a macroevolutionary ratchet whereby environmental selective factors result in the preferential retention of phenotypes derived from heterochronic processes, which in turn reinforces directional heterochronic trends and the proliferation of peramorphic or paedomorphic characteristics. Critically, the distribution of heterochronic weightings among species shows evidence of being influenced by both historical, phylogenetic processes and external ecological pressures. This is most clearly demonstrated by the manner in which the independent occupation of nonmarine environments is accompanied by significant heterochronic trends in both Bellinurina and Austrolimulidae, but manifesting as a paedomorphic trend among bellinurines and a peramorphic trend within austrolimulids. Therefore, while the environment can exert a strong pressure on both phenotype (as expected by the fundamental evolutionary process of natural selection) and the underlying developmental processes that govern phenotype, it can only do so utilizing the available morphological and developmental frameworks. The availability and composition of these frameworks is mediated by contingent, historical factors (as expressed by phylogeny) that limit both the potential for adaptation to certain environmental conditions and the structural or developmental outcome of concerted selective pressure. A well-known example of the former is the fact that vertebrates returning to fully aquatic environments are constrained to breathing air, while a structural example of the latter are the hook-like projections of the beak's tomia in mergansers that are used to aid in gripping their fish prey. The serrations perform a function similar to that of the narrow, curved teeth of gars, which also prey upon small fish. Teeth, however, were lost in the avian stem-lineage; lacking the developmental framework to express teeth, mergansers instead developed modifications to the serrated tomia prevalent in ducks and geese. Xiphosurans demonstrate that such contingent processes can also affect the mechanisms by which developmental shifts occur.
Heterochronic weighting has the proven potential to be an effective method to quantify heterochronic trends within a phylogenetic framework. Comparing the observed heterochronic weightings of clades to randomized distributions permits the discrimination of concerted heterochronic trends from what would be expected under random (nondirectional) character change. The method is readily applicable to any group of organisms that have welldefined morphological characteristics, ontogenetic information, and resolved internal relationships; indeed, in order to test the generality of the observations made for xiphosurans, it is imperative that additional studies be performed in disparate clades. Future work should also aim to apply node-based heterochronic weighting in appropriate groups and seek to apply both tip-and node-based calculations to the same data to further explore the behavior and comparability of both. The combination of this heterochronic metric with ecological affinity data affords easy study of the correlation between developmental changes and environmental shifts as a branch of phylogenetic paleoecology and has the potential to open new avenues into studying the relationship between evolutionary developmental processes and external environmental causal factors.
Acknowledgments
I thank E. Lazo-Wasem (Yale Peabody Museum) for providing photographs of Limulus instars and A. Downey (West Virginia University) for coding assistance that aided in collating randomization results. The concepts and perspectives presented in this paper have been refined over the years through in-depth discussion with my colleagues B. Anderson (West Virginia University), C. Congreve (North Carolina State University), A. Falk (Centre College), A. Manafzadeh (Brown University), and J. Miyamae (Yale University). I am especially grateful to A. Whitaker (University of Kansas) who provided graphical design services to aid in the presentation of this research at the Geological Society of America annual conference and possesses invaluable skills in helping to reassemble 3D printed turtle skulls. I thank D. Bapst (Texas A&M University) and an anonymous referee for their detailed and thoughtful reviews that greatly improved the article and prompted me to clarify aspects of the method, as well as encouraging me to devote further page space to explanations of macroevolutionary phenomena.
==
Domain: Biology Environmental Science Geography
|
[{"added": "2020-05-21T00:08:18.085Z", "created": "2020-05-13T00:00:00.000", "id": "219411769", "metadata": {"extfieldsofstudy": ["Biology", "Geography"], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://www.cambridge.org/core/services/aop-cambridge-core/content/view/83D03C714E53A22BDB5FAE894617C462/S0094837320000172a.pdf/div-class-title-a-new-method-for-quantifying-heterochrony-in-evolutionary-lineages-div.pdf", "pdf_hash": "81980c0dbf67e5ed44484533fa4b39ffa2401d76", "pdf_src": "BioOne", "provenance": "peS2o-0028.json.gz:6511", "s2fieldsofstudy": ["Biology", "Environmental Science"], "sha1": "bc97cbd91b8dc942e321101d30b0ec438208daef", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Numerical anomalies in the dentition of southern fur seals and sea lions ( Pinnipedia : Otariidae )
Cases of dental agenesis, supernumerary teeth and dental losses are presented in three species of South American Otariids: Arctocephalus australis (Zimmermann, 1783), A. tropicalis (Gray, 1872) and Otaria flavescens (Shaw, 1800). For the first time, congenital and acquired dental anomalies were comparatively diagnosed in skull samples from southern Brazil and nearby areas. The skulls and mandibles were accessed in the scientific collection of mammals of the Federal University of Santa Catarina, southern Brazil. Agenesis was found only among maxillary post-canine teeth, especially the distal ones (PC/6), due to an evolutionary trend towards reduction of the number of post-canine teeth in this family. Maxillary and mandibular supernumerary teeth were found in A. australis and A. tropicalis, but their positioning is unrelated to cases regarding phylogenetic and evolutionary implications. Dental losses were found in all species and different stages of alveolar obliteration suggest that this process is common in Otariids and does not affect their survival. The investigation of congenital and acquired dental anomalies in pinnipeds can provide information on dental formula evolution in Pinnipeds and in the phylogenetic relationships among Carnivora.
Teeth are a valuable source of information in studies of mammalian biology, supplying data on the feeding habits, phylogenetic relationships among species, and estimations of age (HOFF & HOFF 1996). The occurrences and implications of dental anomalies in mammals play a major role in studies of comparative osteology, although dental anomalies are still poorly elucidated in otariids.
The first records of numerical anomalies in the Otariidae were published for Callorhinus ursinus (Linnaeus, 1758) and Eumetopias jubatus (Schreber, 1776) (CHIASSON 1955, KUBOTA & TOGAWA 1964), two species from the northern hemisphere. Evolutionary and phylogenetic implications of dental anomalies in pinnipeds were first studied by KUBOTA & TOGAWA (1964) and later by LOUGHLIN (1982).
Although some inventories were published on several Otariidae and Phocidae species from the northern hemisphere (KÖNEMANN & VAN BREE 1997, ABBOTT & VERSTRAETE 2005, CRUWYS & FRIDAY 2006), such inventories from the southern hemisphere were not available until some reports were published considering the ecological and evolutionary context of dental anomalies and pathologies in Arctocephalus G. Saint-Hilaire & Cuvier, 1826 and Otaria Peron, 1816 specimens from Brazil (southern Rio Grande do Sul), Uruguay andArgentina (DREHMER &FERIGOLO 1996, DREHMER et al. 2004).
This paper evaluates numerical anomalies in the dentition of the Otariids Arctocephalus australis (Zimmermann, 1783), Arctocephalus tropicalis (Gray, 1872) and Otaria flavescens (Shaw, 1800) from Southern Brazil and nearby areas, considering anomalies that are both congenital (departures from the typical number of teeth) and acquired in life.
SHORT COMMUNICATION
Numerical anomalies in the dentition of southern fur seals and sea lions (Pinnipedia: Otariidae) (1996). Age categories (young, subadult and adult) were determined through the stage of closure of cranial sutures and by the condilobasal length (SIMÕES-LOPES et al. 1995, DREHMER & FERIGOLO 1997). The evaluation of numerical anomalies was obtained through macroscopic analysis of the syncranium, regarding cases of extra teeth and dental agenesis.
Cases of dental losses in life due to trauma or pathologies were also examined. These cases are characterized by the porosity and irregularity of the bony tissue as a consequence of alveolar resorption, in contrast with the absence of alveoli in the typical cases of dental agenesis (VILÀ et al. 1993, DREHMER & FERIGOLO 1996).
Cases of dental agenesis and supernumerary teeth were registered only in post-canine teeth. Dental agenesis was registered in four specimens of A. australis in the maxillary teeth, in
4 3 1
Mandibular supernumerary teeth (A.australis: UFSC 1274, adult male; A. tropicalis: UFSC 1342, adult female): in UFSC 1274 two small alveoli were observed with little depth and a single alveolar cavity labially displaced between PC/4 and PC/ 5 of both jaws, characterizing a case of symmetrical supernumerary teeth (Fig. 6). Although these teeth were possibly lost during the process of preparation and cleaning of the skull, the pair of alveoli indicated their existence. In UFSC 1342 the existence of supernumerary alveolus was observed between PC/ 3 and PC/4 of the left jaw, smaller in size and oriented to the labial side of the dental row (Fig. 7).
In the specimen UFSC 1017, the partial closure of the lower left post-canine alveolus of PC/5 indicated its previous loss (Fig. 8). In UFSC 1132 we observed the loss of teeth and posterior partial closure of the lower right PC/1 and PC/2. The same situation was observed in UFSC 1323 with upper right PC/5 and PC/6 (Fig. 9). In the specimen UFSC 1228, the closure of upper left PC/5 was almost complete, with minimal traces of its existence. This indicates that the loss probably occurred in the juvenile stage of its life.
In UFSC 1171 there was a multiple loss of lower right PC/ 1, of lower left PC/1, PC/3 and PC/4, and of all incisives, with subsequent partial closure of post-canine alveoli and almost complete resorption of alveolar incisives (Fig. 10). With the partial closure of two consecutive alveoli (PC/3 and 4), the porous bony tissue has become shared between the two alveoli.
Variation in the biological sense comprises differences of any kind, mainly morphological, which exists between individuals of the same species, implying in deviation from a mean or a 'norm'. When the difference is slight, it can be considered a 'normal variation'. On the other hand, when the deviations are more gross and uncommon among individuals, they are considered anomalies (MILES & GRIGSON 1990). As dental formula in mammal orders and genera are quite conserved and well established (NOVACEK 1986), the numerical variation here reported are interpreted as typical cases of dental anomalies.
The dentition of pinnipeds is, in general, considered unstable and liable to variation, mainly in number (KUBOTA & TOGAWA 1964). Dental agenesis in South American Otariids has already been investigated in A. australis and A. tropicalis and in O. flavescens (MILES & GRIGSON 1990, DREHMER & FERIGOLO 1996, DREHMER et al. 2004). According to the authors concerned, the absence of teeth, particularly of the most posterior ones such as PC/5 and PC/6, represents a trend towards reduction of the postcanine teeth in Otariids. Taking into consideration the four cases diagnosed in our study, three of them are related to the absence of PC/6 (two of them are cases of symmetrical agenesis of both PC/6), corroborating this hypothesis. The PC/1 agenesis found in UFSC 1113 is an uncommon case of dental reduction and possibly the first record of this dental anomaly in Otariidae.
According to CHIASSON (1957), the reduction and loss of PC/6 in Callorhinus ursinus occurs due to an evolutionary specialization. Considering that the simplification in shape and the simultaneous reduction in size of a tooth always precede its loss (WOLSAN 1984), the simplified dentition of Otariids must be in a process of numerical reduction along the evolution, but we should not discard the natural variability of the dentition in this group. The process of shortening of the molarization field of the upper jaw was already observed in other Carnivora, such as foxes and some canides, and may represent a trend for this group (SZUMA 1999).
Supernumerary teeth in Otariids possibly represent a reversion to the primitive dentition of terrestrial carnivores, according to the phylogenetic relationships of Pinnipeds (DREHMER & FERIGOLO 1996, DREHMER et al. 2004, CRUWYS & FRIDAY 2006). DREHMER et al. (2004) reported supernumerary teeth in mandibular PC/6, which represents the atavic reappearance of the lower second molar in some specimens of O. flavescens, reconstructing the origins of Pinnipedimorpha evolution in the Oligo/Miocene (BERTA & WYSS 1994, DREHMER et al. 2004). The simultaneous occurrence of two supernumerary teeth between the mandibular PC/4 and PC/5 in A. australis and of one supernumerary between PC/3 and 4 in A. tropicalis seems to be unrelated with atavism. Possibly, these cases are related to mutation or changes in the genetic control of tooth development, without relevant phylogenetic implications (WOLSAN 1984, SZUMA 1999, MILETICH & SHARPE 2003).
Maxillary supernumerary teeth have been described in Otariid species in different positions (MILES & GRIGSON 1990, DREHMER & FERIGOLO 1996, DREHMER et al. 2004, ABBOTT & VERSTRAETE 2005). As no pinniped ancestor has more than four pre-molars and two molars, totalizing the six post-canine teeth (DREHMER et al. 2004), the appearance of maxillary supernumerary teeth in A. tropicalis also does not seem to have any evolutionary or phylogenetic implication.
The reduced size of alveoli and the simplification in shape of the supernumerary teeth observed in this study suggests that they are also typical cases of microdontia, a developmental anomaly already recorded in several mammals (FELDHAMER & STOBER 1993, HOFF & HOFF 1996), although most of the published records are related to humans (e.g. BROOK 1984). Microdont teeth (reduced in size and simplified in shape) have also been recorded in O. flavescens (DREHMER et al. 2004). The appearance of small and cylindrical supernumerary teeth may be caused by the split of some tissue fragment from a tooth germ, originating extra teeth simplified in shape and size (SZUMA 1999).
According to LOUGHLIN (1982), it is plausible that cases of agenesis and supernumerary teeth have an expressionless functional result and that they are morphologically irrelevant to Pinnipeds, considering a trend towards homodonty in the postcanine teeth and their doubtful effectiveness in mastication (MILLER et al. 2007).
The loss of teeth in life is considered to be relatively common in pinnipeds (CHIASSON 1957, DREHMER & FERIGOLO 1996). When a tooth is lost and is not subsequently replaced, the alveolus is filled with bony tissue of porous consistency during the process of alveolar resorption or reossification (VILÀ et al. 1993). The alveolus from which a tooth has been lost becomes shallow and shows clear signs of increase in vascular foramens, as a consequence of the alveolar periosteum reaction (VERSTRAETE et al. 1996).
Loss of teeth in life in pinnipeds occurs due to traumas suffered during defense of territory or while feeding, as well as by complications originating in pathologies such as periodontitis or periapical lesions (DREHMER & FERIGOLO 1996, BRAUNN & FERIGOLO 2004). The fact that in this study these losses were recorded only in adult males reinforces the importance of intraspecific combat on the occurrence of these anomalies. The ingestion of gastroliths by O. flavescens (DREHMER & OLIVEIRA 2003) could also be a cause of dental loss, but it is less plausible that they are mainly responsible for these anomalies, as dental loss also occurs in species that do not ingest gastroliths (as in both species of Arctocephalus), and the ingestion seems to be intentional (DREHMER & FERIGOLO 1996, BRAUNN & FERIGOLO 2004). Post-canine teeth were lost and the alveoli reossify with higher frequency than other teeth in the three species evaluated in this study, although we had observed some cases of alveolar resorption of incisives in O. flavescens.
According to VILÀ et al. (1993), a great percentage of wolves -Canis lupus Linnaeus, 1758 -, truly masticating carnivores, survived for many years despite the dental losses diagnosed, suggesting that this process is not determinant for the survival of populations in this species. The observation of alveoli in different stages of resorption in the Otariids analyzed by this study agrees with this hypothesis, providing evidence that the loss of teeth and its posterior reossification occurs in different life stages of the animals.
Carolina Loch 1, 3 ; Paulo C. Simões-Lopes 1 & César J. Drehmer 2
the number of post-canine teeth in this family. Maxillary and mandibular supernumerary teeth were found in A. australis and A. tropicalis, but their positioning is unrelated to cases regarding phylogenetic and evolutionary implications. Dental losses were found in all species and different stages of alveolar obliteration suggest that this process is common in Otariids and does not affect their survival. The investigation of congenital and acquired dental anomalies in pinnipeds can provide information on dental formula evolution in Pinnipeds and in the phylogenetic relationships among Carnivora. KEY WORDS. Arctocephalus australis; A. tropicalis; dental agenesis; Otaria flavescens; southern Brazil; supernumerary teeth. ZOOLOGIA 27 (3): 477-482, June, 2010 mens were accessed in the scientific collection of mammals of the Departamento de Ecologia e Zoologia, Universidade Federal de Santa Catarina, under the anacronym UFSC (see Appendix). Skulls were prepared according to DREHMER & FERIGOLO
==
Domain: Biology Medicine Geography
|
[{"added": "2018-12-29T02:15:40.159Z", "created": "2010-05-11T00:00:00.000", "id": "83696975", "metadata": {"extfieldsofstudy": ["Biology", "Geography"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.scielo.br/j/zool/a/ZVphqgWtVVNhx5FBgyjrgsG/?format=pdf&lang=en", "pdf_hash": "23dabafd217d3e6c25e32cbb44e520cdcf149204", "pdf_src": "Anansi", "provenance": "peS2o-0028.json.gz:8233", "s2fieldsofstudy": ["Medicine", "Biology"], "sha1": "23dabafd217d3e6c25e32cbb44e520cdcf149204", "year": 2010}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Review of clypeasteroids phylogeny and a case study of Sinaechinocyamus mai ( Taiwanasteridae )
College of Medicine, National Taiwan University, No. 1, Sec. 1, Ren-Ai Road, Taipei 100, Taiwan Department of Geosciences, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 106, Taiwan Department of Life Science, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 106, Taiwan Institute of Biotechnology, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 106, Taiwan This manuscript is submitted to “Terrestrial, Atmospheric and Oceanic Sciences”
Sequencing Technology
The development of evolutionary theory and sequencing technology are coupled with each other. Since 2005, next-generation sequencing (NGS) is one of the most significant technological breakthroughs in the field of molecular biology (Goodwin et al., 2016). With the aid of NGS, the availability of non-model species DNA sequences significantly increases. Therefore, the new field "phylogenomic," using the genomic scale sequence data to reconstruct the phylogenetic relationship between species, has emerged and improved both the depth (more evolutionary distant clades, such as the divergence between kingdoms) and resolution (able to resolve the relationship between closely related groups) of the tree of life (Chan and Ragan, 2013).
In spite of the flourish of NGS technology, the traditional gene markers used in molecular evolution do not entirely step down from the stage. Due to various reasons (i.e., cost, difficulties in data analysis, sample quality), single gene markers remain useful in small-scale analyses. The massive quality of data produced by whole-genome sequencing provides detailed information but is cumbersome in analysis (Cannon and Kocot, 2016). One way to do so is by only sequencing the coding region of the genome. As shown in Table 1., transcriptome analysis represents a balance between the phylogeny resolution whole-genome sequencing (WGS) and the cost and convenient advantage of gene markers (Cannon and Kocot, 2016;Young and Gillung, 2019).
Modern biologists can facilitate paleontology studies in various ways. Through experimental models of modern species, developmental biologists can answer how certain structures possibly formed in fossil species and the potential gene regulatory networks responsible for the evolutionary origin and development of these characters (Thompson et al., 2017). The relationship among fossil species with extant representatives in their corresponding clades can be supported by reconstructing the phylogeny based on sequencing data of extant species. Furthermore, sequencing data is also able to provide evidence of divergence time and correlate with geological events (Coppard and Lessios, 2017). Analyzing the genetic material inside fossil specimens is a field under active development as well (Paabo et al., 2004). Under adequate conditions, ancient DNA (aDNA) can preserve up to more than one million years (van der Valk et al., 2021). Through direct sequencing of DNA of ancient species, researchers can be more certain about the population genetic dynamic and the relationship between fossil and extant species (Orlando and Cooper, 2014). Furthermore, the genetic basis provides an independent dataset for phylogeny inference other than traditional phenotype comparison alone (Orlando and Cooper, 2014). Therefore, the advent of sequencing technology has revolutionized paleontology by adding a new dimension to the study material. < A c c e p t e d M a n u s c r i p t >
Phylogeny of Irregular Echinoids
Echinoidea, commonly known as sea urchins, forms a conspicuous and essential element of many marine benthic communities. They exploit a wide array of marine habitats and contribute the greatest levels of biodiversity in shallow shelf areas. Arisen since the Middle Ordovician in 460 million years ago (Mya), there are more than 1000 living species in the clade of Echinoidea (Kroh, 2020;Kroh and Smith, 2010). The robust endoskeleton structure, known as stereom (Lin et al., In press), which is composed of multi-plated magnesium calcite, allows echinoids to leave a relatively rich fossil record: in total, there are about 10000 fossil species (Kroh, 2020). From paleontology to developmental biology, the extant species and the rich fossil record make the echinoids excellent candidates as valuable study models.
Among the clades in Echinoidea, Irregularia, including species are commonly known as sand dollars (Scutelloida) and sea biscuits (Clypeasteroida), distinguish themselves from irregular echinoids through the unique body plan (Saucède et al., 2007). One of the characteristic features of the Echinodermata phylum is the pentameric symmetry body plan. The larvae of echinoderms are bilaterally symmetrical, while upon metamorphosis, the larval structures typically are lost and yield a pentameric juvenile (Peterson et al., 2000). However, irregular echinoids share a secondary adult bilateral symmetry body plan (Fig. 1), and the establishment of body axis is still largely unknown and intriguing. Furthermore, the phylogeny study regarding the relationship among irregular echinoids suffers from the discrepancy between morphology and molecular evidence (Kroh and Smith, 2010;Mongiardino Koch et al., 2018).
Due to the fact that the taxonomy of Irregularia is still a topic under active discussion, in this article, the definition of Clypeasteroida is used in congruence with Kroh (2020) and Mongiardino Koch et al. (2020) to prevent ambiguity, which does not include the traditional suborder Scutellina. While the term "clypeasteroids" is used to denote the traditional clade consists of Clypeasterina and Scutellina.
Based on the recent total-evidence dated analysis, clypeasteroids possibly originated during the Cretaceous (145-66 Mya) (Mongiardino Koch and Thompson, 2020). By the Middle Eocene (49-37 million years ago), this group already presented in the fossil record worldwide (Pawson, 2007). The principal characters in echinoid fossil morphological analysis are the structure of test, spine, and Aristotle's lantern (Kroh and Smith, 2010;Ziegler et al., 2015). In particular, the lantern structure of clypeasteroids is highly modified compared to other echinoids, and this modification has been hypothesized to be crucial for the epifaunal inhabitant of clypeasteroids (Mooi, 1990a). On the contrary of abundant morphological studies, molecular efforts have lagged on echinoids phylogeny studies. Early attempts to use molecular data to resolve the deep-sea urchin phylogeny relied on just a few single-gene markers, usually ribosomal RNA-coding ones (Smith et al., 2006). The lack of broad sampling of loci across the genome limits the robustness of these phylogenetic analyses due to the genes used might be the outliers of incomplete lineage sorting. When an ancestral species undergoes < A c c e p t e d M a n u s c r i p t > rapid speciation in succession, ancestral polymorphisms may not be fully resolved into different monophyletic lineages, and the individual gene trees may be incongruent with the species tree (Galtier and Daubin, 2008). Other types of biases can also lead to inaccurate phylogenetic topology even in the absence of incomplete lineage sorting. For instance, species exhibiting significantly higher genetic distances to other clades tend to be erroneously clustered together. This phenomenon, known as long-branch attraction, is commonplace in molecular evolution inference, and it may arise from the heterogeneity of substitution rates, inadequate taxon sampling, and base compositional heterogeneity (Qu et al., 2017;Susko and Roger, 2021).
Therefore, although numerous studies have tried to investigate the phylogeny of Irregularia, the relationships among sea biscuits, sand dollars, and other irregular echinoids remain not well understood due to the contradicting results yielded by molecular and morphological data (Kroh and Smith, 2010;Mongiardino Koch et al., 2018). For example, most morphological phylogenies strongly supported the monophyly of clypeasteroids and their origin from a paraphyletic sister group collectively known as "cassiduloids", including extant clades of Echinolampadoida Kroh & Smith, 2010, Cassiduloida Claus, 1880, and other extinct lineages (Kroh and Smith, 2010). On the other hand, the transcriptomic analyses refute the monophyly of Clypeasteroida (Mongiardino Koch et al., 2018). This mismatch remains somewhat unresolved, and further investigation is needed to answer this perplexing inquest.
Body Plan
One of the most conspicuous characteristics of extant echinoids is their pentameral symmetry in the adult stage. Based on the observation of extant echinoderms and the fossil record, it seems the early echinoderms were bilateral in both the larval and adult life stages while the extant species developed secondary pentameral symmetry in adulthood (Peterson et al., 2000). Among many essential genes controlling development, one crucial important player is the homeobox genes, also known as Hox genes. They are a family of transcription factors regulating body plan and axis establishment in embryo development (Lacalli, 2014). A primary recognition criterion for these genes is a sequence of an 180 nucleotides, highly conserved "homeobox," which translates into a sequence of 60 amino acids (the homeodomain motif) that binds to specific DNA or RNA sequences (Mooi and David, 2008).
Hox genes have been found in animals, plants, fungi, and numerous other eukaryotes, and they are taxonomically widespread and highly conserved, suggesting a basal origin on the phylogenetic aspect. Almost all developmental gene networking involves the Hox gene family in metazoans. One of the most conspicuous phenotypic of the Hox cluster is the patterning along the posterior (A/P) axis of bilaterian animals (David and Mooi, 2014). Hox genes are generally arranged into three groups, or classes, on the chromosome: anterior (Hox1 and Hox2, with Hox3 possible categorized into an independent group); medial (Hox4 to Hox8); and posterior (Hox9 to Hox13) (Mooi and David, 2008). In echinoderms, the Hox 9/10 and Hox 11/13a, Hox 11/13b, and Hox 11/13c were designated due to the homologous analysis did not yield a clear one-to-one correlation between echinoderms genes and their chordates counterparts. The anterior class is closer to the 3' end and the posterior closer to the 5' end of the DNA strand. This remarkable linearity in the organization of the genes along chromosomes is the basis for a concept known as "collinearity" (Mooi and David, 2008).
There are two kinds of collinearity observed in Hox gene cluster: temporal and spatial. The genes of the anterior class (3' end) are activated first and expressed earlier in ontogeny than those of the medial class, and the medial ones are expressed earlier than those of the posterior class at the 5' end, and this is temporal collinearity (Mooi and David, 2008). In organisms that grow from anterior to posterior, such as arthropods, temporal collinearity is in turn transcribed into spatial collinearity, in which the first expressed genes at the 3' end of the cluster code for developmental events in the anterior part of the organism, and the last ones at the 5' end for events in the posterior part (Mooi and David, 2008). The schematic diagram for temporal and spatial collinearity of Hox gene cluster is shown in Figure 2.
The Hox gene cluster in Echinodermata lost the collinearity, as shown in Figure 3. Among all echinoderms, the echinoid cluster is by far the most thoroughly investigated, as it has been studied in several species belonging to quite disparate clades. In the Strongylocentrotus purpuratus, one of the most important model species of echinoids with its whole genome sequenced and annotated, the Hox cluster is almost complete besides Hox4, leaving a total of 11 genes (Li et al., 2020). The peculiarity of echinoid Hox clusters is not so much the absence of a member in the series but that the order of the genes themselves along the chromosome is rearranged. Hox1 to Hox3 are located at the 5' end of the cluster in reverse order. This topology is the result of a chromosomal translocation and inversion event.
This observation of unique rearrangement inspired the hypothesis of the body plan of echinoids being a result of a disorganized Hox cluster (Mooi and David, 2008). Since the S. purpuratus is the first sequenced echinoid and the first in the whole phylum Echinodermata, it served as a model species for this highly diverse clade initially (Sea Urchin Genome Sequencing Consortium et al., 2006). However, some recent studies question the significance of the loss of collinearity in Echinodermata species due to a better understanding of the genomic landscape of other Echinodermata species (Byrne et al., 2016). As shown in Figure 3., sequenced samples from the Crinoidea (sea lily), Asteroidea (starfish), and Holothuroidea (sea cucumber) clades show no signs of the same translocation/inversion of S. purpuratus (Li et al., 2020).
Sinaechinocyamus mai
Here, we introduce a recent study to exemplify the importance of incorporating molecular methods in modern taxonomy research. Among 15 extant clypeasteroids recorded in Taiwan shown in Table 2., Sinaechinocyamus mai (Wang, 1984) is one of the most studied sand dollars (Figure 4.). The species was initially described by Chai-Chin Wang in 1984, and the species name is in honor of the Ting-Ying Ma, one of the eminent pioneer geologists in Taiwan and China (Wang, 1984). The original species assignment was Taiwanaster mai.
However, Mooi examined the description of the feature of Taiwanaster and found that
Taiwanaster is the same genus as Sinaechinocyamus Liao, 1979. Therefore, Taiwanaster is an invalid junior synonym, and Taiwanaster mai was reassigned as S. mai (Mooi, 1990b).
Aside from the paleontology research on the fossil record, investigation of the living populations of S. mai began in the early 1990s. The ecology habitat of S. mai is in the intertidal zone along the west coast of Taiwan, from Hsinchu to the mouth of the Tsengwen river (Lee et al., 2019). Since S. mai possesses several characters that do not fit any other clypeasteroids family, Wang proposed a novel superfamily, Taiwanasteritida Wang, 1984, and a new family Taiwanasteridae Wang, 1984 to place S. mai (Wang, 1984). Mainly based on the observation of miniaturized body size, Wang proposed that Sinaechinocyamus was related to other microechinoids in the Asia-Pacific region, such as Fibulariella acuta (Yoshiwara, 1898). On the other hand, Mooi concluded that Sinaechinocyamus is a derived scutelline sand dollar (Mooi, 1990b). More recent morphology-based phylogeny also supports the hypothesis that S. mai is a derived scutelline and places the family Taiwanasteridae as incertae sedis within Scutelliformes (Kroh and Smith, 2010). Based on the morphological study, the extant sister group may be Scaphechinus mirabilis, a widely distributed sand dollar in the northwestern Pacific (Mooi, 1990b). Adult S. mirabilis generally has a body diameter of approximately 7 cm in size, while adult S. mai nearly never exceeds 1 cm (Chen and Chao, 1997).
Miniaturization does not only represent smaller body size but also contributes to essential physiology, behavior, development, and reproductive changes according to the size. There are two main models of paedomorphosis: "progenesis", which miniaturized species has a similar growth rate but growth period ends earlier, and "neoteny", which is characterized by a similar growth period with a slower growth rate (McNamara, 1986). In the case of S. mai, it seems that the predominant model might be the neoteny due to the homogenous decrease of growth rate to 19% of S. mirabilis with a comparable growth period (Chen and Chao, 1997). Further investigation of comparing the developmental process of S. mirabilis and S. mai might give valuable insight into the mechanism of miniaturization of S. mai. Fossil record and paleoclimatology may provide a possible evolutionary explanation of miniaturization.
In this study, we aimed to utilize the transcriptome-based approach to resolve the phylogenetic placing of S. mai, and investigate the monophyly of clypeasteroids through incorporating public available transcriptomic datasets of irregular echinoids.
Sample and library preparation
Live specimens of Sinaechinocyamus mai were collected in their native habitat in Miaoli County, Taiwan (120°39′E; 24°29′N) in May 2018. Total RNA of the sample was extracted using Tri Reagent (Ambion, USA). RNA concentration and quality were determined using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, USA) and an Agilent 2100 Bioanalyzer (Agilent Technologies, USA), which calculates an RNA integrity number (RIN).
Total RNA with A260/A280 = 1.8-2.0 and RIN >8.0. Normalize intact RNA samples to 200 ng/μl with DEPC-treated H2O. Total 1~2 ug RNA was purified by using poly-T oligo-attached magnetic beads. Following purification, the mRNA was fragmented into small pieces using divalent cations under elevated temperatures. The cleaved RNA fragments were reverse transcribed into first-strand complementary DNA (cDNA) using reverse transcriptase and random primers. This was followed by second-strand cDNA synthesis using DNA Polymerase I and RNase H. These cDNA fragments then went through an end repair process, adding of a single "A" base, and then ligating the adapters. The products were then purified and enriched with PCR to create the final cDNA library. Paired-end 75 nucleotide reads from each mRNA library were obtained using NextSeq500 (Illumina Inc., CIC bioGUNE, Spain).
RNA sequencing
We quantified the cDNA with Life Technologies Qubit and Agilent Technologies Tapestation 4200 and used 0.625 nanograms of size-selected cDNA for the sample preparation and the barcoded library prepared by the Chromium Gel bead and library kit (10X Genomics, USA).
Barcoded library sequencing was performed with Illumina Novaseq 6000 sequencer with 2 x 151 paired-end reads. All read-pairs contain a 16-base barcode.
Bioinformatic analysis
The list of the species incorporated in this study is shown in Table 3. The raw reads from the RNA sequencing of Sinaechinocyamus mai in this study and published data in the Sequence Read Archive (SRA) repository underwent quality control and adapter removal using Trimmomatic (Bolger et al., 2014) with default parameters. Reads passed quality control was assembled by Trinity (Haas et al., 2013). Transcriptome-based phylogenetic analysis was then conducted through the Agalma pipeline (Dunn et al., 2013), which sequentially annotated the transcriptome assembly with BLAST (Camacho et al., 2009), aligned gene clusters with MACSE (Ranwez et al., 2011), trimmed alignments with GBlocks (Talavera and Castresana, 2007), and provided preliminary maximum likelihood (ML) phylogeny based on the supermatrix with RAxML (Stamatakis, 2014).
Further ML phylogeny analysis was separately conducted with partition of homologues based on the Agalma pipeline output and without partition in RAxML v. 8.2.12 with 100 replicates, respectively. The substitution model was selected automatically by RAxML (-m PROTGAMMAAUTO).
Transcriptome-based phylogeneomics
The analysis included 22 echinoid species, and Eucidaris tribuloides was designated as the outgroup. In total, the Agalma pipeline yielded 9630 loci comprised 2,574,725 amino acid positions with an actual gene occupancy of 39.2%. The concatenated supermatrix dataset was then used in the transcriptome-based phylogenetic analysis. Both partitioned and unpartitioned dataset was used to infer the phylogeny, and the topology was identical. The phylogeny is shown in Figure 5. The result is congruent with and Koch et al. (2020). The result revealed that the "cassuloids" Conolampas sigsbei is a sister group to the Scutelloida clade. Therefore, the monophyly of traditional clypeasteroids was not supported. As for the newly sequenced S. mai, the result suggests that S. mirabilis is its sister taxa in the sampled dataset, supporting previous studies drawn on morphological analysis.
Monophyly of clypeasteroids
The overall cladogram comparison between previous studies and the result from this work is shown in Figure 6. Mongiardino Koch et al. (2018) firstly used transcriptome-based phylogeny to challenge the morphology-based monophyly of clypeasteroids. In this study, the "cassiduloids" included in the analysis, Conolampas sigsbei, intertwined in the phylogeny relationship of Clypeasteroida and Scutelloida.
In our previous study (Lin et al., 2020), we sequenced the whole mitochondria genome of S. mai and inferred the phylogenetic relationship through the data. The availability of a mitogenome for S. mai sets the stage for exploring the systematic position of this taxon.
According to the mitogenome-based phylogeny, the clades of Echinolampadoida and Scutelloida both receive 100% bootstrap support (Lin et al., 2020). Consistent with the hypothesis raised by Mooi, this result suggests that the presence of Aristotle's lantern in adults was firstly lost in the basal Irregularia, then the re-appeared in the Scutelloida clade (Mooi, 1990a). However, due to the fact that mitogenome data from the Scutelloida clade was unavailable, and this work was unable to support nor refute the monophyly of clypeasteroids.
On the other hand, one analysis published recently tried reconstructing the phylogeny of Echinoidea using the total-evidence dated approach incorporating the fossil morphology evidence and transcriptomic data of extant species (Mongiardino Koch and Thompson, 2020).
< A c c e p t e d M a n u s c r i p t >
By combining the morphological fossil record and molecular data (transcriptome dataset was used for phylogenetic inference in this study), the monophyly of clypeasteroids was refuted again. Therefore, under this phylogenetic theory, the morphological characters shared between Clypeasteroida and Scutelloida (e.g., extremely flattened test and presence of Aristotle's lantern) result from convergent evolution rather than true synapomorphies (Mongiardino Koch and Thompson, 2020).
The transcriptome-based phylogeny analysis presented in this article was consistent with Mongiardino Koch and Thompson (2020), which did not support the monophyly of clypeasteroids. The reason for this inconsistency between mitogenome-and transcriptomebased phylogeny might be primarily due to the limited sampling. As the mitogenome-based analysis did not include Clypeaster spp. from Clypeasteroida, the possible polyphyly status of clypeasteroids cannot be revealed.
Future Perspective
Despite the fact the seemingly solid evidence provided in the transcriptome-based phylogeny, the only representative of "cassiduloids" clade was the extant species Conolampas sigsbei which was resolved as the sister group to Scutelloida. Therefore, one should be careful when interpreting this result, and further investigation sampling more species in the "cassiduloids" clade is still needed for a deeper understanding of the conundrum of the monophyly of clypeasteroids. Nevertheless, these recent works showcase the potential of using molecular data to aid phylogenetic research in the field.
On the other hand, although the role of Hox gene cluster rearrangement in establishing pentameric symmetry of Echinodermata has been questioned based on the due to several sequenced species from Crinoidea (sea lily), Asteroidea (starfish), and Holothuroidea (sea cucumber) clades do not present with rearranged clusters. Due to the lack of available data on irregular echinoids, it is still unclear if the structure of Hox gene cluster contributes to secondary bilateralism in the Irregularia. A further investigation of the high-quality genome assembly within this group will benefit the research of this topic.
CONCLUSION
The technology breakthroughs in the first ten years of the 21st century serve as an important foundation for further investigation in the field of genomics, and investigators of other specialties benefit significantly as well, including paleontologists. Both the traditional morphology-based paleontology and novel genetic methods have their own limitation. However, by combining these two paradigms, further investigation may shed light on the phylogenetic relationships, developmental processes, and the possible mechanism behind the secondary bilateral symmetry of clypeasteroids. Fell (1966) and Mooi (1989). Figure 2. The schematic diagram for the temporal and spatial collinearity of the Hox gene cluster.
Figure legends
Data from Mooi and David (2008). Sinaechinocyamus mai (Wang, 1984) Table 3. The list of species used in the transcriptome-based phylogenetic analysis in this study. The clade designation is based on Kroh (2020) and Horton et al. (2021).
==
Domain: Biology Geology
|
[{"added": "2021-12-31T16:12:05.409Z", "created": "2021-01-01T00:00:00.000", "id": "245576480", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "http://tao.cgu.org.tw/index.php/articles/archive/geology/item/download/3677_eefa91bd45997cfe78e9d0264e89f593", "pdf_hash": "d3394ca720c14f79609795afe00d4677830c4736", "pdf_src": "Anansi", "provenance": "peS2o-0028.json.gz:9407", "s2fieldsofstudy": ["Biology", "Geology"], "sha1": "911b1bfb56dcfed150f5fd6c90afb46aa2b0b799", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
In Vivo Multimodal Imaging of Stem Cells Using Nanohybrid Particles Incorporating Quantum Dots and Magnetic Nanoparticles
The diagnosis of the dynamics, accumulation, and engraftment of transplanted stem cells in vivo is essential for ensuring the safety and the maximum therapeutic effect of regenerative medicine. However, in vivo imaging technologies for detecting transplanted stem cells are not sufficient at present. We developed nanohybrid particles composed of dendron-baring lipids having two unsaturated bonds (DLU2) molecules, quantum dots (QDs), and magnetic nanoparticles in order to diagnose the dynamics, accumulation, and engraftment of transplanted stem cells, and then addressed the labeling and in vivo fluorescence and magnetic resonance (MR) imaging of stem cells using the nanohybrid particles (DLU2-NPs). Five kinds of DLU2-NPs (DLU2-NPs-1-5) composed of different concentrations of DLU2 molecules, QDs525, QDs605, QDs705, and ATDM were prepared. Adipose tissue-derived stem cells (ASCs) were labeled with DLU2-NPs for 4 h incubation, no cytotoxicity or marked effect on the proliferation ability was observed in ASCs labeled with DLU2-NPs (640- or 320-fold diluted). ASCs labeled with DLU2-NPs (640-fold diluted) were transplanted subcutaneously onto the backs of mice, and the labeled ASCs could be imaged with good contrast using in vivo fluorescence and an MR imaging system. DLU2-NPs may be useful for in vivo multimodal imaging of transplanted stem cells.
Introduction
Stem cell transplantation therapy were known as very simple and low invasive medicine in regenerative medicine and could be utilized for treating many serious diseases such as heart, liver, and central nervous system (CNS) disorders. Indeed, stem cell transplantation therapy with somatic stem cells which can transplant with lower invasiveness, such as bone marrow stem cells (BMSCs) [1,2] and adipose tissue-derived stem cells Sensors 2022, 22, 5705 2 of 10 (ASCs) [3,4], has been applied in clinical practice. It is known that the accumulation and engraftment of transplanted stem cells in affected tissues and organs strongly influences the therapeutic efficacy [5,6], and the inflammatory state of the affected tissues and organs is also considered to be similarly affected [7][8][9], but little is known on this subject. Therefore, in vivo real-time imaging of the kinetics of transplanted stem cell behavior, accumulation, and engraftment is essential to ensure maximum therapeutic safety and efficacy of stem cell transplantation. Several methods have been used in clinical practice, such as X-ray computed tomography (CT) and magnetic resonance imaging (MRI); however, these methods struggle to detect transplanted stem cells with high sensitivity when these modalities are used alone. To overcome these issues, multimodal imaging combining fluorescence imaging (FI) and MRI that can detect the small numbers of transplanted stem cells has been developed and has drawn a great deal of attention [10][11][12].
Quantum dots (QDs) have several outstanding fluorescence properties to conventional organic labels such as a high luminance, resistance to photobleaching (long time labeling), wide range of excitation wavelength, and narrow fluorescence wavelength. In particular, QDs which are water-soluble and emit fluorescence in the near-infrared (NIR) region, have attracted attention for their diagnostic applications in the medical field as useful fluorescent probes [13][14][15][16][17][18][19][20]. We previously developed stem cell labeling technology using QDs and in vivo fluorescence imaging technology of transplanted stem cells labeled with QD [4,[21][22][23][24].
Magnetic nanoparticles are known as contrast agents for MRI. Gadolinium (Gd) and superparamagnetic iron oxide (SPIO) nanoparticles are generally used as magnetic nanoparticles in order to increase the contrast of issues in typical imaging studies [25][26][27][28]. Various SPIO nanoparticles have been developed as contrast agents; ferucarbotran (Resovist), an anionic SPIO nanoparticle with a carboxydextran coating, has been successfully applied in the clinical setting as a liver contrast agent [29][30][31][32]. We observed the alkali-treated dextrancoated magnetic iron oxide nanoparticles (ATDM), which are a major component of ferucarbotran (Resovist), and have already applied magnetic nanoparticles to cell labeling and in vivo MRI techniques [33,34].
In this study, we developed nanohybrid particles (DLU2-NPs) composed of DLU2 molecules, three kinds of QDs, and magnetic nanoparticles (ATDM) and addressed the labeling and in vivo fluorescence and MR imaging of transplanted stem cells labeled with the DLU2-NPs.
Animals
C57BL/6 mice were purchased from Japan SLC (Hamamatsu, Japan). The mice were kept in a controlled environment (12 h light/dark cycles at 21 • C) with free access to water and a standard chow diet before sacrifice. All conditions and handing of animals in this study were in accordance with the protocols approved by the Nagoya University Committee on Animal Use and Care.
Isolation and Culture of ASCs
The isolation and culture of ASCs have been reported previously [4]. Female C57BL/6 mice 7-14 months of age were killed by cervical dislocation; adipose tissue specimens in the inguinal groove were isolated and washed extensively with Hank's balanced salt solution or PBS to remove the blood cells. The isolated adipose tissue specimens were cut finely and digested with 1 mL of 1 mg/mL type I collagenase (274 U/mg) at 37 • C in a shaking water bath for 45 min. The cells were filtrated using 250 µm nylon cell strainers and suspended in DMEM/F12 containing 20% FBS, and 100 U/mL of penicillin/streptomycin (culture medium). They were centrifuged at 1200 rpm for 5 min at room temperature and ASCs were obtained from the pellet. The cells were then washed three times by suspension and centrifugation in culture medium and then were incubated overnight in culture medium at 37 • C with 5% CO 2 . The primary cells were cultured for several days until they reached confluence and defined as passage "0". The cells were used for the experiments between passages 2 and 5.
Preparation of DLU2-NPs
DLU2 was dissolved in chloroform/methanol (1/1, v/v) in a volumetric flask and the concentration of DLU2 was adjusted to 10 mg/mL. DLU2 solution was distilled away under nitrogen purging and dried by vacuum drying. The solutions of QDs525 (100 nM), QDs605 (100 nM), QDs705 (100 nM), and ATDM (500 µg/mL) were added to the lipid film composed of DLU2, and then 10 mM HEPES buffer was added. After sonication of the mixture for 30 min, DLU2-NPs (10 mg/mL of DLU2) were obtained.
Measurement of Particle Size and Zeta Potential of DLU2-NPs
DLU2-NPs were diluted 1:100 using PBS, and the particle size and zeta potential were measured by dynamic light scattering (DLS) using the Zetasizer Nano ZS (Malvern Instruments, Ltd., Herrenberg, Germany). These measurements were performed at 25 • C.
Confocal Microscopy Observation of Labeled ASCs
ASCs (1 × 10 5 cells) were seeded in 35 mm glass-bottom dishes with 200 µL of culture medium for 24 h and that was replaced with 200 µL of DLU2-NPs diluted 640-fold in medium. After incubation for 1 or 4 h, ASCs were washed twice with medium. The labeled ASCs were observed with a high-speed multi-photon confocal laser microscope (A1R MP + ; Nikon Corporation, Tokyo, Japan).
Cytotoxicity of DLU2-NPs to ASCs
ASCs (1 × 10 4 cells) were seeded in 96-well plates (BD Biosciences) with 100 µL of culture medium that was then replaced with 100 µL of DLU2-NPs diluted 640-, 320-, 160-, 80-, or 40-fold in medium. After incubation for 4 h, ASCs were washed twice with medium. Viable cells were counted by using a CCK-8. CCK-8 reagent (10 µL) was added to each well, and the reaction was allowed to proceed for 1 h. The absorbance of the sample at 450 nm was measured against a background control using a microplate reader (PLARstar OPTIMA, BGM Labtech, Ortenberg, Germany).
Proliferation test of ASCs Labeled with DLU2-NPs
ASCs (2 × 10 3 cells) were seeded in 96-well plates and labeled with 100 µL of DLU2-NPs diluted 640-or 320-fold in medium in the same way. After incubation for 4 h, ASCs were washed twice with medium. Viable cells were counted using a CCK-8 in the same way.
In Vitro Fluorescence and MR Imaging of ASCs Labeled with DLU2-NPs
ASCs were cultured for several days until they reached confluence and then were replaced with 5 mL of DLU2-NPs diluted 640-, 320-, or 160-fold with medium. After incubation for 4 h, ASCs were washed twice with medium and collected in 1.5 mL centrifuge tubes. The fluorescence images of ASCs with DLU2-NPs in 1.5 mL centrifuge tubes were taken using the IVIS Lumina K Series III (PerkinElmer Inc, Waltham, MA, USA; excitation filter: 460-620 nm, emission filter: 520, 570, and 710 nm long-pass). In vitro MR images of ASCs with DLU2-NPs in 1.5 mL centrifuge tubes were taken using the MR VivoLVA
In Vivo Fluorescence and MR Imaging of ASCs Labeled with DLU2-NPs
ASCs were cultured for several days until they reached confluence, and then were replaced with 5 mL of DLU2-NPs diluted 640-or 320-fold in medium. After incubation for 4 h, ASCs were washed twice. ASCs (3 × 10 6 cells) labeled with DLU2-NPs with 0.2 mL PBS were transplanted into the space of back subcutaneously of C57BL/6 mice. In vivo fluorescence imaging of ASCs labeled with DLU2-NPs images were taken using the IVIS Lumina K Series III (PerkinElmer Inc.; excitation filter: 500-620 nm, emission filter: 520, 620 and 710 nm long-pass). In vivo MRI were taken using the MR VivoLVA (DS Pharma Biomedical Co.). Regarding the spin echo, the images were obtained with repetition time (TR) = 500.0 ms and echo time (TE) = 9.0 ms, or TR = 2000.0 ms and TE = 69.0 ms and field of view = 60 × 60 mm.
Statistical Analyses
Numerical values are presented as the mean ± standard deviation (SD). Each experiment was repeated three times. Statistical significance was evaluated using an unpaired Student's t-test for comparisons between the two groups; p-values < 0.05 were considered statistically significant. All statistical analyses were performed using the SPSS software package.
Properties of DLU2-NPs
The schematic diagram of DLU2-NPs composed of three kinds of QDs (QDs525, 605, 705), magnetic nanoparticles (ATDM), and DLU2 molecules is shown in Figure 1a. The chemical structural formula of the DLU2 molecule consisted of a cationic lipid composed of a polyamidoamine dendron and two alkyl chains was shown in Figure 1b.
The two alkyl chains in the DLU2 molecule are said to have a membrane fusion function, and the presence of an unsaturated bond further promotes membrane fusion. These effects are expected to promote the introduction of quantum dots and magnetic nanoparticles in liposomes into cells [23]. The tertiary amino groups of the polyamidoamine dendron have been reported to adsorb protons, thereby reducing the pH endosomes and accelerating the influx of H + and Cl -, causing Claccumulation and permeation swelling/dissolution [35].
The concentrations of DLU2 molecules, QDs, and ATDM contained in DLU2-NPs are shown in Table 1. The distribution of the particle diameter and zeta potential of DLU2-NPs is shown in Figure 1c,d. The average particle diameter and zeta potential of DLU2-NPs were 135.7 nm and 28.7 mV, respectively. The two alkyl chains in the DLU2 molecule are said to have a membrane fusion function, and the presence of an unsaturated bond further promotes membrane fusion. These effects are expected to promote the introduction of quantum dots and magnetic nanoparticles in liposomes into cells [23]. The tertiary amino groups of the polyamidoamine dendron have been reported to adsorb protons, thereby reducing the pH endosomes and accelerating the influx of H + and Cl -, causing Claccumulation and permeation swelling/dissolution [35].
The concentrations of DLU2 molecules, QDs, and ATDM contained in DLU2-NPs are shown in Table 1. The distribution of the particle diameter and zeta potential of DLU2-NPs is shown in Figure 1c,d. The average particle diameter and zeta potential of DLU2-NPs were 135.7 nm and 28.7 mV, respectively.
ASCs labeling by DLU2-NPs
DLU2-NPs were transduced into ASCs for 4 h of incubation at 37 °C (Figure 2a). The morphology and fluorescence images of ASCs were obtained by confocal microscopy. The
ASCs labeling by DLU2-NPs
DLU2-NPs were transduced into ASCs for 4 h of incubation at 37 • C (Figure 2a). The morphology and fluorescence images of ASCs were obtained by confocal microscopy. The strong fluorescence derived from QDs525, QDs605, and QDs705 was detected at the same site in the cytoplasm of ASCs (Figure 2b-g). No abnormalities in the morphology of labeled ASCs were observed. These data suggest that ASCs can be labeled with DLU2-NPs by simple culture for 4 h without cytotoxicity.
Cytotoxicity of DLU2-NPs to ASCs and the Proliferation Rate of ASCs Labeled with DLU2-NPs
To examine the cytotoxicity of DLU2-NPs to ASCs, ASCs were transduced with various concentrations (640-, 320-, 160-, 80-, or 40-fold diluted solution) of DLU2-NPs for 4 h, and ASCs were incubated for 24 h. Significant cytotoxicity was observed in the ASCs labeled with 160-, 80-, and 40-fold-diluted solution of DLU2-NPs; however, no cytotoxicity was observed in the ASCs labeled with 640-and 320-fold-diluted solution of DLU2-NPs (Figure 3a).
The influence of the DLU2-NPs on the proliferation ability of ASCs was also examined at non-cytotoxic concentrations. The proliferation rates of ASCs labeled with DLU2-NPs in non-cytotoxic concentrations were confirmed to be almost equal to that of normal ASCs (Figure 3b). These data suggest that DLU2-NPs have no cytotoxicity and no effect on the proliferation ability of ASCs at 640-and 320-fold-diluted concentrations.
In Vitro Fluorescence and MR Imaging of ASCs Labeled with DLU2-NPs
To examine the detectable labeling concentration of DLU2-NPs, ASCs labeled with various concentrations of DLU2-NPs were collected in PBS and spun down. The pellets of ASCs in microtubes were then prepared for a fluorescence analysis (Figure 4a). All fluorescence derived from three kinds of QDs (525, 605 and 705) were detected even at 640-fold-diluted concentrations (Figure 4b-d). The ASC pellets were also then prepared for MR imaging in microtubes (Figure 4e). The MR signal of ASCs labeled with DLU2-NPs were detected on T2-weighted imaging even at 640-fold-diluted concentrations (Figure 4f,g). In addition, the MR signal (drawn by yellow dot line) of ASCs labeled with DLU2-NPs was evaluated (Figure 4h). These data suggest that ASCs labeled with DLU2-NPs could be detected with multicolor fluorescence and MR imaging. strong fluorescence derived from QDs525, QDs605, and QDs705 was detected at the same site in the cytoplasm of ASCs (Figure 2b-g). No abnormalities in the morphology of labeled ASCs were observed. These data suggest that ASCs can be labeled with DLU2-NPs by simple culture for 4 h without cytotoxicity.
Cytotoxicity of DLU2-NPs to ASCs and the Proliferation Rate of ASCs Labeled with DLU2-NPs
To examine the cytotoxicity of DLU2-NPs to ASCs, ASCs were transduced with various concentrations (640-, 320-, 160-, 80-, or 40-fold diluted solution) of DLU2-NPs for 4 h, and ASCs were incubated for 24 h. Significant cytotoxicity was observed in the ASCs labeled with 160-, 80-, and 40-fold-diluted solution of DLU2-NPs; however, no cytotoxicity was observed in the ASCs labeled with 640-and 320-fold-diluted solution of DLU2-NPs (Figure 3a). site in the cytoplasm of ASCs (Figure 2b-g). No abnormalities in the morphology of labeled ASCs were observed. These data suggest that ASCs can be labeled with DLU2-NPs by simple culture for 4 h without cytotoxicity.
Cytotoxicity of DLU2-NPs to ASCs and the Proliferation Rate of ASCs Labeled with DLU2-NPs
To examine the cytotoxicity of DLU2-NPs to ASCs, ASCs were transduced with various concentrations (640-, 320-, 160-, 80-, or 40-fold diluted solution) of DLU2-NPs for 4 h, and ASCs were incubated for 24 h. Significant cytotoxicity was observed in the ASCs labeled with 160-, 80-, and 40-fold-diluted solution of DLU2-NPs; however, no cytotoxicity was observed in the ASCs labeled with 640-and 320-fold-diluted solution of DLU2-NPs (Figure 3a).
In Vivo Fluorescence and MR Imaging of Transplanted ASCs Labeled with DLU2-NPs
ASCs (3 × 10 6 cells) labeled with DLU2-NPs were subcutaneously transplanted with PBS onto the back of mice (Figure 5a). All three kinds of QDs (QDs525, QDs605, and QDs705) were observed by in vivo fluorescence imaging system (IVIS Lumina K Series III) (Figure 5b-d). The fluorescence derived from QDs705 was detected with high efficiency, as the fluorescence in the near infrared (NIR) (700-900 nm) known as the biological window. In contrast, the MR signal of ATDM was observed using an in vivo MR imaging system (MR VivoLVA) shown in a yellow-dotted circle (Figure 5e-l). These data suggest that transplanted ASCs labeled with DLU2-NPs in mice could be detected with both fluorescence and MR imaging.
In Vivo Fluorescence and MR Imaging of Transplanted ASCs Labeled with DLU2-NPs
ASCs (3 × 10 6 cells) labeled with DLU2-NPs were subcutaneously transplanted with PBS onto the back of mice (Figure 5a). All three kinds of QDs (QDs525, QDs605, and QDs705) were observed by in vivo fluorescence imaging system (IVIS Lumina K Series III) (Figure 5b-d). The fluorescence derived from QDs705 was detected with high efficiency, as the fluorescence in the near infrared (NIR) (700-900 nm) known as the biological window. In contrast, the MR signal of ATDM was observed using an in vivo MR imaging system (MR VivoLVA) shown in a yellow-dotted circle (Figure 5e-l). These data suggest that transplanted ASCs labeled with DLU2-NPs in mice could be detected with both fluorescence and MR imaging.
Conclusions
In summary, multifunctional nanoparticles (DLU2-NPs) were prepared and examined for their utility in stem cell labeling and in vivo imaging of transplanted stem cells. DLU2-NPs were able to be transduced into ASCs with high efficiency by simple incubation for 4 h. No cytotoxicity of ASCs labeled with DLU2-NPs was noted under certain concentrations of DLU2-NPs. In addition, no effect of DLU2-NPs on the proliferation ability of labeled ASCs was observed. The fluorescence and MR signal of ASCs labeled with DLU2-NPs was detected in vitro. Furthermore, in vivo fluorescence and MR multimodal imaging of transplanted ASCs labeled with DLU2-NPs was achieved using in vivo imaging systems. These data suggest that the development of new multimodal imaging technologies able to detect the dynamics of transplanted stem cells may be imminent. In vivo fluorescence and MR imaging of ASCs labeled with DLU2-NPs. In vivo fluorescence images of ASCs (3 × 10 6 cells) labeled with DLU2-NPs (320-dilute solution) after subcutaneous transplantation onto the back of a mouse, excitation: 500-620 nm, emission: 520-710 nm (a-d). In vivo MR images of labeled ASCs, the transversal images of non-labeled ASCs (e,f), labeled ASCs (g,h), the sagittal images (i,j,k,l), T1-weighted image (e,g,i,j), T2-weighted images (f,h,k,l). (j,l) are enlarged images of (i,k). Yellow arrows and dotted circles indicate the locations where ASCs were transplanted.
Conclusions
In summary, multifunctional nanoparticles (DLU2-NPs) were prepared and examined for their utility in stem cell labeling and in vivo imaging of transplanted stem cells. DLU2-NPs were able to be transduced into ASCs with high efficiency by simple incubation for 4 h. No cytotoxicity of ASCs labeled with DLU2-NPs was noted under certain concentrations of DLU2-NPs. In addition, no effect of DLU2-NPs on the proliferation ability of labeled ASCs was observed. The fluorescence and MR signal of ASCs labeled with DLU2-NPs was detected in vitro. Furthermore, in vivo fluorescence and MR multimodal imaging of transplanted ASCs labeled with DLU2-NPs was achieved using in vivo imaging systems. These data suggest that the development of new multimodal imaging technologies able to detect the dynamics of transplanted stem cells may be imminent. Informed Consent Statement: Not applicable.
Data Availability Statement:
The data presented in this study are available upon request to the corresponding author.
==
Domain: Biology Materials Science
|
[{"added": "2022-08-04T15:05:09.137Z", "created": "2022-07-30T00:00:00.000", "id": "251297206", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1424-8220/22/15/5705/pdf?version=1659169860", "pdf_hash": "567a931c567486c508d8359261f4e3d2c41ed8c3", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:8865", "s2fieldsofstudy": ["Biology", "Materials Science"], "sha1": "83d3b546ae72884e2ced6f0a6c293002239187a3", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Pharmaceutical Nanotechnology: A Therapeutic Revolution
The defi nition of nanotechnology is not yet a consensus in the scientifi c community. Among the most widespread concepts, it can be defi ned as the science that studies nanoscale materials (1 to 1000 nm) involving areas such as materials engineering, energy, biotechnology, physics and pharmacy, among others [1]. It is based on the development of nanostructures, providing revolutionary applications in various sciences [2]. Among the applications of nanotechnology is the pharmaceutical nanotechnology, a developed expertise mainly by pharmacists, engineers and biotechnologists. Regarding to pharmaceutical nanotechnology, it is based on the life sciences, which allows the development of nanostructures capable of promoting innovative drug delivery systems as therapeutic alternatives to various pathologies, as well as biosensors of nanomaterials to perform advanced diagnostics.
Introduction
The defi nition of nanotechnology is not yet a consensus in the scientifi c community. Among the most widespread concepts, it can be defi ned as the science that studies nanoscale materials (1 to 1000 nm) involving areas such as materials engineering, energy, biotechnology, physics and pharmacy, among others [1]. It is based on the development of nanostructures, providing revolutionary applications in various sciences [2]. Among the applications of nanotechnology is the pharmaceutical nanotechnology, a developed expertise mainly by pharmacists, engineers and biotechnologists. Regarding to pharmaceutical nanotechnology, it is based on the life sciences, which allows the development of nanostructures capable of promoting innovative drug delivery systems as therapeutic alternatives to various pathologies, as well as biosensors of nanomaterials to perform advanced diagnostics.
Nanotechnology and pharmaceutical industry
Faced with so many positive aspects offered by this technology, the pharmaceutical industry has been increasingly inserting nanotechnology in its products, based on the concept that innovation moves the world. In addition, nanotechnology has been decisive in the production and optimization of drugs based on potentially promising active principles, but which have limitations that compromise their application. The table 1 exemplifi es some success of nanoformulations transferred to the industry.
Among the biggest problems can be mentioned, high toxicity, degradation of the active ingredient, quick release, non-specifi city, reduced bioavailability and low solubility [3].
To mention, one of the great challenges of the pharmaceutical industry has been the use of pharmaceutical products with low solubility in water and their bioavailability in the therapeutic window. It is known that, of the pool of molecules of pharmaceutical interest under development, 90% have low water solubility, confi gured by the biopharmaceutical classifi cation as class II molecules (low water solubility and high permeability) and class IV (low solubility) water and low permeability) [4,5]. This limitation can be overcome by nanocarriers, opening the door to the development of many new treatments.
Nanocarriers in focus
Nanocarriers are transporters of active ingredients in the nanoscale that have the function of directing substances, increasing bioavailability, reducing toxicity, in addition to modulating the kinetics profi le of the active principle. As an example of nanocarriers, we can mention nanoemulsions, microemulsions and nanoparticles. Nanoemulsions are nanotechnological systems composed of an oily phase and an aqueous phase that is emulsifi ed in the presence of surfactants, which will reduce the surface tension of the phases and, therefore, will allow obtaining nanometric drops in the range of 50 to 500 nm. These systems are thermodynamically unstable and kinetically stable, that is, it is necessary to supply energy to obtain this "stable" system for a time, also called, metastable [6,7]. obtain a microemulsion, since the most comfortable energy state is not the separation of phases, but the microemulsifi ed conformation [6,8].
In the universe of nanocarriers, there are also nanoparticles.
These systems, unlike nanoemulsions and microemulsions that have droplets, presented particles(solid-state). Nanoparticles are colloidal systems at the nanoscale that have been developed as an important strategy for carrying conventional drugs, recombinant proteins, vaccines and, more recently, nucleotides.
Nanoparticles modify the kinetics, body distribution and drug release. Furthermore, they may have specifi c targeting for cells or tissues, optimize pharmacological activity and reduce unwanted side effects [9] There are a variety of nanoparticles, these can be lipid nature, like solid lipid nanoparticles, where we have solid and liquid lipids that will allow the formation of this system, they can be polymeric, where there is a mandatory polymer coating, there are also magnetic nanoparticles that are mostly intended as disease diagnosis systems [10][11][12], among many others. All of these systems mentioned can improve the characteristics of the active ingredients, whether physicochemical, pharmacodynamics and/or pharmacokinetics. It is necessary to understand the needs for choosing the ideal system.
Perspectives
It is clear that nanotechnology represents a key research area to face the pharmaceutical industry's R&D challenges.
Nanotechnology-based medicines have already found success in the industrial scenario. This worldwide trend, had increased in the coming years. However, it is necessary to evaluate the ethical aspects of the impacts of this new technology in the long term with great responsibility.
Nanotechnology represents a real therapeutic revolution, as it will make it possible to multiply the number of active ingredients that are candidates for composing medicines and, therefore, in the long term, the number of therapeutic solutions available for the treatment of diseases.
==
Domain: Biology Materials Science
|
[{"added": "2020-05-13T10:22:15.617Z", "created": "2020-05-05T00:00:00.000", "id": "218606587", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.peertechz.com/articles/IJPSDR-6-127.pdf", "pdf_hash": "7e75d48c6506975a360852fc1ccc23bc77d74046", "pdf_src": "Anansi", "provenance": "peS2o-0028.json.gz:1033", "s2fieldsofstudy": ["Biology", "Materials Science"], "sha1": "7e75d48c6506975a360852fc1ccc23bc77d74046", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Finite element and deformation analyses predict pattern of bone failure in loaded zebrafish spines
The spine is the central skeletal support structure in vertebrates consisting of repeated units of bone, the vertebrae, separated by intervertebral discs (IVDs) that enable the movement of the spine. Spinal pathologies such as idiopathic back pain, vertebral compression fractures and IVD failure affect millions of people worldwide. Animal models can help us to understand the disease process, and zebrafish are increasingly used as they are highly genetically tractable, their spines are axially loaded like humans, and they show similar pathologies to humans during ageing. However, biomechanical models for the zebrafish are largely lacking. Here, we describe the results of loading intact zebrafish spinal motion segments on a material testing stage within a micro-computed tomography machine. We show that vertebrae and their arches show predictable patterns of deformation prior to their ultimate failure, in a pattern dependent on their position within the segment. We further show using geometric morphometrics which regions of the vertebra deform the most during loading, and that finite-element models of the trunk subjected reflect the real patterns of deformation and strain seen during loading and can therefore be used as a predictive model for biomechanical performance.
Introduction
The spine consists of a repeated pattern of motion segments (MSs) of bony vertebrae separated by intervertebral discs (IVDs) that enable movement. Back pain and IVD degeneration affect millions of people worldwide [1,2], and vertebral compression fractures are a frequent feature of osteoporosis [3]. Biomechanical pathologies of the spine are underpinned by genetic, physiological and environmental pathways that together damage IVD, muscle and the bone, changing the mechanics of the system. Animal models, typically rodents, are frequently used to study mechanisms of spinal pathology [4]. However, quadrupeds are disadvantageous for studying the human spine as gravitational load acts perpendicular to their axial skeleton. Zebrafish are increasingly used as a model for human disease, due to their genetic tractability. Unlike quadrupeds, but similar to humans under gravity (figure 1a), their spine is antero-posteriorly loaded as a result of swimming through viscous water [5]. Zebrafish are well established as models for skeletogenesis, pathology and ageing [6], and develop spinal pathologies in response to altered genetics [7] and ageing [8]. However, the biomechanics of the zebrafish spine are comparatively poorly characterized.
Finite-element analysis (FEA) has proven a pivotal tool in the study of biomechanical subjects [9], and offers a method for biomechanically characterizing the zebrafish spine, including intact MSs. This technique digitally models an object of known material properties using a series of linked nodes of known royalsocietypublishing.org/journal/rsif J. R. Soc. Interface 16: 20190430 number and geometry, that can be subjected to a wide variety of forces outputting the predicted geometry, strain and deformation. Results can be validated by comparison with the results of loading experiments in which a sample is loaded ex vivo [10,11]. FEA has been used in zebrafish to test contributions of shape and material properties in joint morphogenesis [12,13] and to study strain patterns in a single vertebra [14].
Here, we describe a novel integrated experimental platform that brings together imaging, modelling and real-world validation to explore the biomechanics of intact zebrafish spinal MSs. We generated an FEA model of the spine, which we validated with a loading experiment using a high-precision material testing stage (MTS) under set loading regimes using micro-computed tomography (µCT). Three-dimensional geometric morphometrics (3D-GM) was used to explore patterns of deformation seen in each vertebra during loading. Comparison of results demonstrated that our FEA model accurately predicted the relative patterns of deformation and strain experienced by real samples loaded ex vivo.
Zebrafish samples
One-year-old, wild-type (WT) zebrafish were fixed in 4% paraformaldehyde and dehydrated to 70% EtOH. MSs were acquired by making two cuts in the trunk, between the morphologically homogeneous vertebrae 18 and 24 of a total of 33 vertebrae [5] (figure 1a-c).
In vitro vertebral loading experiment
Loading experiments were conducted using a custom-built material testing stage (MTS2) in a Bruker SKYSCAN 1272 µCT system. Radiographic visualization of each MS (n = 3) was performed and if required, vertebrae were trimmed to retain three complete vertebrae and associated IVDs (figure 1b-d). Samples were stabilized (anterior-up) in the MTS2 using cyanoacrylate glue. The MTS2 was programmed to perform a sequential series of seven scans at a series of increasing loads (table 1), using 60 keV X-ray energy, 50 W current, 5 µm isotropic voxel size and a 0.25 mm aluminium filter. A total of 1501 projections were collected during a 180°rotation, with 400 ms exposure time. Reconstructions were performed using NRecon (v. 1.7.1.0). Surfaces of vertebrae, muscle and IVDs in each dataset were generated using Avizo (Avizo v. 8; Vizualisation Sciences Group) (figure 1c-e and table 1) and linear measurements of IVDs and MS lengths made using the '3D Measurement' tool. Vertebrae surfaces were further processed in Meshlab (table 2).
Finite-element analysis
An MS surface mesh was created based on a 1-year-old WT specimen µCT scanned using a Nikon XTH 225ST μCT system as described under two conditions: (a) native state and (b) contrastenhanced following 14 day incubation in 2.5% phosphomolybdemic acid [16]. Scan (a) was used to segment vertebrae (V18-V24), and scan (b) to segment IVDs. The resulting binary labels from scans (a) and (b) were saved as 8-bit tiff stacks, manually registered in 3D space in Avizo ('Trackball' tool) and algorithmically combined ('Algebra' tool), creating a single volume of separate materials representing three vertebrae and four IVDs (figure 1d,e and table 2). A 500 µm thick cylinder was created contacting the anterior-most IVD perpendicular to the model axis, to mimic the stainless-steel compressive plate and distribution of forces applied during loading (figure 1f ).
The complete vertebral surface mesh was imported into Simpleware ScanIP (v. 2018.12, Synopsys Inc.) to create an FE model. The model consisted of 1 054 187 linear tetrahedral elements joined at 257 392 nodes comprising four material types: vertebral bone, annulus-fibrosus, nucleus-pulposus and stainless steel (figure 1d-f, table 2). The model was analysed in Abaqus (2018 version). A custom datum coordinate system was created centred on the antero-posterior axis of the model, and a concentrated force applied to the central node of the anterior face of the compressive plate. This loading case was repeated in each of seven steps of a multi-step analysis, with load values matching the increments applied in the MTS (table 1). The model was constrained in two locations using boundary conditions, at the base of the posterior-most IVD (constrained in three axes) and at the top of the compressive plate (constrained in two axes), allowing movement along the model's antero-posterior axis (figure 1f ). Deformed meshes from each step were exported as surface files and analysed using 3D-GM for quantitative comparison between relative and absolute patterns of deformation predicted by FEA and observed in MTS data.
Three-dimensional geometric morphometrics
Three-dimensional geometric morphometrics analysis of vertebral deformation was performed using the 'Geomorph' package for the R statistics software [17]. For each loading experiment, we used the first scan (1 N load) to create a template of 3D coordinates for 22 fixed three-dimensional landmarks (figure 2a-c) linked by 300 surface sliding semi-landmarks (using the 'buildtemplate' function). By assigning the same landmarks in each scan (using the 'digitsurface' function), we compared the first scan with subsequent scans of the same vertebra using generalized Procrustes analysis (allowing semilandmarks to 'slide' in order to remove arbitrary spacing). Resulting shape variables were subjected to principal component analysis (PCA) to identify the principal patterns of variation between scans of the same vertebra, and isolate trends in deformation with increasing compressive load.
Vertebral motion segments fail under loading of 12-16 N at positions of maximum von Mises strain
To test the range of compressive loads that the MS could resist until failure, we subjected an MS to exponentially increasing compressive forces from 1 to 100 N. This specimen failed at 16 N whereupon the central vertebra fractured midcentrum. A primary loading regime between 1 and 16 N was thus established (table 1) for the three primary specimens; occupying the elastic, plastic and failure regions of the compressive loading profile of a typical MS. Failure was considered when at least one vertebral centrum fractured across the axis (e.g. figure 1j,l ). All samples failed between 12 and 16 N upon shallow angle fracture in the central vertebra, with the smallest specimen (specimen 3) failing at the lowest force (figure 1g,h). This is higher than maximum aquatic forces experienced during swim training by Fiaz et al. [5], which reached approximately 9.5 N. Minor differences in mounting orientation created differences in linear deformation between right and left sides, but specimens follow similar patterns. Prior to failure, linear measurements show an increase in IVD antero-posterior thickness ( We found characteristic patterns of deformation and strain in response to compressive loading of zebrafish vertebrae. Three-dimensional results from MTS data follow distinct trends for each vertebrae between the three specimens (figure 2d,i,n), showing consistent dorsoventral compression, and lateral compression that is reversed at higher loads potentially due to elastic rebound of the IVD and fracturing along the zygopophyses that occurs at these loads ( figure 2). This relative pattern is shared between each specimen, although specimen 3 experiences this at lower loads than specimens 1-2, before failing at 12 N. Fractures are observed where the arches and zygopophyses contact the centrum, at loads that precede the failure of the segment (figure 2f,h,k,m,p,r). Comparison with FEA data (blue points in figure 2d,i,n) suggests that the FE model accurately predicts these patterns (figure 2d,i,n), and that patterns of deformation could explain the first signs of damage prior to failure. In both datasets, the anterior vertebra undergoes most deformation, particularly posterior deformation of the arches (figure 2e-h). The central vertebrae and arches show strong torsion (figure 2j-m), increasing through the loading regime leading to the failure of the segment ( figure 1l,o). The posterior vertebra shows the least deformation and is most isotropic in pattern (figure 2or), potentially due to protection offered by the anterior IVDs. Comparison with ex vivo loading of vertebral MSs validates the accuracy of our FEA model for predicting patterns of deformation and strain across these structures. This offers a step towards a digital 'sandbox' approach to modelling the effects of genetic, physiological and morphological properties on the reaction and resistance of vertebral MSs to loading. Inputting specific properties of vertebral samples into a validated FE model will allow their effects on the biomechanics of the spine to be quantitatively tested in silico, allowing the relative contributions of shape and material properties to be explored and empirically tested. This will aid comparison of mechanical performance between different model systems. As an advantage of the zebrafish system is the wealth of mutants modelling human disease genetics [18], comparisons of mechanical performance between genotype and phenotype will be possible. In the longer term, this approach may give insight into biomechanical aspects of spinal pathology, allowing identification of 'at risk sites' in the spine. This could provide a basis for more specific or earlier interventions than those commonly employed.
==
Domain: Biology Materials Science Medicine
|
[{"added": "2019-09-16T18:26:10.211Z", "created": "2019-07-17T00:00:00.000", "id": "203891318", "metadata": {"extfieldsofstudy": ["Biology", "Materials Science", "Medicine"], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://royalsocietypublishing.org/doi/pdf/10.1098/rsif.2019.0430", "pdf_hash": "ec294d3fe1ee3562fca4d5b32d782ac2d2214ea0", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:5861", "s2fieldsofstudy": ["Biology"], "sha1": "1a64f53eb32ef85aa6ac3e7526f0e795ec33b9d3", "year": 2019}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Design of intrinsically disordered proteins that undergo phase transitions with lower critical solution temperatures
Many naturally occurring elastomers are intrinsically disordered proteins (IDPs) built up of repeating units and they can demonstrate two types of thermoresponsive phase behavior. Systems characterized by lower critical solution temperatures (LCST) undergo phase separation above the LCST whereas systems characterized by upper critical solution temperatures (UCST) undergo phase separation below the UCST. There is congruence between thermoresponsive coil-globule transitions and phase behavior. Specifically, the theta temperatures above or below which the IDPs transition from coils to globules serve as useful proxies for the LCST / UCST values. This implies that one can design sequences with desired values for the theta temperature with either increasing (UCST) or decreasing radii of gyration (LCST) above the theta temperature. Here, we show that the Monte Carlo simulations performed in the so-called intrinsic solvation (IS) limit version of the temperature-dependent ABSINTH implicit solvation model, yields a robust heuristic for discriminating between sequences with known LCST versus UCST phase behavior. Accordingly, we use this heuristic in a supervised approach, integrate it with a genetic algorithm, combine this with IS limit simulations, and show how novel sequences can be designed that have LCST phase behavior. These calculations are aided by direct estimates of temperature dependent free energies of solvation for model compounds that are derived using the polarizable AMOEBA forcefield. To demonstrate the validity of our designs, we calculate coil-globule transition profiles using the full ABSINTH model and combine these with the Gaussian Cluster Theory to show that the designed IDPs do show LCST phase behavior.
Introduction
Intrinsically disordered proteins (IDPs) that undergo thermoresponsive phase transitions are the basis of many naturally occurring elastomeric materials 1 . These naturally occurring scaffold IDPs 2 serve as the basis of ongoing design efforts to design thermoresponsive materials 3 . Well-known examples of disordered regions derived from elastomeric proteins 4 , include the repetitive sequences from proteins such as resilins 5 , elastins 6 , proteins from spider silks 7 , titin 8 , and neurofilament sidearms 9 . Elastin-like polypeptides have served as the benchmark systems for the development of responsive disordered proteins that can be adapted for use in various biotechnology settings 10 . The interplay between sequence-encoded intermolecular and chainsolvent interactions combined with the interplay between chain and solvent entropy gives rise to thermoresponsive phase transitions that lead to the formation of coacervates 1 . Here, we show that one can expand the "materials genome" 11 through de novo design strategies that are based on heuristics anchored in the physics of thermoresponsive transitions and efficient simulation engines that apply the learned heuristics in a supervised approach. We report the development of a genetic algorithm (GA) and show how it can be applied in conjunction with multiscale computations to design thermoresponsive IDPs with LCST phase behavior.
Conformational heterogeneity is a defining hallmark of IDPs 12 . Work over the past decadeand-a-half has shown that naturally occurring IDPs come in distinct sequence flavors 12 . Indeed, IDPs can be distinguished based on their sequence-encoded interplay between intramolecular and chain-solvent interactions that can be altered through changes in solution conditions. Recent studies have shown that IDPs can be drivers or regulators of reversible phase transitions in simple and complex mixtures of protein and nucleic acid molecules 13 . These transitions are driven primarily by the multivalence of interaction motifs that engage in reversible physical crosslinks 14 .
IDPs can serve as scaffolds for interaction motifs (stickers), interspersed by spacers. Alternatively, they can modulate multivalent interactions mediated by stickers that are situated on the surfaces 15 of autonomously foldable protein domains 16 .
Thermoresponsive phase transitions arise either by increasing the solution temperature above a lower critical solution temperature (LCST) or by lowering the temperature below an upper critical solution temperature (UCST) 1 Studies of synthetic polymer systems have helped in elucidating the origins of the driving forces for and the mechanisms of LCST phase behavior 22 . A well-known example is poly-Nisopropylacrylamide (PNIPAM) 23 . Here, the dispersed single phase is stabilized at temperatures below ~32˚C by the favorable hydration of amides in the sidechains. Solvation of amides requires that the solvent be organized around the hydrophobic moieties that include the backbone carbon chain and the isopropyl group in the sidechain. The entropic cost for organizing solvent molecules around individual chains increases with increasing temperature. Accordingly, above the LCST of ~32˚C, and for volume fractions that are greater than a threshold value, the system phase separates to form a polymer-rich coacervate phase that coexists with a polymer-poor dilute phase. The driving forces for phase separation are the gain in solvent entropy through the release of solvent molecules from the polymer and the gain of favorable inter-chain interactions, such as hydrogenbonding interactions between amides in the polymer.
Tanaka and coworkers have developed a cooperative hydration approach, inspired by the Zimm and Bragg theories for helix-coil transitions 24 , to model the physics of phase transitions with LCST 25 . Cooperative hydration refers to the cooperative association (below the LCST) or dissociation (above the LCST) of water molecules that are bound to repeating units along the polymer chain 26 . Cooperativity is captured using the Zimm-Bragg formalism by modeling each repeating unit as being in one of two states viz., solvated or desolvated. In the solvated state, the repeating unit has a defined interaction strength with solvent molecules. In the desolvate state, pairs of such repeating units have defined exchange interactions. In addition, desolvation is associated with a gain in solvent entropy. The three-way interplay of direct solvent-chain interactions, the interactions among desolvated pairs of units, and the gain in solvent entropy above the LCST can be captured in a suitable physical framework that can be parameterized to describe system-specific phase transitions. Accordingly, if one has prior knowledge of the interaction energies associated with each repeat unit, one can use the framework of Tanaka and coworkers to design novel sequences with LCST behavior.
An alternative approach, which we adopt in this work, is to leverage the corollary of LCST behavior at the single chain limit 27 . At temperatures that are proximal to the LCST, the system of chain of interest will undergo a coil-to-globule transition in a dilute solution 28 . This is because the chain collapse is a manifestation of the physics of phase separation at the single chain limit. Here, we leverage this connection between phase separation and chain collapse of isolated polymer chains in ultra-dilute solutions to design novel IDPs that are predicted to undergo phase transitions with LCST phase behavior. To do so by using a multi-pronged approach that starts with improved estimates of the temperature dependencies of free energies of solvation of model compounds that mimic amino acid sidechain and backbone moieties. For this, we use free energy calculations based on the AMOEBA forcefield 29 , which is built on a second-generation polarizable model for water molecules and proteins. We incorporate these temperature dependent free energies of solvation into the ABSINTH implicit solvation model and show that thermoresponsive changes to chain dimensions, calculated in the efficient "intrinsic solvation (IS) limit" 30 yields robust heuristics that discriminates sequences with known LCST phase behavior from those that show UCST behavior. We then describe the development of a GA, an adaptation of the GADIS approach, to design novel sequences that relies on all-atom simulations, performed using the ABSINTH model in the IS limit, and learned heuristics as fitness scores. We show that distinct classes of designed sequences emerge from our approach. These are screened to filter out sequences with low disorder scores as assessed using the IUPRED2 algorithm 31 . The resulting set of sequences are analyzed using simulations based on the full ABSINTH model, which show that the designed sequences do undergo collapse transitions above a threshold temperature. The contraction ratio, defined as the ratio of chain dimensions at temperature T to the dimensions at the theta temperature, is computed as a function of simulation temperature is analyzed to extract temperature dependent two-body interaction parameters and athermal three-body interaction parameters that are used in conjunction with the Gaussian Cluster Theory (GCT) 32 to calculate system-specific phase diagrams 28 . The upshot is multiscale pipeline whereby a GA, aided by supervised learning in the form of a derived heuristic and IS limit simulations, leads to the design of novel sequences with predicted LCST phase behavior. Following a post-processing step that selects for sequences with a high confidence of being intrinsically disordered, we combine all-atom ABSINTH-T based simulations with Gaussian Cluster Theory to obtain sequence-specific phase diagrams. These last two steps allow further pruning of the sequence space derived from the designs and provide further confidence regarding the authenticity of the predicted LCST phase behavior. (1)
Results and Discussion
Here, ∆h is the enthalpy of solvation at a reference temperature T0, which is typically set to be 298K, and ∆cP is the molar heat capacity change associated with the solvation process. For simplicity, this is assumed to be independent of temperature 36 .
We build on the approach of Wuttke et al., which leverages the flexibility of the ABSINTH Gibbs-Helmholtz equation, which we then use in ABSINTH-T based simulations to design novel sequences that are predicted to show LCST behavior. Direct calculations of free energies of solvation obviate the need for making a priori assumptions when using experimentally derived values, which become especially problematic for model compounds with ionizable groups.
Results from AMOEBA-based free energy calculations for model compounds: We performed temperature dependent free energy calculations based on the Bennett Acceptance Ratio (BAR) free energy estimator 37 for direct investigation of how ∆µh varies with temperature. These calculations were performed for nineteen different model compounds that mimic the twenty sidechain moieties and the backbone peptide unit. For the free energy calculations, we used the AMOEBA forcefield, which uses atomic polarizabilities and atom-centered multipoles up to quadrupole moments 29 . The AMOEBA forcefield represents the state-of-the-art in modeling protein and peptide units in aqueous solutions using polarizability and higher-order electrostatics without having to resort to quantum mechanical calculations. Importantly, the AMOEBA water model reproduces the temperature-dependent anomalies of liquid water 38 a feature that is directly relevant for extracting temperature-dependent parameters to describe solvation. Additionally, the AMOEBA model yields accurate free energies of solvation of ions 39 and model compounds in aqueous solvents 29,38,40 . Details of the parameterization for model compounds used in this study, and the design of the free energy calculations are provided in the methods section.
Results from temperature dependent calculations of µh for the nineteen relevant model compounds are shown in Figure 1. These panels are grouped by the chemistries of functional groups within the different model compounds. The enthalpy of hydration (∆h) at T0 = 298K and the temperature independent heat capacities of hydration (∆cP) were extracted for each model compound by fitting the calculated temperature dependent free energies of solvation to the integral of the Gibbs-Helmholtz equation -see equation (1). The results are summarized in Table 1.
As expected for hydrophobic hydration 36 Table 1. In the legends we use the three letter abbreviations for each of the amino acids. Here, BB in panel (d) refers to the backbone moiety, modeled using N-methylacetamide, that mimics the peptide unit. * As with the default ABSINTH model, in ABSINTH-T, the rFoS values we used for ionizable residues are offset from the calculated ∆µh by a fixed constant of -30 kcal/mol. This, as was shown in the original work, is required to avoid the chelation of solution ions around ionizable residues. This "feature" remains a continuing weakness of the ABSINTH paradigm and one that we hope to remedy through suitable generalization of the model used in ABSINTH to interpolate between fully solvated and fully desolvated states.
Intrinsic solvation (IS) model, a non-electrostatics version of ABSINTH-T as an efficient heuristic for discriminating IDPs with LCST versus UCST behavior
In the single chain limit, accessible in dilute solutions, polypeptides that show LCST phase behavior undergo collapse above a system specific theta temperature, whereas polypeptides that show UCST phase behavior expand above the system specific theta temperature 1,28 . A GADISlike strategy 21 for de novo design of polypeptide sequences with LCST phase behavior would involve ABSINTH-T based all-atom simulations to evaluate whether an increase in temperature leads to chain collapse. In effect, the fitness function in a GA comes from evaluation of the simulated ensembles as a function of temperature. This becomes prohibitively expensive computationally. Accordingly, we pursued a pared down version of ABSINTH-T, which is referred to as the intrinsic solvation (IS) limit of the model 30 . The IS limit was introduced to set up sequence and composition specific reference models with respect to which one can use meanfield models to uncover how desolvation impacts IDP ensembles 30,49 . In effect, the IS limit helps us map conformations in the maximally solvated ensemble and assess how this ensemble changes as a function of temperature. In the IS limit, the energy in a specific configuration for the sequence of interest is written as: ; The only difference between the full model, see equation (2), and the IS limit is the omission of the Wel term. This increases the speed of simulations by roughly two orders of magnitude. Next, we asked if ensembles obtained from temperature dependent simulations performed in the IS limit could be used to obtain a suitable heuristic that discriminates sequences with LCST versus UCST behavior. These simulations were performed for a set of thirty sequences (see Table S1) that were previously shown by Garcia Quiroz and Chilkoti to have LCST and UCST phase behavior 17 . The results are summarized in Figure 2. As shown in panel (a) of Figure 2, the radii of gyration (Rg), suitably normalized for comparisons across different sequences of Given the range of sequences covered in the calibration based on the IS limit, we pursued an approach whereby we use slopes of vs. T as a heuristic to guide the design of a genetic algorithm to find new sequences with LCST phase behavior. Table S1 in the supporting information. (b) The slope m of the RgN 0.5 vs. temperature profiles. These slopes fall into two distinct categories, one for those with LCST phase behavior (blue) and another for those with UCST phase behavior (red). The gray region corresponds to the values of m that clearly demarcate the two categories of sequences.
GA for the design of IDPs that are likely to have LCST phase behavior
Motivated by previous successes using the GADIS algorithm 21 for designing sequences with bespoke amounts of intrinsic secondary structure contents, we adapted a GA for exploring sequence space to discover candidate IDPs with predicted LCST phase behavior. To introduce the GA and demonstrate its usage, we set about designing novel sequences that are repeats of pentapeptide motifs. We focused on designing 55-mers, i.e., sequences with 11 pentapeptides. To keep the exercise simple, we focused on designing polymers that are perfect repeats of the pentapeptide in question. The GA used in this work is summarized in Figure 3 and the details are described below.
Here, N is the number of amino acids in each sequence, n is the number of replicas used in the simulation, and Ti is the temperature associated with replica i. The slope m was used to select 100 out of the 200 sequences that were chosen at random initially. The picking probability p was based on the following criterion: ; Here, c = 400 in units that are reciprocal to m, and m0 is set to -6.9 ´ 10 - allowed to evolve for multiple iterations until the convergence criteria were met. These include the generation of at least 250 new sequences, each with a value of m being less than -5.0 ´ 10 -3 åK -1 .
For the results presented here, six iterations were sufficient to meet the prescribed convergence criteria. The picking probability p determines the selection pressure encoded into the GA. There needs to be an optimal balance between the two extremes in selection pressure. High selection pressures can lead to early convergence to a local optimum whereas low selection pressures can drastically slow down convergence 51 . The use of a single evolutionary operator can lead to a single sequence becoming the dominant choice. The number of iterations that pass before the emergence of a single sequence is known as the takeover time 51 . High selection pressures lead to low takeover times and vice versa. The issue of a single dominant individual emerging because less of a concern in sequence design given the high dimensionality of sequence space. We tuned the choice of c and m0 to ensure that candidate sequences with putative UCST phase behavior can be part of the offspring, thus lending diversity to sequence evolution by the GA. categories based on their sidechain chemistries i.e., basic residues in blue bars, acidic residues in red bars (although these are not visible since they are not selected), polar residues in green, Pro and Gly in purple, and aliphatic as well as aromatic residues in cyan. Within each group, the bars are sorted in descending order of the mean numbers of occurrences in the designs. Figure 4 quantifies the progress of the GA through each iteration of the design process. The quantification is performed in terms of cumulative distribution functions, which for each iteration will quantify the probability that the emerging sequences have associated slope values that are less than or equal to a specific value. The rightward shift in each iteration is indicative of the improved fitness vis-à-vis the selection criterion, which is the lowering of m.
Panel (a) in
As a final step in the sequence design, we added a post-processing step to increase the likelihood that the designed sequences are bona fide IDPs. Accordingly, we used the disorder predictor IUPRED2 31 to quantify the disorder scores for each of the designed sequences.
IUPRED2 yields a score between 0 and 1 for each residue, and only sequences where over half of the residues in the repeat are above 0.5 were selected as the final set of designs as sequences predicted to have LCST phase behavior. A particular concern with designing sequences for experimental prototyping is the issue of aggregation / precipitation. To ensure that designs were unlikely to create such problems, we calculated predicted solubility scores using the CamSol program 52 and found that all sequences that were selected after the post-processing step also have high solubility scores. This provides confidence that the designed IDPs are likely to show phase behavior via liquid-liquid phase separation above system-specific LCST values without creating problems of precipitation / aggregation. preference for Arg over Lys, which is concordant with the distinct temperature dependent profiles for ∆µh (Figure 1) and the large positive heat capacity of Arg, which is roughly three times larger than that of Lys (Table 1). Interestingly, if we fix the positions of Pro and Gly and select for residues in XPXXG or other types of motifs that are inspired by previous work on elastin-like polypeptides, the design process often converges on repeats that are known to be generators of polypeptides with bona fide LCST phase behavior (data not shown). This observation, and the statistics summarized in Figure 4b indicate that the design process uncovers sequences that are likely to have LCST phase behavior. The actual sequences of the repeats, color-coded by their Hamming distance-based groupings, are shown in Figure 6. There are two features that stand out. First, sequences deviate from being repeats of VPGVG, which is the elastin-like motif. Second, we find that different sequence permutations on identical or similar composition manifolds emerge as candidates for LCST phase behavior. This observation suggests that at least in the IS limit it is the composition of each motif rather than the precise sequence that underlies adherence to the selection pressure in the GA.
Panel (b) in
Interestingly, our observations are in accord with results from large-scale in vitro characterizations of sequences with LCST phase behavior 53 . These experiments show that composition, rather than the precise sequence, is a defining feature of LCST phase behavior -a feature that is distinct from sequences that show UCST phase behavior 3 . Next, we used the estimates of Tq in conjunction with the Gaussian Cluster Theory of Raos and Allegra. We extracted the two and three-body interaction coefficients by fitting the contraction ratio as calculated from simulations using the formalism of the Gaussian Cluster Theory and this yields sequence-specific estimates of B, the two-body interaction coefficient, and w, the threebody interaction coefficient (see panels (a) -(c) in Figure 8). These parameters were then deployed to compute full phase diagrams using the numerical approach developed by Zeng et al., 28 and adapted by others 54 . The results are shown in panels (d) -(f) of Figure 8. The abscissae in these diagrams denote the bulk polymer volume fractions whereas the ordinates quantify temperature in terms of the thermal interaction parameter . Here, which is positive for T > Tq, B is the temperature-dependent two-body interaction coefficient inferred from analysis of the contraction ratio, and nK is the number of Kuhn segment in the single chain, which we set to be five. Note that B is negative for temperatures above Tq. Accordingly, the thermal interaction parameter is positive above Tq as well as the critical temperature Tc. Therefore, comparative assessments of the driving forces for LCST phase behavior can be gleaned by comparing the sequence-specific values of and the volume fraction at the critical point.
It follows that the sequences can be arranged in descending order of the driving forces as (TPTGM)11, (RTAMG)11, and (PTPLV)11, respectively. Importantly, full characterization of the phase behavior using a combination of all-atom simulations and numerical adaptation of the Gaussian Cluster Theory shows that, in general, sequences designed to have LCST phase behavior, do match the predictions (see Figure 8).
Discussion
In this work, we have adapted a GA to design novel sequences of repetitive IDPs that we predict to have LCST phase behavior. Our method is aided by a learned heuristic that was shown to provide clear segregation between sequences with known LCST vs. UCST phase behavior. This heuristic is the slope m of the change in RgN 0.5 vs. T from simulations of sequences performed in the IS limit of the ABSINTH-T model. We use the heuristic in conjunction with IS limit simulations to incorporate a selection pressure into the GA, thereby allowing the selection of sequences that are "fit" as assessed by the heuristic to be predictive of LCST phase behavior.
Here, we presented one instantiation of the GA and used it to uncover 64 novel sequences that can be grouped into four major classes and several minor classes (Figure 6). We then focused on four sequences, one each from each of the four major classes and characterized temperature dependent coil-globule transitions. These profiles, analyzed in conjunction with recent adaptations of the Gaussian Cluster Theory 32 , allowed us to extract sequence-specific values for theta temperatures, temperature dependent values of the two body interaction coefficients, and threebody interaction coefficients. We incorporated these parameters into our numerical implementation 28 of the Gaussian Cluster Theory to calculate full phase diagrams for three sequences. These affirm the predictions of LCST phase behavior and the sequence-specificity in control over the driving forces for thermoresponsive phase behavior.
Our overall approach is aided by the following advances: We used the AMOEBA forcefield 29 to obtain direct estimates of temperature dependent free energies of solvation for model compounds used to mimic sidechain and backbone moieties. These temperature dependent free energies of solvation were used in conjunction with the integral of the Gibbs-Helmholtz equation to obtain model compound specific values for the enthalpy and heat capacity of hydration. The AMOEBA calculations represent the first direct estimates of temperature dependent hydration free energies using a polarizable forcefield, and they help circumvent many of the assumptions that have to be made in dissecting thermodynamic data for whole salts or protonated / deprotonated versions of acidic / basic groups in order to obtain experimentally derived estimates for temperature independent 20 as well as temperature dependent µh values 19 . The only simplifications we make here are (a) the usage of the integral of Gibbs-Helmholtz equation to fit the temperature dependent µh values and (b) assuming that ∆cP is independent of temperature. To test the validity of these assumptions we will need to deploy novel BAR based estimators that allow us to quantify the temperature dependencies of enthalpies and entropies 55 . However, obtaining accurate estimates of the decompositions requires orders-of-magnitude more sampling as shown by Wyczalkowski et al., 55 and we will need to extract second moments of the enthalpies and entropies in order to obtain independent estimates of the temperature dependence of ∆cP.
The methods we present here are a start toward the integration of supervised learning to leverage information gleaned from systematic characterizations of IDP phase behavior and physical chemistry based computations that combined all-atom simulations with improvements such as ABSINTH-T, and theoretical calculations that allow us to connect single chain coil-globule transitions to full phase diagrams 28 . The heuristic we have extracted from IS limit simulations helps with discriminating sequences with LCST vs. UCST phase behavior. These simulations are sufficient for IS limit driven and GA aided designs of sequences that are expected to have LCST phase behavior. This is because composition as opposed to the syntactic details of sequences play a determining role of LCST phase behavior 3 . Recent studies have shown that even the simplest changes to sequence syntax can have profound impacts on UCST phase behavior 56 . This makes it challenging to guide the design of sequences with predicted UCST phase behavior that relies exclusively on IS limit simulations. One will need to incorporate simulations based on either transferrable 5757 or learned coarse-grained models 58 as a substitution for the IS limit simulations.
This approach, although easy to articulate, comes with challenges because one has to be sure that the coarse-grained models afford the requisite sequence specificity without compromising 65 . Of course, the proof of the validity / accuracy of designs and predictions will have to come from experimental work geared toward testing the predictions / designs. These efforts -that leverage high-throughput expression of these de novo sequences in E. coli and in situ characterization of their phase behavior -are underway 66 . Initial experimental investigations suggest that the designs reported here and those that will emerge from application of the methods deployed in this work do indeed show LCST phase behavior. Detailed reports of these experimental characterizations will follow in separate work.
AMOEBA forcefield parameterization for the model compounds of interest
To obtain values of free energies of solvation from AMOEBA simulations, we first derived the AMOEBA force field parameters for the model compounds listed in Table 1 of the main text. The parameters for methane, methanol, ethanol, toluene and p-Cresol are taken from previous work 38 , which is part of the amoeba09.prm parameter file in the released TINKER package 67 . The parameters for other model compounds are derived as follows. The AMOEBA potential energy function is composed of bonded and non-boned terms 29 . Bonded terms include bond stretching, angle bending, bond-angle stretch-bending coupling, out-of-plane bending, and torsional rotation.
Non-bonded terms include van der Waals, permanent electrostatics and induced dipole polarization. The molecular structures were first fully optimized at MP2/6-31G* level of theory 68 .
The molecular structures were first fully optimized at MP2/6-31G* level of theory. Force constant parameters for bonded interactions, van der Waals interactions, and polarizabilities were assigned from the existing Poltype library, and the equilibrium bonded values were from optimized geometry. Based on the optimized geometry for each model compound, we performed single point energy calculations at the MP2/cc-pvtz level of theory to obtain compound specific electron densities. The initial multipole (charge, dipole and quadrupole) parameters were obtained from those electron densities by performing distributed multipole analysis calculation using GDMA program 69 . The atomic dipole and quadrupole moment parameters were further refined to reproduce the MP2/aug-cc-pvtz electrostatic potential (ESP). It is worth mentioning that harmonic restraint was applied to dipole and quadrupole in the ESP fitting, which is the new feature of the potential program in the TINKER software 67 . All the derived parameters involving bonded and non-bonded terms were then collected together. Finally, torsional parameters were obtained by comparing the conformational energy profile of quantum mechanical and AMOEBA based calculations. Along each torsional angle of the molecule, 12 conformers were generated. Torsionrestrained optimization was performed by employing HF method combining with 6-31G* basis set, followed by single point energy calculation using ωB97XD functional 70 with a larger basis function 6-311++G(d,p). All the quantum mechanics calculations were performed using the Gaussian 09 software package 71 . The parameterization procedure has been automated in the Poltype tool ( [URL] up of molecular dynamics simulations using AMOEBA
All the AMOEBA simulations were performed using the TINKER-OpenMM package 72 .
Each model compound was solvated in a cubic water box with periodic boundary conditions. The initial dimensions of the central cell were set to be 30×30×30 Å 3 . Following energy minimization, molecular dynamics simulations were performed using integrators designed for the isothermalisobaric ensemble (NPT) with the target temperature being between 273 and 400 K depending on the temperature of interest and the target pressure being 1 bar. Integrations of the equations of motion were performed using the multiple time step RESPA integrator 73 . The temperature and pressure were controlled using a stochastic velocity rescaling thermostat 74 and a Monte Carlo constant pressure algorithm 75 , respectively. The particle mesh Ewald method 76 , with B-spline interpolation 77 , with a real space cutoff of 7 Å was used to compute long-range corrections to electrostatic interactions. The cutoff for van der Waals interactions was set to be 12 Å. The integrating timestep is set to be 2.0 fs and coordinates are saved every 1.0 ps.
Free energy calculations
We used the BAR 37 method to quantify the free energies of solvation for the model
Supporting Information
Please see supporting information for the sequences shown by Garcia Quiroz and Chilkoti to have UCST and LCST phase behavior that were used for IS limit simulations in this study.
Data Availability Statement
The data that support the findings in this study are available within the article and in the Supporting Information.
==
Domain: Biology Materials Science
|
[{"added": "2020-11-19T09:16:02.703Z", "created": "2020-11-15T00:00:00.000", "id": "227062650", "metadata": {"extfieldsofstudy": ["Biology", "Materials Science"], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "https://aip.scitation.org/doi/pdf/10.1063/5.0037438", "pdf_hash": "8e88deb8a00e5aa3bda8633b60cd48bd962caa87", "pdf_src": "BioRxiv", "provenance": "peS2o-0028.json.gz:2890", "s2fieldsofstudy": ["Materials Science"], "sha1": "5addeae24c41e391e507a158cc29ec3bc016ed38", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
GPCR and IR genes in Schistosoma mansoni miracidia
Schistosoma species are responsible for the disease schistosomiasis, a highly prevalent helminthic disease that requires a freshwater snail as intermediate host. The S. mansoni free-living miracidium must utilize olfaction to find a suitable snail host, and certain types of rhodopsin G protein-coupled receptors (GPCRs) and ionotropic receptors (IRs) have been identified as olfactory receptors in other animal phyla. The Schistosoma genome project, together with the recent availability of proteomic databases, allowed for studies to explore receptors within S. mansoni, some of which may contribute to host finding. We have identified 17 rhodopsin-type GPCR sequences in S. mansoni belonging to four subclasses, including ligand-specific GPCRs (i.e. neuropeptide and opsin). RT-PCR demonstrated the expression of nine out of the 17 GPCRs in the free-living miracidia, each of which have been characterized for homology to S. haematobium. Among the nine GPCRs, two are predicted as Gq-opsins. We also describe the characterization of a Schistosoma-encoded IR based on similarity with other species IR and conservation of IR-like domains. Schistosoma mansoni IR is expressed in miracidia at 3 and 6 h post-hatch. The identification of receptors in S. mansoni miracidia, presented here, contributes not only to further understanding of Schistosoma biology and signal transduction but also provides a basis for approaches that may modify parasite behaviour.
Background
The phylum Platyhelminthes contains prominent endoparasites, including the tapeworms and flukes such as Schistosoma spp. [1]. Schistosomes are responsible for the disease schistosomiasis, the most prevalent and important of the parasitic platyhelminthic diseases of humans. Schistosomiasis occurs in 76 countries, affecting approximately 207 million individuals [2], and causing 280,000 deaths per year in sub-Saharan Africa [3]. This represents a serious disease burden to socioeconomic development.
During the course of its life-cycle, Schistosoma mansoni undergoes distinct stages of differentiation, whilst inhabiting three separate environments -freshwater, a molluscan intermediate host, and a vertebrate definitive host [4]. There are few reports that exist regarding how schistosome species interact with these environments, especially concerning the free-living miracidium stage and its infection of the intermediate snail host. Following hatching from the schistosome egg, a motile miracidium actively seeks its Biomphalaria host, within which it undergoes a series of developmental stages and asexual reproduction [5].
Parasites in general have evolved without many of the mechanisms needed to sustain energy for growth or reproduction without a host, for example, miracidia do not have a gut and rely solely on glycogen stored in their epidermal plate for respiration (aerobic in S. mansoni), and lose their infectivity once stores are depleted [6]. Therefore finding a host within a short amount of time in a potentially large body of water requires the miracidia to have highly adapted sensory mechanisms. Miracidia demonstrate host-seeking behaviors in response to chemosensory cues [1]. However, at 1-3 h post-hatch, miracidia use phototactic and geotactic cues to migrate to snail habitats, and do not respond to host chemosensory cues [7,8]. After 1-3 h post-hatch, host attractant biomolecule(s) can be detected, reported to be non-specific small molecular weight biomolecules, and this was supported by experimental assays showing an increase in miracidia turn-back responses [9]. Further, macromolecular glycoconjugates referred to as miracidia-attracting glycoconjugates, have been implicated following an observed induction of changes to miracidial turn-back responses [10,11]. Overall, these studies indicate that these blood flukes possess the molecular components capable of capturing, and more speculatively, processing environmental signals.
It is believed that elucidation of those molecular components that are critical for miracidial function might eventually lead to novel intervention strategies for schistosomiasis control and elimination [12]. Towards this end, recent studies have slowly unraveled insights into schistosome receptor biology and a broad range of cellular processes, such as interaction, mating and reproduction as well as the host-parasite interplay [13,14]. G-protein coupled receptors (GPCRs) are the largest family of receptors found in eukaryotes, with more than 40 % of all pharmaceuticals targeting their various subfamilies [15]. Due to the large diversity and expansion of GPCRs between species and their ability to respond to a large selection of ligands, selectivity for GPCR-targeted anthelmintic drugs is very promising [16,17]. GPCRs are integral membrane receptors, and respond to a multitude of extracellular ligands to transduce and amplify (or inhibit) intracellular responses involved in metabolism, neuromuscular regulation, endocrine function, vision and olfaction [15]. Key characteristics of GPCRs are 7-transmembrane spanning helical α-chains (which constitute a hydrophobic core domain), an external N-terminus and an intracellular C-terminus [18]. Among the known classes of GPCRs, the rhodopsin-type superfamily accounts for approximately 85 % of all GPCRs within many species [19] and have constituted a target of research for pharmaceuticals with many known antagonists [20][21][22]. Rhodopsin-type receptors are activated by a wide range of stimulants, including light, odorant molecules and neurotransmitters, and play physiological roles in vision and smell.
The availability of whole genome sequencing data has provided a basis for the in silico accumulation and analysis of undiscovered and potentially novel receptors in S. mansoni [23]. This led to the description of 117 GPCR genes belonging to five major families (105 Rhodopsin, 2 Glutamate, 3 Adhesion, 2 Secretin and 5 Frizzled) within the draft S. mansoni genome [17]. In 2011, the S. mansoni draft genome set was systematically upgraded with more than 45 % of predicted genes extensively modified and the total number reduced from 11,807 to 10,852 [24]. Employing comparative genomics, platyhelminth GPCRs have been identified and characterized for S. mansoni and S. haematobium [25]. Of those, the opsins are rhodopsin-type GPCRs that were inferred to be involved in photoreception, typically thought of as lightabsorbing proteins that act as light sensors in animals [26][27][28]. Similar to other GPCRs, opsins have a 7-TM structure but are distinguishable from other GPCRs by a lysine residue in the seventh TM domain that binds to retinal, important for light absorption [29]. Upon light absorption, they can transform photons of light into electrochemical signals via G protein activation [30].
The first stage of eumetazoan animal chemoreception is controlled by chemosensory neurons present within the sensory epithelium, where they express olfactory receptors devoted to binding environmental odorants and transfer this information intracellularly. The accuracy of odor discrimination depends on the specificity with which odorants interact with appropriate olfactory neuronal receptors, which are often rhodopsin GPCRs [31]. The identification of rhodopsin-type GPCRs has been well studied for their role in odor detection in different animals (including humans, mouse, fruit fly, nematode and sea slug), greatly improving our understanding of the molecular mechanism of olfaction in these species [32][33][34][35][36][37]. In contrast, there is limited information on how olfaction works at the molecular level in the platyhelminths, although GPCRs have been identified as important parasite receptors with potential functions in the tegumental matrix of S. mansoni [38,39].
In the animal kingdom, besides members of the superfamily of GPCRs, it has been found that ionotropic receptors (IRs) can be expressed on the olfactory sensory neurons to help confer olfactory specificity through responses to chemosensory cues [40,41]. Unlike GPCRs, the characteristic hallmarks of IRs are their three membrane-spanning segments, a pore-forming domain and a ligand-binding Venus flytrap domain, which seems to interact with olfactory stimuli [42]. In insects, IR92a and IR76b are known to detect small amines and polyamines, respectively [43,44].
In this study, we have used a combination of bioinformatics tools on the improved genomic database to identify S. mansoni GPCR genes, including opsins and putative neuropeptide GPCRs. Importantly, some are expressed in the free-living miracidium, and are possibly involved in host recognition. We also report the characterization of a schistosome-encoded IR. The identification of these receptors not only provides molecular evidence for a potential host recognition strategy in S. mansoni, but also contributes to the understanding of schistosome receptor biology.
Methods
Identification of putative GPCRs within the S. mansoni genome The S. mansoni protein dataset used in this study was based on the improved genome assembly [24], along with expression data provided by the GeneDB (www.genedb.org) and SchistoDB (www.schistodb.net) databases. To these databases, we applied Pfam-based profile searches and identification of TM domains with the goal of identifying receptors belonging to the rhodopsin GPCR family. Specifically, this included two bioinformatic tools to predict TM domains for all proteins, including TMHMM ( [URL]-2.0/) and Phobius ( [URL]/). As TM domains are convenient markers for GPCRs, we only focused on those sequences with 7-TM domains. Next, we applied a Pfam-based profile search using HMMerSearch ( [URL]:// hmmer.org/). Proteins containing putative rhodopsintype GPCR domains were systematically identified by profile hidden Markov model searches using the HMMer package ( [URL]/) and the PFAM model PF00001 (7tm_1). Gene and protein nomenclature was based upon the Schistosoma gene models created from the GeneDB reference (www.genedb.org).
Isolation of S. mansoni miracidia
Livers were obtained from ARC Swiss mice infected with S. mansoni (Puerto Rican strain), under conditions specified by the Australian Department of Agriculture, Fisheries and Forestry (DAFF). A 2-day protocol was used to obtain relatively clean schistosome eggs and miracidia [45]. In brief, the mixture of eggs and mouse liver tissue were incubated with collagenase B, penicillin and streptomycin at 37°C overnight, followed by fractionation using Percoll columns (8 ml Percoll + 32 ml of 0.25 M sucrose in 50 ml tubes). The egg pellets were washed using PBS containing EDTA and EGTA twice on a second Percoll column (2.5 ml Percoll + 7.5 ml 0.25 M sucrose in a 15 ml tube). Purified eggs were transferred into a 200 ml hatching measuring cylinder wrapped completely in light-blocking black tape with the exclusion of the top 4 cm from the lip, thereby producing a light-gradient. The hatching cylinder was topped with pH neutral spring water until above the tape-covered area~1.5 cm and exposed to bright light at 27°C. Eggs were incubated for 3 h post-hatch, and the top 10 ml of miracidia-containing water (MCW) was collected for miracidia isolation; in addition, another collection was performed at 6 h post-hatch. Hatched miracidia were isolated by centrifugation at 8000× g for 1 min at 4°C, and were then washed twice with water. For light microscope examination, 6 h post-hatch miracidia were fixed in 4 % paraformaldehyde on a slide, dried and washed in PBS before photographs were taken using an Olympus BX60 with Nomarski optics and a Nikon Digital Sight DS-U1 camera. For RNA isolation, miracidia were collected at 3 and 6 h posthatch and stored separately in RNAlater.
Reverse-transcription PCR for S. mansoni GPCRs
Total RNA was isolated from S. mansoni miracidia (3 and 6 h post-hatch) using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and RNA quantity and quality were assessed using UV spectrophotometry (NanoDrop ND-1000). First-strand cDNA was generated using random hexamer primers and the Superscript Preamplification System for First-strand Synthesis (Invitrogen). PCR was performed using primers designed (Table 1) on the CLC Genomics Workbench (v6.0; Finlandsgade, Denmark). Amplification of α-tubulin served as an internal control for the amount of RNA from each sample. Samples were heated at 94°C for 5 min and amplified for 30 cycles (95°C for 30 s, 45°C for 50 s, and 72°C for 1 min), followed by a 10 min extension at 72°C. Reverse transcriptase negative controls were included to detect contaminating genomic DNA. The amplified DNA fragments were analyzed by 2.0 % (w/v) agarose gel electrophoresis.
Comparative analysis of S. mansoni and S. haematobium GPCRs
Multiple sequence alignments for non-opsin GPCRs were generated with Molecular Evolutionary Genetics Analysis (MEGA) software version 6.0 [46] with the MUSCLE algorithm [47]. Phylogenetic trees were constructed using the neighbor-joining method with 1000 bootstrap replicates for node support. For opsin GPCRs, a phylogenetic tree was constructed on the MEGA 6.0 platform. ClustalW [48] was used to align the sequences of the predicted proteins and the tree was constructed using the neighbor-joining and maximum-likelihood method, with 1000 bootstrap replicates for node support. Neighbor-joining and maximum-likelihood analysis was performed using no. of differences and Jones-Taylor-Thornton (JTT) method, respectively. Receptor schematic diagrams were prepared using the HMMTOP server version 2.0 ( [URL]/ document.html) [49] and LaTEX TEXtopo package [50].
Identification and reverse-transcription PCR (RT-PCR) of S. mansoni IR
The Drosophila melanogaster IR25a [42] was used for sequence similarity searches using the NCBI tBLASTx search tool, limited to bilateria and the nucleotide dataset, resulting in identification of an EST encoding a potential IR protein within S. mansoni. This protein was loaded into the Pfam database ( [URL]. ac.uk/Tools/hmmer/search/phmmer and [URL]. xfam.org/search), which revealed conserved ligandgated ion channel structure. The presence of recurrent TM domain motifs was searched by TMHMM Server v2.0 ( [URL]/). Multiple sequence alignments were performed using the MUSCLE algorithm [47] with the MEGA software (version 5.1) [46], and the LaTEX TEXtopo package [50] was used to generate schematics showing amino acid conservation.
Isolation of total RNA from S. mansoni miracidia followed by the two-step RT-PCR was performed in a similar manner as describe above for the GPCRs. Primers were designed for S. mansoni IR (sense, 5′-AGT AGA ATG CGT GAA TGG-3′ and antisense, 5′-GTT GCG GTG GTA GTC TTG-3′). Samples were heated at 94°C for 5 min and amplified for 30 cycles (95°C for 50 s, 46°C for 90 s, and 72°C for 60 s), followed by a 10 min extension at 72°C. PCR products were visualized by 2.0 % agarose gel electrophoresis to confirm transcript expression.
Molecular dynamics simulation for S. mansoni IR
The initial conformations of the receptors were built using SWISS-MODEL by sequence alignment of proteins with known 3D structures (template proteins) [51]. The structure with the highest quality estimation, based onQMEAN score, was chosen and subjected to molecular dynamics simulation (MDS) using AMBER version 14. The structure was imported using the LEAP module of AMBER; the sequence segment(s) that was missrepresented (normally at the N-or C-terminus), due to different sequence lengths of the template proteins, was built as a linear structure using LEAP and linked back to the corresponding positions. MDS was fully unrestrained and carried out in the canonical ensemble using the SANDER module. The ff14SB force field [52] was employed. Energy minimisation with 2500 steps was first performed to remove unfavourable contacts. The AMBER structure was then heated to 325 K over 50 ps to avoid being kinetically trapped in local minima [53], then subjected to unrestrained MD simulations at 325 K for the purpose of peptide equilibration. The structural information was sampled every 1 ps (i.e. 10,000 structures were calculated for 10 ns MD simulation). This MD simulation was continued until the root mean square deviation of structures within a reasonable long time range was stable at/less than 3~4 Å. Then a lowest energy structure was determined and considered as the representative of the conformations simulated over this period. Visualization of the systems was effected using VMD software [54].
Results
Putative GPCRs within the S. mansoni genome Using the methodology outlined in Fig. 1a, 98 proteins with 7-TM domains were extracted from the S. mansoni genome-derived protein models, based on TMHMM prediction. By comparison, Phobius prediction led to the identification of 62 proteins with 7-TM domains. Pfam profiling did classify 87 proteins (E-value < 0.0004) as rhodopsintype receptors. All TMHMM, Phobius and HMMer search results can be found in the Additional file 1: Table S1.
In total, 17 genes encoding class A GPCR-like proteins (326 to 585 amino acids) were identified (Fig. 1b and Table 2) belonging to four subclasses (amine, peptide, opsin and orphan). All encode proteins considered as full-length, as determined by the presence of 7-TM domains, putative rhodopsin-type GPCR domains, as well as a methionine start and a stop codon. Table 2 also shows the amino acid identity with the identifiable homologs in the closely related S. haematobium [25]. Of the three new orphan GPCRs described in this paper, (i) Smp_203500 shares significant homology (95 % identity) with an allatostatin-A receptor (GenBank XP_012796783.1), (ii) Smp_204230 shares significant homology (87 % identity) with a S. haematobium hypothetical protein (GenBank XP_012798047.1), and (iii) Smp_178420 also shares significant homology (85 % identity) with a S. haematobium hypothetical protein (Genbank XP_012791982.1). RT-PCR of S. mansoni miracidia, using pooled samples obtained at 3 and 6 h post-hatch, revealed expression of 9 out of the 17 GPCRs (Fig. 1c). Schistosoma mansoni α-tubulin was used as a positive control for the cDNA templates.
Comparative analysis of GPCRs present in S. mansoni miracidia with S. haematobium According to their corresponding sub-classification, phylogenetic trees were constructed for each subclass using the final set of predicted non-opsin GPCRs grouped with the S. haematobium homologs (Fig. 2), confirming the high phylogenetic similarity. Schematic GPCR representations show specific regions of conservation and divergence between species homologs. Most sequence divergence was noted between the orphan receptors Smp_173010 and Sha_107429, specifically within the region containing the TM6 domain through the C-terminus. The amine-type GPCRs showed most variability within the intracellular loop 3 region. Overall, there was very high conservation observed within the peptide-type GPCRs, although some variability is present within the N-termini region. Amongst the S. mansoni rhodopsin-type GPCR genes, two sequences (Smp_104210 and Smp_180030) were predicted as opsin GPCRs. These encode opsin-like receptors that share the greatest degree of conservation to two S. haematobium sequences (Sha_101185 and Sha_101097) ( Table 2). We confirmed the clustering of four distinct ancient bilaterian opsin subfamilies (Fig. 3a), namely the Gq-opsins, ciliary opsins, Go-opsins as well as members of the retinal-photoisomerase subfamily, which includes retinal GPCR (RGR) and retinochrome. As might be expected, Smp_104210 and Smp_180030 grouped with Sha_101185 and Sha_101097 in a pairwise, orthologous manner within the Gq-opsin group. Partial sequence alignment of members of the Gq-opsin subfamilies, specifically within the cytosolic region of the TM 7 domain and C-terminal tail, demonstrates two highly conserved peptide amino acids [Histidine, Proline (H, P)] in the carboxy terminal intra-cellular loop domain that are highly indicative of Gq-opsin families (Fig. 3b, c). This distinctive characteristic is conserved in both S. mansoni Smp_104210 and S. haematobium Sha_101185. No such motif was detected for the other opsin-like GPCR (Smp_180030 and Sha_101097).
Identification of IRs within S. mansoni
A single S. mansoni IR was identified from the genome that encodes a conventional ligand-gated ion channel domain protein (513 aa; 58.7 kDa). This receptor, S. mansoni IR, displays remnants of classical IR motifs at corresponding positions and predicted domains that are critical structural regions responsible for detecting odor ligands and contributing to ligand specificity, including an extracellular two-lobed ligand-binding domain and four features common to all conventional IRs, namely: (i) IR-related motifs with TM stretches, (ii) possession of Pfam domains PF10613 and PF00060, which are specific for the ligand-gated ion channel receptors, (iii) highly-conserved structural features specifically shared amongst the IR family, and (iv) a region surrounding the ligand-binding domain. All are present within the S. mansoni IR receptor showing considerable conservation with the IRs of other species, including the well-studied D. melanogaster.
A representation of the S. mansoni IR compared to six Protostomia species, including Panulirus argus, Helicoverpa assulta, Microplitis mediator, Dendroctonus ponderosae, Schistocerca gregaria and Drosophila melanogaster was used to unify protein structure predictions across species (Fig. 4a). Sequences used for this analysis are provided in Additional file 2: Table S2. All IRs display classical IR motifs at corresponding positions and critical structural regions responsible for binding ligands and contributing to ligand specificity. Figure 4b, c demonstrates the proposed structure model of the Venus flytrap domain of the schistosome IR, and with the putative ligand binding sites. RT-PCR results demonstrate S. mansoni IR expression within the free-living miracidia at both 3 and 6 h post-hatch (Fig. 4d).
Discussion
Schistosome miracidia must find an appropriate host within a very limited time-span, thus it would seem advantageous for them to have evolved finely-tuned molecular strategies allowing for host detection, thereby increasing the likelihood of successful snail infection. GPCRs and IRs are fundamental to chemoreception in many animal species [55][56][57], and we speculate this may also be similar for the Schistosoma miracidia. In this study, we have analyzed the S. mansoni genome to identify putative receptors, and specifically those present within the miracidium, some of which may be important for snail host-finding.
We identified 17 rhodopsin-type GPCRs that belong to amine, peptide, opsin and orphan groups. Among these, the amine group consisted of biogenic amine receptors such as serotonin, dopamine and histamine, that have a prominent role in the flatworm nervous system [58,59]. Of these, experimental validation has been established for the histamine receptor (Smp_043260) [60]. Regarding the other GPCRs identified in this study, three (Smp_203500,Smp_204230 and Smp_178420) have not been described previously, while Smp_173010 was reported by Campos et al. [25] as novel platyhelminthspecific rhodopsin-like orphan family (PROF). This variation can be explained by the different workflow for curation of the final GPCR list, whereby receptors were only taken if they satisfied requirements within all of TMHMM, HMM and Phobius tools. Smp_203500 and Fig. 2 Comparative analysis of S. mansoni and S. haematobium orphan, amine and peptide GPCRs. According to their corresponding GPCR subclasses, phylogenetic trees were constructed using candidate orphan (a-c), amine (e-d) and peptide (f-g) GPCRs identified from the S. mansoni genome compared with the S. haematobium homologs. Schematic representation of GPCR sequences identified in S. mansoni miracidia along with the associated the S. haematobium homologs, showing conserved amino acids. Smp and Sha represent the GPCR sequences from S. mansoni and S. haematobium, respectively PROF receptors Smp_173010 do have some similarity to characterized receptor types, allatostatin and myosuppressin, respectively; however this has not been experimentally validated. Platyhelminthes appear to lack a conventional endocrine system [38], and are therefore heavily reliant on neural signalling via neuropeptides that control vital dynamic physiological and neural functions, such as growth, reproduction, host-seeking olfaction, locomotion, immune evasion and sexual dimorphism [61]. The importance of their peptidergic neural and associated receptor systems holds a promising area of research for new anthelmintic drug targets [16,62,63].
RT-PCR demonstrated the expression of nine out of the 17 GPCRs in the free-living miracidia, suggesting that these receptors are possibly either involved in miracidia host-finding and recognition or required for miracidia metabolism, including the histamine receptor. Representative comparative schematic models demonstrated the divergence in amino acid sequence of these GPCRs and homologs in S. haematobium, suggesting the potential biological differences between these two schistosomes. Notably, PROF receptor Smp_173010 shows relatively large variation in protein sequence within the C-terminal region, thus we speculate that there may be binding of species-specific ligands through this receptor.
The opsin-like GPCR Smp_104210 has been reported as being differentially regulated during the parasite's lifecycle and, in the cercaria, it localizes to organelles found directly below the parasite's epidermis, associated with Yellow shows the predicted ligand binding S1 region, green shows the predicted ligand binding S2 region, and blue demarcates the predicted TM regions. The protein domain structure of conventional IRs in cartoon form is shown below [40,41]. c Schematic representation of all IRs shown in a, demonstrating conserved and invariable amino acids, as well as predicted S1 and S2 regions. d RT-PCR showing expression of IR in S. mansoni miracidia at 3 h and 6 h post-hatch (hph) organelles within the vicinity of the most anterior osmoregulatory flame cells [64]. Based on this morphological description, it likely acts as a photoreceptor responsible for the direct photokinetic behavior of cercaria in response to light [64]. Further, annotation of opsin-like GPCR Smp_180030 has been determined as a result of the analysis of RNA-seq expression profiles [24], yet no role in schistosome photoreception processes has been reported.
As indicated by our phylogenetic analysis, Smp_104210 and Smp_180030 have well-supported clustering with other Gq-opsin genes involved in photosensation that strongly implies that these receptors are Gq-opsins. The detection of both receptors (Smp_104210 and Smp_180030) in light-responsive miracidia, as well as their clustering with other Gq opsins, suggests that they may serve an integral role in host-finding by participating in Schistosoma photokinesis [8,[64][65][66].
We identified a single S. mansoni IR. The IR repertoire throughout protostomes shows substantial variation in size, ranging from three in C. elegans to 85 in the crustacean Daphnia pulex [67,68]. The S. mansoni IR exhibits the typical venus flytrap structure (see Fig. 4b) and shares sequence similarity to characterised IRs of other species. Another type of venus flytrap receptor has been studied in S. mansomi, known as the venus kinase receptor [69]. There are two isoforms of this receptor, one that binds L-arginine (SmVKR1) and another that binds calcium ions (SmVKR2), which are thought to be important for development and reproduction. Like the IR identified in our study, the ligand for the S. mansoni venus kinase receptor is unknown. IRs in the insects are known to bind polyamines [43], yet may also act as thermosensors [70]. We speculate that the S. mansoni IR could play an important role as a chemosensory and/or thermosensory receptor in different life-cycle stages, supported by its observed expression profile in cercariae, schistosomula and adults (GeneDB, version 4.0).
Conclusions
The characterization of GPCRs and IRs in S. mansoni is likely to inform us about their pharmacological profiles and features towards manipulating chemosensorydriven behaviors. Given that S. mansoni IR and at least some of the GPCRs are expressed in the miracidium, we hypothesize they may be dedicated to detect specific odor cues, including responses to odors emitted from Biomphalaria. As these odors are likely instrumental for parasite transmission, GPCRs and IRs may represent proteins against which novel prophylactic therapies can be developed.
Acknowledgments
We acknowledge the help of Ms Mary Duke of QIMR Berghofer with the egg hatching and the miracidia processing and harvesting. This research was undertaken with the assistance of resources from the National Computational Infrastructure (NCI), which is supported by the Australian Government.
Funding
This work was supported by the Australian Research Council (Future Fellowship, FT110100990 to SFC). We thank the University of the Sunshine Coast who provided an internal grant to help support this work (TW). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Availability of data and material
The original S. mansoni protein dataset is available from the GeneDB (www.genedb.org) and SchistoDB (www.schistodb.net) databases.
==
Domain: Biology Medicine
|
[{"added": "2023-01-19T22:11:17.585Z", "created": "2016-10-26T00:00:00.000", "id": "255986432", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://parasitesandvectors.biomedcentral.com/track/pdf/10.1186/s13071-016-1837-2", "pdf_hash": "ae0c1f2124ed412e9c40f793d5fe52634c777b1c", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:5690", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "ae0c1f2124ed412e9c40f793d5fe52634c777b1c", "year": 2016}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Current Situation of Bacterial Infections and Antimicrobial Resistance Profiles in Pet Rabbits in Spain
Simple Summary Rabbits are the second most common specialty pet among households in Europe and the USA. However, research on antimicrobial resistance (AMR) in pet rabbits is very scarce. Therefore, scientific data on AMR in pet rabbits is urgently needed as a guide for veterinarian clinicians to optimize antibiotic use in rabbits for reducing the selection of antibiotic resistance. In addition, antimicrobial stewardship programs should be conducted to educate rabbit owners not to misuse antibiotics on their pets as it may put their own health at risk. This paper aims to provide an overview of the current state of AMR in rabbits attended to in veterinary clinics distributed in Spain to highlight the importance of addressing AMR under the One Health approach. Abstract Research on antimicrobial resistance (AMR) in pet rabbits is very scarce. The aim of this study was to provide an overview of the current state of AMR in rabbits attended to in veterinary clinics distributed in Spain. Records of 3596 microbiological results of clinical cases submitted from 2010 to 2021 were analyzed. Staphylococcus spp. (15.8%), Pseudomonas spp. (12.7%), Pasteurella spp. (10%), Bordetella spp. (9.6%) and Streptococcus spp. (6.8%) were the most frequently diagnosed agents. Enterobacteriaceae, principally Escherichia coli, Klebsiella pneumoniae and Enterobacter cloacae, accounted for about 18% of the cases and showed the highest proportion of multi-drug resistance (MDR) isolates, with 48%, 57.5% and 36% of MDR, respectively. Regarding the antimicrobial susceptibility testing for a number of antimicrobial categories/families, the largest proportion of isolates showing resistance to a median of five antimicrobial categories was observed in P. aeruginosa, Stenotrophomonas maltophilia and Burkolderia spp. In contrast, infections caused by Staphylococcus, Streptococcus spp. and Pasteurella multocida were highly sensitive to conventional antimicrobials authorized for veterinary use (categories D and C). The emergence of AMR major nosocomial opportunistic pathogens such as P. aeruginosa, S. maltophilia and K. pneumoniae in pet rabbits can represent a serious public health challenge. Consequently, collaboration between veterinarians and human health professionals is crucial in the fight against antimicrobial resistance, to optimize, rationalize and prudently use antimicrobial therapies in domestic animals and humans.
Introduction
Antimicrobial resistance (AMR) is a growing global concern, with the emergence of multidrug-resistant bacteria representing a significant threat to human and animal health. The close interaction between pets and their owners can facilitate the transmission of pathogenic bacteria between humans and animals, especially multidrug-resistant (MDR) microorganisms, representing a serious threat for human and animal health. Moreover, MDR infections complicate medical management, lengthen hospital stays and have a big economic impact [1].
Rabbits are the second most common specialty/exotic pet mammals among households, according to the American Veterinarian Association, and they are considered ideal pets for children in the USA and Europe [2]. Currently, rabbits are expanding in other regions, being extremely popular pets in Australia and in Asian countries such as Japan and Singapore [3]. Pet rabbits may also host parasites (Encephalitozoon cuniculi, Cryptosporidium spp., Giardia spp. and Tricostrongylus spp.), viruses (hepatitis E), bacteria (Bartonella spp., Pasteurella spp.) and fungi (dermatophytosis), which can be potential zoonotic pathogens for humans [4]. Elder people and children younger than 5 years, as well as immunocompromised persons and pregnant women, are particularly most susceptible to such pet-induced zoonoses [5]. However, related to AMR bacteria, most of the data published in pets are focused on dogs and cats [6][7][8][9][10][11] and very few are related to other pet species such as rabbits [4,12].
Thus, understanding the prevalence of AMR among pet rabbits is highly necessary from both veterinary and human medicine perspectives. Since the number of antibiotics available in veterinary medicine is limited, and there are many antibiotics contraindicated for oral administration in rabbits because of their toxicity (clindamycin, lincomycin, erythromycin, ampicillin, amoxicillin/clavulanic acid and cephalosporins), it is very important to select the best therapeutic option [13][14][15]. Thus, the use of antibiotics should be based on the results of susceptibility testing and the specific needs of each rabbit case. Empiric treatment should be administered only for urgent cases where the survival of the animal is compromised and should be based on scientific evidence. Therefore, scientific data on AMR in pet rabbits is urgently needed as a guide for veterinarian clinicians to optimize antibiotic use in rabbits for reducing the selection of antibiotic resistance. In addition, antimicrobial stewardship programs will also be conducted to educate rabbit owners not to misuse antibiotics on their pets as it may put their own health at risk. This paper aims to provide an overview of the current state of AMR in rabbits attended to in veterinary clinics distributed in Spain and discuss the potential causes and consequences of this problem under the One Health approach.
Database Collection and Management
Retrospective data on microbiological results of clinical cases of pet rabbits submitted between 2010 and 2021 from Spain and Portugal were analyzed. The database was comprised of 3596 records. These records were provided by a private diagnostic laboratory in Barcelona (Spain), which has had the ISO-9001 quality management system certificate since 1998, and the ENAC (National Accreditation Entity) accreditation according to criteria included in the ISO standard/IEC 17025 defined in technical annexes 511/LE1947 for pharmaceutical toxicology and microbiology tests.
The first step was to filter and categorize the study variables to homogenize all the data for performing subsequent descriptive and quantitative statistical analyzes. The following variables were included in the study: geographical location of the sample; origin of the sample classified in categories as regards the pathological relevance in rabbits (abscesses, dental disease, dermatitis/skin disease, otitis, conjunctivitis, reproductive tract, respiratory tract, urinary tract infections); microbiological result (positive identification or negative/absence of bacterial growth); bacterial species (grouped by genus and species) and the antimicrobial sensitivity results (from the 84 antibiotics included in the study, the antibiotics most conventionally used in veterinary medicine and as a last resort for human medicine were selected).
Based on the lab testing readings, isolates were classified as susceptible, intermediate or resistant. For showing the AST histograms of antimicrobial categories, all isolates that exhibited intermediate resistance were grouped with the susceptible ones. Multidrug resistance (MDR) was defined as resistance to at least 1 agent in ≥3 antimicrobial categories and determined using R version 4.2.0 (R Core Team, 2022) [18], with the AMR package [19], as defined by Magiorakos et al. (2012), where intrinsic resistances were not considered in the analysis [20]. In the definitions proposed for MDR in this study, a bacterial isolate is considered resistant to an antimicrobial category when it is 'non-susceptible to at least one agent in a category' [20].
The most frequent pathogens involved in cases of abscesses (located mainly on the head), dental disease, dermatitis/skin disease, conjunctivitis and otitis were Gram-positive cocci (principally Staphylococcus spp., followed by Streptococcus spp.) and Pseudomonas aeruginosa ( Figure 1). Streptococcus spp. was the primary agent responsible for reproductive tract infections, while Enterococcus spp. was the most frequently responsible for urinary infections. Gram-negative infections caused by P. multocida and B. bronchiseptica (33%), followed by P. aeruginosa (15%), were the most frequent causes of respiratory infections. Additionally, Pasteurella spp. was found in cases of abscesses, dermatitis/skin disease, conjunctivitis and otitis. As regards the AST results, P. aeruginosa was the most prevalent pathogen with the highest levels of AMR, presenting 80% of strains resistant to penicillins, inhibitors of βlactamases (AMC), 1st and 2nd generation (1G/2G) cephalosporins, trimethoprim/sulfonamides and phenicols, and 60% of strains were resistant to 3rd and 4th generation (3G/4G) cephalosporins ( Figure 2). As regards to the MDR profile, 8% (31/381) of P. aeruginosa strains were MDR, but the average number of antimicrobial categories or families that presented resistance was 5 (Table 2, Figure 3).
Other less representative bacteria but with the largest proportion of isolates showing resistance to a median of five antimicrobial categories were Stenotrophomonas spp., specifically S. maltophilia and Burkolderia spp. (Table 2). Both bacterial species were not considered MDR strains because of the intrinsic resistance to several families (Table 2). However, from the clinical point of view, it is interesting to remark that they presented high frequencies of resistance to β-lactams, with special attention to carbapenems (>80% Stenotrophomonas and 50% Burkholderia), also to polymyxins (>75% Burkholderia and 60% Stenotrophomonas) and fluoroquinolones (55% Burkholderia and 48% Stenotrophomonas) ( Figure 2).
As regards the AST results, P. aeruginosa was the most prevalent pathogen with the highest levels of AMR, presenting 80% of strains resistant to penicillins, inhibitors of β-lactamases (AMC), 1st and 2nd generation (1G/2G) cephalosporins, trimethoprim/sulfonamides and phenicols, and 60% of strains were resistant to 3rd and 4th generation (3G/4G) cephalosporins ( Figure 2). As regards to the MDR profile, 8% (31/381) of P. aeruginosa strains were MDR, but the average number of antimicrobial categories or families that presented resistance was 5 (Table 2, Figure 3).
Other less representative bacteria but with the largest proportion of isolates showing resistance to a median of five antimicrobial categories were Stenotrophomonas spp., specifically S. maltophilia and Burkolderia spp. (Table 2). Both bacterial species were not considered MDR strains because of the intrinsic resistance to several families (Table 2). However, from the clinical point of view, it is interesting to remark that they presented high frequencies of resistance to β-lactams, with special attention to carbapenems (>80% Stenotrophomonas and 50% Burkholderia), also to polymyxins (>75% Burkholderia and 60% Stenotrophomonas) and fluoroquinolones (55% Burkholderia and 48% Stenotrophomonas) (Figure 2). Values of minimal inhibitory concentrations (MIC) can be found in Table S1. In general, P. aeruginosa presented the highest levels of MIC90 for a major portion of the antimicrobials tested. The Enterobacteriaceae family, represented principally by E. coli, K. pneumoniae and E. cloacae, showed a high prevalence of AMR to β-lactams: penicillins (>80% K. pneumoniae and E. cloacae), AMC (>80% E. cloacae), 1G/2G cephalosporins (>50% K. pneumoniae and >70% E. cloacae) and 3G/4G cephalosporins (50% K. pneumoniae). Moreover, K. pneumoniae isolates showed resistance to trimethoprim/sulfonamides (50%) and to fluoroquinolones (60%) (Figure 2). Moreover, the percentage of MDR isolates was notable in enterobacteria isolates such as K. pneumoniae (58%), E. coli (48%), Proteus spp. (47%) and E. cloacae (36%) ( Table 2). In addition, the average number of antimicrobial categories presenting resistance was three in almost all enterobacteria, except for K. pneumoniae, in which it was four ( Figure 3).
Finally, Gram-positive cocci (Staphylococcus and Streptococcus) and other Gram-negative bacteria, such as Pasteurella multocida and Trueperella pyogenes, were sensitive to a wide panel of conventional antimicrobial agents, including those classified in categories D and C (Figure 2).
Values of minimal inhibitory concentrations (MIC) can be found in Table S1. In general, P. aeruginosa presented the highest levels of MIC 90 for a major portion of the antimicrobials tested.
Discussion
This study aimed to highlight the importance of addressing AMR in pet rabbits as a crucial step in the fight against antimicrobial resistance more broadly, enhancing the correct use of antibiotics to preserve their efficacy in the future to effectively control bacterial infections in people and pets.
The positive finding of these results is that the most common infections caused by Gram-positive cocci, basically Staphylococcus and Streptococcus spp. involved in abscesses, dental disease, dermatitis/skin disease, conjunctivitis and otitis, presented a low frequency of AMR, being sensitive to antimicrobials of categories D and C according to the EMA [21]. Additionally, Pasteurella (P. multocida), one of the most common bacteria of rabbits which colonizes the upper respiratory tract and the oro-pharynx, was found to be highly sensitive to conventional D and C class drugs. Pasteurella multocida can reside in the nasal flora of asymptomatic rabbits and spread to other sites during grooming, and it is also frequently isolated from abscesses because this bacterium has capsular polysaccharides that resist phagocytosis [22]. In pet rabbits, most abscesses occur around the head and face and are associated with dental disease. Another bacterial agent isolated from abscesses and dental disease was Trueperella pyogenes. This bacterium has been associated with sporadic cases of suppurative disorders in the lungs, liver, spleen and brain of rabbits [23]. Fortunately, and similar to P. multocida, T. pyogenes presented a highly sensitive pattern of AMR in our pet rabbits.
Discussion
This study aimed to highlight the importance of addressing AMR in pet rabbits as a crucial step in the fight against antimicrobial resistance more broadly, enhancing the correct use of antibiotics to preserve their efficacy in the future to effectively control bacterial infections in people and pets.
The positive finding of these results is that the most common infections caused by Gram-positive cocci, basically Staphylococcus and Streptococcus spp. involved in abscesses, dental disease, dermatitis/skin disease, conjunctivitis and otitis, presented a low frequency of AMR, being sensitive to antimicrobials of categories D and C according to the EMA [21]. Additionally, Pasteurella (P. multocida), one of the most common bacteria of rabbits which colonizes the upper respiratory tract and the oro-pharynx, was found to be highly sensitive to conventional D and C class drugs. Pasteurella multocida can reside in the nasal flora of asymptomatic rabbits and spread to other sites during grooming, and it is also frequently isolated from abscesses because this bacterium has capsular polysaccharides that resist phagocytosis [22]. In pet rabbits, most abscesses occur around the head and face and are associated with dental disease. Another bacterial agent isolated from abscesses and dental disease was Trueperella pyogenes. This bacterium has been associated with sporadic cases of suppurative disorders in the lungs, liver, spleen and brain of rabbits [23]. Fortunately, and similar to P. multocida, T. pyogenes presented a highly sensitive pattern of AMR in our pet rabbits.
The zoonotic risk of P. multocida transmission to humans must be considered through bites, scratches or licks of companion animals, with the development of local inflammatory reactions and occasionally the occurrence of abscesses in people [5, [24][25][26]. Moreover, in some patients, principally in immunocompromised people or persons with pulmonary disorders, pasteurellosis may result in more severe pathologies, such as pneumonia, endocarditis, meningitis and sepsis [27,28]. In a recent paper, P. multocida belonging to capsular type A was the type most often detected in humans, and although it was susceptible to the tested antibiotics, in agreement with our AST results, it was equipped with several virulence genes [4]. These findings are of particular interest because rabbits recovered from pasteurellosis very often become asymptomatic carriers of this infection and can represent a risk for the household members, especially for children and elder people [29].
On the other hand, Gram-negative infections caused by P. multocida and B. bronchiseptica, followed by P. aeruginosa, were principally involved in respiratory infections, in agreement with a pervious study conducted in pet rabbits in France [30]. In that study, the authors concluded that marbofloxacin was shown to be a potentially good treatment option for upper respiratory tract disease in pet rabbits. Although the use of fluoroquinolones is the most common therapeutic option in exotic animal medicine, the EMA recommendations appeal for the use of D and C categories in order to preserve the efficacy of critical antimicrobial classes such as fluoroquinolones (category B). For this reason, and considering the AST results of our study, for respiratory infections caused by P. multocida or B. bronchiseptica, trimethoprim/sulfonamides could be a good candidate for treatment in pet rabbits.
Non-fermenting Gram-negative bacilli, such as P. aeruginosa and Acinetobacter baumannii, are among the major opportunistic pathogens involved in the global antibiotic resistance epidemic in human medicine [31]. Data on pet rabbits showed that the antimicrobial treatment of P. aeruginosa can be more complicated, since a high percentage of the isolates presented a resistance profile, including antimicrobials of category B (3G/4G cephalosporins and fluoroquinolones). This pathogen is also found in a wide spectrum of rabbit pathologies (abscesses, dental disease, dermatitis/skin disease, conjunctivitis, otitis and respiratory infections), and the treatment options are very few, limited to carbapenems and polymyxins, which are antimicrobials of category A (reserved for critical use in human medicine), but also to aminoglycosides. Since this former family is classified in category C, aminoglycosides could be the best option for treating pseudomonal infections in rabbits. It is important to note that polymyxins can be highly toxic to rabbits and should be avoided for treatments. However, this antimicrobial class was added in this study for its relevance as a last-resort drug for human medicine.
As regards the Enterobacteriaceae family, E. coli, K. pneumoniae and E. cloacae represented the most frequent species isolated from a large diversity of pathologies. Escherichia coli infections can cause enteritis, sepsis and urinary tract infections in rabbits. Although E. coli was the most prevalent enterobacteria, the frequency of MDR was lower compared to K. pneumoniae, as observed in other pet studies in Spain [9,12]. According to our results, good candidates for treating infections caused by E. coli could be aminoglycosides. On the other hand, K. pneumoniae showed high resistance to most of the antimicrobial classes of conventional use in veterinary medicine, leaving carbapenems as the best therapeutic option even though it is a category A drug. Considering other antimicrobials authorized for veterinary medicine, the best options were aminoglycosides, chloramphenicol or doxycycline, although more than 40% of the isolates presented resistance to these drugs. As a result, the increasing occurrence of K. pneumoniae as a MDR infection and a zoonotic agent represents a real threat to both animal and human health [32,33]. In addition, E. cloacae is another emerging pathogen recognized as a nosocomial bacterium contributing to septic arthritis, skin/soft tissue infections, bacteremia, lower respiratory tract and urinary tract infection, endocarditis, osteomyelitis and intra-abdominal infections in humans [34].
Other less representative bacteria, but with a proportion of resistance to several antimicrobial categories (five as a median), were S. maltophilia and Burkolderia spp. Both bacterial species presented high frequencies of resistant isolates to β-lactams (including carbapenems), as well as to category A (polymyxins) and B (fluoroquinolones) drugs. S. maltophilia is an emerging nosocomial pathogen, with intrinsic resistance to beta-lactams, capable of causing healthcare-associated infections in intensive care units, life-threatening diseases in immunocompromised patients and severe pulmonary infections in cystic fibrosis and COVID-19-infected individuals [35,36].
Lastly, it was interesting to note that 12% of the Enterococcus isolates were resistant to vancomycin, more than 80% to aminoglycosides, around 70% to 1G/2G cephalosporines, half of them to fluoroquinolones and 40% to 3G/4G cephalosporines. With these AMR profiles, the treatment of UTI caused by this bacterium in rabbits can be difficult to plan without a previous susceptibility testing.
Overall, the emergence of AMR strains such as P. aeruginosa, A. baumannii, S. maltophilia and K. pneumoniae in pet rabbits can represent a serious health threat for the owners, since they are among the major opportunistic pathogens with significant contributions to mortality in hospitals worldwide [31,37]. Moreover, these pathogens are designated as urgent/serious threats by the Centers for Disease Control and Prevention and are part of the World Health Organization's list of critical priority pathogens [38].
It is important to remember that the list of antimicrobial therapeutic options for treating bacterial infections in rabbits is not exhaustive and the use of antibiotics should be based on the results of susceptibility testing, the specific needs of each animal case and the risk of toxicity of these drugs in rabbits. However, for urgent cases, when the severity of the clinical process requires immediate antimicrobial therapy with no time for AST analysis, the data reported in the present study can be useful for veterinary practitioners to apply empirical therapy. It is crucial to keep in mind that the best way to proceed for reducing AMR selection is to perform a proper antimicrobial diagnosis with the corresponding AST. Then, antimicrobials with a sensitive result must be prioritized according to the EMA categories, mainly D and C.
Finally, the results of this study provided objective data on the microbiological results in pet rabbits in Spain. The high levels of AMR to critically important antibiotics in human medicine found in pet rabbits are of great concern since potential transmission of resistance genes from rabbits to humans or other pets can occur. Considering that the predominant bacteria in this study are among the top pathogens directly attributed to human deaths due to AMR, it is critical that veterinarians and physicians work together to optimize, rationalize and prudently use antimicrobial therapies in domestic animals and humans under the One Health approach.
==
Domain: Biology Medicine
|
[{"added": "2023-05-17T15:24:31.658Z", "created": "2023-05-01T00:00:00.000", "id": "258733464", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.3390/vetsci10050352", "pdf_hash": "2caaf88fb6e6ece297ede5b6aa5525240daee565", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:9396", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "861aa26ce6b3b921a0e8e8be61a5444463ca0c08", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Genome Editing and Cardiac Arrhythmias
This article reviews progress in the field of cardiac genome editing, in particular, its potential utility in treating cardiac arrhythmias. First, we discuss genome editing methods by which DNA can be disrupted, inserted, deleted, or corrected in cardiomyocytes. Second, we provide an overview of in vivo genome editing in preclinical models of heritable and acquired arrhythmias. Third, we discuss recent advancements in cardiac gene transfer, including delivery methods, gene expression optimization, and potential adverse effects associated with therapeutic somatic genome editing. While genome editing for cardiac arrhythmias is still in its infancy, this approach holds great promise, especially for inherited arrhythmia syndromes with a defined genetic defect.
Introduction
Millions of people around the world experience cardiac arrhythmias at some point during their life [1]. While arrhythmias are harmless most of the time, some heart rhythm disorders can have serious consequences or may result in death. The most common arrhythmia type is atrial fibrillation, which can lead to life-threatening complications such as stroke or heart failure [2]. Though less common, ventricular arrhythmias such as ventricular tachycardia and ventricular fibrillation can be very dangerous and lead to sudden cardiac death [3,4].
Despite advances in screening and preventative treatments, the estimated lifetime risk for premature death due to arrhythmogenic sudden cardiac death remains high (approximately 1 in 9 men and 1 in 30 women) [5]. In part, this can be explained by risk factors including structural heart disease, coronary artery disease, nutrition, and genetic predispositions [6]. While acquired co-morbidities such as coronary artery disease are leading causes of sudden cardiac arrest in older adults, arrhythmias resulting from heritable diseases are the leading causes of sudden cardiac arrest in children and young adults [3].
A growing understanding of the genetics underlying cardiac arrhythmias has enabled new treatment possibilities including the use of cardiac genome editing [7]. Monogenetic diseases associated with cardiac arrhythmias include congenital long QT syndrome (LQTS), short QT interval syndrome, Brugada syndrome, catecholaminergic polymorphic ventricular tachycardia (CPVT), idiopathic ventricular fibrillation, Wolff-Parkinson-White syndrome, arrhythmogenic cardiomyopathy, and other cardiomyopathy syndromes associated with an increased risk of arrhythmias such as dilated cardiomyopathy (DCM), hypertrophic cardiomyopathy (HCM) and hereditary transthyretin amyloidosis (ATTR-CM) [8]. Acquired arrhythmias such as atrial fibrillation or those associated with ischemic heart
Genome Editing Techniques
While gene targeting in embryonic cells has been used for modeling familial cardiac disorders, somatic cardiac genome editing only became feasible following the development of nucleases that can introduce double-stranded breaks (DSBs) at specific locations in the genome. These nucleases can be categorized into four general categories: meganucleases, zinc-finger nucleases (ZFN), transcription activator-like effector nucleases (TALENs), and clustered regularly interspaced short palindromic repeats (CRISPR) associated with nuclease 9 (Cas9) [12]. Notably, however, only the CRISPR/Cas9 system has been used for therapeutic somatic genome editing of the heart. Therefore, we will focus mainly on the CRISPR/Cas9 system for the remainder of this review.
In the CRISPR/Cas9 system, a synthetic RNA spacer sequence is inserted into a scaffold for the Cas9 nuclease [13,14]. This guide RNA (gRNA) sequence allows Cas9 to target genomic sequences of 20 or more base pairs (bp). Cas9 binds DNA at the protospacer adjacent motif (PAM) and cleaves 3 bp upstream of the PAM (Figure 1). In contrast to the amino acid recognition sequences that other endonucleases rely on, the CRISPR/Cas9 spacer sequences are much simpler to design, synthesize, and validate. As such, the field of genome editing overall has largely moved to this platform, and subsequent discoveries in cardiac gene editing have relied on utilizing the CRISPR/Cas9 system. The first publication of somatic genome editing of the heart involved using this system to correct Duchenne's muscular dystrophy (DMD) in a mouse model [15]. Subsequently, there have been several developments in expanding the capabilities and efficacy of the CRISPR/Cas9 systems and these are reviewed in the subsequent sections [16].
The expression of CRISPR/Cas9 with a single gRNA is sufficient to create a DSB at a target genome locus. In cardiomyocytes, this has been shown to cause primarily non-homologous end-joining (NHEJ) repair ( Figure 1) [17]. The lack of fidelity in NHEJ repair, however, results in insertions and deletions (indels) that can result in a frame shift, which-when introduced within a coding sequence-may lead to nonsense-mediated degradation. This method has been employed to model loss-of-function mutations by disrupting a target gene. In addition, this method can be used to disable a toxic or detrimental gain-of-function allele in heterozygous individuals. The feasibility of postnatal gene editing was demonstrated in mice overexpressing a cardiomyocyte-specific Cas9 in combination with adeno-associated viral (AAV) vector-mediated delivery of guide RNAs targeting Myh6, Sav1, and Tbx20 [18]. Following AAV delivery, mosaic disruption of the target genes was seen, and only disruption of Myh6 was sufficient to induce a cardiac phenotype. Despite the limitations of a mosaic knockdown in terms of efficiency, this method has been uniquely beneficial for characterizing a novel loss-of-function arrhythmia disorder that would otherwise be lethal in a complete knockout state [19]. NHEJ editing has also been effective in deleting splice donor or acceptor site sequences in out-of-frame (14)(15)(16)(17)(18)(19) to minimize wild-type allele cleavage. The Cas9 endonuclease introduces a double-stranded break (DSB). Nonhomologous end joining (NHEJ) repairs the mutant allele but introduces insertions and deletions at the mutation site, which can lead to frame-shift mutations and stop codons, resulting in nonsensemediated decay of mutant RNA (top). On the other hand, homology-directed repair (HDR) is less efficient but leads to correction of the mutant allele using a DNA repair template. (B) Cytosine base editors are created by fusing Cas9 nickase (nCas9) or catalytically inactive "dead" Cas9 (dCas9) to a cytidine deaminase. Base editors are targeted to a specific locus using gRNA. They can convert cytidine (C) to uridine (U) within a small editing window (4-7 in the spacer sequence, depending on the editor type), near the PAM site. Uridine is subsequently converted to thymidine (T) through base excision repair. (C) Likewise, adenosine base editors have been engineered to convert adenosine (A) to inosine (I), which is subsequently converted to guanidine (G). (D) Prime editing involves a reverse transcriptase and dCas9 that can produce single-stranded DNA breaks. The 3 -extended single guide RNA contains a primer binding site and a reverse transcriptase template, which is referred to as primer editing guide RNA (pegRNA). Hybridization of the exposed 3 -end to the primer binding site primes the reverse transcription of the nicked DNA strand for the desired edit. CRISPR/Cas9 genome editing can also be used to replace and correct the target sequence with a template through homology-directed repair (HDR) (Figure 1). Although gene correction is attained, HDR is less efficient compared with NHEJ, and a vast majority of edits will contain random insertions or deletions. Given the low efficiency in nondividing cells such as cardiomyocytes, HDR has not been used (yet) for the treatment of cardiac arrhythmias in preclinical animal models. Another CRISPR technique involves using at least one gRNA sequence and a template sequence flanked by the gRNA target sequences in reverse orientation to create a DSB at the target site, followed by insertion of a corrected gene by incorporation of the template DNA through homology-independent targeted integration (HITI) [24]. Due to the limited efficacy of HDR in post mitotic tissue, HITI has not been used to correct arrhythmia phenotypes, although it was used to increase full-length dystrophin expression in a humanized mouse model of DMD with a genomic correction rate of 4-7% in cardiomyocytes [25].
Alternative CRISPR/Cas9-based genome editing techniques have been developed that allow for the introduction of point mutations in the genomic DNA without generating DSBs [26]. By fusing Cas9 nickase (nCas9) or deactivated Cas9 (dCas9) to a deaminase enzyme, the resulting base editor can edit DNA without DSBs, converting C to T, or G to A. Further development of cytosine base editors (CBE) and adenine base editors (ABE) resulted in expansions of possible conversions including C to T, A to G, T to C, and G to A, respectively [26]. Studies revealed an in vitro efficiency of 15-75%, with less than 1% indel formation, although efficiency is much lower in non-dividing cells. Moreover, DNA base editors have some shortcomings, including off-target DNA editing, the generation of bystander mutations, and promiscuous deamination effects in both DNA and RNA. Finally, prime editing uses a fusion protein, consisting of a catalytically impaired Cas9 endonuclease fused to an engineered reverse transcriptase, and a prime editing guide RNA (pegRNA), capable of replacing the target DNA nucleotides without the need for DSBs or donor DNA templates [27].
Using CRISPR/Cas9 Genome Editing to Create Animal and Cellular Arrhythmia Models
The advent of genome editing approaches such as CRISPR/Cas9 has created a whole new set of possibilities to generate animal models for biomedical research using virtually any species. Animal models in which human gene variants are introduced can provide convincing evidence for a disease-causing pathogenic mechanism [28]. While gene overexpression or knockdown strategies may provide useful insights in some cases [29], knock-in strategies that involve introducing the exact genetic variant in the model system are generally superior [30]. The RNA-guided Cas9 nuclease system can be used to easily generate knock-in mouse models [31]. This system has superseded all other systems for genome editing because of its simplicity of use, lower costs, and higher efficiency [32]. These days, most researchers prefer this approach over ES cell-based gene-targeting methods [33]. The CRISPR/Cas9 system allows for the introduction of footprint-free point mutations on various genetic backgrounds. While many of the initial mouse models of monogenetic arrhythmia syndromes were generated using ES cell-based methods, several newer models have been generated using CRISPR/Cas9 gene editing [34,35].
Genome editing can also be used to introduce potential disease-causing variants in patient-derived induced pluripotent stem cells (iPSCs) that can subsequently be differentiated into cardiomyocytes [36]. This approach has been particularly helpful to elucidate whether genetic variants of unknown significance (VUSs) in various genes purported to cause arrhythmias are in fact pathogenic [37]. Various genome editing techniques including CRISPR/Cas9 can introduce or correct genetic variants in iPSC before differentiation into cardiomyocytes, providing high throughput empirical evidence for characterizing novel mutations, their mechanisms, and potential therapies [38]. Many iPSC models of heritable arrhythmias, including LQTS, Brugada syndrome, CPVT, and arrhythmogenic cardiomyopathy, have been generated, as recently reviewed by Yang et al. [39]. These models have been used not only to characterize disease mechanisms but also to test the effectiveness of novel therapeutics [40].
In addition to being a tool for introducing genetic variants into iPSCs, CRISPR/Cas9 also enables efficient correction of inherited variants in iPSCs. By taking somatic cells such as skin fibroblasts or peripheral whole blood samples from patients, patient-derived iPSCs can be generated that include all VUSs present in the patient genome. Generating an isogenic control provides incredibly powerful evidence on whether a specific SNP is likely to be pathogenic. Most studies to this effect have used HDR to generate isogenic controls [41]. However, recent studies have begun to use more advanced techniques such as base editing and prime editing to correct specific variants in iPSCs [42]. The ease with which iPSCs can undergo genome editing has dramatically increased the number of VUS that can be characterized and has also expanded the genetic approaches that can be used to explore disease mechanisms. There are important limitations to the use of iPSC-derived cardiomyocytes (iPSC-CMs). While studies in single iPSC-CMs might yield new insights about certain inherited arrhythmia syndromes, it would be implausible that relevant insights could be gleaned from iPSC-CMs generated from patients with atrial fibrillation or ischemic heart disease. In addition, despite improvements in experimental protocols, iPSC-CMs still exhibit relatively immature and variable phenotypes. Finally, iPSC-CMs often do not have a functional SR, and the intracellular Ca 2+ handling dynamics are quite different from those in freshly isolated CMs from adult animals or human patients [43].
Therapeutic Genome Editing in Preclinical Arrhythmia Models
Somatic genome editing has also been used to correct heart disease in preclinical model systems (Table 1, Figure 2). For example, the CRISPR/Cas9 editing method was employed in correcting the cardiac PRKAG2 syndrome, which is known to cause familial Wolff-Parkinson-White syndrome with cardiomyopathy [44]. PRKAG2 syndrome is an autosomaldominant inherited disease caused by missense mutations in the PRKAG2 gene, which encodes the γ2 regulatory subunit of AMP-activated protein kinase [45]. Altered activity of this AMP-activated protein kinase results in excessive cellular glycogen deposition leading to cardiomyopathy and supraventricular arrhythmias. Systemic administration of adeno-associated virus serotype 9 (AAV9) with gRNA and the CRISPR/spCas9 geneediting system was sufficient to disrupt the mutant PRKAG2 allele in mice while leaving the wild-type allele intact [46]. While the genome-editing efficiency was relatively low (∼6.5% in mice injected at postnatal day (P) 4 and ∼2.6% in mice injected at P42), this treatment strategy reduced preexcitation arrhythmias by 40%, and restored the morphology and function of the heart in mutant mice [46]. ministration of adeno-associated virus serotype 9 (AAV9) with gRNA and the CRISPR/spCas9 gene-editing system was sufficient to disrupt the mutant PRKAG2 allele in mice while leaving the wild-type allele intact [46]. While the genome-editing efficiency was relatively low (∼6.5% in mice injected at postnatal day (P) 4 and ∼2.6% in mice injected at P42), this treatment strategy reduced preexcitation arrhythmias by 40%, and restored the morphology and function of the heart in mutant mice [46]. The diagram shows the plasma membrane with various ion channels including the voltage-gated L-type Ca 2+ channel (LTCC), and K + and Na + channels. The sarcoplasmic reticulum (SR) is shown with the ryanodine receptor type-2 (RyR2)/ Ca 2+ release channel and sarco-/endoplasmic Ca 2+ -ATPase (SERCA2a) with its regulatory subunit phospholamban (PLN). Dystrophin links the plasmalemmal sarcoglycan complex to the sarcomere. The Ca 2+ /calmodulin-dependent protein kinase II (CaMKII) phosphorylates various ion channels and Ca 2+ handling proteins within cardiomyocytes. Gene names that are targets of therapeutic genome editing are indicated in bold and italicized.
Figure 2.
Schematic overview of therapeutic CRISPR/Cas9 targets for arrhythmia treatment. The diagram shows the plasma membrane with various ion channels including the voltage-gated L-type Ca 2+ channel (LTCC), and K + and Na + channels. The sarcoplasmic reticulum (SR) is shown with the ryanodine receptor type-2 (RyR2)/Ca 2+ release channel and sarco-/endoplasmic Ca 2+ -ATPase (SERCA2a) with its regulatory subunit phospholamban (PLN). Dystrophin links the plasmalemmal sarcoglycan complex to the sarcomere. The Ca 2+ /calmodulin-dependent protein kinase II (CaMKII) phosphorylates various ion channels and Ca 2+ handling proteins within cardiomyocytes. Gene names that are targets of therapeutic genome editing are indicated in bold and italicized.
Our research group utilized a similar approach to treat catecholaminergic polymorphic ventricular tachycardia (CPVT) in a preclinical mouse model [47]. CPVT is an autosomaldominant inherited disease caused primarily by missense mutations in the RYR2 gene that encodes the ryanodine receptor type 2 (RyR2) intracellular calcium channel [49]. The mutated channel generates a diastolic calcium leak that can trigger lethal arrhythmias [30]. We found that systemic administration of AAV to deliver gRNA and SaCas9 was sufficient to disrupt the mutant RyR2 allele in a heterozygous mice model while leaving the wild-type allele intact. The editing frequency based on next generation sequencing was found to be around 11% in these RyR2 mutant mice [47]. In addition, the reduced mutant allele mRNA levels were indicative of nonsense-mediated decay. This degree of genome editing reduced the ventricular tachycardia incidence by 100% and normalized channel function at the cellular level [47].
The same NHEJ pathway was induced by CRISPR/Cas9 genome editing in a humanized mouse model of arrhythmogenic dilated cardiomyopathy caused by a truncating variant in phospholamban (PLN) [48]. The PLN-R14del mutation was first discovered in a large Greek family with clinical signs of both dilated cardiomyopathy (DCM) and arrhythmogenic cardiomyopathy (ACM) [50]. Subsequently, many PLN-R14del mutation carriers were found in the Netherlands, where the original mutation is believed to have originated [51]. Mice harboring the R14del mutant human PLN were found to be more susceptible to rapid pacing-induced ventricular tachycardia. In vivo gene editing using AAV9 led to a greater than two-fold reduction in pacing-induced VT incidence and significantly raised the frequency threshold for induction [48].
While gene disruption of dominant-negative mutations is one of the most efficient methods of therapy, other methods of genome editing are needed in the case of haploinsufficiency. Duchenne muscular dystrophy (DMD) is a fatal X-linked recessive disorder that causes progressive neuromuscular weakness and wasting. A quarter of DMD patients die from cardiac causes, and half of these deaths are due to lethal ventricular tachycardia [52]. Abnormal Ca 2+ release from the sarcoplasmic reticulum via RyR2 overactivated by Ca 2+ /calmodulin-dependent protein kinase II (CaMKII) is a major mechanism underlying arrhythmogenesis in mice with DMD [53]. Inherited variants in the DMD gene usually involve single or multi-exon deletions that disrupt the open reading frame (ORF) and introduce a premature stop codon, resulting in a nonfunctional, truncate dystrophin protein [54]. CRISPR/Cas9 has been used to generate indels using NHEJ in the DMD gene to restore the ORF in rodent and large animal preclinical models [20]. Recently, Chemello et al. [42] used an adenine base editor (ABE) delivered using AAV9 as a split-intein system to restore protein expression in a DMD mouse model. By means of a single-swap base pair transition in the dinucleotide splicing motif, exon skipping was accomplished with restoration of dystrophin levels. The on-target editing efficiency ranged from 6.7 to 35.0%, with no notable editing at off-target sites [42]. Whereas the effects of base editing on arrhythmogenesis were not assessed in vivo, complementary studies in human induced-pluripotent stem cell-derived cardiomyocytes revealed that base editing normalized arrhythmic calcium handling deficits [42]. Additional studies are needed to determine whether base editing can also block lethal arrhythmias in vivo in animal models of DMD. Finally, it may be possible to use genome editing for the treatment on non-genetic arrhythmia disorders by targeting key molecular switches that drive disease development. For example, Lebek et al. [55] recently showed that base editing could be used to eliminate oxidation-sensitive methionine residues on CaMKII [56] to confer protection from ischemia/reperfusion injury, which is often associated with ventricular arrhythmias. Since the same CaMKII residues have been implicated in ventricular arrhythmogenesis in DMD [53] and atrial fibrillation [57], similar approaches may hold promise for a wider array of arrhythmia disorders.
Current Challenges of Genome Editing in Arrhythmias and Future Developments
While CRISPR Cas9 genome editing can provide long-term correction to arrhythmia disorders, one major disadvantage of using CRISPR/Cas9 genome editing is that side effects can also be long-lasting. As such, an important question that remains is what amount of correction is needed to prevent arrhythmias. One difficulty in answering this question is that determining the editing efficiency in vivo can be difficult. Whole myocardial tissue includes non-myocyte genomes that may not be as efficiently edited as cardiomyocytes with AAV9. Genome sequencing of RNA transcripts can give a more cardiomyocyte-specific readout, but nonsense-mediated decay from NHEJ will affect these readouts [47]. We have used in vivo AAV9 reporters as a proxy for genome editing, though it requires dual AAV transduction and can, therefore, underestimate the amount of editing [47]. Recent studies have used the ratio of mutant to wild-type alleles as a proxy for determining the efficiency of allele-specific genome editing [48]. Due to the lack of reliability in these methods, reported genome editing efficiencies necessary for preventing arrhythmias have ranged from 6 to 24% of the whole myocardium and less than 50% of myocytes.
One group has used computer modeling of ventricular tissues to determine the editing efficiency needed to prevent arrhythmogenesis [58]. One-, two-, and three-dimensional tissue models were used to simulate early and delayed afterdepolarization-triggered arrhythmias. Stabilizer cells were distributed to simulate cardiomyocyte gene therapy. Due to source-sink relationships in cardiac tissue, a minority of stabilizer cells (20-50%) was sufficient to prevent triggered activity. Clustering stabilizer cells reduced their efficacy, and higher-dimensional models required a greater percentage of gene-edited stabilizer cells. These modeling studies correlated with in vivo gene therapy studies but also reveal the importance of evenly distributed gene delivery.
Even though CRISPR/Cas9 represents a promising approach, this genome editing system still has several limitations and associated risks, making it challenging to use in clinical trials. Delivery methods, editing efficiency, off-target effects, and immunogenicity are several major concerns that must be overcome. The most relevant limitations will each be discussed briefly in the following paragraphs: Delivery methods: The CRISPR/Cas9 genome editing system requires the delivery of a poly II promoter with a gRNA sequence and a poly III promoter with a Cas9 protein in addition to a poly-A tail [59]. While there are many methods of delivering these genes to tissue, including injection of free plasmid, adenovirus, lentivirus, and nanoparticles, AAV vectors are most used for cardiac genome editing [60]. AAVs have the capability to transduce the entire myocardium effectively following intracoronary or intravenous injection without generating a large immune response and so have been the viral vector of choice for preclinical studies of cardiac gene editing. However, a limitation of AAVs is their packaging capacity which is limited to~4.7-4.9 kb depending on the serotype. The most widely used Cas9 homolog, SpCas9 (from Streptococcus pyogenes), is 4.1 kb, and so requires the use of a minimal promoter or second AAV for the expression of gRNA. Other homologs such as SaCas9 (from Staphylococcus aureus) are smaller in size (3.1 kb), which provides the capacity to place gRNA and regulatory sequences into the same vector. Generally, the PAM sequence is dependent on which Cas9 homolog is used, and the smaller-sized homologs tend to have longer PAM sequences, which limits their targeting capability but increases their specificity.
Several groups have been able to expand the packaging capacity by using a second AAV vector with a split Cas9 protein. There are several methods to achieve this including overpacking, DNA homologous recombination, RNA splicing, and split-intein proteins [61]. Of the 13 main AAV serotypes, AAV8 and AAV9 have been used the most for post-natal cardiac genome editing, with AAV9 having the highest tropism to the heart [62]. Even though AAV9 has high tropism to the heart, it also has a great amount of tropism in the liver. As such, several publications focusing on de-targeting the liver have shown a greater cardiac specificity, though not necessarily a greater transduction efficiency [63]. Recently, evolutiondirected in vivo screens of AAV9 capsids engineered to contain multimers have found success in finding muscle tropic AAV9s [64]. These are important for cardiomyopathies associated with dystrophic diseases such as DMD. Interestingly, these capsid variants showed an increased transduction efficiency in the myocardium in comparison with AAV9. One recombinant capsid, AAVrh74, is currently in clinical trial for gene therapy of microdystrophin [65]. Other recombinant capsids such as AAV-MYO-A1 have been used in vivo for genome editing of dystrophin with an eight-fold increase in vector delivery in the heart compared with AAV9 and a significant increase in corrected mRNA following genome editing of dystrophin [66].
Editing efficiency and off-target effects. At the date of this publication, three Cas9 homologs have been used to edit the genome of the heart for therapeutic effect: SaCas9 (1053 amino acids), SpCas9 (1368 amino acids), and Campylobacter jejuni Cas9 (CjCas9, 984 amino acids). Each Cas9 orthologue has a unique PAM that Cas recognizes to bind to DNA. Each homolog has a specific PAM sequence: SaCas9 (NNGRRT), SpCas9 (NGG), and CjCas9 (NNNNRYAC). The specificity of the PAM sequence restricts which DNA sequence can be cleaved, both on and off target. Additionally, work has been performed to modify existing Cas9 homologs to expand nuclease targeting capabilities. For example, the modified KKH SaCas9 (1053 amino acids) with variants in the DNA recognition domain has expanded PAM sequences of NNARRT, NNCRRT, and NNTRRT [67]. Screening of DNA-shuffled short Cas9 homologs called synthetic guided nucleases (1053-1054 amino acids) has identified a potentially more efficient and less restrictive nuclease with an NNGG PAM sequence. These improvements will allow the expanded targeting of Cas9 while still using a single AAV vector system to guide delivery.
The R221K and N394K mutations in SpCas9 have been shown to improve Cas9 nuclease activity by specifically enhancing the ability of SpCas9 to bind to DNA [68]. Furthermore, modifications of the nuclear localization sequences increase the activity of SpCas9 as well as precision editing systems. These and other regulatory elements can be used to enhance the efficiency of cardiac genome editing, provided there is capacity in the gene delivery vector. Another mutant form of Cas9, the D10A nickase, produces a single-strand nick at the target. Combining nickase Cas9 with a pair of gRNAs targeting complementary strands of the target gene greatly reduces off-target DSBs and improves specificity because this approach requires single-strand nicks at both sgRNA target sites to produce a DSB [69].
Reducing immunogenicity. Both Staphylococcus aureus and Steptococcus pyogenes, from which the most common Cas9 proteins (SaCas9 and SpCas9) are obtained, have infected humans for a long time [70]. The majority (58-78%) of humans had anti-Cas9 antibodies in their serum in one recent study [70]. Anti-Cas9 antibodies may lead to a fast degradation of the Cas9 proteins, preventing them from performing the desired genome-editing functions [71]. Strategies to overcome the limits posed by immunogenicity against Cas9 include gene editing treatment early in life and targeting immune-privileged organs. An immune privileged organ can be defined as a site where a graft tissue can be implanted without being rejected by the host due to an immunological reaction. Examples include the eyes, brain, placenta, fetus, and testicles [72]. For the treatment of cardiac arrhythmias, the latter approach could be exploited by performing prenatal or early postnatal delivery of the CRISPR/Cas9 system. For example, Nelson et al. [73] found that an immune response to AAV-CRISPR was minimized in post-natal day 2 neonatal mice.
Conclusions and Future Directions
Cardiac genome editing with various CRISPR/Cas9-based technologies has emerged as a promising new therapeutic modality for inherited arrhythmia syndromes and cardiomyopathies. The CRISPR/Cas9 system has made the creation of cellular and animal models of cardiac arrhythmias more efficient and affordable. The feasibility of therapeutic genome editing using CRISPR/Cas9 has been demonstrated in various preclinical animal models. While innovations in technologies for cardiac gene delivery and targeting have resulted in improvements in editing efficacy in preclinical models, various challenges remain related to delivery methods, editing efficiency, off-target effects, and immunogenicity. The first clinical trials for inherited conditions associated with cardiomyopathies are being initiated, and it is likely that similar studies will commence soon for arrhythmogenic cardiomyopathy or inherited arrhythmia syndromes. The rate of progress in cardiac genome editing has been remarkable given that the discovery of CRISPR/Cas9 as a molecular tool occurred only a little over a decade ago. We can look forward to tremendous progress in genome editing approaches and likely clinical applications in the next decade.
Conflicts of Interest:
Wehrens is a co-founder of Elex Biotech, LLC, a start-up company developing RyR2 modifying drugs for heart disease. Wehrens is also a consultant to Pfizer and Rocket Pharmaceuticals.
==
Domain: Biology Medicine
|
[{"added": "2023-05-13T15:14:31.599Z", "created": "2023-05-01T00:00:00.000", "id": "258649016", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.3390/cells12101363", "pdf_hash": "62b9103c919ddd6804abd57fdac3756f615bb0dc", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:8209", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "f99b02af060db66053330c068b7a097d557354e7", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Tert-Butylhydroquinone Prevents Oxidative Stress-Mediated Apoptosis and Extracellular Matrix Degradation in Rat Chondrocytes
Oxidative stress-induced chondrocyte apoptosis and degradation of the extracellular matrix (ECM) play an important role in the progression of osteoarthritis (OA). In addition, tert-butylhydroquinone (TBHQ) is an activator of the nuclear factor erythroid derived-2-related factor 2 (Nrf2). The present study aimed to determine the effectiveness of TBHQ in preventing the apoptosis of chondrocytes and degradation of the extracellular matrix, induced by oxidative stress, in vitro. Therefore, rat chondrocytes were exposed to 20 μM tert-butyl hydroperoxide (TBHP) for 24 h to establish an oxidative damage model, in vitro. Thereafter, cell viability was evaluated using the Cell Counting Kit-8 assay. Moreover, the level of ROS was determined through 2′,7′-dichlorofluorescein diacetate staining. The mitochondrial membrane potential of chondrocytes was also measured using JC-1. Furthermore, cell apoptosis was assessed through Annexin V-fluorescein isothiocyanate/propidium iodide staining. The study also performed Western blotting and qPCR to evaluate the expression of extracellular matrix components, matrix catabolic enzymes, and changes in signalling pathways. The results showed that 2.5 and 5 μM of TBHQ reduced the TBHP-induced generation of excessive ROS and improved cell viability. Additionally, 2.5 and 5 μM of TBHQ prevented mitochondrial damage and apoptosis in rat chondrocytes. Treatment with TBHQ also increased the mRNA and protein expression levels of aggrecan and collagen II. However, TBHQ reduced the mRNA and protein expression levels of matrix metalloproteinase 3 (MMP3) and matrix metalloproteinase 13 (MMP13) in rat chondrocytes. In addition, treatment with TBHQ enhanced the protein expression levels of Nrf2, NADPH quinone oxidoreductase 1 (NQO-1), and hemeoxygenase-1 (HO-1) in rat chondrocytes. The current study showed that TBHQ was not only effective in protecting against TBHP-induced oxidative stress but also inhibited the apoptosis of rat chondrocytes and degradation of the ECM by activating the Nrf2 pathway. The results therefore suggest that TBHQ holds potential for use in the treatment of OA.
Introduction
Osteoarthritis (OA) is the most common form of arthritis that not only severely affects the patients' quality of life but also causes a substantial clinical and socioeconomic burden. OA affects more than 500 million people worldwide. From 1990 to 2019, the number of people globally suffering from OA increased by 48%. e pathological changes in OA include cartilage degeneration, synovial inflammation, and subchondral bone thickening. Currently, nonsteroidal antiinflammatory drugs (NSAIDs) and tramadol are recommended for the clinical treatment of OA. However, both NSAIDs and tramadol are only used for the short-term relief of symptoms. Safe and effective drugs for preventing or reversing the progression of OA are therefore still lacking [1][2][3].
Notably, chondrocytes are the only population of cells existing in healthy cartilage. e cells are involved in regulating the growth, distribution, and reconstruction of the cartilage matrix. In addition, the increased apoptosis of chondrocytes and the subsequent degradation of the extracellular matrix (ECM), caused by multiple factors, are considered to be signs of cartilage degradation in OA [4][5][6].
Numerous studies have also shown that oxidative stress plays an important role in the occurrence and progression of OA. Moreover, excessive generation of reactive oxygen species induces oxidative damage in chondrocytes, which leads to chondrocyte apoptosis and degradation of the ECM. is in turn causes irreversible damage to the articular cartilage, which contributes to the progression of OA [7][8][9]. erefore, inhibiting the oxidative stress-induced apoptosis of chondrocytes and ECM degradation may be an effective means of preventing cartilage degeneration and progression of OA. e nuclear factor erythroid derived-2-related factor 2 (Nrf2) is a major regulatory factor for antioxidant proteins. Under normal circumstances, the Kelch-like ECH-associated protein 1 (Keap1) binds to Nrf2 in the cytoplasm, hence, inhibiting the activation of Nrf2. However, stimulation through exposure to excessive ROS causes the disassociation of Nrf2-Keap1 and accumulation of Nrf2 in the nucleus.
ereafter, Nrf2 binds to the consensus sequences of the antioxidant response element (ARE), promoting the expression of antioxidant and phase 2 defense enzymes, including hemeoxygenase-1 (HO-1) and NADPH quinone oxidoreductase 1 (NQO-1) [10]. Moreover, the Nrf2 pathway is known to be associated with oxidative stress and may therefore be a potential pharmacological target for oxidative disorders [11][12][13].
Furthermore, tert-butylhydroquinone (TBHQ) is an aromatic organic compound that has been used for a long time as a food preservative. It was also reported that TBHQ is one of the most potent activators of the Nrf2/Keap1/ARE signalling pathway [14]. Additionally, TBHQ was proven to have strong antioxidant effects through the activation of the Nrf2 pathway, in in vitro or in vivo studies of some diseases such as alcoholic cardiomyopathy, acute hepatic injury, and neurodegenerative disease [15][16][17][18]. However, it is still unknown whether TBHQ can inhibit the apoptosis of chondrocytes and ECM degradation, induced by oxidative stress. erefore, the present study attempted to investigate the ability of TBHQ to inhibit the apoptosis of chondrocytes and ECM degradation, induced by oxidative stress. e possible underlying mechanisms were also assessed. e chondrocytes were passaged for the subsequent in vitro studies when the cells reached 80-90% confluence. Moreover, the oxidative stress cell model was induced using tert-butyl hydroperoxide (TBHP), whose concentration (20 μM) was selected based on previous literature [19].
Cell Viability Assay.
Cell viability was measured using the Cell Counting Kit-8 (CCK-8) assay (Beyotime). Briefly, the cells were seeded in 96-well plates at a density of 1 × 10 4 cells/well. ereafter, the cells were treated with 20 μM TBHP and different concentrations of TBHQ for 24 h. After treatment, the cells were incubated with 10 μl of the CCK-8 solution added in 100 μl of serum-free DMEM, at 37°C for 2 h. Afterwards, absorbance was evaluated at 450 nm using a microplate reader (Epoch; BioTek Instruments, Inc.).
Quantification of ROS Production.
e levels of ROS in chondrocytes were evaluated using the ROS sensitive dye, 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA; Sigma-Aldrich). Briefly, chondrocytes were collected and washed thrice with phosphate-buffered saline (PBS, Gibco). ereafter, the cells were stained with 10 μM of DCFH-DA for 30 min. Afterwards, the cells were washed thrice with a serum-free medium to remove the residual extracellular DCFH-DA. Finally, the levels of ROS were evaluated using the BD Accuri C6 Plus flow cytometer (BD Biosciences, Vianen, e Netherlands).
Detection of Mitochondrial Membrane Potential.
e mitochondrial membrane potential of chondrocytes was assessed using a JC-1 staining kit (Beyotime). Briefly, the cells were collected and washed three times with PBS. ereafter, the cells were stained with JC-1 (5 μg/ml) for 20 min at 37°C. e cells were then washed again three times with PBS to remove the residual JC-1. Cells were observed under confocal microscopy (Olympus, FV3000) (magnification, ×200). In addition, the mitochondrial membrane potential of chondrocytes was evaluated using the BD Accuri C6 Plus flow cytometer, and the data were analyzed using the FlowJo software (FlowJo LLC, version 10.6.0).
Determination of Cell Apoptosis.
Annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) staining (BD Biosciences) was used to determine chondrocyte apoptosis. Briefly, the cells were collected and washed thrice with icecold PBS. ereafter, they were incubated with 5u Annexin V-FITC added in 300 μl of binding buffer for 25 min and then stained with 5 μl of PI added in 200 μl of binding buffer for 5 min, at room temperature in the dark. Chondrocyte apoptosis was then evaluated within 30 min using the BD Accuri C6 Plus flow cytometer, and the data were analyzed using the FlowJo software.
Determination of Malondialdehyde (MDA) and Superoxide Dismutase (SOD) Levels.
e levels of superoxide dismutase (SOD) and malondialdehyde (MDA) in chondrocytes were determined using the SOD and MDA assay kits (Beyotime), according to the manufacturer's instructions. Briefly, the cell lysates were treated with a working buffer for 30 min at 37°C. Absorbance was then measured at a wavelength of 523 nm (MDA) or 450 nm (SOD) using a microplate reader. Additionally, total protein concentration was measured using a bicinchoninic acid (BCA) protein assay kit ( ermo Fisher, Waltham, MA, USA) to normalize the MDA and SOD levels.
Real-Time PCR Assay.
Total RNA was extracted from chondrocytes using the RNA simple total RNA kit (Tiangen Biotech, Beijing, China). ereafter, single-stranded cDNA was synthesized from 1000 ng of the extracted RNA using a reverse transcript master mix (TaKaRa, Shiga, Japan). Afterwards, quantitative RT-PCR was performed using the PowerUp SYBR Green Master Mix on the QuantStudio 5 real-time PCR system (Applied Biosystems, Foster City, CA, USA). Moreover, the expression of GAPDH was used to normalize the Ct values. e forward and reverse primer sequences used in this study are given in Table 1.
Statistical Analysis.
Each experiment was independently performed at least three times. Statistical comparisons among multiple groups were then performed using one-way ANOVA followed by Bonferroni's multiple comparison test, in the SPSS software (version 20.0; ITBHQ Corp.). In addition, data were expressed as the mean ± standard deviation, and statistical significance was set at p < 0.05.
Effects of TBHQ on the Viability of Rat Chondrocytes.
e structural formula of TBHQ is shown in Figure 1(a). In this study, the cytotoxic effects of different concentrations of TBHQ (0, 0.625, 1.25, 2.5, 5, 10, 20, 40, 80, and 160 μM) on chondrocytes were evaluated through the CCK-8 assay. e results showed that TBHQ concentrations equal to or greater than 80 μM were cytotoxic to rat chondrocytes (Figure 2(a)). Moreover, 1.25, 2.5, 5, and 10 μM of TBHQ significantly improved the viability of rat chondrocytes exposed to 20 μM TBHP. Notably, 2.5 and 5 μM of TBHQ had the best effects ( Figure 2(b)) and were consequently used in the subsequent in vitro experiments.
TBHQ Prevented Excessive Generation of ROS, Increased the SOD Levels, and Decreased the MDA Levels.
e levels of ROS in rat chondrocytes were evaluated through DCFH-DA staining. Flow cytometry analysis demonstrated that treatment with 25 μM of TBHP increased the levels of ROS in chondrocytes, indicating that the oxidative stress model was successfully established. In addition, 2.5 and 5 μM of TBHQ effectively decreased the levels of ROS (Figures 1(b) and 1(c)). e results demonstrated that TBHQ effectively inhibited excessive generation of ROS as a result treatment with THBP. Furthermore, the levels of SOD and MDA were determined to evaluate oxidative damage. e results showed that treatment with THBQ significantly decreased the levels of MDA but increased the levels of SOD in rat chondrocytes exposed to 20 μM TBHP (Figures 1(f ) and 1(g)). ese suggested that THBQ effectively protected chondrocytes from excessive accumulation of ROS and caused oxidative stress.
TBHQ Prevented Oxidative Stress-Induced Mitochondrial Damage and Cell Apoptosis.
e study further used JC-1 staining to evaluate the mitochondrial membrane potential of rat chondrocytes. e monomers (green fluorescence) and aggregates (red fluorescence) were then quantified through flow cytometry. e results showed that the FITC/ PE intensity ratio was decreased in two TBHQ treatment groups, compared to that in the TBHP treatment category, indicating that TBHQ protected rat chondrocytes from mitochondrial damage induced by oxidative stress (Figures 3(a) and 3(b)). Furthermore, Annexin V/PI staining was used to determine the apoptosis of rat chondrocytes. e findings showed that treatment with 2.5 and 5 μM TBHQ significantly reduced the number of apoptotic chondrocytes, compared to treatment with TBHP.
is suggested that TBHQ protected rat chondrocytes from apoptosis and induced oxidative stress (Figures 1(d) and 1(e)).
TBHQ Ameliorates Oxidative Stress-Induced Matrix
Degradation. Western blotting and qPCR were used to Evidence-Based Complementary and Alternative Medicine 3 Evidence-Based Complementary and Alternative Medicine evaluate the expression of MMP3, MMP13, aggrecan, and collagen II in rat chondrocytes, at the mRNA and protein levels. e results showed that treatment with TBHQ decreased the mRNA and protein expression levels of MMP3 and MMP13 but enhanced the expression of aggrecan and collagen II in rat chondrocytes treated with 20 μM TBHP (Figure 4). e results therefore demonstrated that treatment with TBHQ ameliorates matrix degradation, induced by oxidative stress, in vitro.
TBHQ Activated the Nrf2 Signalling Pathway.
Western blotting was used to measure the expression levels of Nrf2, NQO-1, and HO-1. e results showed that treatment with 2.5 and 5 μM of TBHQ enhanced the expression levels of Nrf2, NQO-1, and HO-1, compared to treatment with TBHP. In addition, the relative expression levels of Nrf2 in the 2.5 μM TBHQ treatment group were higher than those in the TBHP treatment category, although the differences were not statistically significant ( Figure 5). e results suggested that treatment with TBHQ activated the Nrf2 pathway and therefore played an antioxidant role in rat chondrocytes.
Discussion
Osteoarthritis was reported to be strongly associated with increased oxidative stress in chondrocytes. In addition, Evidence-Based Complementary and Alternative Medicine oxidative damage induces the apoptosis of chondrocytes and degradation of the ECM, which finally leads to degeneration of articular cartilage and contributes to the progression of OA [4,9]. Moreover, there are currently no reliable drugs to prevent the progression osteoarthritis [2]. Existing research shows that TBHQ is as an effective antioxidant, although little information exists on its benefits in treating osteoarthritis. e present study therefore explored the ability of TBHQ to prevent the apoptosis of chondrocytes and degradation of the ECM, induced by oxidative stress, in vitro. e findings showed that 2.5 and 5 μM of TBHQ significantly improved the viability of rat chondrocytes treated with 20 μM TBHP. e results from DCFH-DA staining also revealed that 2.5 and 5 μM of TBHQ prevented TBHP-induced excessive production of ROS in chondrocytes. Moreover, treatment with 2.5 and 5 μM of TBHQ reduced the levels of MDA and increased the levels of SOD in rat chondrocytes treated with TBHP. ese findings suggested that TBHQ exerted antioxidative effects, in vitro. Notably, decrease in mitochondrial membrane potential is considered a nearly sign of cell apoptosis. Herein, 2.5 and 5 μM of TBHQ prevented the TBHP-induced decrease in mitochondrial membrane potential and dramatically reduced the rate of apoptosis in chondrocytes treated with THBP. Furthermore, PCR and Western blot analyses showed that 2.5 and 5 μM of TBHQ increased the mRNA and protein levels of aggrecan and collagen II but reduced the mRNA and protein levels of MMP3 and MMP9.
ese findings therefore showed that in addition to preventing cell apoptosis induced by oxidative stress, TBHQ also prevented the degradation of the ECM, caused by oxidative stress. Reactive oxygen species are oxygen-containing free radicals such as O − 2 , H 2 O 2 , and OH − . Additionally, ROS are essential for the maintenance of cellular homeostasis and function [20]. However, excessive accumulation ROS caused by multiple factors may irreversibly damage cells. Increasing evidence also shows that oxidative stress contributes to the pathogenesis of OA. Moreover, dysregulated expression of SOD was observed in the articular cartilages of patients with OA. It was also reported that NO and its derivative could suppress the synthesis of proteoglycan, thus increasing damage to cartilage during the progression of OA [8,21,22]. In the present study, the results showed that 2.5 and 5 μM of TBHQ significantly prevented the TBHP-induced apoptosis of rat chondrocytes and ECM degradation in vitro, suggesting that TBHQ can be used as an effective antioxidant to treat OA.
Additionally, Nrf2 is a master regulator of the intracellular antioxidant response. Notably, dissociation of Nrf2 from the Keap-1/Nrf2 complex and the nuclear translocation of Nrf2 are the essential signalling steps for the activation of Nrf2. In addition, the downstream molecules, NQO-1 and HO-1, are often activated to exert antioxidant effects after Nrf2 translocates to the nucleus. e clinical significance of Nrf2 is that it may be used as a pharmacological target, thereby, benefiting patients [11,12,23]. Furthermore, chondrocytes are the only resident cell types and the major producers of ROS in articular cartilage. Increased levels of apoptotic chondrocytes were also observed in patients with OA, suggesting that the apoptosis of chondrocytes is crucial in the pathogenesis of OA. Moreover, increase in oxidative stress is positively correlated with collagen TBHQ significantly increased the protein expression levels of Nrf2 and its downstream molecules, NQO-1 and HO-1. is suggested that TBHQ activated the Nrf2 pathway in rat chondrocytes, in vitro. * * P < 0.05 and * * * * P < 0.01. n.s, not significant.
degradation, indicating that ROS is an important factor affecting catabolism of the cartilage matrix. Oxidative stress-induced apoptosis of chondrocytes also promotes the expression of matrix-degrading proteases and reduces extracellular ECM synthesis, thus causing ECM degradation. erefore, preventing the apoptosis of chondrocytes and ECM degradation, induced by oxidative stress, may be a potential therapeutic strategy for OA. In addition, TBHQ is an aromatic organic compound that serves as an Nrf2 activator. It was also reported that TBHQ can protect various cells and organs against oxidative damage by activating Nrf2 signalling. Moreover, the antioxidant properties of TBHQ have been confirmed in patients or animal models of certain diseases [14,16,18,[24][25][26][27][28]. e present study demonstrated that TBHQ activated the Nrf2 pathway and upregulated HO-1 and NQO-1 to exert antioxidative effects in rat chondrocytes, consistent with previous reports on other diseases. However, it is not clear whether the Nrf2 pathway is the only signalling pathway through which TBHQ exerts its antioxidant effects. Further research is therefore needed to uncover other possible pathways or validate the present findings.
In summary, the current study showed that TBHQ effectively protected against TBHP-induced oxidative stress, thus inhibiting the apoptosis of rat chondrocytes and degradation of the ECM, by activating the Nrf2 pathway. However, the study only focused on cartilage degeneration. More studies are therefore needed to clarify the effects of TBHQ on synovial inflammation and subchondral bone remodeling. Moreover, further in vivo research is required to establish the effectiveness TBHQ in the treatment of OA. Nonetheless, the study suggests that TBHQ has potential for use in the treatment of OA.
Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.
==
Domain: Biology Medicine
|
[{"added": "2021-12-12T17:23:52.211Z", "created": "2021-12-08T00:00:00.000", "id": "245085372", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://downloads.hindawi.com/journals/ecam/2021/1905995.pdf", "pdf_hash": "2cb76c2d96f19fb02cd08859280cf5a0fd1f2c5a", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:1695", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "cc276cc08fb2d5357df6504d973f32876fef0652", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Exploring the Molecular Mechanisms of Eucommia Ulmoides Extract in Treating Atherosclerosis Based on Network Pharmacology and Molecular Docking Technology
: [Objective] This study aimed to investigate the mechanism of action of Eucommia ulmoides extract in treating atherosclerosis using network pharmacology and molecular docking technology. [Methods] The researchers screened the chemical active ingredients of Eucommia ulmoides using the TCMSP platform and collected the corresponding targets of these active ingredients using the Swiss Target Prediction platform. They searched for relevant targets of atherosclerosis in the DisGeNET, GeneCards, TTD, OMIM, Drug Bank, and Pharmgkb databases and obtained common targets between the two using the Venny platform. These common targets were then used to construct a protein interaction network and a core target network using Cytoscape 3.8.0, and topological parameters were obtained. GO and KEGG pathway enrichment analyses were performed using DAVID, and the results were visualized using bioinformatics and Omicshare platforms. The researchers also constructed and analyzed the "active ingredient-target-disease" network model using Cytoscape 3.8.0 software. Finally, AutoDock and Pymol software were used for molecular docking of important components and core targets to predict their binding ability. [Results] The researchers screened 27 active ingredients from Eucommia ulmoides, and found 206 common targets between atherosclerosis and active ingredients. These targets were mainly related to the Protestations in cancer, Prostate cancer, and Endocrine resistance pathways, among which AKT1 had the highest affinity for β-sitosterol. [Conclusion] This study suggests that multiple active ingredients in Eucommia ulmoides extract may exert various effects in the treatment of atherosclerosis. The findings provide a scientific basis for further development of clinical applications of Eucommia ulmoides extract.
Introduction
With the improvement of living standards and changes in lifestyle, atherosclerosis (AS) has become one of the main causes of cardiovascular disease, posing a serious threat to people's health. AS is an arterial disease based on endothelial injury and characterized by lipid deposition, and is one of the main causes of cardiovascular disease. The occurrence and development of atherosclerosis is a complex process involving multiple biochemical and molecular biology mechanisms. Currently, clinical methods for treating AS mainly include lifestyle changes, lipid-lowering drugs, antioxidants, and vasodilators [1] , but these methods have problems such as significant side effects and poor efficacy. Therefore, finding a safe and effective treatment method is very important.
Eucommia ulmoides (Du Zhong in Chinese) is a precious Chinese herbal medicine with a long history of use and medicinal value. It is known as the "southern ginseng" and contains rich polyphenols and flavonoids, with effects such as strengthening tendons and bones, nourishing the liver and kidneys, invigorating the spleen and stomach, and anti-aging [2] . In recent years, studies have shown that Eucommia ulmoides extract has a certain therapeutic effect on atherosclerosis, but its specific molecular mechanism is not clear.
Network pharmacology can systematically analyze the interaction between drugs and biomolecules, predict the mechanism and indications of drugs, and analyze their side effects and interactions. Molecular docking technology can simulate the interaction between drug molecules and biomolecules, and predict the optimization direction of drug molecules. The combination of network pharmacology and molecular docking technology can provide comprehensive and accurate support for drug development. Therefore, this study will use network pharmacology and molecular docking technology to explore the molecular mechanism of Eucommia ulmoides extract in the treatment of atherosclerosis, providing new ideas for further research on the molecular mechanism of Eucommia ulmoides extract in the treatment of atherosclerosis.
Screening of active ingredients and target proteins of Eucommia ulmoides
We searched the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) ( [URL]) for active ingredients of Eucommia ulmoides with an oral bioavailability (OB) greater than or equal to 30% and drug-likeness (DL) greater than or equal to 0.18. Then, we used PubChem database ( [URL]/) and Open Babel software to obtain the SMILES structure of these active ingredients and uploaded them to the Swiss Target Prediction platform ( [URL]/) to collect the corresponding target proteins of these active ingredients. We standardized the gene information of these active ingredients through UniProt database ( [URL] set the species as human to obtain the potential target proteins of these active ingredients.
Acquisition of target genes related to atherosclerosis
We searched for target genes related to atherosclerosis using six databases, including DisGeNET, GeneCards, TTD, OMIM, Drug Bank, and Pharmgkb, which contain disease-related gene information. After merging and removing duplicate targets, we obtained the relevant targets for atherosclerosis.
Determination of the common targets of active components of Eucommia ulmoides and atherosclerosis
We used the above steps to obtain the potential target proteins of active components and disease-related target proteins of atherosclerosis. Then, we uploaded them to Venny2.1.0( [URL]/) to draw a Venn diagram to determine the common targets. The intersection of the obtained targets represents the potential targets of active components of Eucommia ulmoides that may act on the related targets of atherosclerosis.
Construction of protein interaction network and core target network
We imported the shared targets of Du-zhong active ingredients and atherosclerosis into the STRING platform Version 11.0, and constructed a Du-zhong protein interaction network for the treatment of atherosclerosis using Cytoscape software. We used its plugin MCODE to obtain the core target network.
GO functional enrichment analysis and KEGG pathway enrichment analysis
GO analysis includes three parts: biological process (BP), cellular component (CC), and molecular function (MF), which collectively describe the functions of gene products. We used the online tool DAVID to perform GO enrichment analysis on the 206 potential targets of Duzhong extract for the treatment of atherosclerosis, and obtained corresponding entries based on three plugins (GOTERM_BP_DIRECT, GOTERM_CC_DIRECT, GOTERM_MF_DIRECT). We imported the common target genes into the DAVID database ( [URL] GO and KEGG pathway enrichment analysis, and selected the biological processes and pathways with P<0.05 based on three plugins (GOTERM_BP_DIRECT, GOTERM_CC_DIRECT, GOTERM_MF_DIRECT) and sorted them by the number of involved targets from high to low. We used R studio for image visualization processing, drew GO functional enrichment and KEGG pathway enrichment diagrams, and finally combined with the KEGG database to draw key pathway mechanism diagrams.
Construction and analysis of the "Active ingredient-Target-Disease" network model
We imported the information of traditional Chinese medicine, chemical components, their corresponding target proteins, and pathway information into Cytoscape 3.8.0software to construct the "Active ingredient-Target-Disease" network.
"Active ingredient-core target" molecular docking
The core target protein was selected from the Uniprot database, and the best protein structure was downloaded from the PDB ( [URL] based on the protein's tertiary structure characteristics. The 2D structure of the active ingredient was obtained from PubChem and converted to a 3D structure mol2 file. Then, in the PyMOL ( [URL], the ligands and non-protein molecules in the protein were removed, and the protein was processed by adding hydrogen, adding charge, and merging non-polar hydrogen, and saved as a PDBQT file. Finally, molecular docking was completed using AutoDock Tools 1.5.6 and AutoDock Vina software, and the results were visualized. We collected 28 effective chemical active ingredients from the TCMSP database, as shown in Table 1. A total of 1,110 target proteins corresponding to the active ingredients were collected through the Swiss Target Prediction platform. The target genes were converted to standard gene names using the UniProt database, and genes without "homo sapiens" and "reviewed" UniProt IDs were removed, resulting in 506 potential target proteins for Eucommia ulmoides ingredients. Then, we screened 2,698 gene proteins through the DisGeNET, GeneCards, TTD, OMIM, DrugBank, and PharmGKB databases, and after removing duplicate target genes, 2,088 genes remained. We used Venny 2.1 to take the intersection and obtained 206 common targets, as shown in Figure 1.
Analysis of protein-protein interaction network and core target network of Du Zhong active ingredients and atherosclerosis
We imported the 206 shared gene targets into the STRING database and obtained 489 protein-protein interaction information (see Figure 2). Then, we imported the protein-protein interaction information into Cytoscape and used network topology analysis to select the targets whose node degree was greater than the average value of 6.62 as the core targets for Du Zhong treatment of atherosclerosis. We obtained 117 core targets and used the CytoNCA plugin to obtain the core target network (see Figure 3). The size and color of the nodes represent the degree value, with larger nodes indicating higher degree values and darker colors indicating stronger correlation with the mechanism of Du Zhong treatment for atherosclerosis. Among them, SRC, AKT1, HSP90AA1, MAPK3, PIK3CA, MAPK1, ESR1, PTPN11, EGFR, and others may be related to the treatment of atherosclerosis by Du Zhong extract.
GO functional enrichment analysis and KEGG pathway enrichment analysis
The GO analysis results show that the biological processes of Du Zhong extract in treating atherosclerosis mainly include regulation of inflammatory response, response to molecules of bacterial origin, response to lipopolysaccharides, vascular processes in the circulatory system, and cellular response to chemical stress. The cellular components mainly include membrane rafts, membrane microdomains, and membrane regions. The molecular functions mainly include steroid binding, protein tyrosine kinase activity, and nuclear receptor activity (see Figure 4 for details). The KEGG results indicate that Du Zhong extract mainly regulates signaling pathways such as Proteoglycans in cancer, Prostate cancer, and Endocrine resistance, acting on atherosclerosis (see Figure 5 for details).
"Active ingredient-target-disease" network model construction
The results of the model show that active ingredients can act on multiple targets. For example, in Eucommia ulmoides, both chlorogenic acid and quercetin, as well as β-sitosterol, can act on atherosclerosis through multiple targets (see Figure 6 for details). The core targets SRC and AKT1 were docked with their corresponding active ingredients to predict their binding ability. The smaller the binding free energy, the greater the affinity between the receptor and the ligand, and the more stable the conformation (see Table 2). The results were visualized using PyMOL software (see Figure 7). The results showed that all docking binding energies were less than 0, indicating that the molecular docking between all core targets and active ingredients had relatively small binding free energies and good affinity, with the highest affinity observed between β-sitosterol and AKT1.
Discussion and analysis
Du Zhong, as a traditional Chinese medicine, has the effects of warming the kidney and assisting yang, activating blood circulation and dredging collaterals in the theory of traditional Chinese medicine. In recent years, more and more studies have shown that Du Zhong extract also has certain pharmacological effects in the treatment of atherosclerosis. However, the mechanism of Du Zhong extract in the treatment of atherosclerosis is still unclear. Network pharmacology and molecular docking technology provide powerful tools for exploring the molecular mechanism of Du Zhong extract in the treatment of atherosclerosis.
In this study, we explored the molecular mechanism of Eucommia ulmoides extract in treating atherosclerosis through network pharmacology and molecular docking techniques. Our results indicated that Eucommia ulmoides extract contains multiple active components, and network pharmacology is a research method based on systems biology, computer science, and network science. By constructing a drug-target-disease network, we investigated the multi-target effects and complex mechanisms of Eucommia ulmoides extract. We found that several active components of Eucommia ulmoides extract, such as naringenin, quercetin, and β-sitosterol, can act on multiple targets including SRC, AKT1, HSP90AA1, MAPK3, and PIK3CA to exert various biological effects. Among them, targets such as SRC, AKT1, and HSP90AA1 are closely related to the occurrence and development of atherosclerosis. SRC is a non-receptor tyrosine kinase that is involved in smooth muscle cell proliferation and migration [3] . In addition, SRC plays an important role in regulating macrophage function, particularly in mediating inflammatory and immune responses. The adhesion, migration, and inflammation of macrophages are important links in the development of atherosclerosis. SRC is involved in regulating the inflammatory signaling pathway mediated by all members of the Toll-like receptor (TLR) family, indicating its potential role in macrophage inflammation in atherosclerosis [4] . AKT1 can control cell signal transduction, energy metabolism, immune function, and thrombosis formation by regulating multiple signaling pathways [5] . HSP90AA1 is a heat shock protein that is a member of the HSP90 family, and it is associated with the process of lipid metabolism and cholesterol transport, which also play important roles in the occurrence and development of atherosclerosis [6][7] . A study found [8] that inhibiting the expression of HSP90 can reduce lipid accumulation and the expression of cholesterol transport proteins, thereby reducing the degree of atherosclerosis. This provides a new explanation for the pharmacological activity of Eucommia ulmoides extract. Furthermore, we found through molecular docking that proteoglycans in cancer, prostate cancer, endocrine resistance, platelet activation, and the Rap signaling pathway may be related to the occurrence and development of atherosclerosis. Therefore, the multi-target effects of Eucommia ulmoides extract can regulate various biological effects and thus play a therapeutic role in the treatment of atherosclerosis.
It should be noted that these predictions and docking results only serve as a theoretical basis for guiding subsequent experimental research and must be further validated through experimental verification. Some researchers have attempted to extract active compounds from Eucommia ulmoides extract, such as flavonoids and glycosides, and conducted pharmacological studies [9] . One study showed that the Eucommia ulmoides glycoside in the extract can inhibit the expression of inflammatory mediators such as IL-6, TNF-α, and NF-κB [10] . In addition, studies have also shown that certain components in Eucommia ulmoides extract can affect the activity of enzymes such as cytochrome P450, thereby affecting drug metabolism and toxicity. This suggests that the active ingredients in Eucommia ulmoides extract may affect the pathogenesis of atherosclerosis through multiple pathways, including inhibiting cholesterol synthesis and TG degradation, and reducing inflammation [11][12] .
Overall, Eucommia ulmoides extract may affect the pathogenesis of atherosclerosis through multiple pathways. The application of network pharmacology and molecular docking technology provides researchers with an effective screening tool to identify active compounds in Eucommia ulmoides extract and predict their role in the pathogenesis of atherosclerosis. However, the predicted results must be experimentally verified for feasibility. In addition, further in-depth studies are needed on the pharmacological properties and toxicity of active compounds in Eucommia ulmoides extract in order to better utilize them in the treatment of related diseases such as atherosclerosis. Dulate these pathways and contribute to the treatment of atherosclerosis.
Figure 1 :
Figure 1: The intersection of the targets of action of Eucommia and the targets of atherosclerosis
Figure 2 :
Figure 2: Protein PPI network between active constituents of Eucommia globulus -atherosclerosis shared target proteins
Figure 3 :
Figure 3: Core target network among the target proteins of Du Zhong active components and atherosclerosis.
Figure 4 :
Figure 4: GO function enrichment chart
Figure 6 :
Figure 6: Network model of "active ingredients-targets-diseases" : Docking of piceatannol with SRC; B: Docking of piceatannol with AKT1; C: Docking of piceatannol with HSP90AA1;D: Docking of Quercetin with SRC; E: Docking of Quercetin with AKT1; F: Docking of Quercetin with HSP90AA1;G: Docking diagram of β-sitosterol with SRC; H: Docking diagram of β-sitosterol with AKT1; I: Docking diagram of β-sitosterol with HSP90AA1.
Table 1 :
Basic information on the 28 active chemical constituents in Eucommia globulus
Table 2 :
Combined capability component analysis
==
Domain: Biology Medicine
|
[{"added": "2023-04-26T15:02:41.757Z", "created": "2023-01-01T00:00:00.000", "id": "258328503", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "http://www.clausiuspress.com/assets/default/article/2023/04/24/article_1682344519.pdf", "pdf_hash": "e9904e7918ff9b3c4993ff71f65e77781cd7888e", "pdf_src": "Anansi", "provenance": "peS2o-0028.json.gz:4191", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "71027855c5294a61de257e5a2677b4fb7976dd5f", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Combination of VP3 and CD147-knockdown enhance apoptosis and tumor growth delay index in colorectal tumor allograft
Cancer therapies that kill cancer cells without affecting normal cells is the ultimate mode of treating cancers. The VP3, an avian virus-derived protein, can specifically initiate cell death through several signal transduction pathways leading to apoptosis. In cancer, chemoresistance and cell survivability implicate the cell surface protein, CD147. In this study, transfection of VP3 and silencing of CD147 genes was achieved through the treatment of tumors with pVIVO1-GFP/VP3 (VP3), psiRNA-CD147/2 (shCD147/2), and their combination of CT26 colon cancer cell-induced in mice. The effectiveness of tumor-treatment was ascertained by electrophoresis, TUNEL assay, and flow cytometry analysis. While histopathological and biochemical analysis were used as toxic side effect identification. The tumor growth delay index (TGDI) after treatment with VP3, shCD147/2, and their combination treatments increased by 1.3-, 1.2-, 2.0- and 2.3-fold respectively, over untreated control. The VP3-shCD147/2 combination treatment was more efficacious then either VP3 or shCD147/2 alone in the retardation of mouse CT26 colorectal cell tumor allograft. The antitumor effect of the combination treatment is the result of synergistic effects of VP3 and shCD147/2 on the tumor cells resulting in apoptosis. Thus, the study shows that combination of VP3 and shCD147/2 treatment can be developed into a potential approach for anticolorectal cancer treatment regimen.
Background
Colorectal cancer is the third most common cancer cases after lung (1.82 million) and breast (1.67 million) cancers [1]. Within the next 15 years approximately 1.4 million new cases of colorectal cancer are expected to occur with an estimated death of 693 000 that would account for 8.5 % of all cancer deaths [2]. Poor survival rate of colon cancer patient is partly due to poor understanding of the disease and its progression, invasion, migration and metastasis [3].
Basigin/CD147, a transmembrane glycoprotein of the immunoglobulin superfamily, is expressed widely on many cell types and highly expressed in various tumor cells [4]. CD147 play an important role in proliferation, angiogenesis, invasiveness and metastatic activity of malignant melanoma [5]. Increased expression of CD147 was shown to correlate with enhanced tumor progression and poor prognosis in different cancers [6][7][8]. Thus, an attractive way to curb tumor progression is through suppression of the CD147-dependent cell proliferation, invasion and metastasis by RNAi-mediated silencing [5,9] and eventually induce cell apoptosis due to detachment of anchorage-dependent cell from the surrounding extracellular matrix [10,11].
In the development of anti-cancer compounds, apoptosis is the preferred mode of cancer cell death. Viral protein of the avian anemia VP3/Apoptin has a positively charged C-terminus that is reported to induce apoptosis selectively on transformed and tumor cells, leaving normal cells intact [12,13]. VP3 in the murine tumor model was shown to be anti-tumorigenic, mostly through induction of apoptosis [14,15]. The ability of VP3 in inducing p53-independent apoptosis has been demonstrated in more than 70 tumor cell lines [16]. Although simultaneous VP3, interleukin-24 [17], interleukin-18 [18], upregulations and survivin downregulation [19] seemed to show greater anti-tumor activity than VP3 alone, the combined effect of VP3 and shRNA on CD147 affecting tumor growth and progression is yet to be investigated. In this study the combined effect of pVIVO1-GFP/VP3 and psiRNA-CD147/2 was examined in the attempt to discover a new therapeutic approach for colorectal cancers.
Animals
Female, 5 to 6 week-old BALB/c mice were obtained from Institute Medical Research (IMR, Malaysia) and were acclimatized for a week prior to use. All mice were kept in individually ventilated cages (IVC) with constant rotation rate of 70 air-changes/h. Mice were fed on sterilized commercial diet, given water ad libitum and subjected to 12 h light and dark cycle. The study was performed with approval of the Institutional Animal Care and Use Committee, Universiti Putra Malaysia (UPM/FPV/PS/3.2.1551/AUP-R103).
Animal colon cancer model
The mice were anesthetized with 40 mg ketamin plus 8 mg xylazine/kg bwt intraperitoneally and placed on 37°C warming pad. Cell suspension containing 1 × 10 6 CT26 cells in 0.2 mL sterile PBS were subcutaneously injected on the right flank of the mice with minimal trauma. The mice were observed on alternate days for tumor development and palpable tumors were measured. Treatments of the mice were instituted when the tumors reached sizes of approximately 50 mm 3 or ≥200 mm 3 . Each control and treatment group comprise of 3 mice.
Measurement of tumor growth and evaluation of antitumoral effect
Tumor volume was determined by measuring the greatest length and width using calipers, and calculated by using the following formula [21]: Evaluation of antitumoral effect was determined according to Sanceau J. et al. [22]. Individually relative tumor volume (RTV) was defined as follows: where V x is the volume (mm 3 ) at a specific time and V 1 is the volume at the beginning of treatment. Treatment efficacy was expressed as the percentage of tumor growth inhibition (TGI) as follows: where T and C is the mean RTV of treated and control group at the time of sacrifice, respectively. Tumor growth delay (TGD) was determined as the time required for the tumor volume to reach 10-fold over the initial volume. Tumor growth delay index (TGDI) was calculated as follows: where TGD T and TGD C is the mean TGD of the treated and control group, respectively.
Protocol II: Combination treatment
Mice with tumor size of ≥200 mm 3 were treated intratumorally with 100 μg of treatment in 70 μl of sterile PBS according to the following regimens: Control groups were either a) non-treated, b) received 3 doses of 100 μg pVIVO1-GFP/LacZ or c) 3 doses of 100 μg psiRNA-h7SKzeo. Treatment mice received either a) 3 doses of 100 μg of pVIVO1-GFP/VP3 or b) 3 doses of 100 μg of psiRNA-CD147/2. In combination therapy, mice received either a) 3 doses of 50 μg of pVIVO1-GFP/VP3 in combination with 3 doses of 50 μg of psiRNA-CD147/2 or b) 3 doses of 100 μg of pVIVO1-GFP/VP3 in combination with 3 doses of 100 μg of psiRNA-CD147/2, representing low and high dose treatments, respectively. In combination study, mice received pVIVO1-GFP/VP3 and psiRNA-CD147/2 treatments alternately, while in control and single treatment, mice received doses at alternate days. Tumor growth was examined on alternate days for 25 days post-treatment. Blood was collected from all mice prior to sacrifice and tumor tissues were fixed either in 10 % neutral buffered formalin for hematoxylin and eosin staining and immunohistochemical analysis or flash-frozen in liquid nitrogen and stored at -80°C for molecular analysis. The serum creatinine, blood urea nitrogen (BUN), alkaline phosphatase (ALP), alanine transaminase (ALT) and aspartate transaminase (AST) were determined spectrophometrically using standard commercial kits (Roche, Swizerland).
DNA fragmentation analysis
In this analysis the genomic DNA (gDNA) from frozen tumor tissues were isolated using DNAzol (Molecular Research Centre, Inc, USA) in accordance with manufacturer's protocol. Briefly, 50 mg of tumor tissue was rinsed with PBS. DNAzol-tumor tissue homogenization was performed using pre-cleaned pestle and mortar. The homogenate was then centrifuged for 10 min at 10 000 × g to sediment the remaining insoluble tissues. Thereafter, the viscous supernatant was transferred to new microcentrifuge tube and apoptotic DNA fragments were precipitated using 100 % absolute ethanol at 6 000 × g for 6 min. After centrifugation, the DNA pellet was rinsed 2 times with 70 % ethanol by inverting a few times and sediment at 1000 × g for 1 min. Finally, DNA pellet was air dried and resuspended in sterile dH 2 O. DNA fragmentation was determined by 2 % agarose gel electroporation in 1 × TBE buffer and run at 80 V for 45 min. The DNA was stained with ethidium bromide and visualized under UV transilluminator. Apoptotic cells were appeared as a ladder pattern while necrotic as a smear pattern on the gel. Intact genomic DNA appeared as a band at the top of the lane.
Terminal deoxynucleotidyl transferase-mediated nick end-labeling assay
Apoptotic endonucleases cleave DNA to produce fragments with 3'-OH groups that can be detected on tumor sections stained with FragEL DNA Fragmentation Detection Kit-Klenow Enzyme (Calbiochem, USA) and recorded digitally using light microscopy (Nikon Elipse TE2000-S, Nikon, Japan) at × 200 magnification. Approximately, 5-10 random images were taken for each group (n = 3). Briefly, fixed tumor tissues were dehydrated, cleared, infiltrated and paraffin embedded. Tissue sections of 4 μm were prepared using rotary microtome and mounted onto glass slides, deparaffinized, rehydrated and treated according to manufacturer's procedure. Apoptosis was determined by stained nuclei with brown color after labelled with DAB. Tumor sections were counterstained with methyl green. TUNEL-positive cells were counted and analyzed using Image J software (ImageJ 1.43u, USA), and the apoptotic index (AI) was calculated as percentage of TUNEL-positive cells per total number of cells.
Flow cytometry
To further ascertain that the treatment caused tumor cell death through apoptosis rather than necrosis, the tumor cells were subjected to flow cytometry after staining with annexin V-FITC and propidium iodide. The technique allows for differentiation between living, apoptotic, and necrotic cells. Apoptotic cells were further differentiated into those in early and late apoptosis. This method detects the translocation of the negatively charged phospholipid phosphatidylserine (PS) on cell membrane surface during the early stages of apoptosis. Single cell suspensions were subjected to flow cytometry following Annexin V-FITC and propidium iodide (PI) staining using the ApopNexin™ FITC Apoptosis Detection Kit (Chemicon, USA). For single cell preparation, tumor tissues were placed on sterile petri dish and washed 3 times with PBS. Tumor tissues were cut into small pieces (1-2 mm 3 in size) and then carefully disintegrated with fine forceps in 2 ml of PBS. Cells were then transferred into a 15 ml conical centrifuge tube and resuspended gently and rapidly in 10 ml of ice-cold PBS. The cells suspension was then centrifuged at 170 × g for 1 min (4°C) to sediment the remaining tissue fragments. The supernatant containing single cells was transferred into a new 15 ml conical centrifuge tube and centrifuged at 170 × g for 10 min. The supernatant was discarded and pellet was resuspended in ice-cold PBS at a concentration of 1 × 10 6 cells/mL and kept on ice. Then, tumor cells suspensions were centrifuged to remove PBS and resuspended in ice-cold 1× Binding Buffer (10 mM Hepes/NaOH; pH 7.4, 140 mM NaCl and 2.5 mM CaCl 2 ) at 1 × 10 6 cells/mL. 200 μL of cells suspension were aliquoted in polystyrene round-bottom tube and stained with 3 μL of Annexin V-FITC. 2 μL of 100× PI solutions were added to the Annexin V-FITC-labelled cells and the suspension incubated at room temperature in the dark for 15 min and analyzed immediately using the Becton Dickinson FACS Calibur equipped with Cell-Quest Pro software. Cells labelled as FITC + /PI − are in early apoptosis; cells labelled as FITC − /PI + are necrotic or broken; cells labelled as FITC + /PI + are either in late apoptosis or secondary necrosis; and cells negatively labelled as FITC − /PI − are viable.
Statistical analysis
The results are expressed as mean ± standard error of the mean. The data were analyzed by either Student's paired t-test or ANOVA followed by Tukey multiple comparison post hoc test. The P value of <0.05 was considered significant.
Colon cancer mice model
A small tumor mass of 48.41 ± 1.28 mm 3 was palpable at day 10 day post-implantation. The tumor was allowed to grow achieving a size of 210.9 ± 7.26 mm 3 by 15 days post-implantation.
Complete regression by VP3 or shCD147/2 treatment in small size tumor In mice with tumor size approximately 50 mm 3 , VP3 as well as shCD147/2 treatments showed regression in tumor volume from 47.0 ± 3.5 and 48.4 ± 2.2 mm 3 before treatment to 7.1 ± 1.1 and 31.1 ± 7.5 mm 3 on the day 7 post-treatment, respectively (Fig. 1). By that time the tumor volume in untreated control mice increased to 336.5 ± 9.2 mm 3 and continued to increase rapidly reaching a volume of 4 000 mm 3 by day 20. In mice treated with either VP3 or shCD147/2, the tumor remained smaller than 50 mm 3 and finally regressed completely by day 30 post-treatment.
VP3-shCD147/2 combination treatment increased tumor growth delay index (TGDI)
In mice with tumor size of ≥200 mm 3 , both low and high dose VP3 plus shCD147/2 combination treatments caused more significant (p < 0.05) antitumor effects than either VP3 or shCD147/2 alone (Fig. 2). In combination treatments, the tumor size decreased from the initial size of 200.9 ± 14.2 and 211.3 ± 16.6 mm 3 to 150.3 ± 25.1 and 166.1 ± 24.6 mm 3 on day 3 post-treatment for low and high dose, respectively. Treatments with either VP3 or shCD147/2 did not reduce tumor size, on day 3. In fact the tumor increased slightly in size from 216.5 ± 13.8 to Relative tumor volume, which is the size of tumor at a given time compared with the initial size, is shown in Fig. 3. The growth began slowly and began to accelerate from approximately 12 days post-treatment. However, in untreated mice the tumor grew rapidly reaching much higher relative volume than in treated mice at day 25 post-treatment. Tumors treated with VP3, shCD147/2, low and high dose combination undergo 40.0, 45.2, 51.1 and 60.3 % of growth inhibition, respectively. TGDI, indicating treatment efficiency is calculated as the delay in days taken by the treated tumors to reach a 10-fold RTV divided by the delay in the control group. The TGDI of tumors treated with VP3, shCD147/2, low and high dose combination treatments increased by 1.3-, 1.2-, 2.0-and 2.3-fold respectively, from the initial volume. Thus, the results showed that combination VP3 and shCD147/2 treatments were more effective than VP3 or shCD147/2 alone. Fig. 2 Representative photographs of tumors taken at day-1 before treatment and day-3 after treatment. C UT = Untreated control, C LacZ = pVIVO1-GFP/LacZ control, C zeo = psiRNA-h7SKzeo control, T VP3 = Treated with pVIVO1-GFP/VP3 (VP3), T shCD = Treated with psiRNA-CD147/2 (shCD147/2), T V-shCD(50) and T V-shCD(100) = Treated with low and high dose of VP3-shCD147/2 combination respectively. Circles show significantly size reduction in combinative treatment
Biochemical analysis of colon cancer mice model after treatment
Serum liver enzymes and kidney function parameter concentrations were estimated to determine safety of plasmids pVIVO1-GFP/VP3 (VP3) and psiRNA-CD147/ 2 (shCD147/2) as therapeutic compounds. Except for AST, neither liver enzymes nor kidney function parameters showed significant (p > 0.05) difference between treatments and untreated control (Fig. 4). However, AST, which in this case reflects muscle integrity were significantly (p < 0.05) higher in mice tumor treated with LacZ and zeo.
VP3 overexpression and knockdown of CD147 induced a cellular morphologic change in CT26 tumor cells
Under H&E staining, tumor tissues in all mice treated with VP3, shCD147/2 or their combination showed typical features of apoptosis to include interstitial spaces, apoptotic bodies, and dark nuclei. In contrast, control, LacZ-and zeo-treated mice did not show similar cellular morphology. Tumor of treated mice also showed fewer mitotic events than those of the controls. Tissue section from tumors treated with shCD147/2 also showed numerous blood vessel ruptures (Fig. 5).
VP3, shCD147/2 and combinations treated tumors characterized by DNA laddering
Treatment with VP3, shCD147/2 or their combination at 72 h post-treatment lead to high intensity laddering indicating apoptotic activity (Fig. 6). This observation was most obvious in tumors treated with 100 μg VP3-shCD147/2 combination. The gels from untreated control, LacZ-and zeo-treated tumors showed smeared bands suggesting complete DNA lysis indicating necrosis.
Enhanced apoptotic events in VP3-shCD147/2 combination treated tumors
Apoptotic endonucleases cleave nuclear DNA to produce fragments with 3'-OH groups that can be detected on tumor sections. Apoptotic cells were observed as dark brown nuclear staining while viable cells stained green color (Additional file 1: Figure S1). Tumors treated with VP3, shCD147/2 and their combinations showed numerous TUNEL-positive cells indicating apoptosis (Fig. 7a). There were more than 60 % apoptosis in the treated tumors compared to <1 % in the untreated and control tumors. The apoptotic index (AI) of tumors treated with VP3 and shCD147/2 were 63.9 ± 4.0 and 62.1 ± 4.2 % respectively, while in the 50 and 100 μg VP3-shCD147/2 combination treatments, the AI was 74.7 ± 0.4 and 92.1 ± 3.5 % respectively (Fig. 7b).
Discussion
In this study, transfection of colorectal tumors with VP3 gene in combination with psiRNA-CD147/2-induced CD147 silence as cancer gene therapy for colorectal cancers in the CT26 colorectal cancer cell-induced mouse model was investigated. The CT26 cell line is a rapidgrowing grade IV carcinoma that can readily undergo metastasis [23]. For that reason, the CT26 mouse tumor is one of the most extensively used model in the investigation of colorectal carcinomas [24]. Intra-tumoral administration of pVIVO1-GFP/VP3 was shown to cause significant reductions in tumor size in these mice [25,26]. The viral-vectored VP3 has also been shown to cause regression and complete remission of the xenograft of human hepatomas grown in mice [27].
To determine the therapeutic effect of VP3 protein on CT26 tumors, a sustainable and tumor-inducible GRPpromoter was used to enhance the VP3 expression in a targeted cell population [28][29][30]. When the tumor was treated with recombinant pVIVO1-GFP/VP3 there was complete regression of tumor showing that recombinant plasmids harboring VP3 can be anti-tumorigenic. Partial regression of CT26 tumors can be induced by CD147 silencing as shown in mice with human colon cancer xenograft [31] and can be achieved with psiRNA- Fig. 4 Serum liver enzymes and kidney function parameters in treated mice with CT26 colon cancer cell-induced tumor. a Liver enzymes; ALT = alanine transaminase, ALP = alkaline phosphatase, AST = aspartate transaminase and b kidney function parameters; creatinine and urea. Data are mean ± SEM. *P < 0.05 compared to the untreated group. UT = Untreated control, LacZ = pVIVO1-GFP/LacZ, zeo = psiRNA-h7SKzeo, VP3 = pVIVO1-GFP/VP3 and shCD147/2 = psiRNA-CD147/2 CD147/2. Although pVIVO1-GFP/VP3 and psiRNA-CD147/2 are both effective antitumor agents, our study showed that pVIVO1-GFP/VP3 is superior to psiRNA-CD147/2. Further, when pVIVO1-GFP/VP3 was used in combination with psiRNA-CD147/2, the antitumor effect was enhanced. This observation suggests that pVIVO1-GFP/VP3 and psiRNA-CD147/2 act synergistically in causing tumor regression. It is proposed that the synergistic effect is attributed to the tumor CD147 silencing causing inhibition of tumor cell proliferation and invasion, and proapoptotic VP3 gene transfected into the tumor cells through the use of pVIVO1-GFP/ VP3 [32].
Induction of apoptosis is the mode of cell death targeted by most antitumor agents. Treatments with pVIVO1-GFP/VP3, psiRNA-CD147/2 and their combination were shown to cause apoptosis of CT26 mouse tumor cells. The antitumor effect of pVIVO1-GFP/VP3 and psiRNA-CD147/2 was rapid and remained constant for the period of the study in the case of pVIVO1-GFP/VP3 treatment or eventually waned when psiRNA-CD147/2 was used. When pVIVO1-GFP/VP3 and psiRNA-CD147/2 were administered as combination treatment, apoptosis of tumor cells was slow to occur; however, after 25 days the combination in fact killed the majority of tumor cells. The mode of tumor cell death was apoptosis and this was supported by histopathology, where tumor tissues showed abundance of apoptotic features. On the contrary, there was an abundance of mitotic features in the untreated and control tumor tissue indicating rampant tumor growth. One of the effects of psiRNA-CD147/2 is the triggering of indirect endothelial damage in the tumor tissues causing collapse of tumor vasculature. The net effect is poor blood flow and deprivation of oxygen supply to the tumors tissue culminating in tumor cell death.
The effectiveness of tumor-treatment was also ascertained by electrophoresis, TUNEL assay, and flow cytometry analysis. Normally, apoptotic DNA cleavage produced a signature pattern with high and low molecular weight fragments [33][34][35]. The presence of multiples of 180 to 200 bp DNA fragments indicated that treatment with pVIVO1-GFP/VP3, psiRNA-CD147/2 and their combination had caused apoptosis of tumor cells. CD147 knockdown eventually sensitized tumor cells to anoikis, which is a form of apoptosis induced by the detachment of anchorage-dependent cells from the surrounding extracellular matrix [10,36]. In this study, the intensity of DNA ladder of the tumor cells treated with pVIVO1-GFP/VP3 was equivalent to that produced by those treated with psiRNA-CD147/2. However, the DNA ladder intensity was higher in tumors treated with pVIVO1-GFP/VP3-psiRNA-CD147/2 combination.
Another method used to assess for apoptosis is the in situ terminal deoxynucleotidyl transferase biotin-dUTP nick end labeling assays (TUNEL). The TUNEL assay was purposely used to detect nuclear DNA fragmentation by identifying generation of nicks and breaks in the DNA strands [37]. On the other hand, this assay was used in quantifying and comparing the number of TUNEL-positive cells between tumor samples; VP3-, shCD147/2-and combination-treated. The highest extent of apoptosis indicate by AI was in tumors treated with pVIVO1-GFP/VP3-psiRNA-CD147/2 combination compared to tumors from VP3 or shCD147/2 treated alone.
Flow cytometry analysis allowed the sensitive detection of apoptotising cells. The apoptosis percentage in the pVIVO1-GFP/VP3 treated tumor was sustained at day-3 and day-25 due to the GRP promoter sustainable effect. Meanwhile, apoptosis percentage in the psiRNA-CD147/ 2 treated tumor was high at day-3 and then declining at day-25 because vasculature rupture at the first few days causes anoikis which is interpreted as late-apoptosis; however the treatment becomes less effective with time. The apoptosis percentage in the combinatively pVIVO1-GFP/VP3-psiRNA-CD147/2 treated tumor was markedly increased compared to individually treated samples at day-25 post-treatment. The effects of shCD147/2 are triggering indirect damage to the pre-existing tumoral endothelium, results in collapse of the vasculature inside solid tumors. Thus, the tumor cells is deprived of oxygen supply or blocked from blood flow, which consequently leads to enhancement of pro-apoptosis induction by pVIVO1-GFP/VP3. The pVIVO1-GFP/VP3-psiRNA-CD147/2 combination treatment seemed to synergise the effects of pVIVO1-GFP/VP3 and psiRNA-CD147/2 by intensifying antitumor effect in prolonged treatment. Tumor growth is exponential in the early stages, then becomes less aggressive, and plateaus at the late stages of the disease [38]. Since psiRNA-CD147/2 is very effective early and pVIVO1-GFP/VP3 has a consistent effect throughout tumor development, the pVIVO1-GFP/VP3-psiRNA-CD147/2 combination treatment would be the more efficacious antitumor regimen than either pVIVO1-GFP/VP3 or psiRNA-CD147/2 alone.
Chemotherapy is plagued with side-effects, thus new drugs or therapeutic regimens require toxicity testing. In this study, the effect of pVIVO1-GFP/VP3, psiRNA-CD147/ 2 and their combination on the liver and kidneys were ascertained by determining serum ALT, ALP, AST, urea and creatinine concentrations. With the exception of slightly elevated AST concentration, all blood biochemical parameters were normal indicating that the liver and kidneys were not affected by the treatments. Increase in AST may be associated with some muscle damage or increased muscular activities that are not associated with the toxic effect of the treatments. Thus, pVIVO1-GFP/VP3, psiRNA-CD147/2 or their combination is generally nontoxic and safe to be in mice, but proper Pharmacokinetics (PK) analyses are required to confirm prior to clinical work in humans.
Conclusions
The antitumor effect of the combination treatment is the result of synergistic effects of VP3 and shCD147/ 2 on the tumor cells resulting in apoptosis. Based on these studies, we conclude that the pVIVO1-GFP/ VP3-psiRNA-CD147/2 combination therapy is potentially effective and safe regimen for treating colorectal cancers.
==
Domain: Biology Medicine
|
[{"added": "2023-01-16T14:14:05.003Z", "created": "2016-07-13T00:00:00.000", "id": "255831982", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1186/s12885-016-2530-8", "pdf_hash": "27c07fa4aa5735a5f3dff3a5c86773c42ee1d952", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:3322", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "27c07fa4aa5735a5f3dff3a5c86773c42ee1d952", "year": 2016}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Ixodes scapularis dystroglycan-like protein promotes Borrelia burgdorferi migration from the gut
The causative agent of Lyme borreliosis, Borrelia burgdorferi, is transmitted by Ixodes ticks. During tick feeding, B. burgdorferi migrates from the tick gut to the salivary glands from where transmission to the host occurs. B. burgdorferi-interacting tick proteins might serve as vaccine targets to thwart B. burgdorferi transmission. A previous screening for B. burgdorferi-interacting Ixodes scapularis gut proteins identified an I. scapularis putative dystroglycan protein (ISCW015049). Here, we describe the ISCW015049’s protein structure and its cellular location in the tick gut in relation to B. burgdorferi migration. Secondly, in vivo B. burgdorferi–tick attachment murine models were performed to study the role of ISCW015049 during B. burgdorferi migration and transmission. In silico analysis confirmed that ISCW015049 is similar to dystroglycan and was named I. scapularis dystroglycan-like protein (ISDLP). Confocal microscopy of gut tissue showed that ISDLP is expressed on the surface of gut cells, is upregulated during tick feeding, and is expressed significantly higher in infected ticks compared to uninfected ticks. Inhibition of ISDLP by RNA interference (RNAi) resulted in lower B. burgdorferi transmission to mice. In conclusion, we have identified a dystroglycan-like protein in I. scapularis gut that can bind to B. burgdorferi and promotes B. burgdorferi migration from the tick gut. B. burgdorferi exploits tick proteins to orchestrate its transmission to the host. B. burgdorferi is able bind to an I. scapularis dystroglycan-like protein (ISDLP). Inhibition of ISDLP in ticks results in lower B. burgdorferi transmission to mice. ISDLP is a potential target to prevent Lyme borreliosis. B. burgdorferi exploits tick proteins to orchestrate its transmission to the host. B. burgdorferi is able bind to an I. scapularis dystroglycan-like protein (ISDLP). Inhibition of ISDLP in ticks results in lower B. burgdorferi transmission to mice. ISDLP is a potential target to prevent Lyme borreliosis.
Introduction
In the USA, Ixodes scapularis is the vector of Borrelia burgdorferi, the causative agent of Lyme borreliosis [1,2]. B. burgdorferi colonization of the tick gut can occur when uninfected Ixodes larvae acquire B. burgdorferi when feeding on a B. burgdorferi-infected animal [3]. B. burgdorferi anchors itself to the tick gut wall by expressing outer surface protein A (OspA), which binds to the tick receptor OspA (TROSPA) [4]. When a B. burgdorferi-infected I. scapularis nymph feeds on a vertebrate host, B. burgdorferi becomes metabolically active, changes its outer surface proteins, and migrates from the tick gut to the salivary glands [5]. Motility Electronic supplementary material The online version of this article (doi:10.1007/s00109-015-1365-0) contains supplementary material, which is available to authorized users. of B. burgdorferi appears not to be essential for exiting the gut, as described in a newly proposed model called Badherence-mediated migration^by Dunham-Ems et al. [6]. Dunham-Ems and colleagues observed that during tick feeding, B. burgdorferi spirochetes initially replicate in the lumen of the gut and remain non-motile. After approximately 24 h, B. burgdorferi spirochetes transition into aggregates at the basal lamina of the gut. From here, a small percentage of B. burgdorferi penetrate the gut, followed by migration via the hemolymph to the salivary glands into the skin of the host [6].
The close interaction of B. burgdorferi with tick gut epithelial cells suggests potential interactions between the tick gut proteins and B. burgdorferi proteins that might be critical for B. burgdorferi growth in the tick gut and its egress from the gut-a critical step for successful transmission to the vertebrate host. B. burgdorferi-interacting tick gut proteins might thus be a vaccine targeted to prevent spirochete migration from the gut and preempt transmission. The advantage of using tick gut proteins as antitick vaccines is that migration of B. burgdorferi can be targeted early in tick feeding-even before spirochetes have been transmitted to the host. Of note, a vaccine against Lyme disease is currently not available for humans [7]. Since the last decade, tick salivary gland proteins and tick gut proteins have become a target of vaccine development as the tick plays a central role in B. burgdorferi transmission [8]. Immunization against salivary gland proteins introduced into the skin that facilitate tick feeding provided (partial) protection against B. burgdorferi transmission [9][10][11] as well as immunization against tick gut proteins [4,12]. One limitation of vaccine targeting tick gut proteins is that gut proteins might not provide an anamnestic response, since they are not presented to the host during a tick bite. Nonetheless, the future development of cocktail vaccines combining tick gut proteins that facilitate spirochete migration from the gut and salivary antigens that facilitate survival at the bite site might provide a robust impairment of B. burgdorferi transmission by simultaneously targeting spirochete egress from the gut and survival at the bite site. Recently, we used a yeast surface display (YSD) approach to screen for B. burgdorferi-interacting tick gut proteins and identified three putative B. burgdorferi-interacting gut proteins: ISCW008121, ISCW015049, and ISCW015135 [12]. One of them, ISCW008121, was shown to be a transmembrane fibronectin domain-containing I. scapularis protein that enabled B. burgdorferi adherence to the basal lamina of the gut to facilitate transmission [12]. In this study, we report the characterization of ISCW015049 and examine its vivo role in the context of B. burgdorferi transmission and assess its ability to serve as a transmission-thwarting vaccine.
Animal experiments
The rabbit immunized against recombinant I. scapularis dystroglycan-like protein (rISDLP) and mice used in the RNA interference (RNAi) experiments were housed and handled under the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The animal experimental protocol was approved by the Yale University's Institutional Animal Care and Use Committee (protocol number 2008-07941, approval date: 31 March 2014). All animal infection experiments were performed in a biosafety level 2 animal facility, according to the regulations of Yale University. In addition, the mice that were used for immunization experiments were housed and handled under the approval of the Animal Care and Use Committee of the University of Amsterdam (DIX103179).
Ticks
I. scapularis nymphs and larvae were obtained from a tick colony at the Connecticut Agricultural Experiment Station in New Haven CT, USA, and ticks were maintained as described earlier [13]. Ticks were allowed to feed to repletion and RNA isolated from guts and salivary glands using TRIzol (Invitrogen, CA, USA) as described earlier [22]. Complementary DNA (cDNA) was synthesized using the iScript RT-PCR kit (Bio-Rad, CA, USA) and analyzed by quantitative PCR for the expression of tick actin and B. burgdorferi flab using gene-specific primers (Online Resource 1) and the iQ SYBR Green Supermix (Bio-Rad, Hercules, CA, USA).
Identification of the full-length transcript of ISCW015049 and purification of rISDLP
First-strand cDNAwas synthesized from the total I. scapularis gut RNA using a 3′-RACE adapter. The RLM-RACE kit was used to identify the sequence at the 3′-end and 5′-end according to the manufacturer's instructions (Invitrogen, CA, USA). The identification of the sequence at the 5′-end was performed using ISDLP-specific primers (Online Resource 1). The fulllength sequence was assembled using the Web-based software SMART [10] ( [URL]). Purification of rISDLP was performed as described previously [8].
Confocal microscopy
Confocal microscopy to detect native ISDLP was performed as previously described [12]. Briefly, guts from nymphal ticks (B. burgdorferi-infected or uninfected) were dissected and fixed in 4 % paraformaldehyde (PFA). Washed guts were incubated with rabbit anti-rISDLP antibody and bound antibodies detected using fluorescein isothiocyanate (FITC)-labeled affinity purified goat anti-rabbit IgG antibody (Sigma, MO, USA) and nuclei stained with propidium iodide or with TO-PRO-3 iodide (Invitrogen, CA, USA). Control guts were incubated with IgG purified from rabbit anti-ovalbumin sera. Stained guts were visualized under a Zeiss LSM 510 Confocal microscope.
Pixel intensity quantification
Pixel intensities in the tetramethylrhodamine (TRITC) channel (as a measure of anti-rISDLP serum binding to tick gut ISDLP) or in the FITC channel of confocal images were quantified using the ImageJ 1.47t software. Confocal images of four individual guts were examined in each control and experimental group, and mean pixel intensities representing the average intensity of pixels in the region of interest were obtained in five different regions of each tick gut.
Immunization of rabbits and mice against ISDLP
Four-to six-week old New Zealand white rabbits were immunized subcutaneously with 30 μg of rISDLP or ovalbumin in complete Freund's adjuvant (CFA) and boosted twice with 30 μg of rISDLP or ovalbumin at weeks 3 and 6 in incomplete Freund's adjuvant (IFA). Test bleeds were obtained from ear veins 2 weeks after the final boost, and reactivity to recombinant rISDLP and ovalbumin was assessed by Western blot. Rabbits were euthanized, and serum was obtained by cardiac puncture. Polyclonal IgG was purified from the sera using the Melon Gel IgG purification kit (Thermo Scientific, IL, USA).
For immunization of mice, animals were immunized with 10 μg of rISDLP or ovalbumin in CFA and boosted twice with 10 μg of rISDLP or ovalbumin at weeks 2 and 4 in IFA. To address the role of rISDLP in B. burgdorferi transmission, eight B. burgdorferi N40-infected nymphs were placed on each immunized mouse. Nymphs were allowed to feed to repletion. Salivary glands and guts were dissected and combined in pools of two to three ticks for quantitative reverse transcription polymerase chain reaction (RT-PCR) as described earlier [9]. DNA was isolated from skin punch biopsies at 7, 14, and 21 days and from heart and joints 21 days post-tick detachment, and Borrelia burden was assessed by quantitative PCR as described [9].
RNAi silencing of isdlp in B. burgdorferi-infected I. scapularis nymphs RNAi silencing of isdlp in ticks was performed as described before [9] using primers specific for isdlp with a T7 promoter sequence (Online Resource 1). Double-stranded (ds) isdlp double-stranded RNA (dsRNA) was synthesized using the MEGAscript RNAi kit (Ambion/Invitrogen, CA, USA). ds isdlp RNA or ds gfp RNA (5 nl, 3×10 12 molecules/ml) was injected into the anal pore of Borreliainfected nymphs as described earlier [9]. dsRNAinjected ticks were allowed to feed until repletion and weighed to assess feeding efficiency, and guts and salivary glands were dissected for messenger RNA (mRNA) isolation and quantitative RT-PCR as described above. B. burgdorferi burden in mice was assessed by quantitative PCR as described earlier [9].
Statistical analysis
The significance of the difference between the mean values of the groups was analyzed using a non-parametric two-tailed Mann-Whitney test or a two-tailed Student's t test with the Prism 5.0 software (GraphPad Software, San Diego, CA, USA), and p≤0.05 was considered significant.
Full-length ISCW015049 encodes a potential transmembrane dystroglycan-like protein
The protein sequence present in the YSD screening yeast colony BClone 3^-matched amino acids (aa) 1 to 201 of the protein annotated on VectorBase as Bputative dystroglycan, ISCW015049.^The complete sequence of ISCW015049 was confirmed from the I. scapularis gut extract using 3′-end and 5′-end RLM-RACE. Using mRNA from guts of fed I. scapularis nymphs, the start and stop codons of ISCW015049 were identified and the complete transcript of 2904 bp was sequenced (Fig. 1a). Identification of the full sequence revealed that the full length of ISCW015049 was 968 aa long, since one fragment of ISCW015049 that annotated as an intron (www. vectorbase.org) was actually found to be part of the ISCW015049 transcript. In silico analysis showed that fulllength ISCW015049 has a potential transmembrane domain and is similar to dystroglycan, a widely distributed protein involved in the linkage between the extracellular matrix and the cytoskeleton [14]. The protein sequence of fulllength ISCW015049 was 27.0, 28.7, and 26.7 % identical to Drosophila melanogaster dystroglycan (NP_725523.3), Homo sapiens dystroglycan (AA81779.1), and Mus musculus dystroglycan (NP_001263423.1), respectively. Online programs for protein modeling (SMART) predicted that ISCW015049 has three dystroglycan-type cadherinlike (CADG) domains and two dystrophin-associated glycoprotein 1 (DAG1) domains (Fig. 1b). The 201 aa region identified in the YSD screen was predicted to be located on the extracellular region, which consists of one CADG domain and a C-terminal domain of αdystroglycan (Fig. 1b, c). Based on the similarity with dystroglycan, full-length ISCW015049 is henceforth referred to as the I. scapularis dystroglycan like protein (ISDLP) and has been submitted to GenBank (accession number KR782315).
Production of recombinant ISDLP and confirmation of binding to B. burgdorferi
We expressed the full-length protein transcript of ISDLP in a D. melanogaster expression system. Binding of purified recombinant ISDLP (rISDLP) (Fig. 1d) to B. burgdorferi was confirmed by an ELISA-based binding assay. We observed a dose-dependent increase of rISDLP binding to B. burgdorferi membrane extract compared to the control tick protein rTSLPI (Fig. 1e). ; and a C-terminal domain of α-dystroglycan that was identified at aa 111-236. c A 3D model of ISDLP was created using the Phyre2 software ( [URL]2) [26]. Six hundred seventy-six residues (70 %) were modeled at >90 % accuracy. d Purified Drosophila-expressed recombinant full-length ISDLP electrophoresed on SDS 10 % polyacrylamide gel and stained with Coomassie blue. e ELISA assessment of dose-dependent binding of ISDLP to B. burgdorferi membrane protein extract-coated plates compared to rTSLPI, a tick protein that is known not to bind to B. burgdorferi
Localization of native ISDLP in the tick gut by confocal microscopy
To study the expression and protein localization of native ISDLP, we generated antibodies against rISDLP by immunizing a rabbit with rISDLP. Unfed and partially fed (24 and 48 h) B. burgdorferi and uninfected I. scapularis guts were collected, fixed with PFA, and probed with anti-ISDLP rabbit serum or anti-ova rabbit serum as a control. Binding of antibodies in the gut was visualized with immunofluorescence confocal microscopy. Comparing the binding of anti-rISDLP antibodies at the different time points by mean pixel intensity using the ImageJ software showed that ISDLP expression increases significantly during tick feeding (Fig. 2a-c, e). Furthermore, increased binding was observed in B. burgdorferi-infected ticks compared to uninfected ticks. In line with our computer-based modeling of ISDLP (Fig. 1), Z-stack imaging suggested that ISDLP is represented both on the cell surface and in the cytosol (Fig. 2d).
Immunization against rISDLP does not prevent B. burgdorferi transmission
To test whether immunization against rISDLP would prevent B. burgdorferi transmission, we immunized eight mice against rISDLP and eight mice against ovalbumin. We achieved good IgG titer levels against rISDLP in the sera of mice that were immunized against rISDLP after immunization with complete Freund's adjuvants and two boosters with incomplete Freund's adjuvant (Fig. 3a). Two weeks after the second IFA boost, we placed eight ticks per mouse which were allowed to feed until repletion, ranging from 3 to 5 days. No difference was found in post-feeding tick weight compared to the ovalbumin group (Fig. 3b), nor did we observe a difference in B. burgdorferi migration to the salivary glands or in B. burgdorferi transmission to the host by RT-qPCR (Fig. 3c). Based on qPCR analysis, B. burgdorferi loads in skin tissue from the tick bite site (ears) as well as deeper tissue were similar between the rISDLP-immunized animals and the ovalbumin-immunized animals (Fig. 3d, e).
isdlp RNAi reduces B. burgdorferi migration to the salivary glands and transmission to the murine host It is likely that in the active immunization experiment, the function of ISDLP in relation to B. burgdorferi migration and transmission remained unaffected by murine anti-ISDLP antibodies present in the gut during tick feeding. To circumvent the use of antibodies to provide an insight in the role of ISDLP during B. burgdorferi transmission, we performed another experiment in which isdlp expression in ticks was silenced by RNAi. We injected double-stranded (ds) isdlp RNA or ds gfp RNA as a control in B. burgdorferi-infected I. scapularis nymphs before placing four to five ticks on C3H/ H3N mice. The decrease of isdlp expression in the tick gut was confirmed by RT-qPCR (Fig. 4a). No difference in tick engorgement weights was observed between isdlp and the control ticks, indicating that reduced isdlp expression does not influence successful tick feeding (Fig. 4b). In contrast with the active immunization, significantly lower B. burgdorferi loads were detected in the salivary glands (Fig. 4c). Furthermore, qPCR on DNA from skin tissue showed significantly lower B. burgdorferi numbers in mice on which isdlp silenced ticks fed compared to the control group (Fig. 4d). However, B. burgdorferi loads were not significantly different at a later time point (t=14 days) or in deeper tissues such as joint and heart, indicating that B. burgdorferi growth in the murine host is not affected by the reduced expression of isdlp in the tick (Fig. 4e).
Discussion
During tick feeding, while the tick gut adapts to cope with the uptake of blood, B. burgdorferi becomes metabolically active, replicates, and binds to hypertrophic and differentiating gut cells in order to cross the gut barrier [6,15]. The molecular mechanisms that direct the growth and migration from the tick gut and entry into salivary glands are only beginning to unfold [5]. B. burgdorferi has been shown to bind to host and tick proteins to facilitate its survival and dissemination [16]. To better understand vector-B. burgdorferi interactions, we performed a YSD screening to identify I. scapularis gut proteins that interact with B. burgdorferi. We identified four B. burgdorferi-interacting tick proteins of which one, Ixofin3D, has been previously described [12]. Here, we characterize one of the other three proteins, referred to as ISDLP, and assess its role in B. burgdorferi transmission.
Computer-based protein structure and function predictions showed that ISDLP is similar to the conserved transmembrane protein dystroglycan. Recombinant ISDLP binds to B. burgdorferi and is abundantly expressed on the surface of gut epithelial cells during tick feeding, which was in accordance with previous assessment of ISDLP expression by RT-qPCR [12]. The function of ISDLP for I. scapularis has not yet been described. In other organisms, dystroglycan is part of the dystrophin-associated protein complex and is cleaved post-translationally into two subunits, αand β-dystroglycan, that together form the dystroglycan complex [14]. The dystroglycan complex can bind to the extracellular matrix by binding to laminin. Furthermore, studies have shown that the dystroglycan complex is involved in cell adhesion-mediated signaling, tissue remodeling, and cell polarity and that βdystroglycan is involved in MAPK signaling [17,18]. The functional role of ISDLP on cell metabolism, cell signaling, or tissue remodeling during tick feeding remains to be defined.
RNAi-mediated decrease in the expression of ISDLP reduced B. burgdorferi transmission to the murine host. While B. burgdorferi load in the gut was not altered, B. burgdorferi load in the salivary glands was significantly reduced, suggesting that ISDLP might have a role in B. burgdorferi migration from the gut. There have been no previous reports on B. burgdorferi interactions with human dystroglycan, which is identified as a receptor for a number of viruses as well as for Mycobacterium leprae [19,20]. B. burgdorferi is known to bind to extracellular matrix proteins such as decorin and fibronectin among others [15,21], and we speculate that human dystroglycan could be a ligand for B. burgdorferi in humans Fig. 2 ISDLP is a membrane-bound protein increasingly expressed in the gut during tick feeding. Gut nuclei and ISDLP were stained with TO-PRO-3 (blue) and anti-rISDLP rabbit serum (TRITC, red), respectively. a-c Confocal microscopy of guts of unfed (24 and 48 h) and fed uninfected and B. burgdorferi-infected I. scapularis nymphs. Magnification ×20. Guts stained with anti-ovalbumin IgG (TRITC, red) served as antibody control. d A Z-stack (magnification ×63) of a B. burgdorferi-infected gut after 48 h of feeding gut. e Mean pixel intensities of regions of interest in the TRITC channel (representing anti-rISDLP rabbit serum binding to ISDLP) of the confocal images obtained in a-c, as measured by the ImageJ software. Each data point represents one region of interest. The error bars represent mean±SEM, and the mean values significantly different in a two-tailed non-parametric Mann-Whitney test are indicated by an asterisk (p<0.05) or by three asterisks (p<0.0001) Fig. 3 Immunization against ISDLP does not have an effect on B. burgdorferi transmission to murine skin. Mice were actively immunized with rIxofin3D-PF or ovalbumin. Eight B. burgdorferi N40 ticks/mouse were placed and fed until repletion. Mice were sacrificed after 14 days of B. burgdorferi infection. a Mean IgG titer in the serum from animals vaccinated against rISDLP, diluted to 1:10 2 to 1:10 7 on ELISA-coated plates with rISDLP. The error bars represent mean± SEM. The cutoff for titer was calculated as an OD value of ova-immune serum+3 SD. b Engorgement weights of ticks post-feeding. Each data point represents one tick. A tick was considered female when >3.5 mg. c RT-qPCR assessment of B. burgdorferi burden in tick guts and salivary glands. d, e qPCR assessment of B. burgdorferi burden in the murine skin at 7 days and in the skin, bladder, and heart at 14 days posttick feeding. The experiment has been performed once with eight mice per group or other vertebrates. The specific binding partner of B. burgdorferi that binds ISDLP and the mechanism by which ISDLP promotes B. burgdorferi migration and transmission remain to be understood. Although silencing of ISDLP by RNAi did not impair tick feeding, it cannot be excluded that the effect on B. burgdorferi transmission is the result of ISDLP-mediated processes in the feeding gut, e.g., altered gut tissue remodeling or a reduced barrier.
Based on our findings that ISDLP interacts with B. burgdorferi and that ISDLP is expressed during tick feeding, we speculated that ISDLP could be a target to impair transmission. Tick gut antigens could be useful to target B. burgdorferi migration from the gut and derail transmission early in the process, i.e., Bnipping it in the bud.^While RNAimediated interference of ISDLP expression decreased B. burgdorferi transmission, active immunization against ISDLP did not impair B. burgdorferi migration to the salivary glands and did not reduce transmission to the murine host. There are several explanations for the discrepancy between our RNAi experiment and immunization experiment. While RNAi-mediated silencing is initiated prior to and during tick feeding, sufficient antibody uptake by the tick from the host might take more than 24-36 h coincident with the arrival of blood meal into the tick gut [22]. Thus, delayed entry of antibodies might allow B. burgdorferi to exploit the gut ISDLP and continue its migration from the gut. In order to prevent B. burgdorferi egress from the gut, anti-ISDLP antibodies have to significantly neutralize ISDLP's interaction with B. burgdorferi. Immunofluorescence microscopy suggests that ISDLP is ubiquitously represented on the tick gut. Affinity of the antibody binding as well as amounts of antibody that enter the gut would determine the successful Fig. 4 Silencing of ISDLP expression by RNA interference results in decreased B. burgdorferi burden in the salivary glands and in murine skin. Double-stranded isdlp (ds isdlp) or ds gfp as a control was injected through the anal pore 3 h prior to B. burgdorferi-infected I. scapularis challenge (five ticks/mouse, five mice per experiment). Mice were sacrificed after 14 days. a RT-qPCR assessment of isdlp expression in the gut. b Engorgement weights of ticks post-feeding. Each data point represents one tick. A tick was considered female when >3.5 mg. c B. burgdorferi burden in tick guts and salivary glands post-feeding. d, e qPCR assessment of B. burgdorferi burden in the murine skin at 7 days and in the skin, bladder, and heart at 14 days post-tick feeding. The error bars represent mean±SEM, and the mean values that were significantly different in a two-tailed non-parametric Mann-Whitney test are indicated by an asterisk (p≤0.05). The pooled results of two independent mouse experiments are shown abrogation of B. burgdorferi-ISDLP interaction. In addition, other possibilities that impaired a protective effect of anti-ISDLP antibodies could be the inaccessibility of the protective epitope within the tick or the inability of the generated antibodies to block the interaction between ISDLP and B. burgdorferi. The latter possibility is supported by in vitro observations that binding of ISDLP to B. burgdorferi could not be blocked by anti-ISDLP antibodies (Online Resource 2). More research to identify specific regions of ISDLP that interact with B. burgdorferi as well as the B. burgdorferi ligand that interacts with ISDLP would be informative for effectively blocking ISDLP-B. burgdorferi interaction through antibodies.
While several tick proteins with pharmacological functions critical for tick feeding and B. burgdorferi transmission have been identified [23], vaccine targeting has been confounded by the functional and structural paralogy of the tick transcriptome [24]. Recent efforts have increased our understanding of tick-B. burgdorferi interactions that facilitate B. burgdorferi migration within the tick. Vaccines targeting these B. burgdorferi-interacting tick proteins provide another avenue to interrupt B. burgdorferi transmission. It is becoming evident that B. burgdorferi exploits multiple tick proteins to temporally and spatially orchestrate its migration from the gut, entry into salivary glands, and transmission to the host. Elucidation of B. burgdorferi-interacting tick proteins that facilitate the various aspects of transmission would help design an optimal combination of vaccine targets that would provide a synergistic impairment of transmission to the vertebrate host.
==
Domain: Biology Medicine
|
[{"added": "2022-11-19T15:12:37.277Z", "created": "2015-11-23T00:00:00.000", "id": "253636987", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://link.springer.com/content/pdf/10.1007/s00109-015-1365-0.pdf", "pdf_hash": "29c07b6bb24380ff328ca916a8a252eb72cf8b51", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:7259", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "29c07b6bb24380ff328ca916a8a252eb72cf8b51", "year": 2015}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Unbiased PCR-free spatio-temporal mapping of the mtDNA mutation spectrum reveals brain region-specific responses to replication instability
The accumulation of mtDNA mutations in different tissues from various mouse models has been widely studied especially in the context of mtDNA mutation-driven ageing but has been confounded by the inherent limitations of the most widely used approaches. By implementing a method to sequence mtDNA without PCR amplification prior to library preparation, we map the full unbiased mtDNA mutation spectrum across six distinct brain regions from mice. We demonstrate that ageing-induced levels of mtDNA mutations (single nucleotide variants and deletions) reach stable levels at 50 weeks of age but can be further elevated specifically in the cortex, nucleus accumbens (NAc), and paraventricular thalamic nucleus (PVT) by expression of a proof-reading-deficient mitochondrial DNA polymerase, PolgD181A. The increase in single nucleotide variants increases the fraction of shared SNVs as well as their frequency, while characteristics of deletions remain largely unaffected. In addition, PolgD181A also induces an ageing-dependent accumulation of non-coding control-region multimers in NAc and PVT, a feature that appears almost non-existent in wild-type mice. Our data provide a novel view of the spatio-temporal accumulation of mtDNA mutations using very limited tissue input. The differential response of brain regions to a state of replication instability provides insight into a possible heterogenic mitochondrial landscape across the brain that may be involved in the ageing phenotype and mitochondria-associated disorders.
Background
Ageing is characterised by diverse molecular, physiological, and behavioural changes, the mechanistic onset of which is poorly understood. Time-dependent accumulation of mitochondrial DNA (mtDNA) mutations through a vicious cycle of reactive oxygen species (ROS) production and oxidative damage to mtDNA has been proposed as a possible driver [1], causing ageinginduced mitochondrial dysfunction. However, accumulating evidence suggests that mtDNA is not associated with oxidative damage [2,3], despite the presence of other oxidative damage in mitochondria [4,5]. Though it has been highly debated whether mtDNA mutations are cause or effect of ageing [3], mtDNA mutations alone are adequate to drive ageing as seen in the mutator mouse, an ageing model expressing a proof-readingdeficient mitochondrial DNA polymerase Polg, Polg D257A [6]. Evidence suggests the involvement of specific mtDNA mutations in oxidative stress [7,8], indicating that the 37 genes encoded by mtDNA (13 proteins, 22 tRNAs, and 2 rRNAs) as well as the non-coding control region (NCR) have different contributions to mitochondrial dysfunction. As the NCR interacts with the inner mitochondrial membrane [9] and several proteins [10][11][12] to form the nucleoid structure and regulate transcription and replication, mitochondria may be highly dependent on the NCR for proper function. What remains unclear is the exact spatio-temporal accumulation of mtDNA mutations between individual organs and even more importantly, the heterogeneity with which mtDNA mutations may accumulate across an organ.
We have developed a variation of the mutator mouse that expresses an alternative proof-reading-deficient Polg, Polg D181A , under the control of the CaMKIIα-promoter, resulting in forebrain neuron-specific expression of the transgene [13]. This model can be used to examine the neuron-specific mitochondrial response to replication instability arising from lack of proof-reading of Polg, and we have previously demonstrated the accumulation of dysfunctional mitochondria in these mice [14,15].
Up until now, purified mitochondria from whole organs [16] or some form of either partial or nearly fulllength PCR amplification of mtDNA [17][18][19] have been required for next-generation sequencing. Especially in highly heterogeneous tissue such as the brain, the cellular composition, metabolic profile, and local environment could potentially harbour regional mutational hotspots that may contribute to both the ageing phenotype and various disorders, but such limited tissue regions have been too small to study without PCR amplification. PCR amplification not only introduces a bias to the study of mtDNA, it also makes it difficult to identify rearrangements of mtDNA such as deletions and duplications unless full-length amplification of mtDNA is performed.
In this study, we implemented a method to prepare mtDNA for next-generation sequencing without PCR amplification prior to library preparation using DNA extracted from small brain dissections from mice during ageing. We show that mice accumulate single nucleotide variants (SNVs) and deletions with ageing across all brain regions in a largely homogenous manner. However, the expression of Polg D181A causes a highly brain region-specific increase in the ageing-induced accumulation of SNVs and deletions. Our data demonstrate a previously undescribed bimodal distribution of deletion sizes across both genotypes. In addition, we demonstrate a Polg D181A -dependent and ageing-induced accumulation of NCR-containing multimers. In all, our unbiased approach to map the full spatio-temporal mtDNA mutation spectrum provides an unprecedented method to gain insight into brain-wide mitochondrial heterogeneity.
Results
Implementation of a method for isolation of mtDNA from small tissue samples To investigate the full mtDNA mutation spectrum in small mouse brain regions and avoid the inherent bias present in PCR amplification, we implemented a method for the enzymatic depletion of nuclear DNA (nDNA) from total DNA [20] extracted from brain tissue. We enzymatically depleted nDNA from total DNA by treatment with exonuclease V, an enzyme that targets the free ends of linear DNA essentially leaving circular mtDNA intact [20]. We used mtDNA-enriched samples directly for library preparation for next-generation sequencing (Fig. 1a).
We allowed for split-read mapping to identify deletions with BBMap [21] (Fig. 1b) using a custom mtDNA reference composed of two mm10 MT references in tandem (dMT). In a two-round mapping approach, we removed residual nDNA-derived sequencing reads, especially due to the presence of nuclear mitochondrial DNA segments (Numts), i.e. mtDNA-like sequences in the nuclear genome. Due to the circularity of mtDNA (Fig. 1c), deletions may span the "ends" of the mtDNA reference, that is linear in nature, which will interfere with deletion calling. By using dMT, we circumvented this and were able to reliably detect variants at any position in mtDNA. As sequencing reads generated by Nextera are well known to exhibit GC bias in the first bases of the read, we trimmed these bases and excluded an additional 5 bp at the read ends during variant calling (see the "Methods" section). Based on this, we have no reason to believe that variant calling is influenced by transposase sequence bias. For identification of mtDNA variants, we sampled the sensory cortex (COR), caudate putamen (CP), dorsal raphe (DR), nucleus accumbens (NAc), paraventricular thalamic nucleus (PVT), and substantia nigra (SN) (Fig. 1d) during mouse ageing and confirmed nDNA depletion by qPCR before library prep and sequencing (Fig. 1e).
Ageing-related accumulation of SNVs and deletions across all brain regions
We initially mapped the ageing-related changes in mtDNA mutations across the brains of 10-, 50-, and 80week-old wild-type (WT) mice (Fig. 1f). As expected, the number of SNVs in 10-week-old mice was very low but rose on average 10-fold in 50-week-old mice and remained relatively unchanged at 80 weeks (Fig. 1f, left).
Similarly, deletions also reached a plateau at 50 weeks after increasing 2.5-3-fold from 10 weeks (Fig. 1f, right). The plateau reached in both SNVs and deletions at 50 weeks indicates a restriction in the load of mtDNA mutations. This may be imposed by loss of mitochondria function, thus limiting its propagation or triggering mitophagy. Alternatively, selective replication of mtDNA molecules may keep the mutation load from further increasing. In all, both SNVs and deletions appeared homogenously accumulated across the examined brain regions during ageing, but may be influenced by different pathological settings.
Polg D181A expression causes brain region-specific SNV accumulation with ageing Having established that our method could be used to map ageing-induced mtDNA mutations, we turned to Fig. 1 Ageing increases the load of both SNVs and deletions in mtDNA across all brain regions. a Schematic illustration of the workflow from mouse to prepared library. Briefly, brain regions of interest were rapidly sampled and total DNA was extracted. Linear DNA was enzymatically degraded by exonuclease (ExoV), and non-linear DNA is purified and used for library preparation. FL: full-length mtDNA molecule, Δ: mtDNA molecule with deletion. b Overview of the analysis workflow to optimise mtDNA variant detection. Shortly, after quality filtering, reads were mapped to mm10 without the mitochondrial chromosome (MT). Unmapped reads were then re-mapped to a modified MT reference (dMT: two MT references in tandem) and variants called. c Overview of mouse mtDNA. Green: rRNA encoding genes; blue: protein-coding genes; red: tRNAencoding genes; orange: non-coding region (NCR). d Schematic showing the areas isolated as the cortex (COR), caudate putamen (CP), dorsal raphe (DR), nucleus accumbens (NAc), paraventricular nucleus of the thalamus (PVT), and substantia nigra (SN). e DNA stored before and after ExoV digestion was subjected to qPCR to determine the relative levels of three mtDNA and three nuclear targets before and after digestion (shown for two different mice, A and B). Mouse C was treated as A and B but without the addition of ExoV. Bars show the mean of target signals and the standard deviation is indicated. †: nDNA after ExoV treatment was not detected or only detected at a very low level by qPCR and may not be visible in the bar plot. f Dot plot illustrating the age-dependent increase in the load of SNVs (left) and deletions (right) across the investigated brain regions (as indicated by the colour legend). All samples have been normalised to the mean of the variants at 10 weeks. Grey diamonds indicate the mean of all regions at the indicated age, and the 95% confidence interval is shown. Three-way ANOVA showed age, not region or animal, significantly (p < 0.01) contributed to SNV and deletion levels. Tukey's test was used post hoc to determine p values between each age group our Polg D181A model mice to investigate the brain region-specific influence of proof-reading deficiency. We sequenced mtDNA from the six brain regions of interest from Polg D181A mice and identified the SNVs in each region at 10, 50, and 80 weeks of age (Fig. 2a).
In young mice, there was no change in SNV levels between WT and Polg D181A in 10-week-old mice. At 50 and 80 weeks, we observed a significant increase in SNV levels which was especially prominent in COR, NAc, and PVT (Fig. 2a). This demonstrated that the heterogeneity of the mitochondrial response to proof-reading deficiency is present across an organ and not only between organs [6,16,22,23].
Importantly, we found no relationship between the expression of the Polg D181A transgene in the investigated brain regions and the level of detected SNVs (Additional file 1: Fig. S5c).
SNVs are excessively shared between brain regions
We wondered whether the increase in SNVs with both ageing and Polg D181A expression was affecting the same positions in mtDNA across brain regions. Indeed, looking in 10-bp non-overlapping intervals, we found an increase in shared SNV positions with ageing which was enhanced by Polg D181A expression (Fig. 2b). The overlap between individual animals was most prominent in PVT and NAc (Fig. 2c). As Polg D181A expression introduced a shift of the SNV distribution to the right side of the distribution plot (i.e. towards the NCR) compared to WT (Fig. 2d), we examined where shared SNVs were located. Fig. 2 SNVs heterogeneously accumulate across brain regions in Polg D181A mice and cause mtDNA position-specific mutational patterns. a Dot plot illustrating the age-dependent increase in the load of SNVs in Polg D181A mice across the investigated brain regions (as indicated by the colour legend) normalised to the mean of WT samples at 10 weeks. Grey diamonds indicate the mean of WT-derived brain region samples for reference (same as in Fig. 1f). Red diamonds indicate the mean of Polg D181A -derived brain region samples and the 95% confidence interval is shown. Three-way ANOVA (age, region, and animal) of Polg D181A -derived samples showed that age significantly contributed to SNV levels (p values of post hoc Tukey's test are shown). Three-way ANOVA showed a significant contribution of all variables (age, genotype, region). p values of post hoc Tukey's test comparing WT and Polg D181A at each age are shown. For region contribution, we found a significant contribution of COR, NAc, and PVT to SNV levels in Polg D181A mice using a linear model for main effects. b SNVs were counted in 10-bp non-overlapping bins for WT (grey) and Polg D181A (red) mice at 10, 50, and 80 weeks, and the number of regions with SNV in each bin calculated. Note that in the case that one region has more than one SNV in a bin, it is only counted as one instance of an SNV. The overlap was visualised for non-overlapping bins ("1"), bins shared across two or three regions ("2-3"), and bins shared across four to six regions ("4-6"). c SNVs were counted in 10-bp nonoverlapping bins for WT (grey) and Polg D181A (red) mice at 50 weeks, and the number of individual animals with SNVs in each bin calculated. Note that in the case that one animal has more than one SNV in a bin, it is only counted as one instance of an SNV. The overlap was visualised for non-overlapping bins ("1"), bins shared across two or three animals ("2-3"), and bins shared by four or more animals ("≥ 4"). d Cumulative percentage of SNVs detected in each examined brain region (thin lines) for both WT (grey) and Polg D181A (red) at 10, 50, and 80 weeks old. Bold lines indicate the smooth conditional mean for each genotype. e The relative average SNV allele frequency for each region for WT (grey) and Polg D181A (red) mice at 10, 50, and 80 weeks as indicated shown as boxplots. p values of two-sided t tests are shown. f SNVs across brain regions were pooled for each genotype at each age and divided into 100-bp bins across the mtDNA reference and the allele fraction for SNVs in each bin summed and normalised (i.e. highest peak set to 1). Grey areas indicate mtDNA regions where peaks are found across all variables (α), peaks that are ageing-dependent (β), and ageing-induced Polg D181A -dependent peaks (γ) We found that shared SNV positions were significantly different from non-shared SNV positions in Polg D181A mice (t test, p < 1 × 10 −5 ) but not in WT (t test, p = 0.254) when looking across all ages and shifted towards the 3′ region (i.e. towards the NCR) (Additional file 1: Fig. S1a).
The increase in shared SNVs was accompanied by a significant increase in SNV frequency with Polg D181A expression ( Fig. 2e) which was driven by high frequency SNVs in specific mtDNA regions (Fig. 2f). These regions appeared highly context-dependent, i.e. peaks that are found across all samples ("α" on Fig. 2f), only in very aged mice ("β"), or are Polg D181A -specific ("γ"). This was mimicked in the Pearson correlation, where most brain regions from 50-and 80-week-old Polg D181A mice form a distinct cluster and most samples from 10-week-old mice form a distinct cluster (Additional file 1: Fig. S1b) and we found specific SNV hotspots in 10-week-old animals independent of genotype (Additional file 1: Fig. S1c). In addition, there was a significant overlap of the specific positions at which SNVs are present in COR, NAc, and PVT (the brain regions most sensitive to Polg D181A expression) at both 50 and 80 weeks in Polg D181A mice (Additional file 1: Fig. S1d, right). For WT mice, the overlap is only pronounced at 80 weeks and p values do not reach similar levels of significance (Additional file 1: Fig. S1d, left).
We found an increase of SNVs in the NCR and complex III genes (Additional file 1: Fig. S1e), while transitions and transversions (Additional file 1: Fig. S1f) were comparable to those of previous studies, and we saw no indication of either ageing-or Polg D181A -induced oxidative mutations [19,24], together with no change in the types of mutations (Additional file 1: Fig. S1g).
Together, these data demonstrated a brain regionspecific ageing-dependent Polg D181A -induced mtDNA SNV spectrum, where COR, NAc, and PVT are regional hotspots. In addition, certain mtDNA positions are highly sensitive to SNVs and seem to function as context-dependent mutational hotspots.
Ageing-dependent Polg D181A -induced deletions accumulate in the same brain regions as SNVs We next turned our attention to the influence of Polg D181A expression on the accumulation of deletions. We found a significant ageing-induced accumulation of deletions in both 50-and 80-compared to 10-week-old Polg D181A mice, but we observed no significant differences between WT and Polg D181A at any age ( Fig. 3a). However, deletion accumulation in response to Polg D181A showed a prominent region specificity. While CP, DR, and SN Polg D181A SNV levels were only slightly elevated compared to WT mice, COR, NAc, and PVT showed a very high accumulation of deletions at 50 and Fig. 3 Accumulation of deletions induced by Polg D181A expression are brain region-specific and ageing-dependent. a Dot plot illustrating the agedependent increase in the load of SNVs in Polg D181A mice across the investigated brain regions (as indicated by the colour legend) normalised to the mean of WT samples at 10 weeks. Grey diamonds indicate the mean of WT-derived brain region samples for reference (same as in Fig. 1f).
Red diamonds indicate the mean of Polg D181A -derived brain region samples and the 95% confidence interval is shown. Three-way ANOVA (age, region, and animal) of Polg D181A -derived samples showed that age significantly contributed to deletion levels (p values of post hoc Tukey's test are shown). p values of three-way ANOVA (age, genotype, region) with post hoc Tukey's test are shown for each age group. For region contribution, we found a significant contribution of COR, NAc, and PVT to deletion levels in Polg D181A mice using a linear model for main effects. b Chord diagrams indicating the deletions accumulated at 10, 50, and 80 weeks in DR and PVT from WT and Polg D181A mice. Data is normalised pr. brain region, and the width of each gene indicates the summed allele fraction of deletions spanning the indicated gene(s). The colour of the chord indicates the gene in which the breakpoint 5′ position is located. Plots were made using circlize 80 weeks. We found a significant difference in deletion levels between these regions compared to the other regions in Polg D181A when pooling data from 50-and 80week-old mice (p = 0.002, one-way ANOVA). Similar to SNVs, we found no indication that expression levels of the Polg D181A transgene were the major driver of deletion levels in the Polg D181A mice (Additional file 1: Fig. S5c).
The differences across brain regions with ageing of WT and Polg D181A mice can be appreciated by chord diagrams showing the span of all deletions at each time point (Fig. 3b). Where PVT showed both an ageinginduced and a clear ageing-dependent Polg D181A -induced accumulation of deletions, DR only showed an ageinginduced accumulation of deletions, highlighting the brain region-specific mtDNA sensitivity to a setting of replication instability. Pearson correlation indicated similarities in the deletions found with ageing of Polg D181A mice (Additional file 1: Fig. S2a), indicating that Polg D181A expression induces a specific landscape of mtDNA deletions.
Deletions share characteristics independent of genotype
The positions at which deletions start and end are termed breakpoints and based on the co-occurring deletions between brain regions from Polg D181A mice, we hypothesised that breakpoints must be shared between different samples. We looked in 100-bp bins along the mtDNA and found that shared breakpoints cluster in very distinct locations (Additional file 1: Fig. S2b). Some shared breakpoints are age-and genotype-independent (~5 kb) whereas others are genotype-dependent (~15 kb). Sizes of deletions themselves follow a bimodal distribution independent of age and genotype and can be roughly divided into those < 100 bp and those > 1 kb, with few observations in the intermediate range (Fig. 4a). Even though the number of deletions in 10-week-old animals is low, they still follow this distribution, though the fraction of very small deletions is high compared to aged animals.
Molecular determinants of deletions
A previous study has suggested that the majority of mtDNA deletions in Parkinson's patients occur at direct repeats [17], a proposed [25] though highly debated [26] feature of human mtDNA deletions. To investigate the influence of direct repeats in breakpoint formation in the mouse brain, we identified direct repeats ≥ 8 bp in mtDNA (Additional file 1: Fig. S2c). After pooling deletions per genotype, we identified the direct repeat pair with the shortest average distance from the 5′ and 3′ breakpoints of WT and Polg D181A deletions as well as for in silico generated, deletion length-matched deletion libraries for each genotype (see the "Methods" section). The shortest average distance was shorter for experimentally derived deletions than randomly generated deletions for both WT and Polg D181A mice (Fig. 4b), but there was no difference between WT and Polg D181A (t test, p = 0.905). This indicates that direct repeats may contribute to at least a part of the identified deletions.
We found this to be the case at 10 and 50 weeks but not 80 weeks (Additional file 1: Fig. S2d), as deletions are significantly closer to direct repeats than the in silico deletion libraries for both WT and Polg D181A mice. Restriction of sequence similarity to direct repeats is a rigorous criterion. We therefore calculated the sequence The shortest average distance from 5′ and 3′ deletion breakpoint pairs to a direct repeat pair in the mitochondrial genome for the observed deletions (darker colour) and a random in silico generated deletion length-matched library (lighter colour) for both WT (left, in grey) and Polg D181A (right, in red) using pooled data from all ages and brain regions examined for each genotype. p values of two-sided t tests are shown. c Needle identity score calculated in a ± 10 bp window at the 5′ and 3′ deletion breakpoints as a function of deletion size after pooling of WT and Polg D181A samples. Correlation for each age is indicated by the full lines and correlation data indicated in the same colour code identity scores in a 20-bp window surrounding all 5′ and 3′ breakpoints (i.e. 10 bp on each side of the breakpoint). We found a negative correlation between sequence identity score and deletion length in 50-and 80-week-old Polg D181A mice (Fig. 4c) and further saw a significant difference between the identity scores of deletions < 100 bp and > 100 bp at 50-and 80-week-old Polg D181A as well as WT mice (Additional file 1: Fig. S2e). These data imply a differential contribution of non-direct repeat sequence similarity to short and long deletions.
Abundant NCR multimers are exclusive to Polg D181Aexpression As expected, the sequencing coverage exhibited some variability, likely associated with a slight sequence specificity of the transposase used for library preparation [27][28][29]. However, in specific brain regions from the 50and 80-week-old Polg D181A mice, we observed an increased coverage in the 15 kb+ region including at least a part of the NCR (Additional file 1: Fig. S3a). Localised increased coverage is often thought to be associated with duplicated regions. As mitochondrial DNA is circular, it is not possible to distinguish small duplications from very long-range deletions (VLRDs) (Additional file 1: Fig. S3b). We therefore wondered if our data of mtDNA deletions could support the presence of multimers. By classifying VLRDs as deletions > 15 kb, we found that VLRDs are specifically enriched in NAc and PVT from 50-and 80-week-old Polg D181A mice (Fig. 5a) and enriched in the 15 kb+ region (Fig. 5b and Additional file 1: Fig. S3c), supporting the idea that mtDNA multimers including at least part of the NCR accumulate in a brain region-specific and Polg D181A -dependent manner with age. We also found an increase in discordant reads in 50-week-old Polg D181A mice, which further supports the presence of genomic rearrangements such as multimers ( Fig. 5c and Additional file 1: Fig. S3c).
These putative multimers appear to form in a quite restricted region of mtDNA as their 5′ and 3′ "breakpoints" (indicating the end and the start of the duplicated sequence, respectively), accumulate at rather discrete positions (Fig. 5d) spanning a region with a low conservation score across mammals (Fig. 5d, bottom panel). Previous data suggested the presence of multimers in the brain from the mutator mouse [16]. We used the same PCR approach as Williams et al. and validated the Polg D181A -specific presence of multimers (Additional file 1: Fig. S4a, top and middle panel). An alternative PCR setup that would only yield a product in the presence of multimers confirmed these results (Additional file 1: Fig. S4a, bottom panel) and subsequent data also indicated the presence of inversions (Additional file 1: Figs. S4b,c).
Together, these data support the presence of highly brain region-specific ageing-induced Polg D181Adependent multimers which are highly specific to a partial NCR-containing segment of mtDNA specifically in NAc and PVT.
Direct repeats may be involved in NCR multimer formation
Multimers can be formed by several mechanisms. One mechanism is by strand slipping during replication which may be influenced by the local environment surrounding the NCR, which is known to interact with the inner mitochondrial membrane [9]. Another mechanism is mediated by the DNA sequence surrounding the start and end positions of the multimer region. In support of the idea of strand slipping, we find VLRDs in the 15 kb+ region to be closer to direct repeats compared to multimers in other parts of the mtDNA (Fig. 5e), though the overall sequence similarity surrounding breakpoints is not different (Fig. 5f). SNVs were enriched near VLRD 5′ breakpoints as well as~7 kb upstream with a mean distance of 75 ± 462 bp to the nearest SNV (Additional file 1: Fig. S5a). Fifteen percent of VLRD breakpoints co-position with SNVs, a number which is not influenced by discordant reads. SNVs were not enriched within the putative multimeric region (Additional file 1: Fig. S5b).
Transgene expression level does not drive variants
The expression of transgenes are often not similar across tissues, which is also true for Polg D181A expression [14]. To confirm that the expression differences were not driving the differences we observed in the accumulation of mutations in response to Polg D181A expression, we evaluated the expression levels of endogenous Polg and transgenic Polg D181A . Importantly, we were interested in the relative expression levels of the two transcripts, as endogenous and transgenic Polg will be competing for access to mtDNA during replication. As presented in previous sections, we find no correlation between mtDNA mutation levels and relative Polg D181A /Polg levels at any age (Additional file 1: Fig. S5c). Together, this demonstrates that transgene expression levels were not the major driver of brain region specificity to proofreading deficiency in mitochondria.
mtDNA variants cluster together along genomic regions
Throughout our analysis of the mutation spectrum of mtDNA from both WT and Polg D181A mice, it became increasingly clear that different types of variants often were found in specific mtDNA regions. To further investigate this, we plotted all variants analysed-SNVs, deletions, VLRDs (i.e. multimers)-across mtDNA in a circular plot (Fig. 6a). Visual inspection of this plot showed that different types of variants are enriched in the vicinity of each other. We found a strong, positive correlation between SNV and deletion load at the gene level which is independent of ageing and genotype (Fig. 6b), indicating positional sensitivity to the accumulation of mutations which may reveal underlying genomic instability in specific regions or be caused by higher order structures.
Discussion
Here we report a brain-wide spatio-temporal map of the mtDNA mutation spectrum in WT and proof-readingdeficient mice expressing Polg D181A under the CaMKIIαpromoter. Using a PCR-free approach to enrich mtDNA for next-generation sequencing, we were able to study small tissue dissections while minimising bias to the analysis. Using this approach, we found that (1) the ageing-induced increase in SNVs and deletions is largely brain region-independent and reaches saturation at 50 weeks; (2) Polg D181A expression specifically increase SNV and deletion levels in COR, NAc, and PVT; (3) ageing increases the number of shared SNVs, a feature that is enhanced in Polg D181A mice and SNVs are prominent in the NCR; (4) deletions follow a bimodal size distribution independent of age and genotype; and (5) Polg D181A induces NCR-containing multimers specifically in NAc and PVT in an ageing-dependent manner.
Deletions have been described in e.g. the brain [17,[30][31][32][33][34], muscle [35][36][37], and heart [30,37]; however, all these studies suffer from limitations in either the amount of tissue required for input, a priori bias to the analysis, or limited temporal insight. Early studies tend to argue that deletions are mainly major arc deletions between the two origins of replication; however, we do not find any indication of specific accumulation of common mtDNA deletion between the origins of replication (Figs. 3b and 4a), similar to the mutator mouse [16]. Instead, we identified a highly diverse deletion spectrum ranging from 1 bp to 15 kb with a tendency for very short (< 10 bp) deletions to preferentially accumulate across all ages (Fig. 4b). The bimodal distribution of deletion lengths (Fig. 4b) was quite surprising. MtDNA deletions may be subject to two opposite working mechanisms: (1) shorter molecules, caused by larger deletions, finish replication faster, in principle leading to their rapid accumulation [23,38]; (2) mtDNA undergo purifying selection [39], leading to the preferential loss of deleterious molecules (e.g. larger deletions) [40]. This bimodal distribution might be explained by the combination of these two mechanisms. However, the positive selection hypothesis has been disproved [41], and random genetic drift as the major contributor to the accumulation of short mtDNA species [42] was also recently questioned [43]. Furthermore, the presence of large deletions will naturally restrict the presence of smaller deletions. The description of a tight link between transcription and replication has increased interest in the positive selection idea [44], but the mechanisms remain unknown. However, the fact that young mice have a higher proportion of very small deletions (Fig. 4b) could favour a hypothesis including positive selection. mtDNA deletions have been proposed to be prevalent between direct repeats [45,46] in both normal ageing and disease, though this is not always the case [47,48]. As such, deletions are grouped into type I (flanked by direct repeat) and type II (not flanked by direct repeat) [49]. Type I deletions are hypothesised to be formed by polymerase slippage during replication [50]. The discrepancy between human disease cases [51] and mouse models regarding direct repeats may relate to the different involvement of direct repeats in deletion formation between long-lived and short-lived mammals [26]. Type II deletions colocalize with 2D and 3D mtDNA structures [49] and occur spontaneously during replication either through strand slipping [48], Polg stalling [52], or repair of double-stranded breaks [53]. The fact that Polg D181A expression increases deletion load without affecting the distribution of deletion sizes remains elusive. We speculate that Polg D181A -induced replication instability would increase the frequency of deletion-mediating events without effecting their overall properties or characteristics, but increased SNVs in itself do not seem to increase deletion rate [54].
Studies on duplications in mtDNA have previously focused on the D-loop [55][56][57], and we similarly found duplications to be a brain region-and NCR-specific ageing-induced Polg D181A -dependent event (Fig. 5a). NCR duplications could be caused by Polg stalling during replication [58]. Data indicates that the D-loop may not be an ideal region to use as a control in the estimation of mtDNA copy number [59], which could be related to the propensity of duplications in this region and may be partly due to direct repeats (Fig. 5d). In addition, the D-loop is often chosen for primer design for longrange PCR, why such long-range PCR is not suitable for rearrangement analysis without a priori knowledge of the multimer landscape which may be highly dependent on the animal model, the disease progression, or the tissue investigated. The D-loop has been proposed to directly interact with the inner mitochondrial membrane and through this interaction mediates protein recruitment and mitochondrial structure [12]. Multimers of NCR may therefore influence mitochondrial function in other ways than the classical view of energy production and mtDNA copy number.
We observed no differences in the types of SNV with ageing in neither WT nor Polg D181A mice indicating that age-induced oxidative damage is not a major driver of mutations in mtDNA but rather mutations are caused by the accumulation of Polg errors in both WT and Polg D181A mice. Oxidative damage of mtDNA causes 8oxo-dG which was argued to result in G>T transversions [60]; however, we find that G>T only take up a small fraction of the identified SNVs, which could indicate that Polg is able to correctly incorporate T opposite 8oxo-dG under oxidative conditions [61]. In support of this, oxidative damage is limited in mutator mice [6,22,62], data which can likely be extrapolated to our model. Instead, C>T and T>C are the major identified SNVs (Additional file 1: Fig. S1e). The major base interpretation mistake by Polg is T-GTP mispairing [63][64][65][66], though A>G can also occur due to deamination of adenosine. C>T is generally associated with cytosine deamination, though the exact link to Polg function is not clear. Overall, the mutation spectrum identified here is similar to that of the mutator mouse [19] as well as mtDNA analysis of ChIP-seq-derived data [24].
We found that SNVs tend to cluster in hotspots which display genotype and ageing characteristics (Fig. 2d). We found another Polg D181A -specific mtDNA mutation trait, the presence of multimers (Fig. 5a). The presence of several Polg D181A -dependent mutation traits compared to WT should caution the use of mtDNA mutationinducing mouse models to describe ageing processes. We cannot testify to the tissue specificity of these observations, but at least the heart from mutator mice also seems to harbour genotype-specific multimers [16]. These mutations may not reflect the naturally occurring ageing phenotype adequately.
An unanswered question remains: what is the cause of regional sensitivity of proof-reading deficiency? As mitochondria are highly dynamic organelles that constantly undergo fission and fusion, the rates of these processes influence both replication and turnover of mtDNA. This is especially true in neurons, where mitochondria can roughly be divided into those found in the soma and those at the synapse with a constant transport of mitochondria along the axon [67,68]. In addition, Parkin has been shown to protect SN neurons from the accumulation of mtDNA mutations in the mutator mouse [69], and it is likely that Parkin or other proteins influence similar processes in different brain regions. As the total level of Polg also appears to influence the propagation of deleterious mtDNA molecules [70], the transgenic expression of Polg may influence this regulation.
Elucidation of the cause and effect relationship between mtDNA mutations and ageing is not so straightforward. In mice, there is evidence that mtDNA SNVs themselves are adequate to induce premature ageing [71,72], though the SNV load in these models is several fold higher than that observed in aged humans [73], and heterozygous Polg D257A mice that also have elevated SNV levels do not show signs of premature ageing [6]. Human data suggest the ageing-dependent accumulation of mtDNA deletions [74], but whether this is a cause or consequence of ageing is not clear. Interestingly, mice lacking Mgme1, the mitochondrial exonuclease, accumulate deletions but do not show premature ageing [75], indicating the requirement of SNVs, not deletions, for premature ageing, likely by affecting the functionality of mitochondrial proteins or functional RNAs.
In all, our data provide a novel view of the spatiotemporal accumulation of mtDNA mutations by providing a method for investigating the full mutation spectrum from very limited tissue dissections. The differential response across brain regions to a state of replication instability provides insight into a possible heterogenic mitochondrial landscape across the brain that may help explain the specificity of neuropsychiatric disorders in individuals with mitochondrial disease as well as neurological changes associated with ageing. Appreciating the tissue and region specificities of the mitochondrial genome in terms of copy number variations [59,76], mutations [14], or gene expression [77] is pivotal to understand changes in mitochondrial dynamics in ageing and disease states.
Conclusions
We provide a novel unbiased spatio-temporal mapping of the full mtDNA mutation spectrum of discrete regionspecific dissections of mouse brain using an approach that does not require PCR amplification of mtDNA prior to library preparation. We demonstrate that both single nucleotide variants (SNVs) and deletions accumulate homogeneously across the examined brain regions during ageing from 10 to 50 weeks but see no further increase towards 80 weeks. In mice expressing proof-reading-deficient Polg, Polg D181A , the mitochondrial response to this state of replication instability is highly brain region-specific, and we demonstrate that the paraventricular thalamic nucleus and nucleus accumbens are mutational hotspots. The increased mutation load in ageing Polg D181A mice compared to wildtype is only moderately associated with changes in mutation characteristics. Polg D181A also induces an ageingdependent accumulation of non-coding control-region multimers, a feature that appears almost non-existent in wild-type mice. Our data show that unbiased sequencing of mtDNA from small tissue dissections can contribute to our understanding of the heterogenous mtDNA regulatory processes in ageing and disease states.
Animals
All animal care and experimental procedures were in accordance with the guidelines for proper conduct of animal experiments published by Science Council of Japan and approved by RIKEN Wako Animal Experiment Committee. All CaMKIIα-Polg D181A transgenic mice used were heterozygotes. Animals were bred as described previously [14,78]. In brief, mutant male CaMKIIα-Polg D181A mice were mated with wild-type (WT) C57BL/6J female mice. Genotyping was performed using genomic DNA isolated from tail biopsies as described [14].
mtDNA enrichment
Each sample was incubated at 37°C for 16 h in 90 μL DNA buffer (10 mM TrisHCl pH 8.0, 0.1 M NaCl, 1% SDS) and 10 μL Proteinase K (Roche, #03115828001) and treated with RNase A for 10 min at room temperature. Total DNA was extracted using 1 vol AMPure XP beads (Beckman Coulter, #A63881). A small aliquot of the purified DNA (1/10 vol) was stored for qPCR and the remainder used for mtDNA enrichment. Total DNA was exonuclease treated for 36 h at 37°C with interval shaking (1000 rpm) (18 μL total DNA, 3 μL NEBuffer 4 (10x), 4 μL ExoV (NEB, #M0345L), 6 μL ATP (10 mM)). Additional 1 μL NEBuffer 4 (10x), 2 μL ExoV, 6 μL ATP (10 mM) was added and incubated at 37°C for 16 h. DNA was extracted using 0.4 vol AMPure XP beads. A small aliquot of the purified DNA (1/10 vol) was saved for qPCR. We note that the initial crude extraction of total DNA was unsuitable for the cerebellum due to the high lipid content of this tissue. For samples with high fibre density (e.g. NAc), we also periodically experienced some issues with solubility which was solved with either increased incubation time or increased buffer volume.
Library preparation and sequencing
Up to 0.5 ng DNA was prepared for sequencing using the NexteraXT DNA kit (Illumina, #FC-131-1024). Libraries were quantified using KAPA Universal kit (Kapabiosystems, #UKK4824) or Qubit dsDNA HS Assay (Invitrogen, #Q32851) and library size estimated by BioAnalyzer HS DNA chip (Agilent, #5067-4626). Libraries were pooled to 2 nM and sequenced on the Illumina MiSeq with 150-bp paired end reads (Illumina, #MS-102-2002). Note that the same number of PCR cycles for library amplification was used independent of start DNA input and that this did not lead to low complexity libraries or affected library size distribution as evaluated by BioAnalyzer.
Sequencing analysis
Sequencing data was filtered, trimmed, mapped, and variant called using the BBTools suite (38.07) [21] using Clumpify (dedupe), FilterByTile (default), BBDuk (first: ktrim = r k = 23 mink = 11 hdist = 1 tbo tpe minlen = 100 ftm = 5 ordered (using adaptor resource provided with BBTools); second: k = 27 ordered qtrim = r trimq = 8 (using sequencing artefacts and phiX sequences provided with BBTools)), BBMap (vslow k = 11 secondary = t minratio = 0.55 tipsearch = 300 maxindel = 160000 ambig = all rescuedist = 30000 qtrim = lr), and CallVariants (rarity = 0.005 minallelefraction = 0.005 ploidy = 100 minedistmax = 5 border = 5 minquality = 10 minqualitymax = 10 minscore = 10) and postfiltered in R for minimum supporting reads (SNVs 4, deletions 2), sequencing depth (SNVs 50, deletions 25), and quality score (SNVs 20, deletions 10). Initially, reads were mapped to mm10 (ensembl) without the mitochondrial chromosome (MT) (using BBMap perfect mode) and unmapped reads were re-mapped to a modified version of MT consisting of two tandem MT sequences (in principle, a "double" MT chromosome, which we termed dMT) in order to call long-range deletions. We note that deletion calling is sensitive to kmer length chosen during mapping. Whereas a longer kmer showed more reproducible results, shorter kmers resulted in better detection sensitivity of deletions. Thus, we chose the shortest kmer value within the higher kmer lengths that gave reproducible results. The same number of aligned reads was used for each sample for variant analysis using random sampling with BBTools Reformat to avoid bias due to sequencing depth. Because of this approach, it is not necessary to normalise the read depth by the number of total reads. We filtered deletions to be a maximum of 15 kb based on our previous data indicating the presence of~2 kb mtDNA molecules in the Polg mice [13]. For the analysis of multimers, only deletions with a length > 15 kb were retained, so no variants were included in both analyses.
The in silico deletion length-matched libraries were generated by randomly sampling n mtDNA positions (where n is the number of deletions in either WT or Polg D181A samples), and these positions were used as deletion start positions. Deletion lengths (of either WT or Polg D181A deletion libraries) were randomly shuffled and assigned to each randomly sampled mtDNA position, and the sum of these denoted the end position of the in silico generated deletions.
Statistical tests used are indicated in individual figure legends and were performed in R using base functions. To test the influence of ageing, genotype, animal, and/or region on the accumulation of mtDNA variants, we performed two-or three-way ANOVA. For pairwise testing, a two-sided t test (Welch's) was performed. For pairwise testing of non-normal distributed data (as evaluated by Shapiro-Wilk's method) that does not fulfil the central limits theorem, Wilcoxon testing was performed. All p values were Bonferroni corrected. Circular plots were made using the R package circlize [81].
To detect direct repeats in mtDNA, we split MT into 100-bp non-overlapping fragments and composed a local blast database (makeblastdb -in 100bp_fragments.fa -in-put_type fasta -dbtype nucl -out blastdb_100bp) against which we blasted the mtDNA sequence (blastn -task blastn-short -num_descriptions 500000000 num_alignments 500000000 -ungapped query mtDNA.fa -db blastdb_100bp -word_size 5 -evalue 1e300 -outfmt 6out mtblast). Putative direct repeats were post filtered to include maximum 1 mismatch per 4 nucleotides and for a minimum size of 8 bp due to the high frequency of shorter direct repeats.
PCR and cloning
Total DNA was prepared from two separate littermate pairs (female) 35-37 weeks old as described above. PCR was performed using Tks Gflex (Takara, #R060A) with 5 ng total DNA/reaction. PCR was run using 1 cycle: 94°C 2 min; 30 cycles: 94°C 15 s, X°C 20 s, 68°C Ys; 1 cycle: 68°C 7 min, where X and Y were optimised for each primer set used. PCR products were run on 1% agarose gels in TAE buffer, and DNA was visualised using ethidium bromide. Primer sequences and primerspecific PCR conditions can be found in Additional file 2: Table S2. PCR products were purified from 1% agarose gels using Wizard SV Gel and PCR Clean-Up System (Promega), cloned (Invitrogen, #45-1641), and sequenced (BigDye Terminator v3.1, Applied Biosystems).
RNA analysis
Tissues from 4 pairs of female Polg mice and littermate controls (10-12 weeks) were collected as described above but were immediately stored in TRIzol after dissection instead of snap-freezing. RNA was purified using the Direct-zol RNA Microprep Kit (Zymo Research, #R2060). One hundred nanograms of random hexamers (Invitrogen) was annealed (85°C, 5 min) to 100 ng RNA. Reverse transcription was carried out with 100 U M-MLV Reverse Transcriptase (Invitrogen, #28025013) in a 13-μL volume supplemented with 10 mM DTT and 1 mM dNTP (NEB, #N0447S) at room temperature, 10 min, then at 37°C, 60 min. cDNA was used for qPCR in technical triplicates. qPCR was performed as above. Primer sequences can be found in Additional file 2: Table S3.
==
Domain: Biology Medicine
|
[{"added": "2023-01-16T15:15:17.077Z", "created": "2020-10-23T00:00:00.000", "id": "255827931", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1186/s12915-020-00890-5", "pdf_hash": "5aaeb56822a525bcb007d5e212458491ed38186e", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:7621", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "5aaeb56822a525bcb007d5e212458491ed38186e", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Structural insights into a 20.8-kDa tegumental-allergen-like (TAL) protein from Clonorchis sinensis
Survival of Clonorchis sinensis, a cause of human clonorchiasis, requires tegument proteins, which are localized to the tegumental outer surface membrane. These proteins play an important role in a host response and parasite survival. Thus, these proteins are interesting molecular targets for vaccine and drug development. Here, we have determined two crystal structures of the calmodulin like domain (amino acid [aa] positions 1–81) and dynein light chain (DLC)-like domain (aa 83–177) of a 20.8-kDa tegumental-allergen-like protein from Clonorchis sinensis (CsTAL3). The calmodulin like domain has two Ca2+-binding sites (named CB1 and CB2), but Ca2+ binds to only one site, CB1. The DLC-like domain has a dimeric conformation; the interface is formed mainly by hydrogen bonds between the main chain atoms. In addition, we have determined full-length structure of CsTAL3 in solution and showed the conformational change of CsTAL3 induced by Ca2+ ion binding using small-angle X-ray scattering analysis and molecular dynamics simulations. The Ca2+-bound form has a more extended conformation than the Ca2+-free from does. These structural and biochemical analyses will advance the understanding of the biology of this liver fluke and may contribute to our understanding of the molecular mechanism of calcium-responsive and tegumental-allergen-like proteins.
Signalling by calcium ions is important in living system such as parasites. The most common related in calcium signalling motif is the EF-hand motif which is the best characterized in calmodulin 34 . Several antagonist of calmodulin, chlorpromazine (CPZ), Trifluoperazine (TPZ) and Phenothiazine (PTZ), were used in the treatment psychotic disorders [35][36][37] . Moreover, the tegumental proteins, such as SmTAL1,2,3 and CsTALs, is localized in host-interactive layer that has accessibility of selecting target molecules for vaccines and drugs 38 . Thus, the tegumental proteins are one of the most interesting molecular targets for development of vaccines and drugs 32,39 .
In this work, we determined 2.6 Å crystal structure of the DLC-like domain (amino acid [aa] positions 83-177) and 1.3 Å crystal structure of the calmodulin like domain (aa positions 1-81) of CsTAL3. Furthermore, we present the full-length structure of CsTAL3 in solution state and its conformational change upon Ca 2+ binding using small-angle X-ray scattering (SAXS) analysis. Our results should improve the understanding of the biology of liver flukes and may contribute to the development of new vaccines and drugs against clonorchiasis.
Results and Discussion
Overall structure of DLC-like domain of CsTAL3. At first, we tried crystalizing full-length CsTAL3 (aa 1-184), but the crystal structure contained only the DLC-like domain (aa 83-177). The interesting thing is that similar results were reported for SmTAL2 and FhCaBP2 27,40 . Both proteins belong to the TAL protein family of the class of fluke proteins that consist of a calmodulin like domain (or N-terminal domain) and a DLC-like domain (or C-terminal domain) as in CsTAL3. We also confirmed that CsTAL3 is completely cleaved into two domains in constant buffer condition (20 mM Tris/HCl, pH 7.5, 100 mM NaCl, 1 mM DTT) after ~20 days at 20 °C with various Ca 2+ ion concentration ( Supplementary Fig. 1). The cleavage mechanism of the flexible linker of these proteins shows instability of proteins and may be a general property in vivo 27 . As a result, the selenium-methionine-derivatized crystal of the DLC-like domain of CsTAL3 (aa 83-177) diffracted to 2.8 Å resolution and was found to belong to space group P2 1 2 1 2 with six protomers in the asymmetric unit. The initial phase determination and model building were accomplished by the SAD method with anomalous signals of 18 selenomethionines. The native crystal of the DLC-like domain of CsTAL3 diffracted to 2.6 Å resolution and belongs to space group C222 1 with three molecules per asymmetric unit. For determining the structure of the N-terminal domain (calmodulin like domain) of CsTAL3, later, we also attempted to crystallize only this domain (aa 1-81), and next, the resulting crystal diffracted to 1.3 Å resolution (see Table 1).
The monomeric structure of the DLC-like domain consists of four anti-parallel β-strands that are packed with each other and an extended loop protruding from β-sheets; the other face of the β-sheets is packed with two α-helices (Fig. 1a) Fig. 2). The dimeric interface information calculated by PISA 42 is that the dimeric interface area is on average ~1044 Å 2 (17.4%) at the total solvent-accessible area of 6033 Å 2 , and the solvation free energy gain upon formation of the interaction is on average −15.2 kcal/mol (Fig. 2b) Cα atoms are aligned, Z-score is 11.9). The five residues (G147 to T152) of each protomer in the extended loop interact with five residues (V′142 to D′146) of the neighboring protomer in the β 2 -strand via a pair of hydrogen bonds. Side chains of these strands also contribute to the hydrophobic interaction (Fig. 2b). Nonetheless, despite the conservation of the structural characteristics, the amino acid sequences of the dimeric interface are not conserved relative to the other DLC families (Fig. 1c). These findings suggest that dimeric interactions of the DLC-like domain are determined only by hydrogen bonds of the main chain 27 . Protein partners of LC8 interact with the extended β-sheet of the LC8 homodimer with backbone hydrogen bonds and side chain interactions. In the structure of the LC8 complex with peptide of Nek9 (PDB: 3ZKE, 3ZKF), the peptide interacts with the hydrophobic groove of the LC8 dimer; this groove is composed of β1, β3, β4, and α2′ [43][44][45] . The DLC-like domain of CsTAL3 also contains a hydrophobic groove (Fig. 2c). The superimposition of LC8 with the peptide and DLC-like domain shows a similar conformation (Fig. 2d). This result suggests that CsTAL3 may interact with its binding partner proteins in a similar manner.
Overall structure of Calmodulin like domain of CsTAL3. Due to cleavage of the full-length protein, we had grown a crystal of the calmodulin like domain (aa 1-81) of CsTAL3. The crystal diffracted to 1.3 Å resolution and belongs to space group P4 1 with one molecule per asymmetric unit. The initial phase determination and model building were carried out by the SAD method with an anomalous signal of one selenomethionine. The molecule shows structural similarities with the calmodulin like domain of a family of soluble Ca 2+ sensor Kv-channel-interacting proteins (KChIPs) and of the short Ca 2+ -binding mitochondrial carrier (SCaMC) with r.m.s. deviation 1.8 Å and 2.8 Å, when 60 Cα and 62 Cα atoms are aligned, and DALI server Z-scores 8.2 and 8.1 46 , respectively ( Supplementary Fig. 3). The structure of the calmodulin like domain is composed of five α-helices. The α1 to α4 helices are classical EF-hand motifs where two helix-loop-helix structures and two short antiparallel β-sheets (β1 and β2) are connecting the Ca 2+ -binding loops (Fig. 1b). Ca 2+ -binding motif 1 (CB1, residues 12-23) is a Ca 2+ -binding loop that contains 12 partially conserved residues starting with N-terminal aspartate and ending with C-terminal glutamate as in the EF-hand motif of other calmodulin like proteins ( Fig. 1b and c). One Ca 2+ ion binds to CB1 via D12, D14, T16, V18, E23, and a water molecule (positions X, Y, Z, −Y, −Z, and −X, respectively) in a geometrical pattern of a pentagonal bipyramid. Other residues bind a Ca 2+ ion via their side chain carboxyl groups, but V18 (-Y position) binds to a Ca 2+ ion via its main-chain carbonyl oxygen atom (Fig. 3a). Recently, structure of SmTAL3 was predicted that does not bind a Ca 2+ ion according to various biochemical experiments such as by limited proteolysis, native gel electrophoresis, differential scanning fluorimetry, and dot blots with radioactive calcium ions 25,47 . However, the CB1 of SmTAL3 sequences is highly conserved relative to CB1 of other calmodulin like proteins as SCaMC, KChIP1 and CsTAL3 (Fig. 1c). Although the -Y position sequence different, this is not a problem for the Ca 2+ -binding property because a Ca 2+ ion is bound only by the main-chain carbonyl oxygen atom of the -Y position residue. Considering this, we propose the possibility of Ca 2+ -binding property in the CB1 of SmTAL3.
Residues of Ca 2+ -binding motif 2 (CB2, aa 46-57) are predicted to be D46, D48, T50, S52, and T57 from the sequence alignment with the EF-hand motif of KChIP1 and SCaMC (Fig. 1c). Moreover, CB2 structure appears to be similar to that of CB1, but a Ca 2+ ion is absent in our structure. A significant difference is that the -Z position of CB1. This position is glutamate in the other EF-hand motif, but the -Z position of CB2 is threonine (T57). Although threonine is also a polar amino acid, it is not accessible to the Ca 2+ -binding region (Fig. 3b).
To confirm that the dimeric interaction of CsTAL3 is affected by the conformational change of the backbone folding with Ca 2+ binding in CB1, we performed AUC analyses. The c(s) distribution of CsTAL3 shows the presence of a single species with a sedimentation coefficient (s 20,w ) of 3.3 ± 0.1 S (without CaCl 2 ) and 3.2 ± 0.1 S (+5 mM CaCl 2 ). The molecular weight of the single species corresponds to ~50 kDa with and without a Ca 2+ ion ( Fig. 3c and d). These results suggest that the dimeric form of CsTAL3 is stable and its dimerization state is not affected by the Ca 2+ binding in CB1. The c(s) distribution peak shape of without CaCl 2 is much sharper than that of with CaCl 2 . These results proposed that CsTAL3 structure changes upon Ca 2+ ion binding and the c(s) peak shape reflects the CsTAL3 structural changes.
Determination of the full-length structure of CsTAL3 in solution.
To analyze the conformational change of CsTAL3 induced by Ca 2+ binding, we performed SAXS measurements. Scattering intensity I(q) was obtained in the protein concentration range 1.9 to 5.2 mg/mL. The Guinier plot indicated that the protein solution used in the SAXS analysis did not contain any aggregates (Fig. 4c). Estimated molecular mass of CsTAL3 was approximately range ~50 to ~60 kDa, indicating that CsTAL3 exists as a homodimer in solution (Fig. 4e). I(q) was slightly but unambiguously changed in a q-range < 0.15 Å −1 (Fig. 4b). The radius of gyration (R g ) of the Ca 2+ -bound form was larger than that of the Ca 2+ -free form (Fig. 4e). Distance distribution function P(r) of CsTAL3 showed a broader distribution with a shoulder ~55 Å, which is a characteristic of noncompact flexible proteins composed of two domains (Fig. 4d). There are a slightly broader shoulder of P(r) and larger R g in the Ca 2+ -bound form than those of in the Ca 2+ -free form. These results indicate that the distance between the calmodulin like domain and DLC-like domain in the Ca 2+ -bound form seems to be slightly larger than that in the Ca 2+ -free form.
BILBOMD is rigid body modeling with molecular dynamics simulations; it generates the minimal ensemble for the best agreement with the experimental scattering data 48 . Although full-length crystal structure was not obtained here, we determined the full-length structure of CsTAL3 in solution and analyzed the conformational change of CsTAL3 induced by Ca 2+ binding by means of BILBOMD. Residues 1-80 of the calmodulin like domain and residues 88-177 of the DLC-like domain were defined as fixed, while residues 81-87 (the unstructured portion between the two domains) were defined as flexible in our BILBOMD analysis. Despite setting large R g ranges (20-50 Å) for molecular dynamics simulations, each resulted in calculation of R g only between 30 and 34 Å. Thus, we estimated that flexibility is limited between the DLC-like domain and calmodulin like domain. The best-fit models of the Ca 2+ -bound form and Ca 2+ -free form have R g ≈ 32 Å. The theoretical scattering profile for each model is in good agreement with the experimental scattering data (χ 2 = 1.8 and 2.6). According to the results of the bead modeling, linear length of the Ca 2+ -bound form is ~157 Å, which is longer than ~140 Å of the Ca 2+ -free form. In the molecular dynamics simulation models, the distance between the two domains corresponds to the more extended conformation in the Ca 2+ -bound form than in the Ca 2+ -free form (Fig. 5). This result indicates that the calmodulin like domain and flexible linker undergo a conformational change upon Ca 2+ binding. These modeling results are consistent with AUC results. Thus, we propose that the structure of CsTAL3 changes to the extended conformation upon Ca 2+ binding; this conformational change may play a role in parasitic worms. This structural information should improve the understanding of the unique Ca 2+ -binding tegumental proteins and may facilitate the development of new drugs or vaccines. Thus, more research is needed on the exact function and mechanism of action of this protein in vivo to understand the physiological processes of parasitic worms.
Materials and Methods
Protein expression and purification. The gene of full-length (aa 1-184) CsTAL3 (UniProt ID: Q2PMV7) was cloned into pRSET-b (Merck Millipore, GE). The recombinant vector was transfected into Escherichia coli BL21(DE3) and B834(DE3) (Merck Millipore, GE). The cells were grown at 37 °C in the Luria-Bertani medium and in the M9 minimal medium containing 60 μg/ml L-selenomethionine with 100 μg/ml ampicillin up to optical density (at 600 nm) of 0.6. The protein expression was induced by the addition of 1 mM isopropyl-D-thiogalactopyranoside at 18 °C with incubation for 18 h. The cell pellet was resuspended in ice-cold lysis buffer consisting of 20 mM Tris-HCl (pH 7.5), 100 mM NaCl, 5 mM CaCl 2 , and 2 mM β-mercaptoethanol. After sonication and centrifugation for 1 h at 17364 × g and 4 °C, the supernatant was loaded onto a HiTrap Chelating HP column (GE Healthcare, USA). The recombinant protein was eluted using a linear gradient of 1 M imidazole added to lysis buffer. The fractions was incubated at 4 °C for a hour in the buffer included 5 mM EGTA to remove any bounded calcium ions. Further purification was conducted on a HiLoad 16/600 Superdex 200 prep-grade column (GE Healthcare, USA) with a buffer consisting of 20 mM Tris-HCl (pH 7.5), 100 mM NaCl, ±5 mM CaCl 2 , and 1 mM dithiothreitol. The fractions containing the purified protein were pooled and concentrated to 20 mg/mL using an Amicon Ultra Centrifugal Filter (Merck Millipore, GE), which was stored at −80 °C prior to crystallization trials.
The calmodulin like domain (aa 1-81) of CsTAL3 was cloned into pET-21a (Merck Millipore, GE). It was also transfected, expressed, and purified using the same protocols and buffering conditions as described above for the full-length CsTAL3 protein.
Crystallization and data collection. Initial protein crystal screen of full-length CsTAL3 was carried out by the sitting-drop vapor diffusion method at 22 °C. After a month, microcrystals were obtained in 200 mM MgSO 4 and 20% polyethylene glycol 3350. These crystallization conditions were optimized by the hanging-drop vapor diffusion method at 22 °C. The suitable crystals of native and selenium-methionine-derivatized version were grown in 170 mM MgSO 4 and 21% polyethylene glycol 3350. The crystals were frozen in liquid nitrogen with 15% (w/v) ethylene glycol as a cryoprotectant. The selenium-methionine-derivatized crystals were grown under the same conditions and in the same cryoprotectant as the native crystals were. X-ray diffraction data were collected using a wavelength 1.1000 Å on beamline BL-1A at the Photon Factory (Tsukuba, Japan).
Crystals of the calmodulin like domain of CsTAL3 were prepared by the same protocol as we used for full-length CsTAL3. The initial crystals were obtained in 100 mM sodium acetate and 3 M NaCl. The suitable crystals of the native and selenium-methionine-derivatized version for X-ray diffraction analysis were grown in 90 mM sodium acetate and 3.25 M NaCl. The crystals were frozen in liquid nitrogen with 4.8 M NaCl as a cryoprotectant. X-ray diffraction data on the native and selenium-methionine-derivatized crystals were collected using a wavelength 0.9800 Å on beamline BL-17A at the Photon Factory (Tsukuba, Japan) and on beamline 5C-SBII at the Pohang Light Source (Pohang, Korea), respectively. The raw data were indexed, integrated, and scaled using the HKL2000 software suite 49 . Crystallographic statistics of data collection are provided in Table 1.
Structure determination. The initial phases were obtained from the selenium-methionine single-wavelength anomalous dispersion (SAD) dataset using AutoSol in software package PHENIX 50 . Further structure was determined by molecular replacement based on the initial model of selenium-methionine data using PHASER in PHENIX 51 . The model building and refinement were performed using the Coot 52 and PHENIX 51 . The structure was validated with MolProbity 53 . The statistics of structure refinement are provided in Table 1. The coordinates and structure factor of the DLC-like domain and calmodulin like domain were deposited in the Protein Data Bank with the accession codes 5X2D and 5X2E, respectively. Structural analysis. The structure-based sequence alignment was generated using Clustal Omega 54 and ESPript 55 . Root mean square (r.m.s.) deviation and Z-score of structure alignment were calculated using the DALI server 46 . The dimeric interface area and free energy of dissociation were calculated in PISA 42 . All images of the crystal structure were generated using PyMol 41 . Analytical ultracentrifugation (AUC) experiment. The experiments were conducted at 20 °C using an Optima XL-I analytical ultracentrifuge (Beckman Coulter, USA) with an An-50 Ti rotor. For sedimentation velocity experiments, cells with a standard Epon two-channel centerpiece and sapphire windows were used. The sample (400 μL) and reference buffer (420 μL) were loaded into the cells. The rotor temperature was equilibrated at 20 °C in the vacuum chamber for 1-2 h prior to the startup. The sedimentation velocity experiment was conducted at protein concentrations of 2.5 and 0.6 mg/mL. Changes in the concentration gradient were monitored with a Rayleigh interference optical system at 10-min intervals during sedimentation at 50 × 10 3 rpm. Partial specific volume of the protein, solvent density, and solvent viscosity were calculated from standard tables using the SEDNTERP software 56 . The resulting scans were analyzed using the continuous distribution c(s) analysis module in the SEDFIT software 57 . Sedimentation coefficient increments of 100 were used in the appropriate range for each sample. The frictional coefficient was allowed to float during fitting. The weighted average sedimentation coefficient was obtained by integrating the range of sedimentation coefficients in which peaks were present. The values of the sedimentation coefficient were corrected to 20 °C in pure water (s 20,w ). The c(s) distribution was converted into c(M), a molar mass distribution.
Small Angle X-ray Scattering (SAXS). SAXS measurements were performed at 20 °C on a BioSAXS-1000 system (Rigaku, Japan) mounted on a MicroMax007HF X-ray generator (Rigaku, Japan). The PILATUS 100k detector, at a sample-to-detector distance of 482.8 mm, was used to measure scattering intensities. Sample solutions in 20 mM Tris-HCl pH 7.5 with 100 mM NaCl were used for SAXS measurements. The samples containing 5 mM CaCl 2 were used for analysis of the Ca 2+ -bound form. Circular averaging of the scattering intensities was carried out by means of the SAXSLab software (Rigaku, Japan) to obtain one-dimensional scattering data I(q) as a function of q (q = 4πsinθ/λ, where 2θ is the scattering angle, and the X-ray wavelength λ = 1.5418 Å). To check the interparticle interference, I(q) data were collected at different protein concentrations (1.9, 3.3, 4.0, and 4.5 mg/mL for the Ca 2+ -free form; 2.0, 3.2, 4.2, and 5.2 mg/mL for the Ca 2+ -bound form). To estimate molecular mass of CsTAL3, SAXS measurements of standard proteins (5.8 mg/mL glucose isomerase [172 kDa], 1.6 mg/ mL BSA [66 kDa], 5.0 mg/mL ovalbumin [43 kDa], and 2.7 mg/mL hen egg lysozyme [14 kDa]) were carried out under the same conditions. Exposure time was 2 h for CsTAL3, BSA, and lysozyme and 0.5 h for glucose isomerase and ovalbumin. All SAXS data were analyzed with the software applications embedded in the ATSAS package 58 . The radius of gyration R g and forward scattering intensity I(0) were estimated from the Guinier plot of I(q) in a smaller-angle region of qR g < 1.3 59 . The distance distribution function P(r) was calculated by means of the GNOM software 60 , where the experimental I(q) data were used in a q-range from 0.011 to 0.303 Å −1 . The maximum particle dimension D max was estimated from the P(r) function as the distance r for which P(r) = 0 60 . Bead-modeling software DAMMIF 61 was used to generate an ab initio model. Ten individual runs of DAMMIF were conducted and averaged with DAMAVER 62 . BILBOMD was used for rigid body modeling by molecular dynamics simulations and minimal ensemble model generation 48 . Initial models of full-length CsTAL3 for BILBOMD analysis were generated with Coot 52 . The collected SAXS data and data statistics are provided in Fig. 4.
==
Domain: Biology Medicine
|
[{"added": "2023-02-17T14:41:54.373Z", "created": "2017-05-11T00:00:00.000", "id": "256938496", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41598-017-02044-0.pdf", "pdf_hash": "4cd122df0f4a740746157f261ef9fb5ed9de3b7d", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:3025", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "4cd122df0f4a740746157f261ef9fb5ed9de3b7d", "year": 2017}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Phosphoglucomutase 1 contributes to optimal cyst development in Toxoplasma gondii
Toxoplasma gondii is a ubiquitous parasite of medical and veterinary importance; however, there exists no cure for chronic toxoplasmosis. Metabolic enzymes required for the production and maintenance of tissue cysts represent promising targets for novel therapies. Here, we use reverse genetics to investigate the role of Toxoplasma phosphoglucomutase 1, PGM1, in Toxoplasma growth and cystogenesis. We found that disruption of pgm1 did not significantly affect Toxoplasma intracellular growth and the lytic cycle. pgm1-defective parasites could differentiate into bradyzoites and produced cysts containing amylopectin in vitro. However, cysts produced in the absence of pgm1 were significantly smaller than wildtype. Together, our findings suggest that PGM1 is dispensable for in vitro growth but contributes to optimal Toxoplasma cyst development in vitro, thereby necessitating further investigation into the function of this enzyme in Toxoplasma persistence in its host.
Introduction
Toxoplasma gondii is an obligate intracellular protozoan responsible for toxoplasmosis in humans and other warm-blooded animals. Infections occur mostly from consuming contaminated water, food, or undercooked meat from chronically infected animals [1]. Bradyzoites inside tissue cysts are released into the gastrointestinal tract where they invade enterocytes and convert to tachyzoites inside a parasitophorous vacuole (PV). Tachyzoites replicate rapidly, eventually lysing out of the host cell to disseminate throughout the body. In response to stressful stimuli, they convert back to bradyzoites which remain encysted in the brain and skeletal muscles for life [2].
Chronic toxoplasmosis is incurable and parasite reactivation life-threatening, particularly for the immunocompromised [3].
Here, we used the CRISPR/Cas9 gene-editing system [16] to disrupt pgm1 in a cyst-forming Toxoplasma Open Access BMC Research Notes *Correspondencestrain. Our data show that this mutation did not prevent intracellular replication or the completion of the lytic cycle. While both strains could produce amylopectincontaining cysts, we found that pgm1-defective cysts are significantly smaller than the parental cysts. Together, our findings corroborate previous reports that PGM1 is dispensable for Toxoplasma viability and demonstrate that the enzyme contributes to optimal cyst development in vitro.
Replication assay
Freshly released parasites were centrifuged at 1500 rpm for 10 min and washed once with 1XPBS. Confluent HFFs on glass coverslips were infected with 1.2 × 10 5 parasites in cDMEM for 24 h. The number of parasites per vacuole was determined by immunofluorescence microscopy, as previously described [18], following staining with mouse α-SAG1 and rabbit α-GRA7 obtained from the Boothroyd lab. Immunostaining and visualization are further described below.
Plaque assay
WT and Δpgm1 tachyzoites were syringe-lysed through a 27G needle and passed through a 5 µm filter. Confluent HFFs were infected with 250 parasites in cDMEM and incubated at 37 ºC with 5% CO 2 for 10 days undisturbed. Following methanol fixation and crystal violet staining, plaque numbers and sizes were determined using a stereoscope (Leica EZ4) and ImageJ version 1.52A (National Institutes of Health) [19,20].
Tachyzoite-to-bradyzoite differentiation
Tachyzoites were induced to differentiate into bradyzoites in HFFs as previously described [21]. Briefly, confluent HFFs on glass coverslips were infected with 4.8 × 10 4 parasites for 3 h in cDMEM before replacing the medium with Switch Medium (RPMI 1640 supplemented with 1% fetal bovine serum, 100 U/ml penicillin, and 100 µg/ml streptomycin, 10 mg/mL HEPES, pH 8.2). Parasites were incubated for 4 days at 37 ºC with ambient CO 2 and the medium was changed every 24 h to maintain alkaline conditions.
Immunostaining fluorescence assay and amylopectin staining
Infected monolayers were washed with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde (Electron Microscopy Sciences) for 15 min at room temperature (RT). Cells were permeabilized with 0.2% or 0.4% Triton X-100 for 20 min and incubated for 1 h in 3% Bovine Serum Albumin (BSA; Fisher Scientific) in PBS. Primary antibodies diluted in 3% BSA/PBS (mouse α-SAG1 1:10,000, rabbit α-GRA7 1:1000) were added to the monolayers, when indicated, and incubated overnight at 4 ºC. Unbound antibodies were washed away with three 5 min washes in 1XPBS. The cells were then stained with secondary antibodies in 3% BSA/PBS (Goat α -Mouse 546 or Goat α-Rabbit 488 at 1:5000) for 45 min at RT. Dolichos biflorus Agglutinin (DBA; Vector Laboratories) was used at 1:100 to detect the cyst wall. After washing as described above, the coverslips were mounted with VECTASHIELD Mounting Medium containing DAPI (Vector Laboratories). Amylopectin was stained with Periodic Acid Schiff (PAS; Fisher Scientific) according to the manufacturer's guidelines.
Immunofluorescence images were obtained using an inverted microscope (Leica DM IL LED) with 100 × oil immersion objective. The number of parasites (SAG1positive) inside individual vacuole (GRA7 +) from randomly selected fields was determined from direct count under the microscope. The areas of plaques and cysts, both selected from random fields of view, were determined using ImageJ version 1.52A and 1.53, respectively [19,20].
Statistical methods
Statistical analyses were performed using GraphPad Prism version 8.4.3. A p-value ≤ 0.05 was considered a statistically significant difference between groups.
Toxoplasma phosphoglucomutases are upregulated during chronic infection in mice
Comparative transcriptomic and proteomic analyses [22] revealed that Toxoplasma expresses stage-specific proteins which enable the parasite to survive and to be efficiently transmitted between hosts. We mined the transcriptional data from Pittman et al. [12] available on the commonly used Toxoplasma Informatics Resources database (ToxoDB) [10,23] to specifically identify metabolic enzymes involved in gluconeogenesis and glycolysis that are significantly upregulated at least 2 folds in chronic vs. acute infection. Of the 422 genes upregulated in chronic infection, our analysis revealed 21 that are specifically associated with carbohydrate metabolism (Fig. 1A, B, Additional file 1). As expected, these genes include well-known glycolytic isoenzymes involved in tissue cyst formation, such as lactate dehydrogenase 2 (ldh2) [24] and enolase 1 (eno1) [25]. Interestingly, unlike ldh1/ldh2 and eno1/eno2 which are expressed in a stage-dependent manner, both PGM isoforms (pgm1 and pgm2) were upregulated 6.4 and 3.1 folds, respectively, in the chronic stage, 28 days post-infection (dpi) [12]. Transcriptional analyses of gene expression at 28, 90, and 120 dpi from Garfoot et al. [26] indicate that unlike pgm2 whose expression remained similar up to 120 dpi, pgm1 transcripts further increased from 28 to 120 dpi. Together, this analysis strongly suggests that transcriptional regulation of pgm1/pgm2 may be critical for the development and/or maintenance of tissue cysts in mice. Furthermore, the increased expression of PGM1 during chronic infection and its enzymatic activity at the intersection of energy storage and production pathways, namely glycolysis and amylopectin metabolism, warrant determining the role of this enzyme during Toxoplasma growth and differentiation.
Disruption of pgm1 does not hinder parasite growth in vitro
To determine the contribution of PGM1 to Toxoplasma growth, we used the CRISPR-Cas9 gene-editing system to create an insertional mutant Me49ΔhxgprtΔpgm1 (Δpgm1) by introducing a hxgprt selection cassette at the pgm1 locus [16] (Fig. 2A, B, Additional file 1: Figure S1).
We assessed the intracellular growth of Δpgm1 parasites vs. WT 24 h after infection of HFFs in glucose replete growth medium. SAG1-positive parasites inside GRA7positive vacuoles were enumerated. We found similar numbers of Δpgm1 vacuoles with either 2, 4, or ≥ 8 parasites as WT (Fig. 2C). Likewise, no significant differences in plaque numbers and sizes were observed 10 days after infection (Fig. 2 D, E). Thus, as previously reported for Toxoplasma RH strain [15,27], our data indicate that PGM1 is dispensable for Toxoplasma intracellular growth and lytic cycle in vitro, albeit in glucose-rich conditions.
pgm1-defective parasites produced smaller amylopectin-containing cysts in vitro
Given the upregulation of pgm1 in chronic infection, we tested whether disruption of pgm1 would impede tissue cyst formation. We induced tachyzoites to differentiate into bradyzoites in nutrient-poor, alkaline conditions in ambient CO 2 [21]. After 4 days, we stained the monolayers with Dolichos biflorus agglutinin (DBA) to detect the cyst wall and Periodic Acid Schiff (PAS) to visualize amylopectin [8]. Both WT and Δpgm1 parasites produced PAS-positive cysts (Fig. 3A), suggesting that PGM1 is not essential for amylopectin accumulation during stage conversion in vitro. However, further studies are required to determine any differences in the relative amount of this polysaccharide between WT and Δpgm1 cysts. Interestingly, Δpgm1 cysts were on average ~ 4060 pixels 2 smaller than WT (p = 0.0362 by Mann-Whitney test, Fig. 3C). Together, our results indicate that although PGM1 is not required for stage conversion and amylopectin storage, the enzyme contributes to optimal cyst development in vitro.
Discussion
PGM1 is one of two PGM isoforms differentially expressed in Toxoplasma [10,12,28]. In this study, we showed that disruption of pgm1 in a cyst-forming Toxoplasma strain did not prevent intracellular growth or completion of the lytic cycle in glucose-replete conditions, corroborating previous studies in non-cyst forming Type I tachyzoites [15,27]. Our observation that tachyzoites lacking pgm1 could differentiate into bradyzoites in the absence of glucose further supports the nonessential role of PGM1 and PGM1-dependent glucose-6-phosphate production in tachyzoites as suggested by Imada et al. [29]. Interestingly, PGM1 has been implicated in Ca 2+ -dependent microneme secretion in tachyzoites [11,13,15], and thus, like functionally characterized PGMs in other organisms [9,30], it may play an unconventional role during Toxoplasma development.
Additionally, the absence of pgm1 did not abrogate amylopectin biosynthesis and storage, probably due to functional compensation with PGM2. While both pgm1 and pgm2 transcripts are higher in bradyzoites than tachyzoites [28], the proteins share only 25% homology. PGM2 has a significantly lower enzymatic activity than PGM1 [29]. Interestingly, Saha et al. [15] demonstrated that PGM2 didn't compensate for the deletion of PGM1 in the context of Ca 2+ -regulated microneme secretion in tachyzoites.
Although glycolysis is not required for tachyzoite viability, it is critical for tissue cyst formation and pathogenesis in mice [31]. Parasites lacking hexokinase, the first enzyme in glycolysis that catalyzes the phosphorylation of glucose to glucose-6-phosphate, produce smaller cysts in vitro [31]. This phenotype was recapitulated in pgm1-defective parasites, further supporting the importance of glycolytic intermediates during cystogenesis. While the bradyzoite burden of PGM1-deficient cysts and their infectivity remain to be determined, it is plausible that the parasites inside these mutant cysts have decreased resistance to proteases and are less infectious following oral infection, as previously shown for Bradyzoite Pseudokinase 1 (BPK1) mutants [32]. Because the absence of PGM1 does not significantly alter the replication rate of tachyzoites, it is conceivable that the bradyzoite burdens in the mutant and wildtype cysts be comparable. This assertion is supported by Watts et al. who showed that cyst size is not a strong predictor bradyzoite burden [33].
Overall, this study suggests that PGM1 is not critical for Toxoplasma growth and differentiation; however, it is required for optimal cyst maturation, which is critical for the establishment of chronic Toxoplasma infections. Future studies are needed to parse out the interplay and diverse activities of Toxoplasma PGMs Fig. 2 Disruption of pgm1 and growth assays. A Schematic representation of disruption of pgm1 using CRISPR-Cas9 gene-editing system for nonhomologous insertion of the hxgprt selectable marker cassette. The dotted line represents the region in the first exon of pgm1 targeted by the small guide RNA (sgPGM1). B Image of DNA gel electrophoresis of PCR1-3 performed using DNA from wildtype (WT) and mutant (Δpgm1) to demonstrate integration of the hxgprt expression cassette at the pgm1 locus. The expected product for PCR1 (212 bp) was obtained only for WT while products for PCR2 (813 bp) and PCR3 (1185 bp) were amplified only with Δpgm1 DNA. C Intracellular growth. HFFs were infected with 1.2 × 10 5 WT or Δpgm1 parasites for 24 h in cDMEM. Monolayers were fixed and stained with antibodies raised against SAG1 (tachyzoite surface marker) and GRA7 (PV marker). Intracellular parasites were enumerated in at least 20 vacuoles/strain/experiment, N = 3 independent experiments; error bars = standard error of the mean; p-value was determined by Chi-square test. D Total numbers of plaques counted 10 days after infection of HFFs with 250 WT or Δpgm1 parasites. E Plaque areas were determined for 85 WT and 109 Δpgm1 plaques using Fiji/ImageJ in pixels 2 , N = 3 replicates/strain in a single experiment, error bar = standard deviation; ns: p-value > 0.05 by nonparametric Mann-Whitney test and understand how they affect central carbon metabolism and developmental differentiation in this ubiquitous parasite. PGMs are among several metabolic enzymes whose transcripts are significantly upregulated during chronic infection with Toxoplasma. While PGM1 was our initial focus, future work will evaluate the contributions of other poorly characterized glycolytic enzymes identified in our bioinformatic search. Similar to PGM1, these enzymes may be critical to Toxoplasma biology and serve as potential therapeutic targets against chronic toxoplasmosis.\===
Domain: Biology Medicine. The above document has
* 2 sentences that start with 'Together, our findings',
* 2 sentences that start with 'The number of parasites',
* 2 sentences that end with 'the lytic cycle',
* 2 paragraphs that start with 'Toxoplasma gondii is',
* 2 paragraphs that end with 'optimal cyst development in vitro'. It has approximately 2072 words, 104 sentences, and 29 paragraph(s).
|
[{"added": "2023-01-20T14:40:34.149Z", "created": "2022-05-21T00:00:00.000", "id": "256019673", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1186/s13104-022-06073-5", "pdf_hash": "195233d550cc06fc9138795974add0d6e01a6db8", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:5511", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "195233d550cc06fc9138795974add0d6e01a6db8", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Enhanced exosome secretion in Down syndrome brain - a protective mechanism to alleviate neuronal endosomal abnormalities
A dysfunctional endosomal pathway and abnormally enlarged early endosomes in neurons are an early characteristic of Down syndrome (DS) and Alzheimer’s disease (AD). We have hypothesized that endosomal material can be released by endosomal multivesicular bodies (MVBs) into the extracellular space via exosomes to relieve neurons of accumulated endosomal contents when endosomal pathway function is compromised. Supporting this, we found that exosome secretion is enhanced in the brains of DS patients and a mouse model of the disease, and by DS fibroblasts. Furthermore, increased levels of the tetraspanin CD63, a regulator of exosome biogenesis, were observed in DS brains. Importantly, CD63 knockdown diminished exosome release and worsened endosomal pathology in DS fibroblasts. Taken together, these data suggest that increased CD63 expression enhances exosome release as an endogenous mechanism mitigating endosomal abnormalities in DS. Thus, the upregulation of exosome release represents a potential therapeutic goal for neurodegenerative disorders with endosomal pathology.
Introduction
A dysfunctional endosomal pathway and abnormally numerous and enlarged early endosomes are found within vulnerable Alzheimer's disease (AD) neurons early in life [7]. These characteristics are also seen in Down syndrome (DS) [10], a genetic disorder caused by trisomy of human chromosome 21 that leads to earlyonset AD [55]. Early endosomal abnormalities correlate with developmental brain abnormalities and intellectual disabilities in DS patients [35,36,54]. Dysfunction within the endosomal pathway, which occurs in several neurodegenerative disorders [53] in addition to AD and DS, results in the accumulation of material in neuronal endosomes with subsequent neuronal vulnerability and degeneration [39]. Early endosomal changes have also been observed in vitro in fibroblasts derived from DS individuals [8] and in neurons of well-established mouse models of DS [9,26,44]. In this study we have utilized the trisomic mouse model Ts[Rb (12.17 16 )]2Cje [52] (hereafter called Ts2), which presents phenotypic and pathological features [26,29,52] similar to the extensively studied Ts65Dn mouse model [9,15,20,22], but has a genetic configuration with advantages in breeding [26,32,52].
Early endosomes are the first vesicular compartment along the endocytic pathway, which internalizes cargoes for delivery to late endosomes/multivesicular bodies (MVBs) for either degradation in lysosomes or for release into the extracellular space via exosomes. Exosomes are extracellular vesicles (EVs) formed as intraluminal vesicles (ILVs) by inward invagination of the membrane of MVBs, which are released into the extracellular space upon fusion of MVBs with the plasma membrane [12]. ILVs formation is regulated by the endosomal sorting complexes required for transport (ESCRT) machinery as well as by an ESCRT independent system that includes proteins from the tetraspanin family [11]. While most tetraspanins are present in the plasma membrane, CD63 is uniquely enriched in the membrane of MVBs [42]. Intracellular trafficking of MVBs towards the plasma membrane is regulated by several Rab GTPases and SNARE proteins [17]. Among them, rab35 likely plays a role in the docking of MVBs to the plasma membrane [24].
We investigated whether endosomal abnormalities affect the exosome secretory pathway in DS and explored the molecular mechanism underlying changes in exosome secretion in vivo in the brain of DS patients and the Ts2 murine model, and in human DS fibroblasts grown in vitro. The results support the hypothesis that in neurodegenerative disorders with endosomal-lysosomal dysfunction, such as DS, exosome secretion serves as a disposal mechanism of toxic material that is abnormally accumulated in endosomal compartments. Given that endosomal abnormality causes neuron degeneration [39], mitigation of this pathology by enhanced exosome release can be expected to be protective and a target for therapeutic intervention.
Materials and methods
Mice Ts[Rb (12.1716)]2Cje (Ts2) [52] and normal disomic (2N) littermates were studied at 3, 8, 12 and 24 months of age. Both females and males were used for all analyses. All animal procedures were performed following the National Institutes of Health guidelines with approval from the Institutional Animal Care and Use Committee at the Nathan S. Kline Institute for Psychiatric Research.
Human brain tissues
Postmortem samples of Brodmann area 9 (BA9) obtained from human DS and control subjects (Table 1) were kindly provided by Dr. Jerzy Wegiel, Director, Brain Bank for Developmental Disabilities and Aging, Institute for Basics Research in Developmental Disabilities, Staten Island, New York. Tissue accession and use protocols were approved by the Nathan S. Kline Institute.
EVs isolation from brain tissue
EVs were isolated from frozen samples of cortical human brain region BA9 and from the right murine hemibrains (without the cerebellum and the olfactory bulbs). In each experiment, EVs were simultaneously isolated from a brain of either a DS patient or a Ts2 mouse and from an age-matched 2N control. Brain EVs were isolated and purified as we have previously described [40,41]. Briefly, frozen brain tissues were treated with 20 units/ml papain (Worthington, Lakewood, NJ) in Hibernate A solution (HA, 3.5 ml/sample; BrainBits, Springfield, IL) for 15 min at 37°C. The brain tissue was gently dissociated in 6.5 ml of cold HA supplemented with protease inhibitors, centrifuged at 300 g for 10 min at 4°C to discard the cells, and the supernantant was sequentially filtered through a 40 μm mesh filter (BD Biosciences, San Jose, CA) and a 0.2 μm syringe filter (Corning Life Sciences, Teterboro, NJ). The filtrate was sequentially centrifuged at 4°C, at 2000 g for 10 min and 10,000 g for 30 min to discard membranes and debris, and at 100,000 g for 70 min to pellet the EVs. The EVs pellet was resuspended in 60 ml of cold PBS (Thermo Fisher Scientific), and centrifuged at 100,000 g for 70 min at 4°C. The washed EVs pellet was resuspended in 2 ml of 0.95 M sucrose solution and inserted inside a sucrose step gradient column (six 2-ml steps starting from 2.0 M sucrose up to 0.25 M). The sucrose step gradient was centrifuged at 200,000 g for 16 h and fractions were collected from the top of the gradient. The fractions were diluted in cold PBS and centrifuged at 100,000 g for 70 min. Sucrose gradient fraction pellets were resuspended in 30 μl of cold PBS.
Quantification of EVs levels in the brain
Protein levels from the sucrose gradient purified fractions b, c and d, were quantified using the BCA Protein Assay Kit (Thermo Fisher Scientific). Acetylcholine esterase (AChE) activity assay, commonly used to quantify exosomes, was also performed on fractions b, c, and d, as we have previously described [40,41]. Total EVs protein levels and EVs AChE activity were normalized to total brain sample protein content. AChE activity was additionally normalized to total EVs protein.
In-gel digestion
EVs samples (10 μg protein for each sample) isolated from the brains of three Ts2 mice and three littermate controls were run into a 4% SDS-PAGE gel until all proteins entered the gel and stacked together. After staining by EZ-Run™ Protein Gel Staining Solution (Fisher BioReagents™, Boston, PA) the gel bands were excised, reduced, alkylated, and digested in-gel with sequence grade modified trypsin (Promega, Madison, WI) as described [48]. Peptides were extracted by 50% acetonitrile in 5% formic acid, dried by vacuum centrifugation, and desalted using StageTips [43] packed with C18 beads.
MS data processing and analysis
The data analysis was performed with MaxQuant [13] software (Version 1.2.7.0, Max Planck Institute of Biochemistry) supported by Andromeda [14] to search a UniProt Mouse fasta database with 50,316 protein entries. Mass tolerance was set to 7 ppm for peptide masses and 20 ppm for HCD fragment ion masses with carbamidomethylation as a fixed modification and protein N-terminal acetylation and methionine oxidation as variable modifications. Up to two missed cleavages were allowed while requiring strict trypsin specificity and minimum sequence length of seven. Peptides and proteins were identified with a false discovery rate (FDR) of 1%. GO term enrichment analysis was done using Panther ( [URL] of mRNA levels
qPCR was performed on microdissected mouse hippocampi with RNA purified using the miRNAeasy mini kit (Qiagen, Germantown, MD). Taqman qPCR primers for CD63 (Mm01966817_g1) and rab35 (Mm01204416_ml) (Life Technologies) were utilized with samples assayed on a real-time qPCR cycler (7900HT, Life Technologies) in 96-well optical plates with coverfilm as described previously [2-4, 18, 29]. The ddCT method was employed to determine relative gene level differences between Ts2 and 2N mice [1,3,4,29]. Succinate Dehydrogenase Complex Flavoprotein Subunit A (Sdha, Mm01352360_m1) qPCR products were used as a control housekeeping gene. Negative controls consisted of the reaction mixture without input RNA and reaction mixture without Superscript III enzyme. PCR product synthesis was modeled as a function of mouse genotype, using mixed effects models to account for the correlation between repeated assays on the same mouse [37].
Cell culture and interference RNA (siRNA) depletion EVs isolation from fibroblast-conditioned media and quantification of exosome secreted levels EVs were isolated from conditioned media of DS and diploid controls fibroblasts. Cell culture media were replaced with DMEM supplemented with 10% FBS, which was depleted of EVs by ultracentrifugation at 100,000 g for 70 min, 100 units/ml of penicillin and 100 μg/ml streptomycin (Thermo Fisher Scientific), and 2 mM GlutaMAX (Thermo Fisher Scientific). The conditioned media were collected every 24 h and fresh medium was added to the cells over 3 days. EVs were isolated from the conditioned media as previously described [50]. Briefly, conditioned media were centrifuged at 300 g for 10 min. The supernatant was sequentially centrifuged at 2000 g for 10 min, at 10,000 g for 30 min, and finally at 100,000 g to pellet the EVs. The EVs were resuspended in cold PBS and then centrifuged at 100,000 g. The washed EVs pellet was resuspended in PBS and lysated in 2X radioimmune precipitation assay (RIPA) buffer. Equal volumes of EVs lysates were loaded on the electrophoresis gel. Exosome levels were quantified by intensity of the bands obtained by Western-blot analysis of the exosomal markers CD63, Alix, and TSG101 using the ImageJ software (NIH).
Morphometric analyses of endosomes
DS and 2N control fibroblasts grown on glass coverslips were fixed with cold 2% PFA in PBS for 20 min, subsequently blocked with 10% FBS for 1 h at room temperature, and incubated overnight with the anti-EEA1 antibody (1:200, Cat# 07-1820, EMD Millipore) at 4°C. Cells were incubated with the secondary antibody Alexa 568 (1:500, Cat# A11036, Thermo Fisher Scientific) for 1 h at room temperature. Confocal microscopy was carried out using a LSM 510 microscope (Carl Zeiss). For the quantification of EEA1 staining, digital images were taken at ×63 magnification with comparable background intensities among conditions. The number of EEA1-positive endosomes and the endosomal area per cell were analyzed for each fibroblast in captured images taken in a single plane of focus by the ImageJ morphometry software (NIH) by a genotype-and condition-blinded observer. Individual fibroblasts and EEA1-positive endosomes were outlined and area measured. For each condition, a total of 20 fibroblasts per experiment were assessed at random from several fields.
Statistical analyses
Data are presented as mean ± SEM. Unpaired, twotailed, Student's t-test statistical analysis were used to compare differences in EVs levels in DS brains and 2N controls, Ts2 brains and age-matched 2N controls, and to compare protein levels in Ts2 EVs, DS brain homogenates and DS fibroblast cell lysates to 2N controls. qPCR results were analyzed using one-way ANOVA and posthoc analysis (Neumann-Keuls test; level of statistical significance was set at p < 0.05). Unpaired, two-tailed, Student's t-test statistical analyses were used to determine in vitro differences in exosome secretion between untreated 2N with DS cells, and within the same line transfected with CD63 siRNA as compared to cells transfected with negative control siRNA. Endosomal changes in 2N and DS cells were assessed by one-way ANOVA followed by Tukey post-hoc multiple comparison test.
Results
Higher levels of EVs in the brain of DS patients and Ts2 mice compared to diploid controls Brain EVs were isolated as previously described [40,41] from DS patients and age-matched normal controls (Table 1), and from the right hemibrain of 3-, 8-, 12and 24-month-old Ts2 mice and 2N littermates. Separation of the EVs on a sucrose gradient resulted in 7 fractions, from a, the least dense, to g, the densest fraction, and Western-blot analysis showed that fractions with densities higher than 1.07 and lower than 1.17 (fractions b, c and d) were immunoreactive to Flotillin-1 and Flotillin-2, lipid raft proteins found in EVs, and established exosomal markers (Fig. 1a). Quantification of the exosome-enriched EVs fractions b, c, and d was performed by measuring the total protein content in the fractions normalized to total protein content in the brain tissue. In the samples of the frontal cortex of DS patients we found higher EVs levels compared to 2N controls (DS/2N ratio = 1.39, p = 0.022) (Fig. 1b). A similar increase in EVs levels was found in the brain extracellular space of the DS mouse model Ts2 at 12 (Ts2/2N ratio = 1.20, p = 0.0054) and 24 (Ts2/2N ratio = 1.29, p = 0.048) months of age compared to littermate controls, but not in younger, 3-(Ts2/2N ratio = 1.08, p = 0.31) and 8-month-old (Ts2/2N ratio = 1.18, p = 0.16) mice (Fig. 1c). We also measured the levels of exosome-enriched EVs by quantifying the activity of AChE, a protein that is specifically sorted into exosomes [27,46]. The AChE activity measurements were normalized to total protein content in the brain tissue and the results supported the finding of DS-induced higher levels of exosomes with a trend in the brain extracellular space of DS patients (DS/2N ratio = 1.29, p = 0.14) (Fig. 1d), and conclusively in 12-(Ts2/2N ratio = 1.26, (Fig. 1e). No significant differences in the levels of AChE activity were found in the brains of 3-(Ts2/2N = 1.08, p = 0.28) and 8-month-old (Ts2/2N = 1.12, p = 0.34) Ts2 mice compared to 2N littermates (Fig. 1e). Additionally, AChE activity was normalized to EVs protein content instead of total protein in the brain to estimate the AChE levels per EV. AChE activity levels per EV in the brains of DS patients were not different from 2N controls (DS/2N ratio = 0.96) (Fig. 1f ), similar to the levels of AChE per EV in the brains of 3-(Ts2/2N ratio = 1.00), 8-(Ts2/2N ratio = 0.95), 12-(Ts2/2N ratio = 1.05) and 24-month-old (Ts2/2N ratio = 1.06) Ts2 mice compared to age-matched 2N controls (Fig. 1g). Altogether these results argue that higher number of exosome-enriched EVs loaded with similar levels of AChE are present in the brain extracellular space of DS patients and 12-and 24-month-old Ts2 mice compared to 2N controls.
Higher content of CD63 and rab35, regulators of ILVs generation and release, in Ts2 brain EVs as compared to 2N To identify candidate proteins whereby a change in expression could be mechanistically linked to increased levels of exosome-enriched EVs in DS, we performed label-free liquid chromatography-mass spectrometry (LC-MS)-based proteomics on EVs (pooled from the sucrose gradient fractions b, c, and d) isolated from the brains of 12-month-old Ts2 and 2N littermates. We identified 1587 and 1659 proteins in 2N and Ts2 EVs samples, respectively, with a large overlap in protein identification between biological replicates: 1252 common proteins in the samples from three 2N mice (Fig. 2a), and 1363 common proteins in the samples from three Ts2 mice (Fig. 2b). 1549 proteins were common to both genotypes (Fig. 2c). Gene ontology (GO) enrichment analysis of the common proteins identified by LC-MS showed that the most abundant components were proteins characteristic of vesicles, extracellular vesicles, extracellular organelles, and exosomes (Fig. 2d). Among the top 100 proteins that are often identified in exosomes (Exocarta, exosome database available online; [URL]/) 70 were identified in this study (Additional file 1: Table S1). Thus, using an unbiased approach, this analysis confirmed that vesiclerelated proteins are highly enriched in brain EVs preparations, with specific enrichment of exosomal proteins. The LC-MC analysis revealed a trend for higher levels of CD63 and rab35 in Ts2 EVs compared to 2N that was confirmed and found to be significantly different by Western-blot analysis (Fig. 2e).
Higher protein expression levels of CD63 and rab35 in DS brains
To determine whether the differences found in EVs protein levels reflect differences in protein expression levels within the cell leading to changes in exosome secretion, we analyzed the expression levels of CD63 and rab35 in human brain homogenates by Western-blot. Higher expression levels of CD63 (DS/2N ratio = 1.53; p = 0.030) and rab35 (DS/2N ratio = 1.85; p = 0.034) were observed in the frontal cortex of human DS patients as compared to 2N controls (Fig. 3a). In contrast, no significant differences were detected in the levels of the exosomal markers Alix (DS/2N ratio = 1.12; p = 0.66) and TSG101 (DS/2N ratio = 1.03; p = 0.88) between DS and 2N controls (Fig. 3b).
Enhanced exosome secretion by DS fibroblasts
Given the endosomal abnormalities observed in cultured human forearm skin fibroblasts isolated from DS patients [8], similar to the brains of DS patients [10] and Ts2 mice [26], we hypothesized that exosome release into the media will be enhanced in order to alleviate the endosomal pathology. Western-blot analysis of EVs secreted into cultured media was performed with antibodies to CD63, rab35, Alix, and TSG101. Quantification of the bands and normalization to total cell proteins revealed that the levels of CD63 (DS/2N ratio = 1.70, p = 0.023), Alix (DS/2N ratio = 1.60, p = 0.030) and TSG101 (DS/2N ratio = 1.56, p = 0.022) were higher in DS EVs compared to 2N controls (Fig. 4a). The rab35 protein was not detected in fibroblasts EVs, potentially due to low content. We also studied the levels of these proteins in cell lysates and found that the expression levels of CD63 (DS/2N ratio = 2.30, (See figure on previous page.) Fig. 1 Higher levels of exosome-enriched EVs in the brains of DS patients and of Ts2 mice as compared to age-matched diploid controls. a Representative Western-blots of EVs isolated from human brain tissue and purified on a sucrose step gradient column. The sucrose gradient fractions b, c and d showed the presence of the exosomal proteins Alix and CD63, and the EVs proteins Flotillin-1 and Flotillin-2. b Quantification of total protein levels of EVs isolated from the brain extracellular space of DS patients, normalized to brain tissue protein levels, showed higher EVs levels compared to controls. c Higher EVs levels were also found in the brain extracellular space of 12-and 24-month-old Ts2 mice compared to 2N littermates. No significant differences were found in total EVs protein levels of 3-and 8-month-old Ts2 mice compared to controls. Similar results were obtained when AChE activity levels were measured in EVs isolated from the brain extracellular space of DS patients (d) and Ts2 mice (e) as compared to 2N controls when normalized to brain tissue protein content. AChE activity levels normalized to EVs protein content were not different between brains of DS patients (f) and Ts2 mice (g) compared to 2N controls. EVs levels are presented as trisomic to 2N ratio. Student t-test, n = 5 (DS and 2N human brains), n = 4 (3-and 24-month-old), n = 5 (8-month-old), and n = 7 (12-month-old) brains of Ts2 and 2N mice (*p < 0.05; **p < 0.01; ***p < 0.001) p = 0.0016) and rab35 (DS/2N ratio = 1.56, p = 0.024) were higher in DS than in 2N fibroblasts (Fig. 4b), consistent with the observation in DS postmortem brain (Fig. 3a). However, the levels of Alix (DS/2N ratio = 0.62, p = 0.13) and TSG101 (DS/2N ratio = 0.95 p = 0.754) were not significantly different in the cell lysates obtained from DS as compared with 2N fibroblasts (Fig. 4c). Thus, the higher EVs levels of Alix and TSG101 in the media cultured by DS fibroblasts are due to higher levels of exosome secretion by DS fibroblasts as compared to 2N cells.
Higher mRNA levels of CD63 but not rab35 in Ts2 brains
To determine whether differences in CD63 and rab35 protein levels were due to differences in mRNA levels, qPCR analyses were performed on RNA extracted from the hippocampus of 12-month-old Ts2 and 2N littermates. Upregulation of CD63 RNA expression was observed in Ts2 brains as compared with 2N controls (29% higher p = 0.0286; ddCt values expressed as mean ± SEM were 2.3558 ± 0.1966 in 2N and 3.0382 ± 0.1954 in DS; n = 5 per genotype). No significant difference was found for rab35 transcript levels.
A role for CD63 in exosome secretion as a mechanism to modulate endosomal pathology We induced a CD63 loss of function in fibroblasts to investigate the role played by CD63 in the regulation of exosome secretion. Our small interfering RNA (siRNA) strategy targeting CD63 transcripts successfully knocked down CD63 levels in 2N and DS fibroblasts (Fig. 5a). The reduction in CD63 expression was 77.68 ± 5.1% and Fig. 2 Proteomic analysis of exosome-enriched mouse brain EVs. A Venn diagram shows the overlap between biological replicates within each genotype for 2N (a) and Ts2 (b) EVs. c 1549 proteins were common to both genotypes. 91 and 18 proteins were unique to Ts2 and 2N samples, respectively. d GO analyses for components of the 1549 proteins common to both genotypes revealed enrichment of extracellular vesicles and exosomal proteins in the EVs preparations. P-values for each cellular category are shown on the right. e Representative Western-blots with anti-CD63 and anti-rab35 antibodies of the 2N and Ts2 sucrose gradient EVs fractions and corresponding quantification. Student t-test, n = 7 (*p < 0.05) 73.03 ± 3.9% for 2N and DS cells, respectively. The exosomes secreted into the medium over the course of three days after the transfection were quantified by Westernblot analysis (Fig. 5b). No significant changes were found in exosome secretion by 2N cells following CD63 knockdown as evidenced by similar levels of Alix (CD63/control siRNA ratio = 1.01, p = 0.37) and TSG101 (CD63/ control siRNA ratio = 1.07, p = 0.25) (Fig. 5c). The levels of CD63 in 2N EVs showed a trend to decrease (CD63/ control siRNA ratio = 0.70, p = 0.08), which was expected due to CD63 knockdown. In contrast, reducing CD63 expression led to lower levels of exosomes secreted into the cell culture media by DS fibroblasts as seen by reduced levels of exosomal Alix (CD63/control siRNA ratio = 0.82, p = 0.02), TSG101 (CD63/control siRNA ratio = 0.73, p = 3.7 × 10 −7 ) and, as expected, CD63 (CD63/control siRNA ratio = 0.56, p = 4.8 × 10 −6 ) (Fig. 5d). When the expression levels of Alix and TSG101 were normalized to EVs protein content, no changes were detected in the levels of Alix (CD63/control siRNA ratio = 1.04, p = 0.69) and TSG101 (CD63/ control siRNA ratio = 0.95, p = 0.76) in EVs from DS cells in which CD63 was reduced compared to controls, indicating that the loading levels of Alix and TSG101 per EV are similar after CD63 knockdown in DS cells. Therefore, the lower intensity of the bands for Alix and TSG101 shown in Fig. 5b is the result of lower number of EVs and rules out the possibility that CD63 knockdown is responsible for a change in EVs composition.
To determine whether changes in exosome secretion affect endosomal compartments, endosomal morphology was studied in 2N and DS fibroblasts using early endosome antigen 1 (EEA1) immunocytochemistry (Fig. 5e). As previously demonstrated [8], the number (Fig. 5f) and area (Fig. 5g) of endosomes were significantly higher in DS cells compared to 2N. While CD63 knockdown did not affect endosomes in 2N cells ( Fig. 5f-g), the number of endosomes was higher (Fig. 5f) and a trend to increased endosomal area (Fig. 5g) was observed in DS fibroblasts in which CD63 expression was reduced compared to DS cells transfected with negative control siRNA.
Discussion
Different types of EVs have distinct intracellular origin, including plasma membrane-derived microvesicles, apoptotic vesicles, and MVBs-derived exosomes (reviewed in [28,30,51]). We found higher levels of EVs in the brain extracellular space of DS patients and Ts2 mice compared to controls. Proteomic analysis confirmed that exosomal proteins were highly enriched in the brain EVs preparations, and identified higher levels of the exosomal marker CD63 and the exosome-related protein rab35 in brainderived Ts2 EVs compared to 2N. These LC-MS data suggest that the higher levels of EVs found in the brain extracellular space of DS patients and Ts2 are due to higher levels of exosomes. The brains of DS patients at advanced ages likely have significant AD neuropathology, not present in age-matched control brains, raising the possibility that enhanced exosome secretion is due to AD Table 1) and quantification showing the overexpression of CD63 and rab35 in DS brains compared to 2N. b No differences in the levels of Alix or TSG101 were detected in homogenates of human DS brains compared to 2N controls. β-actin was blotted as an internal control for loading. Student t-test, n = 5 (*p < 0.05) neuropathology. However, we have previously reported that Tg2576 mice, overexpressing APP, do not secrete more exosomes than their littermate controls at an age when amyloid pathology has fully developed [40], suggesting that AD neuropathology is not causing higher exosome secretion. Further, we found that DS fibroblasts secrete more exosomes into the cell culture media than 2N cells. This suggests that the higher exosome levels found in vivo is a consequence of enhanced exosome secretion rather than altered exosome stability or less exosome clearance in the brain extracellular space.
Once early endosomal cargoes are delivered to late endosomes there are two possible fates, either lysosome degradation or exosome release. The endosomal function under conditions of increased early endosomal drive in DS [8] would require a corresponding increase in either or both of these pathways. Since we measured a statistically enhanced exosome secretion in the brain of 12-month-old and older Ts2 mice and the endosome enlargement phenotype is observed in neurons of 4-month-old mice [26], we hypothesize that enhanced exosome secretion constitutes a delayed cellular response designed to lower the size and number of endosomal compartments in DS by shedding more endosomal content into the brain extracellular space (Fig. 6). A similar mechanism of exosome release was suggested for the cell to partially overcome the accumulation of free cholesterol within late endosomes/lysosomes in the Niemann-Pick Type C disease [49].
The expression levels of the ESCRT proteins Alix and TSG101 did not differ in the brains of human DS patients and in DS fibroblasts as compared with 2N controls. These data suggest that the ESCRT machinery is not the trigger behind enhanced exosome secretion. We found that CD63 is overexpressed in DS brains and in DS fibroblasts compared to 2N controls. Similarly, a recent study reported higher protein levels of another member of the tetraspanin family, tetraspanin-6, in the brains of AD patients, and in vitro experiments related tetraspanin-6 overexpression to the generation of more b Elevated expression levels of CD63 and rab35 and (c) no differences in the levels of Alix and TSG101 in lysates of DS fibroblasts compared to 2N controls, as shown by the representative Western-blots and corresponding quantification. β-actin was blotted as an internal control for loading. Student t-test, n = 5 independent experiments (*p < 0.05; **p < 0.01) Fig. 5 Effect of CD63 knockdown on exosome secretion and endosomal pathology in DS cells. 2N and DS fibroblasts were transfected with either CD63 or negative control siRNAs. a CD63 knockdown was confirmed by Western-blot analysis of cell lysates. b Over 3 days, exosomes were collected from the cell culture media and quantified by Western-blot analysis for the exosomal markers CD63, TSG101, and Alix. c No significant changes were observed in exosome release by 2N cells following CD63 silencing compared to controls. d DS fibroblasts in which CD63 was silenced showed decreased release of exosomes as seen by lower levels of exosomal TSG101 and Alix as compared to control DS cells. Student t-test, n = 4 independent experiments (*p < 0.05; ***p < 0.001). e Early endosomes were immunolabeled with an anti-EEA1 antibody of transfected 2N and DS cells (calibration bar = 20 μm). f No significant changes in the endosomal number were detected in CD63-reduced 2N fibroblasts compared to control 2N cells, while a significant increase in number of endosomes was observed in DS cells following CD63 knockdown. g No significant differences were found in the area occupied by endosomes in 2N and DS cells after knocking down CD63, however DS fibroblasts showed a trend for an increase. Note that the number and area occupied by endosomes in DS fibroblasts is significantly higher than in 2N under basal (control-siRNA) conditions (f, g). Area is expressed in pixels per cell. One-way ANOVA followed by Tukey post-hoc multiple comparison test, n = 4 independent experiments (**p < 0.01; ***p < 0.001; ****p< 0.0001) exosomes [19]. Moreover, we report that CD63 transcription is enhanced in the brains of Ts2 mice. In contrast, higher levels of protein expression, but not mRNA was found for rab35, a regulator of exosome secretion [24]. We suggest that higher rab35 protein levels may be downstream of the induction of CD63, in response to ILVs accumulation by releasing excess MVBs load into the extracellular space. Upregulation of CD63 gene rules out the possibility that higher CD63 protein levels in DS is due to its accumulation as a consequence of endosomal pathology. We then investigated whether enhanced exosome secretion by CD63 has a role in alleviation of endosomal pathology. DS fibroblasts secreted more exosomes into the cultured media compared to 2N cells and CD63 knockdown reduced exosome secretion by the DS cells. In support of an inter-relationship between exosomal release and endosomal pathology, CD63 knockdown also caused changes in endosomes of DS fibroblasts, characteristic of the endosomal abnormalities reported in DS patients [10], DS fibroblasts [8], and in DS mouse models [9,26]. These data argue that partially blocking exosome release in cells with endosomal pathology aggravates the intracellular accumulation of endosomal membranes. In contrast, in normal cells without endosomal pathology, silencing CD63 did not affect exosome release or endosome accumulation, implying that exosome release can be regulated in 2N cells even under changes in expression of proteins involved in exosome generation, while DS cells lose this capability. Therefore, these data suggest that CD63 is involved in the formation of more ILVs in MVBs when the system is compromised, similar to the finding that CD63 overexpression is associated with higher exosome release in fibroblasts from patients with systemic sclerosis [38]. It cannot be ruled out that in the diseased brain, this protective mechanism of the exosome secretory pathway to relieve DS neurons of accumulated endosomal contents might be outweighed by the propagation of toxic material by exosomes throughout the brain. The ability of exosomes to promote the spread of the disease has emerged as a general mechanism of propagation of neurodegenerative disorders [23,56] and it was recently reported that neuronal exosomes obtained from blood of DS patients have elevated levels of amyloid-β peptides and phosphorylated-tau [21]. Further, reducing exosome secretion has been suggested as a potential therapeutic intervention for AD [6,16]. However, blocking exosome production resulted in exacerbated behavioral and pathological defects in a transgenic mouse overexpressing a mutant form of the toxic protein TDP-43 [25], negating inhibition of exosome release in neurodegenerative disorders as a therapeutic approach.
Conclusions
In summary, the data presented here show a role for exosomes in the regulation of endosomal function in DS, implicate CD63 in driving exosome release in the DS brain, and suggest that this is a protective mechanism to alleviate the endosomal pathology. Since the naturally occurring enhancement of exosome release in the brain of DS patients is not sufficient to alleviate the endosomal pathology, which is already observed in fetuses and worsens with age, therapeutic approaches that enhance exosome secretion even further may be beneficial. Indeed, apart from pathological proteins, exosomes naturally transport cargoes with therapeutic properties [31,33,34,45] and can be engineered to target neurons while carrying specific molecules for therapeutics [5]. Increasing our knowledge of brain exosome secretion and its potential therapeutic effects is necessary to provide new insights into the mechanisms of disease and may help to develop novel therapeutic strategies for neurodegenerative disorders with accumulation of toxic material in endosomes, like DS and AD.
==
Domain: Biology Medicine
|
[{"added": "2023-01-31T14:34:36.717Z", "created": "2017-08-29T00:00:00.000", "id": "256405980", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1186/s40478-017-0466-0", "pdf_hash": "d1ea3152eadba9059ba5f1887a8a9c9599ffddc7", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:8655", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "d1ea3152eadba9059ba5f1887a8a9c9599ffddc7", "year": 2017}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Molecular epidemiological study of hepatitis B virus in blood donors from five Chinese blood centers
Although the genetic variability of hepatitis B virus (HBV) in HBV-infected patients has been extensively studied, reports on genotypes, subtypes and mutations in the S region of HBV strains from Chinese blood donors are limited. In this study, 245 blood samples from HBsAg-positive blood donors were collected from five geographically diverse blood centers in China. The S region of HBV was amplified, and the HBV genotype and subtype were determined. The amino acid sequences of the S region were aligned, and mutations related to the failure of immunization and HBsAg detection were determined. Of the 245 samples, 228 (93 %) were genotyped successfully. We found that genotypes B, C, D and A accounted for 58.8 %, 21.9 %, 6.6 % and 3.95 % of the isolates, respectively. The distribution of HBV antigen subtypes was as follows: adw (67.6 %), adr (23.3 %) and ayw (8.7 %). Mutations were present in 39 (17.1 %) of 228 samples in the major hydrophilic region (MHR) of the S region. This study demonstrated that HBV genotype/subtype B/adw was the most frequent strain circulating in HBV-infected Chinese blood donors, followed by C/adr. The occurrence of MHR mutants in HBV-infected blood donors and the potential failure to detect some of them in collected units poses a threat to transfusion safety.
Introduction
Hepatitis B virus (HBV) is an aetiological agent of acute and chronic liver disease, including fatal fulminant hepatitis, cirrhosis and hepatocellular carcinoma, which is one of the most common human cancers and causes of death worldwide [1]. It has been estimated that more than 2 billion of the global population have been infected with HBV. Of these, approximately 360 million people are chronically infected, and an estimated 500,000 to 700,000 people die from complications of HBV infection each year worldwide [2]. HBV infection is highly endemic in China, with 5.84 % prevalence of HBV surface antigen (HBsAg) in the population of 1-59 years of age in 2007 [3].
HBV demonstrates remarkable genetic variability. Historically, using hepatitis B surface antigen (HBsAg) subtyping techniques based on specific antibodies, the genetic variability of the HBV envelope allowed classification into For the NHLBI Retrovirus Epidemiology Donor Study-II (REDS-II), International Component. ten serologic subtypes, designated ayw1, ayw2, ayw3, ayw4, ayr, adw2, adw3, adw4, adrq?, and adrq- [4]. More recently, HBV has been classified into eight genotypes (A to H, in the order of discovery) with each genotype differing from the others by 8 % nucleotide divergence in the complete genome and 4 % in the sequence of the S gene [4]. Currently, HBV subtypes can be deduced from amino acid sequences at positions 122, 127, 134, 159, 160, 177 and 178 [5]. HBV genotypes have a distinct geographical distribution [6,7]. For genotype A, subgenotype A2 is found mainly in northwestern Europe, North America and in Australians of European origin, while the more prevalent A1 is found in East and South Africa and along the coast of the Indian Ocean; genotypes B and C are prevalent in Southeast Asia, China and Japan; genotype D is spread worldwide, but it is predominant in the Mediterranean region and the Middle East; genotype E is almost entirely restricted to Africa; genotype F is found in Central and South America; genotype G has been reported in France and North America; and genotype H predominates in Central America. Recently, a ninth genotype (I) was tentatively proposed for HBV strains detected in Laos and a tenth genotype (J) was proposed for a HBV strain detected in a Japanese HCC patient [8,9]. In China, genotypes C and B are predominant in patients with chronic liver disease, followed by genotypes D, E and A [10].
Besides genotype diversity, HBV S-gene mutations have been reported to affect nearly all amino acid positions of the major immunogenic region, the ''a'' determinant, which spans residues 124-147 of HBsAg. These mutations can lead to a false negative result when testing for HBsAg in blood donations and thus impact blood safety, especially in countries, including China, where screening of antibody to hepatitis B core antigen (anti-HBc) or HBV DNA is not routinely done [10].
Because the prevalence and sequelae from chronic HBV infections are very high in China, the Chinese government has taken effective measures to establish a universal infant immunization program to prevent HBV infections since 1992. Since then, hepatitis B vaccine coverage has reportedly increased from 30.0 % for newborns in 1992 to 93.4 % in 2005 [11]. However, nearly 20 years after the vaccination policy was put into effect in China, no comprehensive study of the HBV genotype prevalence or S-gene mutations in Chinese blood donors has been conducted. Little is known about mutations in the MHR region of HBV from blood donors. In this study, we analyzed the HBV genotype/subtype and mutations in the S region from voluntary blood donors who attempted to donate in five geographically diverse Chinese blood centers that participated in the Retrovirus Epidemiology Donor Study-II (REDS-II) International-China Program. The REDS-II International Program was developed to improve international blood safety by studying infectious and non-infectious risks of blood transfusion and mechanisms to improve the successful recruitment of low-risk volunteer blood donors. All blood donors participating in the REDS-II International China studies were provided with information on the objectives and nature of the studies, and they gave their written consent before enrolling in the studies. All REDS-II study protocols were reviewed and approved by the IRB at Johns Hopkins University and Chinese Academy of Medical Sciences prior to the implementation of the protocols.
Samples
Five blood centers in China including the Yunnan Kunming Blood Center (Kunming, Yunnan), Urumqi Blood Center (Urumqi, Xinjiang), Luoyang Blood Center (Luoyang, Henan), Mianyang Blood Center (Mianyang, Sichuan), and Liuzhou Blood Center (Liuzhou, Guangxi) participated in the NHLBI-sponsored REDS-II International-China Program. Blood samples from donors who attempted to donate blood at these five blood centers between 2008 and 2010 and consented to participate in the REDS-II program were tested twice by ELISA for HBsAg using ELISA kits made by two different manufactures (The kit produced by Shanghai Kehua was used by the Kunming, Urumqi and Luoyang blood centers for the firstround ELISA, and the kit from Xinchuang was used by the other two blood centers for the first-round detection. For the second-round ELISA, the kit produced by Jinhao was used by the Urumqi blood center, the kit from Xinchuang was used by the Kunming and Luoyang blood centers, and the kit from Abbott was used by the other two blood centers. Samples that were reactive with either one or both of these screening ELISA kits were tested again using a third ELISA kit (Monolisa HBsAg ULTRA Assay; Bio-Rad, France) as a surrogate measure for confirmation of HBV infection. Samples confirmed by the third ELISA testing were included in this study. The reactive wholeblood samples were collected in two ethylenediaminetetraacetate (EDTA)-K2 (with separator gel) vacuum tubes (Greiner, Kremsmünster, Austria) at the blood-collection sites. Plasma was separated from the RBCs by centrifugation at 4°C. Samples were then frozen and shipped on dry ice to the Institute of Blood Transfusion (IBT), Chinese Academy of Medical Sciences. To avoid repeat freezing and thawing, all of the plasma transported to IBT was divided into 1.5-mL tubes and stored at -70°C.
The basic demographic information about the donors was obtained from the REDS-II China database [12]. From 2008 to 2010, a total of 3,240 HBsAg-reactive samples were collected, and up to 10 samples were selected randomly from each blood center every half year. Each month, the first two HBsAg-reactive samples were selected at each center. If more than ten samples were selected in a half year for the first-round selection, only the first ten samples were included in the study. If fewer than ten samples were selected within a half year for the first-round selection, additional samples were selected from the months when more than two reactive samples were collected. Because either no samples or fewer than ten HBsAg-reactive samples were collected in some half years from some centers, only 245 HBsAg-positive samples were selected for this study.
Viral DNA extraction and amplification of the S region HBV DNA was extracted from 200 ll plasma using a QIAampÒ DNA Blood Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions, and 50 ll of eluted DNA was stored at -70°C until use. The HBV S region was amplified by nested PCR using a thermal cycler (Veriti, Applied Biosystems) with BS1 (nt 203-221, 5 0 -GCGGGGTTTTTCTTGTTGA-3 0 ) as the sense primer and BS2 (nt 788-769, 5 0 -GGGACTCAAGATGTTG TACAG-3 0 ), and BS3 (nt 712-731, 5 0 -AAGCCCTAC-GAACCACTGAA-3 0 ) as the antisense primers for the firstround and second-round PCR, respectively. The first-round PCR was performed with Taq DNA polymerase (Tiangen) in a total volume of 40 lL, with the following reaction variables: predenaturation at 95°C for 5 minutes, followed by 35 cycles of 15 seconds of denaturation (95°C), 30 seconds of annealing (53°C), and 30 seconds of extension (72°C), with a final extension at 72°C for 5 minutes. The cycling conditions of the second-round PCR were the same as the first-round PCR but using 2 lL of the first-round PCR product as template. PCR cycling was performed on a thermal cycler (Veriti, Applied Biosystems). To avoid false positive results, comprehensive procedures were followed to prevent sample cross-contamination, and test results were accepted as valid only when obtained in duplicate.
DNA sequencing, phylogenetic analysis and multiplex PCR
The PCR products were analyzed by electrophoresis in 1.5 % agarose gels and purified using a commercially available kit (NucleoSpin Extract II kit, Macherey-Nagel GmbH & Co. KG, Düren, Germany) according to the manufacturer's instructions. Purified products were used as templates in cycle sequencing reactions. Nucleotide sequences were determined from both strands using primers BS1 and BS3, and the resulting sequences were read directly with a genetic analyzer (ABI 3730, Applied Biosystems). For the single sequence with no heterogeneous sites, the HBV genotypes were determined by phylogenetic analysis with a panel of reference sequences from genotypes A to H retrieved from GenBank. The GenBank accession numbers for the reference strains were AB0 76678, AF536524, AJ012207, and AB194951 for genotype A; AB073851, AB106885, AB010290, AB205119, AB073 834, AY293309, AF121245, AB033554, AB031267, AB21 9426, and AB219427 for genotype B; AB074047, AF2239 61, AB031262, AB014374, AB014360, AB026814, X75 656, AB105172, GQ377635, Y18855, AB241110, AB2 41109, and AF241410 for genotype C; AY161157, AF151735, AB078033, AB090269, AJ132335, AY902776, AB048701, AB033559, DQ315779, and DQ315780 for genotype D; AB091255 and AB194948 for genotype E; AB036909 and AY090458 for genotype F; AB064311 and 064313 for genotype G; and AY090457 for genotype H. Nucleotide sequences were multiply aligned using the CLUSTAL_X (version 1.83) program [13]. The alignments were then used to construct phylogenetic trees for each subalignment using the neighbor-joining method implemented by the MEGA program [14]. The statistical validity of the neighbor-joining trees was assessed by bootstrap resampling with 1000 replicates. If the amplified sequence had heterogeneous sites identified by sequencing both strands, multiplex PCR was used for genotyping as reported previously [15].
Analysis of subtypes and mutations in the S region
The distribution of the HBV subtypes was deduced from the amino acid sequences at positions 122, 127, 134, 159, 160, 177 and 178 [16]. The presence of amino acid mutations in the antigenic loop was analyzed from amino acid 103 to 173 of HBsAg and compared to a consensus sequence from the same genotype. Mutations related to genetic polymorphisms were noted, but the focus was placed on mutations reported to be associated with vaccine escape, diagnostic failure, treatment failure with HBsAg immune globulins, or resistance to antiviral therapy [17,18].
Statistical analysis
Data were expressed as mean ± SD and percentages, as appropriate, with 95 % confidence intervals (95 % CI). Comparisons between groups were analyzed by chi-square test or Fisher's exact test for categorical variables and by Student's t-test for quantitative variables. P-values below 0.05 were considered significant. All statistical analysis was performed using SPSS software for Windows 10.0 (SPSS, Chicago, IL).
Results
A total of 228 (93.1 %) of the 245 randomly selected HBsAg-positive samples were successfully genotyped. Of these, 146 samples were sequenced directly, and the mean donor age was 28.3 ± 8.8 years. Eighty-two samples were genotyped by multiplex PCR, and the mean donor age was 32.0 ± 10.2 years. The donors from the second group were significantly older than those from the first group (p = 0.004). The sequences from the second group samples were found by direct sequencing to have heterozygous loci, which accounted for 36.0 % of the successfully genotyped samples. Overall The geographic distribution of HBV genotypes varies as shown in Fig. 1 (p \ 0.0001). HBV genotypes B and C were predominant in four of the five blood centers (Fig. 1). Genotype B was the most common genotype in all blood centers except for the Luoyang Blood Center, where the prevalence of genotype C was higher than that of genotype B. In Urumqi, however, a higher proportion of genotype D than C (11.63 % vs. 6.98 %) was found. The proportion of genotype D was significantly higher in Urumqi Blood Center (11.63 %) and Luoyang Blood Center (13.33 %) than in the other three blood centers (all \ 3 %). The prevalence of genotype A in Urumqi was also higher than in the other blood centers (9.30 % vs. 1.79 % in Kunming, 1.67 % in Luoyang, 4.17 % in Liuzhou, and 4.44 % in Mianyang,). In Liuzhou Blood Center, no subjects with genotype D or with a genotype mixture were found.
The HBsAg subtype distribution in this study was as follows: adw (67.6 %, 95 % CI: 61.9 %-74.1 %), adr (23.3 %, 95 % CI: 17.8 %-28.7 %) and ayw (8.7 %, 95 % CI: 5.1 %-12.5 %) ( Table 2). The HBV isolates from 150 of 155 (96.8 %) subtype adw subjects belonged to subtype adw2, while only four (1.8 %) belonged to adw3, and the subtype from one adw subject could not be differentiated Fig. 1 Distribution of HBV genotypes in five blood centers. In this figure, A, B, C and D represent genotypes A, B, C and D, respectively, and ''mixture'' refers to a mixture of at least two different genotypes. Genotype B was the most common genotype in all blood centers except for the Luoyang Blood Center, where the prevalence of genotype C was higher than that of genotype B between adw2 and adw3. All of the isolates from adr subjects belonged to subtype adrq?. The isolates from the ayw subjects were roughly evenly distributed among the subtypes ayw1, ayw2, and ayw3. The distribution of the HBV subtype varies geographically as shown in Table 1 (p \ 0.0001). As with genotype, the subtype distribution in the Luoyang region is most notably different than other regions, where no statistical difference was found in the HBV subtype distribution among the other four centers (p = 0.07). In this study, eight of nine (88.9 %) genotype A isolates belonged to subtype adw2, and one to adw3. For genotype B, 123(91.8 %) viruses belonged to subtype adw2, six to ayw1, three to adw3 and one to adrq?. For genotype C, 47 (94.0 %) were adrq?, and three each belonged to subtypes adrq? and adw2. For genotype D, six (40.0 %) belonged to ayw2, eight (53.3 %) to ayw3, and one to adw2 ( Table 2). With the exception of the Luoyang Blood Center, where the proportion of adw2 and adrq? was almost equal, subtype adw2 was predominant in all of the blood centers. The data are shown in Table 1. No subjects with subtype adw3 were found in Luoyang Blood Center, while one subject was found in each of the other four blood centers.
Demographic characteristics of donors associated with HBV genotypes and subtypes
The overall distribution of HBV genotypes (p = 0.25) and subtypes (p = 0.27) were not significantly different between men and women ( Figs. 2A and 3A). All nine subjects with genotype A were of the Han ethnic background; however, no statistical difference was found in the genotype distribution between Han donors and those from minority ethnic groups (Fig. 2B, p = 0.57). Figure 3B shows no statistical difference in the proportion of HBV subtypes between the Han and the minorities (p = 0.64). There was a statistically significant temporal difference in genotype distribution (Fig. 2C, p = 0.04). There appears to be a slight increase in the prevalence of subtype adr over time. There was not a statistically significant temporal difference in subtype distribution (Fig. 3C, p = 0.21). Most of the donors with genotype A were found in the 18-25 age group, and no donors infected with this genotype were found in the age group older than 40 (Fig. 2D); however, there was no statistical difference in the proportion of HBV genotypes by age (p = 0.29). Also, there was no statistical difference in the proportion of HBV subtypes by age (Fig. 3D, p = 0.52). Prevalence and characteristics of mutations in the S region Subsequent sequencing results revealed the prevalence and characteristics of the central MHR mutations among the blood donors in this study. Thirty-nine out of 228 (17.1 %) samples were found to have mutations. Fifteen (6.6 %) samples were identified with mutations that have a potential impact on the detection of HBsAg. Thirty-four of these 39 mutated strains had only one amino acid mutation in the studied region, and five had multiple mutations. There was a statistically significant difference in mutations by center (p = 0.02). The mutations were less frequent in donors at Luoyang Blood Center than at Kunming Blood Center, Liuzhou Blood Center and Mianyang Blood Center (Table 3). The mutations also varied by genotype (p = 0.02), where 31 of the viruses with MHR mutations belonged to genotype B ( Table 3). The proportion of MHR mutations varied by subtype (Table 3, p = 0.001), where only 2 of 53 (3.8 %) subjects with MHR mutations belonged to subtype adrq?, and the proportion of MHR mutations was higher among subtype adw (19.4 %) and higher again among subtype ayw (35.0 %). One isolate was found with the G145R mutation, which is one of the most important mutations in the HBV S region [19]. In addition, a four-amino-acid insertion (TNRT) between amino acid position 114 and 115 was found in an isolate from the Kunming Blood Center. The alignment of amino acid sequences of the central major hydrophilic region of the 39 samples is shown in Fig. 4.
Discussion
Molecular epidemiological studies provide valuable information to help understand the prevalence and characteristics of HBV genotypes and mutations. However, most of the molecular epidemiological studies of HBV in China have focused on patients, not on blood donors, who are generally asymptomatic and healthy individuals. Knowledge of the HBV genotypes and mutations in HBsAgpositive donors is important for developing effective donor screening for HBV. Studying healthy donors also provides a window into the molecular epidemiological evolution of HBV in the general population, and therefore, screening healthy blood donors should be a part of the comprehensive surveillance program. We filled this gap with a multicenter study of the geographical and demographic distribution of genotypes, subtypes and mutations in the S region of the HBV genome in donors from five blood centers, where the blood donors were geographically, socially, economically and ethnically diverse.
Although sequencing of the whole genome or S region is the gold standard for determining the HBV genotype, it is difficult to detect genotype mixtures by this method. In our study, a multiplex PCR assay was used to detect genetic mixtures in samples with heterogeneous sequences. Using this method, we detected 20 samples with genotype mixtures.
Genotypes C and B were the major HBV genotypes endemic in Mainland China. As shown in our study, HBV genotype B is more prevalent than genotype C in blood donors, while other studies have reported that genotype C has a higher prevalence in patients than genotype B. This difference may be due to the greater virulence of HBV genotype C than B [19][20][21]. Genotype C is associated with the development of cirrhosis and hepatocellular carcinoma as well as a lower response rate to interferon therapy. It also has a lower rate of seroconversion from HBeAg to anti-HBe and a higher HBV DNA level compared to genotype B [22]. Individuals infected with HBV genotype B may be more likely to remain asymptomatic and become part of the donor pool, posing a substantial threat to the safety of the blood supply.
Genotypes of HBV are generally subtype specific, although some subtypes are heterogeneous. In general, subtype adw is usually found in genotypes A and B, while adr occurs in genotype C [22]. In this study, we found that all of the genotype A strains and most of the genotype B HBV strains belonged to subtype adw, and most of the genotype C strains belonged to subtype adr, which is similar to previously published reports [22].
HBsAg is the main target for virus neutralization, either by natural or vaccine-induced anti-HBs. The basic working model is that of a protein with four transmembrane helices in which several residues at the N-and C-termini and a central major hydrophilic region (MHR) from residues 103-173 are exposed at the surface of viral particles. The MHR is composed of five regions: HBs1, HBs2, HBs3, HBs4 and HBs5. Mutations in this region occurring naturally or under immunization pressure could affect HBsAg test results or an individual's reaction to vaccination [17]. In our study, about 17 % of samples were found to have mutations, and most of the mutations were located at loci that might be related to the failure of immunization and HBsAg detection. The mutation rate was lower than those reported elsewhere, including Japan [23] (24 %), Korea [24] (50 %), France [25] (28 %) and Spain (40 %) [26]. However, the lower mutation rate might also be due to differences in the length of the fragment that was analyzed. It is also possible that our mutation rate is an underestimation, because this study only included HBsAg-positive donors, while there may be HBsAg-false-negative donors who were missed by the screening due to the mutations in the ''a'' determinant region. In addition, the sequences were cloned, and only one clone was selected for sequencing for the samples that showed heterogeneity by direct sequencing. This approach may miss some mutations and decrease the rate of detection of mutations. The most important and best-documented mutation in MHR was G145R, which is also the most critical substitution to prevent HBsAg detection [27]. We detected this mutation in only one sample from the Urumqi Blood Center. Since G145R is associated with false negative HBsAg screening, it may be reasonable to assume that there may be infected donors with this mutation who were not included in our study because they had a false negative HBsAg screening result.
There are several potential limitations to this study. Due to low viral loads or mutations in the primer-recognition sites, a total of 17 (6.9 %) of the 245 selected HBsAgpositive samples were not genotyped successfully. Although it is possible that some of these 17 samples may be false positive for HBsAg, new primers and a more sensitive genotyping method may be useful in further studies. Because complete genome sequencing is not applicable for routine investigations, we have limited our analysis to a part of the S gene that has been described as reliable for genotyping [7]. Sequencing of a larger region will be necessary to identify subgenotypes accurately. In our study, the genotype of an isolate from Urumqi Blood Center could not be determined by sequencing because of the low bootstrap value with any genotype from A to G, but it was genotyped successfully as genotype B using multiplex PCR. However, complete genome sequencing is needed to clarify the genotype of this subject.
This study is the first multi-center study of HBV genotypes, subtypes and mutations in voluntary Chinese blood donors from representative regions throughout China. The study results were analyzed in the context of donors' demographic and geographic characteristics. We have identified viral envelope mutants in this study that may compromise HBsAg detection in HBV-infected donors. However, HBV with envelope mutations can be detected by the nucleic acid test (NAT), which currently is not included as a test for routine blood donor screening. Our data suggest that the addition of the NAT to the routine donor screening process would minimize false negative HBV results due to S region mutations and would further increase blood safety in China. Meanwhile, large-scale molecular epidemiological studies of HBV in Chinese donors as well as in the general population will help to understand the diversity and distribution of HBV mutants and are essential to reduce the risks of transfusion-transmitted HBV.\===
Domain: Biology Medicine. The above document has 3 sentences that start with 'In this study', 2 sentences that start with 'The distribution of', 2 sentences that start with 'Nucleotide sequences were', 2 sentences that start with 'Genotype B was the most', 2 sentences that start with 'The proportion of', 2 sentences that start with 'There was a statistically significant', 2 sentences that start with 'In our study', 2 sentences that end with 'in this study', 2 sentences that end with 'shown in Fig', 2 sentences that end with '( Table 2)'. It has approximately 4141 words, 174 sentences, and 30 paragraph(s).
|
[{"added": "2022-11-29T15:19:11.461Z", "created": "2012-06-06T00:00:00.000", "id": "254052070", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://link.springer.com/content/pdf/10.1007/s00705-012-1331-x.pdf", "pdf_hash": "220c4ec04b53321e9b8d35c9e708a49b79ee4190", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:8957", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "220c4ec04b53321e9b8d35c9e708a49b79ee4190", "year": 2012}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
The Related Mechanisms Predicted through Network-Based Pharmacological Analysis and the Anti-Inflammatory Effects of Fraxinus rhynchophylla Hance Bark on Contact Dermatitis in Mice
Fraxinus rhynchophylla Hance bark has been used to treat patients with inflammatory or purulent skin diseases in China, Japan, and Korea. This study was undertaken to determine the mechanism responsible for the effects of F. rhynchophylla and whether it has a therapeutic effect in mice with contact dermatitis (CD). In this study, the active compounds in F. rhynchophylla, their targets, and target gene information for inflammatory dermatosis were investigated using network-based pharmacological analysis. Docking analysis was conducted using AutoDock Vina. In addition, the therapeutic effect of an ethanolic extract of F. rhynchophylla (EEFR) on skin lesions and its inhibitory effects on histopathological abnormalities, inflammatory cytokines, and chemokines were evaluated. Finally, its inhibitory effects on the nuclear factor-kappa B (NF-κB) and mitogen-activated protein kinase (MAPK) signalling pathways were observed in RAW 264.7 cells. In our results, seven active compounds were identified in F. rhynchophylla, and six were associated with seven genes associated with inflammatory dermatosis and exhibited a strong binding affinity (<−6 kcal/mol) to prostaglandin G/H synthase 2 (PTGS2). In a murine 1-fluoro-2,4-dinitrobenzene (DNFB) model, topical EEFR ameliorated the surface symptoms of CD and histopathological abnormalities. EEFR also reduced the levels of tumour necrosis factor (TNF)-α, interferon (IFN)-γ, interleukin (IL)-6, and monocyte chemotactic protein (MCP)-1 in inflamed tissues and inhibited PTGS2, the nuclear translocation of NF-κB (p65), and the activation of c-Jun N-terminal kinases (JNK) in RAW 264.7 cells. In conclusion, the bark of F. rhynchophylla has potential use as a therapeutic or cosmetic agent, and the mechanism responsible for its effects involves the suppression of inflammatory mediators, nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor (IκB)-α degradation, the nuclear translocation of NF-κB, and JNK phosphorylation.
Introduction
Fraxinus rhynchophylla Hance (family Oleaceae) is a deciduous tree native to Mǎnzhōu (northeast China) and the Korean peninsula, and its bark has been used to treat patients with inflammatory or purulent skin diseases in China, Japan, and Korea [1]. F. rhynchophylla bark is a 'heat-clearing herb' according to the theory of traditional medicine that reduces
Seven Active Compounds Were Found in F. rhynchophylla
In total, 21 compounds were found (Supplementary Materals S1), and seven satisfied the ADME criteria as active compounds (Table 1).
Forty-Seven Related Targets Were Found in F. rhynchophylla
Targets of the seven compounds were checked using the TCMSP database. The compounds were found to have 47 targets and 70 interactions. Of the seven compounds, β-sitosterol was the most associated with target genes (37 genes), followed by sitogluside (17 genes), 8-hydroxypinoresinol (six genes), fraxin (five genes), esculin (two genes), and sinapaldehyde glucoside (two genes). In addition, six of the active compounds targeted COX-2 ( Figure 1).
Forty-Seven Related Targets Were Found in F. rhynchophylla
Targets of the seven compounds were checked using the TCMSP database. The compounds were found to have 47 targets and 70 interactions. Of the seven compounds, βsitosterol was the most associated with target genes (37 genes), followed by sitogluside (17 genes), 8-hydroxypinoresinol (six genes), fraxin (five genes), esculin (two genes), and sinapaldehyde glucoside (two genes). In addition, six of the active compounds targeted COX-2 ( Figure 1).
All Six Active Compounds Showed Strong Binding Affinity with PTGS2
The binding affinities between PTGS2 and the six active compounds were predicted with molecular docking analysis (Table 2). These six compounds showed strong binding affinity with PTGS2 with binding energies of less than −6 kcal/mol. The interactions between the six compounds and PTGS2 are shown in Figure 3. Table 2. Binding affinities for interactions between PTGS2 (COX-2) and the six active compounds.
All Six Active Compounds Showed Strong Binding Affinity with PTGS2
The binding affinities between PTGS2 and the six active compounds were predicted with molecular docking analysis ( Table 2). These six compounds showed strong binding affinity with PTGS2 with binding energies of less than −6 kcal/mol. The interactions between the six compounds and PTGS2 are shown in Figure 3.
EEFR Ameliorated Surface Symptoms and Inhibited Skin Thickening in CD Mice
Repeated stimulation with DNFB induced the skin lesions of CD such as scaling, excoriation, erythema, and skin roughness ( Figure 4A,B), and these symptoms were alleviated by EEFR ( Figure 4A). EEFR (600 μg/day) significantly reduced skin surface scores ( Figure 4B). The mean skin thickness was significantly higher in the CTL group than in the NOR group, and EEFR significantly prevented skin thickening ( Figure 4C). DNFB stimulation elevated erythema indices significantly in the CTL group, and the topical application of EEFR at 600 μg/day significantly reduced erythema indices ( Figure 4D). Mice in all experimental groups had similar melanin indices ( Figure 4E).
EEFR Ameliorated Surface Symptoms and Inhibited Skin Thickening in CD Mice
Repeated stimulation with DNFB induced the skin lesions of CD such as scaling, excoriation, erythema, and skin roughness ( Figure 4A,B), and these symptoms were alleviated by EEFR ( Figure 4A). EEFR (600 µg/day) significantly reduced skin surface scores ( Figure 4B). The mean skin thickness was significantly higher in the CTL group than in the NOR group, and EEFR significantly prevented skin thickening ( Figure 4C). DNFB stimulation elevated erythema indices significantly in the CTL group, and the topical application of EEFR at 600 µg/day significantly reduced erythema indices ( Figure 4D). Mice in all experimental groups had similar melanin indices ( Figure 4E).
EEFR Prevented DNFB-Induced Histopathological Abnormalities
DNFB stimulation induced remarkable epidermal hyperplasia, hyperkeratosis, and immune cell infiltration ( Figure 5A). Treatment with 600 µg/day EEFR reduced epidermal hyperplasia significantly ( Figure 5B). Large numbers of infiltrating immune cells were observed around the dermis and blood vessels, and this phenomenon was also significantly reduced by EEFR treatment at 180 or 600 µg/day ( Figure 5C).
EEFR Prevented DNFB-Induced Histopathological Abnormalities
DNFB stimulation induced remarkable epidermal hyperplasia, hyperkeratosis, and immune cell infiltration ( Figure 5A). Treatment with 600 μg/day EEFR reduced epidermal hyperplasia significantly ( Figure 5B). Large numbers of infiltrating immune cells were observed around the dermis and blood vessels, and this phenomenon was also significantly reduced by EEFR treatment at 180 or 600 μg/day ( Figure 5C). . Skin surface scores were used to assess symptom severities (B). Weights of 5 mm diameter skin samples were measured using a micro-balance (C). Erythema and melanin indices were determined using a dermospectrophotometer (D,E). EEFR, ethanolic extract of Fraxinus rhynchophylla bark; DEX, dexamethasone. Results are presented as means ± standard deviation (SD)s. ## p < 0.01 and ### p < 0.001 vs. NOR; * p < 0.05 and ** p < 0.01 vs. CTL.
x FOR PEER REVIEW 8 of 16
EEFR Suppressed Lipopolysaccharide (LPS)-Induced Increases in Pro-Inflammatory Cytokine
Expression and COX-2, and the LPS-Induced Activation of NF-κB and JNK in RAW 264.7 Cells LPS increased the mRNA expression of TNF-α, IL1-β, IL-6, and IL-8 in RAW 264.7 cells, and these increases were reduced by pre-treatment with EEFR. In addition, COX-2 expression was also reduced by EEFR ( Figure 7A). As shown in Figure 7B,C, LPS treatment resulted in the degradation of nuclear factor of kappa light polypeptide gene enhancer in Bcells inhibitor (IκB)-α in the cytoplasm and increased p65 (the active form of NF-κB) levels in the nuclear fractions. These effects of LPS were effectively blocked by EEFR treatment. Furthermore, EEFR suppressed the LPS-induced phosphorylation of JNK ( Figure 7D). ol. Sci. 2023, 24, x FOR PEER REVIEW 9 of 16
EEFR Suppressed Lipopolysaccharide (LPS)-Induced Increases in Pro-Inflammatory Cytokine Expression and COX-2, and the LPS-Induced Activation of NF-κB and JNK in RAW 264.7 Cells
LPS increased the mRNA expression of TNF-α, IL1-β, IL-6, and IL-8 in RAW 264.7 cells, and these increases were reduced by pre-treatment with EEFR. In addition, COX-2 expression was also reduced by EEFR ( Figure 7A). As shown in Figure 7B,C, LPS treatment resulted in the degradation of nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor (IκB)-α in the cytoplasm and increased p65 (the active form of NF-κB) levels in the nuclear fractions. These effects of LPS were effectively blocked by EEFR treatment. Furthermore, EEFR suppressed the LPS-induced phosphorylation of JNK ( Figure 7D).
Discussion
Network-based pharmacology and molecular docking analysis are now frequently used to predict the efficacies of herbal medicines and formulas and have considerably reduced the time, effort, and cost of research in this area. Accordingly, we used networkbased pharmacological analysis to identify compounds with the potential to replace corticosteroids.
Discussion
Network-based pharmacology and molecular docking analysis are now frequently used to predict the efficacies of herbal medicines and formulas and have considerably reduced the time, effort, and cost of research in this area. Accordingly, we used network-based pharmacological analysis to identify compounds with the potential to replace corticosteroids.
In silico-determined binding affinities provide indications of the strengths of interactions between molecules. In the present study, AutoDock Vina docking scores were used to assess the binding affinities, according to which a score of <−5 kcal/mol indicates stable binding [14]. Of the seven target genes, PTGS2 (COX-2; PDB-ID: 5f19; a major inflammatory mediator) showed the highest association with the six active compounds, and thus, we used molecular docking analysis to predict the binding affinities between PTGS2 and these compounds. The results obtained confirmed that all six active compounds stably bound to PTGS2 (Table 2 and Figure 3), which suggested they act as anti-inflammatories by reducing the activity of COX-2.
Acute CD is characterized by itching, vesicles, erythema, and scaling, whereas longstanding lesions are typified by skin thickening with lichenification, pigmentation, and scaling [7,15], and in the present study, the repeated application of EEFR ameliorated scaling, erythema, and skin thickness increases in the DNFB-induced mouse model of CD (Figure 4).
The most common histological feature of acute CD is epidermal hyperplasia. In addition, immune cell infiltration has been reported in the dermis and epidermis of CD patients [16]. In the present study, EEFR inhibited epidermal hyperplasia, immune cell infiltration, and hyperkeratosis ( Figure 5). Considering the results shown in Figures 4 and 5, these findings imply that EEFR can prevent skin thickening by inhibiting epidermal hyperplasia and ameliorate scaling by suppressing hyperkeratosis. However, EEFR had no observable effect on melanin indices ( Figure 4E), indicating it does not reduce hyperpigmentation, which is a problem associated with chronic dermatitis.
The pro-inflammatory cytokine TNF-α can stimulate keratinocytes and cause hyperplasia and hyperkeratosis in the epidermis and also influence the progression of inflammation, especially that associated with immune cells, by inducing the expression of adhesion molecules on the surfaces of endothelial cells and keratinocytes [17]. IFN-γ also contributes to the inflammatory response and with TNF-α can activate keratinocytes and T-cells and promote Th1 shifted reactions [18]. Moreover, IFN-γ directly induces immune cell infiltration into epidermal tissue, epidermal hyperplasia, and hyperkeratosis [19]. IL-6 signalling is related to inflammatory infiltration by controlling the expressions of inflammatory chemokines and adhesion molecules [20] and has been reported to promote keratinocyte proliferation and lead to inflammatory skin disorders [21]. On the other hand, MCP-1, an important mediator of immune-mediated skin diseases, can accelerate monocyte and macrophage migration and infiltration [22].
In our study, EEFR effectively lowered TNF-α, IFN-γ, IL-6, and MCP-1 levels in inflamed mouse tissues ( Figure 6) and in RAW 264.7 cells ( Figure 7A). In particular, EEFR effectively inhibited epidermal hyperplasia and hyperkeratosis ( Figure 5). These findings suggest that EEFR ameliorates epidermal hyperplasia and hyperkeratosis by inhibiting the production of pro-inflammatory cytokines and chemokines through the prevention of keratinocyte accumulation. In addition, EEFR significantly prevented immune cell infiltration ( Figure 5C), which we believe was due to its inhibitory effects on TNF-α, IL-6, and MCP-1 production. Taken together, these results imply that EEFR can suppress the production of TNF-α, IFN-γ, IL-6, and MCP-1, and thus, inhibit epidermal hyperplasia, hyperkeratosis, and immune cell infiltration. These histopathological changes finally improved the skin symptoms of CD and suppressed skin thickening.
As shown in Table 2 and Figure 3, COX-2 showed the highest association with the six active compounds of F. rhynchophylla. Our PCR results showed EEFR significantly suppressed LPS-induced COX-2 overexpression ( Figure 7B). These results mean that the anti-inflammatory mechanism of EEFR is related to the inhibition of the COX-2 pathway.
The NF-κB and MAPK signalling pathways play central roles in cytokine production and can be activated by inflammatory cytokines and chemokines produced by various immune cells [23]. Recently, Silvia et al. reported that the ERK, p38, and JNK signalling pathways can induce intercellular adhesion molecule (ICAM)-1 expression in keratinocytes [24]. Our results showed that EEFR prevented LPS-induced IκB-α degradation in the cytoplasm, nuclear p65 accumulation, and JNK phosphorylation in RAW 264.7 cells ( Figure 7D) and inhibited TNF-α induced ICAM-1 expression in keratinocytes (Supplementary Materials S2). These observations suggest that the anti-inflammatory mechanisms initiated by EEFR are involved in the regulation of ICAM-1 expression via inhibition of the NF-κB and JNK pathways.
Classically activated (M1) macrophages play a role in direct damage to the epidermis and recruitment of other inflammatory cells and are found in large numbers in allergic CD skin lesions [25]. In addition, dermal macrophages are key modulators in contact hypersensitivity responses [26]. RAW 264.7 cells, a mouse macrophage cell line, used in in vitro study are commonly used in research studies to investigate the immune and inflammatory responses. While the relationship between macrophages and CD is complex and not yet fully understood, this means that RAW 264.7 cells can be a useful tool for investigating the immune and inflammatory responses for identifying potential therapeutic targets for the treatment of CD.
Recently, esculetin from F. rhynchophylla was reported to attenuate atopic skin inflammation induced by the house dust mite or 2,4-dinitrochlorobenzene [27]. In addition, Chen et al. reported that esculetin ameliorated psoriasis-like skin disease in mice [28]. We found that esculin, one of the standard materials of EEFR, did not affect skin lesion thickness or color in CD mice (Supplementary Materials S3). This means that the anti-inflammatory efficacy of F. rhynchophylla is closely related to the active ingredients such as esculetin rather than esculin, and also means that additional in vitro or in vivo studies are needed to confirm the results predicted by the in silico analysis.
As shown in Figure 4, dexamethasone (DEX) did not improve skin surface symptoms including the erythema index, but significantly reduced the skin thickness. This phenomenon is thought to be a result of corticosteroid-induced skin atrophy [29]. In addition, topical corticosteroids play a limited role in the treatment of CD. In our previous study using DNFB animal models, DEX did not improve skin symptoms and significantly reduced the skin thickness [30].
The spleen/body weight ratio is considered an indicator of systemic immune function [31], and in some cases, spleen shrinkage is considered an indicator of general immune suppression. We confirmed DEX reduced the spleen/body weight ratios in CD mice and that EEFR had no effect (Supplementary Materials S4), which implies that the antiinflammatory mechanism of EEFR differs from that of corticosteroids, particularly in terms of systemic immune suppression. Compounds in F. rhynchophylla were confirmed using TCMSP ( [URL]. com/tcmsp.php (accessed on 11 January 2022)), which is widely used in network pharmacological analysis. The ADME properties oral bioavailability (OB) and drug likeness (DL) were used for active compound screening. The following criteria were used: OB ≥ 20 (%) and DL ≥ 0.18. TCMSP was used to verify the target information, and the DisGeNET platform ( [URL] (accessed on 12 January 2022)) was used to identify inflammatory dermatosis-related genes. The official gene names of the targets were checked using UniProt ( [URL]/ (accessed on 17 January 2022)) [32].
Docking Analysis
Docking analysis was performed as described previously [34], and the binding affinities of the active compounds and target proteins were examined using AutoDock Vina (The Scripps Research Institute, La Jolla, CA, USA) [35]. The dried bark of F. rhynchophylla (200 g) was extracted using a standard method, as we previously described [30], and yielded 6.5 g of lyophilized powder (EEFR, yield 3.25%). A sample of EEFR was also deposited in the School of Korean Medicine, Pusan National University (Voucher no. MH2017-0014). The fingerprint of EEFR is provided in the supplementary data (Supplementary Materials S5).
Animals
Male 6-week-old Balb/c mice were purchased from Samtaco (Osan, Republic of Korea). The mice were housed under specific pathogen-free conditions under a 12 h light/dark cycle with free access to standard rodent food and water. All animal experiments were approved beforehand by the animal care and use committee of Pusan National University and conducted according to institutional guidelines (PNU-2015-0979).
Induction of CD and Experimental Design
CD was induced using DNFB in Balb/c mice, as we previously described [30]. Briefly, the mice were randomly divided into six groups, namely, a normal (NOR) group of nontreated mice (n = 6); a control (CTL) group of non-treated CD mice (n = 8); three EEFR groups of CD mice treated with 60, 180, or 600 µg/day of EEFR for six consecutive days (n = 8/group); and a dexamethasone (DEX)-treated CD group treated with 150 µg/day of DEX for six consecutive days (n = 8). EEFR and DEX were dissolved in 70% ethanol and diluted in vehicle (AOO, acetone: olive oil, 4:1). The EEFR and DEX solutions were applied to the shaved backs of Balb/c mice. The experimental schedule is summarized in the supplementary data (Supplementary Materials S6).
Skin Surface Scores and Thickness Measurements
Skin lesions were observed using a digital camera (Olympus, Tokyo, Japan). Skin surface scores were assessed by summing the skin lesion scores for scaling, excoriation, erythema, and skin roughness. The scores were evaluated using a modified version of Amano's method [36]. Skin tissues were cut into 5 mm diameter pieces, and the thicknesses were measured using vernier calipers (Mitutoyo, Kanagawa, Japan).
Erythema and Melanin Indices
The erythema and melanin indices were determined using dermo-spectrophotometer measurements (Cortex Technology, Hadsund, Denmark) obtained at three different points on the skin surface per mouse.
Histopathological Examinations
Skin tissues were fixed in 10% (v/v) formaldehyde, embedded in paraffin, sectioned at 4 µm, and stained with hematoxylin and eosin. The stained tissue slides were observed for histological changes under a light microscope (Carl Zeiss AG, Oberkochen, Germany) at ×100.
Evaluations of Epidermal Hyperplasia and Immune Cell Infiltration
To evaluate epidermal hyperplasia, the vertical distances between the basal lamina and the outer stratum granulosum were measured. Three random measurements were made per slide using the Zen program (ZEIZZ, Jena, Germany). To evaluate immune cell infiltration, cells were enumerated using a cell-counting grid (1024 × 1024 µm) in four randomly selected, non-overlapping regions per slide. Immune cells were defined as macrophages, polymorphonuclear leukocytes, lymphocytes, eosinophils, plasma cells, and giant cells [37].
Isolation of Total RNA and Reverse Transcription Polymerase Chain Reaction (RT-PCR)
The total RNA was extracted from cells using TRIzol (Invitrogen, Carlsbad, CA, USA). Complementary DNA (cDNA) was synthesized as described previously [38]. The PCR cycling conditions were 95 • C for 5 min, followed by 20-28 cycles of 95 • C for 30 s, 55 • C for 30 s, and 72 • C for 40 s. The primer sets used are shown in Supplementary Materials S7. The target gene expressions were normalized versus GAPDH.
Western Blot Analysis
Protein isolation and Western blotting were performed as previously described [36]. Blots were developed using an enhanced chemiluminescence system (SuperSignal ® West Femto, Thermo Scientific, Rockford, IL, USA).
Statistical Analysis
Data were analysed using Kruskal-Wallis tests followed by Dunn's comparison test. Prime 5 for Windows version 5.01 (GraphPad Software Inc., La Jolla, CA, USA) was used for the analysis. The results are presented as means ± standard deviations, and statistical significance was accepted for p values < 0.05.
Conclusions
This study identifies six active compounds (β-sitosterol, sitogluside, 8-hydroxypinoresinol, fraxin, esculin, and sinapaldehyde glucoside) and seven target genes (CHRM3, HTR2A, JUN, PIK3CG, SLC6A4, TGFB1, and PTGS2) related to the anti-inflammatory effects of F. rhynchophylla on inflammatory dermatosis. In addition, all six compounds stably bound PTGS2. This study describes the anti-inflammatory effects of EEFR on DNFB-induced CD in mice. According to the results of this study, EEFR effectively prevented epidermal hyperplasia and hyperkeratosis in DNFB-treated mice by inhibiting the production of TNFα, IFN-γ, IL-6, and MCP-1, and thus, ameliorated scaling, erythema, excoriation, and skin thickening. Finally, EEFR inhibited COX-2 expression and the activation of the NF-κB and JNK pathways in RAW 264.7 cells. Taken together, our findings indicate that F. rhynchophylla bark has potential use as a corticosteroid alternative for the treatment of inflammatory dermatoses and that its anti-inflammatory effects are associated with the suppressions of pro-inflammatory cytokine and chemokine levels and the NF-κB and JNK pathways.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement:
The data presented in this study are available upon request from the corresponding author.
==
Domain: Biology Medicine
|
[{"added": "2023-03-26T15:07:15.490Z", "created": "2023-03-23T00:00:00.000", "id": "257752708", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1422-0067/24/7/6091/pdf?version=1679642121", "pdf_hash": "0bc8bd93362e6a44e1deb8eabc4db1e89fae393a", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:1073", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "786e0b68dbc61bea31b2a0231cb05580eb7683a4", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Mitochondria-associated regulation in adipose tissues and potential reagents for obesity intervention
Introduction A systematic review analysis was used to assess the profile of mitochondrial involvement in adipose tissue regulation and potential reagents to intervene in obesity through the mitochondrial pathway. Methods Three databases, PubMed, Web of Science, and Embase, were searched online for literature associated with mitochondria, obesity, white adipose tissue, and brown adipose tissue published from the time of their creation until June 22, 2022, and each paper was screened. Results 568 papers were identified, of which 134 papers met the initial selection criteria, 76 were selected after full-text review, and 6 were identified after additional searches. A full-text review of the included 82 papers was performed. Conclusion Mitochondria play a key role in adipose tissue metabolism and energy homeostasis, including as potential therapeutic agents for obesity.
energy through non-fibrillatory thermogenesis by mitochondria (3). Adipose tissue is highly plastic and its restriction can drive pathological consequences such as obesity and metabolic diseases. As a risk factor for several metabolic diseases, obesity contributes to inflammation, insulin resistance, type 2 diabetes, non-alcoholic fatty liver disease and cardiovascular disease (4).
Mitochondria are organelles found in eukaryotes that are essential for energy metabolism and cellular homeostasis. Mitochondria have unique enzymes and systems that contribute to the citric acid cycle, fatty acid oxidation and oxidative phosphorylation (5). The mitochondria are essential to the differentiation of adipocytes (adipogenesis) and to major adipocyte functions (6). White adipocyte mitochondria are elongated and thin, which are involved in the production of ATP. Mitochondria of brown adipocytes have more quantity and larger body size compared to white adipocytes, and UCP1 located in the inner mitochondrial membrane can cause proton leakage across the inner membrane of the mitochondria, thus transforming the electrochemical energy into heat (7). The pathological expansion of the adipose tissue expansion is accompanied by the downregulation of mitochondrial oxidative pathways and changes in mitochondrial shape and number, ultimately leading to cell death (8)(9)(10). Mitochondrial dysfunction has deleterious effects on important adipocyte biology processes (lipid metabolism, adipocyte differentiation, insulin sensitivity, and thermogenesis), leading to metabolic diseases such as obesity and type 2 diabetes (11). Recent studies have shown that mitochondrial function can be improved by adding some herbal extracts and natural compounds in the diet, thereby inducing browning of WAT and adaptive thermogenesis of BAT to maintain metabolic homeostasis.
In this review, we focus on the specific role of mitochondria in adipocytes, summarize the mechanisms of mitochondrial regulation of adipose tissue, and discuss potential therapeutic agents from the perspective of treating obesity, which provide new insights for clinical treatment.
Search strategy
The reporting items were created in strict accordance with the systematic review statement. All articles were retrieved from the PubMed, Embase and Web of Science databases. We used search words such as 'Mitochondria', 'Obesity', 'Adipose Tissue, White', and 'Adipose Tissue, Brown' in conjunction with Boolean operators. Within each concept, we combined subject words and free words with the 'OR' Boolean operator, and the four concepts were combined with the 'AND' Boolean operator. The specific search process can be found in the supplementary materials.
Study selection
We screened the titles and abstracts of each paper, and articles that were repeated and irrelevant were removed. The inclusion criteria were the mechanism of mitochondrial regulation of adipose tissue and potential reagents for obesity intervention through the mitochondrial pathway. Articles were excluded according to the following criteria: review articles, editorials, commentaries, conference abstracts, case reports, articles written in other languages, letters to the editor and articles not relevant to the main topic. A 'PRISMA' flow chart was used to document the selection process ( Figure 1) (12).
Mitochondria regulate adipocyte thermogenesis
Adipose tissue is one of the significant features in maintaining systemic energy homeostasis and insulin sensitivity, not only as a storehouse of excess energy substrates but also as a metabolic health sensor and regulator of energy storage and expenditure. Thermogenesis in adipose tissue is activated in a state of metabolic overload to rapidly utilize excess nutrients. BAT disrupt electron transport in the respiratory chain by activating UCP1 on the mitochondria, thereby preventing the production of ATP and converting energy into heat (13). In addition, mitochondria provide energy for thermogenesis due to efficient proton gradient generation during catabolic processes and electron transport chain (ETC) function (14). It has been reported that individuals with obesity have reduced adipocyte UCP1 expression and that activation of UCP1 improves obesity and metabolic complications. Therefore, targeting mitochondria to activate UCP1 is a strategy to treat obesity (15).
Sirtuin 3 (SIRT3) is a mitochondrially localized deacetylase (16) that belongs to the sirtuin family. Sirtuins are evolutionarily conserved deacetylases whose activity is dependent on NAD+ and have a variety of physiological functions, including the regulation of cell proliferation, DNA repair, antioxidant activity and mitochondrial energy homeostasis. One study reported that caloric restriction activated SIRT3 expression in WAT and BAT, and in BAT from genetically obese mice, SIRT3 was downregulated along with genes related to mitochondrial function, suggesting that SIRT3 activates mitochondrial function and contributes to adaptive thermogenesis in BAT (17,18). The peroxisome biogenesis factor Pex16 (19), the angiogenic factor VEGF-A (20), disulfide-bond-A oxidoreductase-like protein (DsbA-L) (21), and PR domaincontaining 16 (PRDM16) (22) may also regulate thermogenesis by regulating mitochondrial function.
The overexpression of hypoxia-inducible factor-1a (HIF-1a) in adipose tissue inhibits thermogenesis and cellular respiration in BAT and promotes weight gain in mice, which is associated with a reduction in oxygen consumption in BAT, while the decrease in oxygen consumption may be mediated by a reduction in mitochondria (23). Carbohydrate response element-binding protein (ChREBP) is one of the major transcription factors regulating lipogenesis (24). One study showed that overexpression of ChREBP-b reduced the expression of genes involved in mitochondrial biogenesis, autophagy and respiration, leading to a bleached phenotype of BAT (25), which suggested that ChREBP-b is a negative regulator of thermogenesis in BAT. Bone morphogenetic protein 7 (BMP7) has also been shown to upregulate UCP1 and increase adipocyte thermogenesis (26). The absence of the membrane-associated estrogen receptor G proteincoupled receptor 30 (GPR30) may promote BAT mitochondrial uncoupling of respiration (27), suggesting that it is a negative regulator of thermogenesis and contributes to reduced obesity. Optic atrophy protein 1 (OPA1) (28), Brain-derived neurotrophic factor (BDNF) (29), also plays an important role in the adaptive thermogenesis of BAT.
Fat burning relying on adaptive thermogenesis has emerged as a viable strategy to reduce obesity, and the activation of mitochondrial-localized UCP1 and related molecules targeting mitochondria becomes a potential driving force to execute this strategy. Therefore, mitochondria are an essential organelle for maintaining adipocyte metabolic homeostasis.
Activation of WAT browning and BAT whitening
Some WATs exhibit a BAT phenotype when exposed to certain stimuli, which is called "the browning of WAT". WAT browning produces beige adipocytes, which exhibit UCP1-dependent thermogenesis, and fibroblast growth factor 21 (FGF21) was shown to play an important role in this thermogenesis (30). WAT browning has been found to suppress diet-induced obesity and improve systemic energy metabolism in many animal models (31, 32). Therefore, WAT browning has been investigated as an alternative therapy to BAT thermogenesis. Exercise is a major driver of fat browning. In this sense, the impact of myokines, which are factors secreted by the contracting muscle, on fat browning has provided a molecular mechanism to explain the benefits of exercise on weight loss and metabolic disease prevention. In this sense, several myokines act as positive (FNDC5/irisin, FNDC4, BAIBA and meteorin-like) and negative (myostatin) regulators of fat browning (33-37).
The role of mitochondria in WAT browning cannot be ignored. As an important factor of mitochondrial function, the expression of peroxisome prolilerators-activated receptor g coactivator l a (PGC1a) was significantly increased in WAT of protein kinase Cb (PKCb) deficiency on profound obesity, double knockout (DBKO) mice (38). OPA1, a key protein for mitochondrial fusion, promotes browning of white adipocytes (39). Human white adipocyte mitochondrial activity is regulated by the ubiquitin carrier protein 9/microRNA-30a axis, which is involved in controlling white adipocyte browning (40). MiR-337-3p in adipocytes inhibits Twist1, a negative feedback regulator of BAT metabolism, and enhances adipocyte browning (41). It has also been found that the small molecule compounds RepSox (42), Endonuclease G (EndoG) (43), milk fat globule membrane (MFGM) and its components phosphatidylcholine (PC) (44), bone morphogenetic protein-4 (MBP4) (45), cannabinoid receptor type 1 (CB1R) (46), and E2F transcription factor 1(E2F1) (47) can be involved in the induction of WAT browning. Electroacupuncture (EA) has also been shown to remodel WAT to BAT by deacetylating SIRT-1-dependent peroxisome proliferator-activated receptor gamma (PPARg) and regulating the PGC1a-TFAM-UCP1 pathway in order to induce mitochondrial biogenesis (48). These appears to be potential strategies for the treatment of obesity through WAT browning.
The "whitening" of BAT is closely related to obesity-related BAT dysfunction. It is worth mentioning that the process of BAT whitening associated with obesity is reversible by cold exposure and bariatric surgery (49, 50). It is possible that inflammation in whitened BAT contributes to the typical inflammatory state found in obesity (49). Shimizu, I et al. (51) showed that vascular rarefaction results in mitochondrial dysfunction and loss in BAT, which is a noteworthy causal factor in the whitening of BAT in mice models and could affect obesity and obesity-linked diseases. Activation of mTOR signaling downregulated PGC1a, a key activator of mitochondrial biosynthesis, and nuclear respiratory factor 1 (NRF1), an important transcriptional regulator, and downregulated Mitofusin 2 (Mfn2) and OPA1, genes involved in mitochondrial dynamics (52), which may have led to the "whitening" of BAT, suggesting that inhibition of the mTOR signaling pathway is a potential therapeutic avenue for obesity.
Obesity is associated with reduced WAT browning and BAT thermogenesis, and some transcriptional regulators and signaling pathways can regulate WAT browning and BAT thermogenesis by improving mitochondrial quality control, suggesting that targeting mitochondrial regulation is an important entry point for obesity prevention and treatment ( Figure 2).
Mitochondrial regulation of glucolipid metabolism in subjects with obesity
Mitochondrial dynamics in adipocytes may play a key role in initiating systemic metabolic dysregulation. Nutrient overload promotes hypoxia in BAT, which leads to whitening through mitochondrial dysfunction and loss, subsequently leading to impaired systemic glucose metabolism (53). Bean and coworkers (54) identified that the OPA1 gene regulates insulin sensitivity and adipose tissue functions, and controlled OPA1 overexpression in mice could reduce body weight, improve glucose metabolism and insulin sensitivity, reduce fat accumulation and promote browning of white adipocytes. Mfn2 is a gene that promotes mitochondrial fusion and mitochondrial-endoplasmic reticulum interactions. When Mfn2 is knocked down in adipocytes, adult mice consume more food, gain more fat, and have impaired glucose metabolism in standard diets (55). Mfn2 also plays a key role in the regulation of brown adipose tissue thermogenesis favouring mitochondria to lipid droplet interactions (56). A dysfunctional mitochondrial system in WAT is implicated in obesity-related insulin resistance. OPA1 deletion completely prevented the increase in adiposity and improved insulin fate sensitivity in mice fed a high-fat diet (57). MLX interacting protein-like (MLXIPL) is a transcriptional regulator. MLXIPLdeficient mice are resistant to excessive lipid accumulation and heat-induced mitochondrial degradation in brown adipocytes (58). This result suggested that knockout of MLXPIL may be a potential therapeutic target for obesity-related metabolic diseases. Takaya et al. (59) found that the expression of UCP1 was increased in BATderived cultured preadipocytes and their local transplantation reduced inguinal fat pad weight. This finding suggests that local transplantation of BAT-derived preadipocytes may increase energy expenditure and thus reduce obesity. In the tricarboxylic acid cycle, pyruvate is transported to mitochondria by pyruvate carrier 1 (MPC1) to be oxidized to acetyl coenzyme A. Inhibition of MPC1 inhibits pyruvate transport and thus activates fatty acid oxidation (60), this suggests that MPC1 may be an important regulator of mitochondrial energy metabolism.
Mitochondria play a key role in the regulation of glucose utilization and lipid metabolism in adipocytes. By regulating mitochondrial biogenesis and mitochondrial dynamics in adipocytes, the development of obesity and metabolic diseases can be improved.
Other mitochondrial regulation approaches in obesity
Mitophagy is a process that selectively removes damaged mitochondria through a specialized form of autophagy and is essential for mitochondrial quality control (mitochondrial QC) and metabolic homeostasis. The mitochondrial autophagy receptor Fundc1 is a newly defined mitophagy receptor, and mice lacking Fundc1 develop more severe obesity and insulin resistance when fed a high-fat diet (HFD), and disruption of Fundc1 leads to impaired mitophagy and mitochondrial quality in WAT (61). PGC1a is essential for maintaining energy homeostasis. Overexpression of PGC1a in epicardial adipose tissue induces mitosis and improves mitochondrial function, resulting in improved adipose tissue quality (62). The deletion of mitochondrial transcription factor A (TFAM) in adipose tissue results in reduced mtDNA copy number, mitochondrial dysfunction, increased mitochondrial oxidation and positive metabolic effects (63,64), suggesting that adipose tissue Mitochondria monitor the thermogenesis of brown and activate the browning of WAT. Sirtuin 3 (SIRT3); cAMP response element-binding protein (CREB); peroxisome proliferator-activated receptor-g cofactor 1a (PGC1a); uncoupling protein-1 (UCP1); optic atrophy 1 (OPA1); hypoxia-inducible factor-1a (HIF-1a); peroxisome proliferator-activated receptor gamma (PPARg); sirtuin 1 (SIRT1); AMP-activated protein kinase (AMPK); mitochondrial transcription factor A (TFAM). mitochondrial biosynthesis regulation may be a potential therapeutic target for the treatment of obesity.
The expression of PTEN-induced kinase 1 (PINK1), a protein involved in mitochondrial autophagic clearance, was upregulated in the WAT of HFD mice (65). Deletion of PINK1 induced BAT dysfunction, suggesting that the regulation of mitochondrial autophagy contributes to the "whitening" of adipose tissue during the development of obesity (66). The transition from beige to white adipocytes was associated with decreased mitochondria and increased autophagy (67), uncovering a mechanism by which autophagy-mediated mitochondrial clearance controls the maintenance of beige adipocytes, thus providing an opportunity to combat obesity.
Intercellular mitochondrial transfer can support the survival of cells with impaired metabolism. Rohatgi et al. (68) found that adipocytes and macrophages employ intercellular mitochondrial transfer as an immunometabolic crosstalk mechanism to regulate metabolic homeostasis, which is impaired in obesity. Borcherding et al. (69) showed the existence of a potential direct dietary mechanism on the basis of the above studies that could largely ameliorate mitochondrial translocation.
3.5 Mitochondrial targeting as a strategy to treat obesity 3.5
.1 Herbal extracts and natural compounds
Increasing energy expenditure is a common approach to obesity prevention, and activating BAT may be a potential strategy against obesity. Beta vulgaris has been shown to have an anti-obesity effect by increasing UCP1 during nonshivering thermogenesis in brown adipocytes (70). Liu Z et al. (71) studied the effects of sesamol on disorders of adiposity and fat-related metabolism in mice fed a Western diet. They found that sesamol reduced WAT and BAT mass and adipocyte size by improving the expression of mitochondria-related genes, including PGC1a and UCP1. Jung Y et al. (72) investigated the anti-obesity effects of vanillic acid (VA) in vivo and in vitro and found that VA increased mitochondrial and thermogenesis-related factors such as UCP1 and PPARg-1 in BAT and primary cultured brown adipocytes from mice, suggesting the potential of VA as a thermogenesis-activated anti-obesity agent.
Stimulating browning of white adipose cells helps to limit obesity and related metabolic disorders. Lactobacillus plantarum dy-1 (LFBE) can suppress obesity by enhancing thermogenic processes in BAT and browning of adipose tissue in the epithelium using a UCP1-dependent mechanism of activation (73). Averrhoa bilimbi can also induce adipocyte browning and enhance mitochondrial activity due to upregulation of UCP1 (74). Kang J et al. (75) studied the effects of secoisolariciresinol diglucoside (SDG) on WAT browning and found that SDG increased UCP1, PGC1a and PRDM16 in WAT and BAT in mice, as well as mitochondrial biogenesis and activation, suggesting that SDG is a potential candidate for ameliorating obesity and other metabolic disorders. Lycopene (LYC), one of the major carotenoids in tomatoes, has been used preclinically and clinically in the treatment of obesity and type 2 diabetes. LYC can induce browning and enhance mitochondrial respiration in white adipocytes and improve glucose and lipid metabolism by upregulating PPARg (76).
The whitening of BAT during obesity and aging promotes metabolic disorders and related diseases. Gao P et al. (77) determined that the inhibitory effect of capsaicin on HFDinduced obesity and BAT whitening was dependent on the involvement of SIRT3, which could mediate the beneficial effects of capsaicin on attenuating reactive oxygen species production, increasing mitochondrial activity and limiting HFD-induced mitochondrial calcium overload.
Regulation of mitochondrial biogenesis is also a potential avenue for the treatment of obesity. The ethanolic extracts of both rutin and Platycodon grandiflorum (PG) can provide benefits for obesity by increasing the expression of genes involved in mitochondrial biogenesis (78,79). a-Lipoic acid (a-LIP) is a naturally occurring antioxidant that promotes mitochondrial biogenesis and brown-like remodeling in cultured white subcutaneous adipocytes from donors with obesity (80). Marqueś and colleagues (81) found that resveratrol has potential therapeutic effects in improving mitochondrial biogenesis (Table 1).
Some compounds can also activate mitochondrial biogenesis in skeletal muscle. Resveratrol was shown in 2006 to affect skeletal muscle mitochondrial biogenesis and thus metabolic homeostasis by decreasing PGC1a acetylation and increasing PGC1a activity (88). As a thermogenic tissue, skeletal muscle is also capable of regulating energy expenditure. When UCP1 is absent or non-shivering thermogenesis is affected, skeletal muscle generates heat through shivering and is supplied with energy by carbohydrates and lipids (89). Thermogenesis in skeletal muscle is dependent on the homologue of UCP1, uncoupling protein 3 (UCP3). 5,3'-Triiodo-L-thyronine (T3) was reported to induce UCP3 expression to regulate skeletal muscle thermogenesis (90). It is interesting to note that these compounds also work through similar pathways in BAT. Recruitment of BAT and skeletal muscle bear very high degree of similarities (91). Both organs are highly vascularized, neuralized, and contain abundant mitochondria (92). Studies by Bal et al. (93) have shown that skeletal muscle thermogenesis can be activated to compensate for the absence of BAT, suggesting that the two thermogenic systems can be functionally complementary. It has been shown that PRDM16 can control the bidirectional transformation of brown adipocytes and skeletal muscle cells (94), which may represent two different potential therapeutic targets to expand the thermogenic capacity of adipose tissue.
Traditional Chinese medicine and other alternative medicines
Zhang et al. (95) found that the natural antioxidant Lycium could enhance UCP1 expression, upregulate PGC1a and induce browning in white adipocytes, as well as enhance glucose uptake and oxidative utilization, lipolysis and fatty acid oxidation in 3T3-L1 adipocytes, and the application of Lycium is a promising strategy to combat obesity and obesity-related metabolic disorders. Huangqi San (HQS) is a traditional Chinese medicine formula, and Hao M et al. (96) found that it could significantly increase the number of mitochondria, increase the expression of UCP1 and PGC1a in BAT, and improve metabolic disorders and lipid deposition in hyperlipidemia in obese rats after 13 W intravenous injection of HQS, suggesting that HSQ is an effective drug for the treatment of hyperlipidemia with obesity. Tanshinone IIA (TAN2A) is a major active ingredient of the traditional Chinese medicine tanshinone, and tanshinone 20 (TAN20) is a derivative of TAN2A. Ma L et al. (97) showed that both TAN2A and TAN20 were able to increase mitochondrial content in adipose tissue, increase energy expenditure and reduce body weight, thereby improving insulin sensitivity and metabolic homeostasis in mouse models of obesity and diabetes. It has also been shown (98) that administration of tanshinone 1 (TAN1) prevents HFD-induced obesity in mice, which was associated with enhanced expression of brown adipocyte-related genes in WAT and BAT, and that TAN1 also led to increased mtDNA content and lipolysis.
Drugs with therapeutic potential can combat obesity by affecting the differentiation of white adipocytes. Ravaud C et al. (99) showed that HIV protease inhibitors (PIs) can reduce the expression of UCP1, improve mitochondrial function in brown adipocytes and regulate thermogenesis and are promising potential drugs against obesity. Empagliflozin can increase mitochondrial biogenesis and fusion, improve their function and promote browning of 3T3-L1 adipocytes (100), FAM134B improves adipocyte differentiation by enhancing mitophagy (101), and melatonin drives WAT browning (102), which are also promising anti-obesity treatments.
Targeting mitochondria as a therapeutic strategy for metabolic diseases is being extensively investigated as a new potential approach. In conclusion, the possibility of reducing obesity and metabolic diseases by mediating mitochondrial biogenesis or mitochondrial dynamics to increase thermogenesis, activate WAT brownout, and promote lipid metabolism and fatty acid oxidation is being progressively demonstrated.
Conclusion
The importance of mitochondria in energy metabolism and cellular homeostasis has been extensively noted and studied. Since the rediscovery of active brown and beige adipocytes in humans a decade ago, more and more research has focused on the regulation of adipocyte function by mitochondria. We summarized the searched articles and showed that mitochondria are involved in regulating the thermogenesis of BAT and increasing energy expenditure; in lipid homeostasis of adipose tissue and regulating glucose metabolism; and are also able to participate in inducing browning of WAT while promoting the whitening of BAT. These are key therapeutic targets for obesity-related metabolic diseases. The regulation of mitochondrial biogenesis, mitochondrial autophagy and mitochondrial translocation are also potential therapeutic opportunities for obesity. In fact, a series of studies (87) are working on the interpretation of drugs or potential drugs for the treatment of obesity through these mitochondrial pathways.
In conclusion, mitochondria-targeted drugs for adipocytes are very promising. Therefore, understanding the molecular mechanisms underlying adipocyte mitochondrial dysfunction and the pathogenesis of obesity-related metabolic diseases is crucial for the development of new therapeutic approaches. However, future studies need to elucidate the mechanisms by which mitochondria undergo metabolic disorders to better account for their pathogenic role in metabolic diseases.
Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material. further inquiries can be directed to the corresponding authors.
Acknowledgments
Ying Ning, Yan He, Yingxiu Mei and Yue Jin are acknowledged for their assistance in the writing of this article.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The
==
Domain: Biology Medicine
|
[{"added": "2023-07-02T05:09:48.404Z", "created": "2023-06-16T00:00:00.000", "id": "259302837", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.frontiersin.org/articles/10.3389/fendo.2023.1132342/pdf", "pdf_hash": "ed9d5441d30d98d42dfc064e73dc93ddf428fa1a", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:1963", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "ed9d5441d30d98d42dfc064e73dc93ddf428fa1a", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Reintroduction of DJ-1 in Müller Cells Inhibits Retinal Degeneration in the DJ-1 Deficient Retina
The eye is continuously under oxidative stress due to high metabolic activity and reactive oxygen species generated by daily light exposure. The redox-sensitive protein DJ-1 has proven to be essential in order to protect retina and retinal pigment epithelium (RPE) from oxidative-stress-induced degeneration. Here, we analyzed the specific role of Müller cell DJ-1 in the adult zebrafish retina by re-establishing Müller-cell-specific DJ-1 expression in a DJ-1 knockout retina. Loss of DJ-1 resulted in an age-dependent retinal degeneration, including loss of cells in the ganglion cell layer, retinal thinning, photoreceptor disorganization and RPE cell dysfunction. The degenerative phenotype induced by the absence of DJ-1 was inhibited by solely expressing DJ-1 in Müller cells. The protective effect was dependent upon the cysteine-106 residue of DJ-1, which has been shown to be an oxidative sensor of DJ-1. In a label-free proteomics analysis of isolated retinas, we identified proteins differentially expressed after DJ-1 knockout, but with restored levels after Müller cell DJ-1 re-insertion. Our data show that Müller cell DJ-1 has a major role in protecting the retina from age-dependent oxidative stress.
Introduction
DJ-1 is a multifunctional and ubiquitously expressed protein encoded by the park7 gene [1]. It is highly recognized as a general protector of oxidative stress through regulating antioxidant and anti-apoptotic gene transcription, as well as several distinct pro-survival pathways [2,3]. DJ-1 has a role in mitochondrial homeostasis and dynamics [4], in part through chaperone-mediated autophagy of damaged mitochondria [5], and was recently also shown to have a crucial role in mitochondria-ER interaction function [6].
DJ-1 function can be regulated through post-translational modifications of a highly conserved cysteine residue (Cys106) [7], which is recognized as an oxidative sensor. These modifications include both cysteine oxidation [7] and nitrosylation [8]. It should be noted, though, that some functions of DJ-1 also seem Cys106-independent [9,10].
The loss of DJ-1 causes an age-dependent increase in retinal abnormalities in DJ-1deficient mice, including molecular and structural abnormalities, loss of photoreceptors, outer retina thinning and visual dysfunction [11,12]. DJ-1 seems to act similarly to an oxidative sensor in the retina; by inducing retinal oxidative stress, DJ-1 levels are upregulated [13].
Müller cells are the main glial cell in the vertebrate retina. They span the entire retina thus enabling them to interact with all retinal neurons. Müller cells are responsible for metabolic support of retinal neurons and obtaining retinal homeostasis [14]. Their neuroprotective function also comprises the release of antioxidants and neurotrophic factors [15]. Hence, understanding the basis of neuron-glial crosstalk is highly important. Müller cells are also of interest in regenerative therapies, as they are the stem cells of the retina [16].
Astroglial DJ-1 has proven to have a neuroprotective function in the brain by inhibiting oxidative-stress-induced death of dopaminergic neurons [17,18]. It was, therefore, of interest to explore whether Müller cell DJ-1 expression could have a similar role in the retina.
Zebrafish are highly valuable in vivo models due to their short regeneration time and accessibility for transgenesis. We have previously established a CRISPR/Cas9-based DJ-1-deficient zebrafish model [19]. Our first attempt was to elucidate if this model showed similar retinal degeneration as observed in rodents. By using glial fibrillary protein (gfap) promotor driven expression, it is possible to drive Müller-cell-specific expression in the retina [20,21]. Thus, by re-inserting DJ-1 under the control of the gfap promotor in the DJ-1 null background, we could study the role of Müller cell DJ-1 expression.
Loss-of-function mutation in DJ-1 causes an early onset form of familiar Parkinsonism [22]. Visual disorders are common in parkinsonism, but not specifically associated to any specific DJ-1 mutation [23]. Retinal thinning, however, has been suggested as an early biomarker of Parkinson's disease as the more accessible eye can be a window to early pathologies [24]. The antioxidant function of DJ-1 is suggested to play a general role in several ocular neurodegenerative diseases [25]. Giving the accessibility and potential in transgenesis of zebrafish, modeling DJ-1 function in retina may be highly valuable in a translational perspective.
Here, we show that DJ-1 loss-induced structural retinal changes and changes in protein profiles can be inhibited by the retinal selective DJ-1 expression in the Müller cells and that this effect is dependent on the redox sensitive Cys106 residue.
Materials and Methods
Zebrafish were used to elucidate the function of Müller cell DJ-1 in retinal protection to oxidative stress. Transgenic lines with Müller specific expression of DJ-1 were established in a retinal DJ-1 null background. The function of Müller DJ-1 was studied by using a combination of morphological analysis and protein profiling by mass spectrometry.
Animal Maintenance
Animals were housed in the zebrafish facility located in the Department of Biological Science at the University of Bergen. The facility is run according to the European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes. Adult zebrafish were maintained at 26-28 • C, with a 14/10 light cycle and were fed twice daily.
Founder fish were outcrossed with wild-type and progeny embryos (F1) collected. Stable lines were expanded from single F1 founders, and eGFP expression in transgenic animals was examined by using a fluorescent Zeiss SteREO Lumar microscope.
Eye Sectioning, Toluidine Blue Staining and Image Analysis
The eyes of adult zebrafish (3-18 months) were fixed in 4% paraformaldehyde for 48 h at 4 • and were washed and rehydrate in (50%, 70% and 96%) EtOH. Eyes were pre-infiltrated for 4 h, at room temperature, in 50%(96%) EtOH/50% base liquid (Kulzer Technovit ® 7100: 64-7090-03, Kulzer GmbH, Wehrheim, Germany). They were then infiltrated in preparation solution overnight at room temperature (see user instruction). Eyes were oriented in the polymerization solution at room temperature and left overnight. Then 2 µm sections were prepared by using a Leica Microtome (RM 2165, Leica Biosystems, Nussloch, Germany). Sections were dried before and after staining with 1% toluidine blue. The sections were mounted in DPX (Sigma:06522, Merck, Damstadt, Germany). Images were obtained by using Zeiss Axio Scan. Z1.
All image processing and measurements were performed in ImageJ2 version 2.1.0/1.53c. Selections containing the retina were created for each image, and binary images were created from these selections. Retinal thickness was then measured by using the "Average thickness" plugin (part of MorphoLibJ [26]). The number of cells in the ganglion cell layer was counted manually from sagittal eye sections, including n.opticus. Statistical analysis was performed in Prism 9.2.0 for macOS (GraphPad Software, San Diego, US) using one-way ANOVA and Tukey's multiple comparisons test.
Cryo-Sectioning
Zebrafish eyes were fixed in 4% paraformaldehyde for 24 h, at 4 • C, and cryoprotected overnight in 25% (OCT) compound medium in 25% sucrose solution. Eyes were oriented in a plastic mold in 100% OCT, cooled down and stored at -80 • C for sectioning [27]. Then 10 µm serial sections were prepared at −25 • C, using a Leica CM1800 cryostat.
In Situ Hybridization
Plasmids containing gstp1 (cb356) cDNA fragments were purchased from ZIRC ( [URL], 10/September/2020). Sense and antisense RNA probes were synthesized from the amplified template, using the recommended polymerases and Digoxigenin-labeled ribonucleotides (Roche: Cat. No. 11093274910). In situ hybridization for gstp1 transcript (sense and antisense) was performed on 10 µm serial sections (n = 6) of adult zebrafish eyes. Tissue pretreatment and in situ hybridization steps were performed as described previously [28].
Transmission Electron Microscopy
Resected eyes were washed in 0.1 M sodium cacodylate buffer and fixed in 1.5% glutaraldehyde in 0.1 M Na-cacodylate buffer, pH 7.4, overnight. Eyes were then washed in 0.1 M sodium cacodylate buffer and post-fixed in 1% osmium tetroxide in 0.1 M sodium cacodylate buffer for 60 min. Eyes were washed twice for 15 min in 0.1 M sodium cacodylate buffer and dehydrated in graded solutions of ethanol. Eyes were then embedded in plastic, using Agar 100 resin, cut into 60 nm thin sections (Reichert Ultracut S Ultra microtome, Leica Biosystems, Nussloch, Germany) and stained in uranyl/lead. Sections were examined by using a Jeol JEM-1230 transmission electron microscope at the UiB Core Facility for Imaging.
Isolation of Retina for Mass Spectrometry
Retinas were sampled from age-matched animals (9 months) between 9:00 and 10:00 a.m. Animals were first euthanized with MS-222 in facility water and thereafter transferred to an ice bath. The cornea was cut, and the lens was taken out in situ under a stereo microscope. The eye was thereafter lifted out, and the n.opticus was cut. The retina was then collected by pushing back the sclera, using two forceps, and immediately frozen in liquid nitrogen.
Label-Free Mass Spectrometry
Tryptic peptides 0.5 µg were analyzed on Ultimate 3000 RSLC system (Thermo Scientifi, Sunnyvale, California, CA, USA) connected online to a QExactive HF mass spectrometer (Thermo Scientific, Bremen, Germany). The sample was loaded and desalted as previously described in Frøyset et al. [29].
The peptides were separated during a biphasic ACN gradient from two nanoflow UPLC pumps with flow rate 250 nL/min on a 25 cm analytical column (PepMap RSLC, 25 cm × 75 µm ID EASY-spray column, packed with 2 µm C18 beads, Thermo scientific, Waltham, MA, USA). Solvent A and B were 0.1% FA (vol/vol) in water and 100% ACN, respectively. The gradient composition was 5% B during trapping (5 min), followed by 5-7% for 0.5 min, 7-22% B for the next 44.5 min, 22-35% over 15 min and 35-80% B over 5 min. Elution of very hydrophobic peptides and conditioning of the column were performed during a 7-min isocratic elution with 80% B and 10-min isocratic conditioning with 5% B. The total length of the LC run was 90 min.
MS spectra were acquired as described in Reference [29], but with minor/some changes. The instrument control was through Q Exactive HF Tune 2.9 and Xcalibur 4.1. MS spectra were acquired in the scan range 375-1500 m/z with resolution R = 120,000 at m/z 200, automatic gain control (AGC) target of 3e6 and a maximum injection time (IT) of 100 ms. The 12 most intense eluting peptides above intensity threshold 40,000 counts, and charge states 2-5 were sequentially isolated to a target value (AGC) of 1e5 and a maximum IT of 118 ms in the C-trap, and isolation with maintained at 1.2 m/z (offset of 0.3 m/z), before fragmentation in the HCD (Higher-Energy Collision Dissociation) cell. Fragmentation was performed with a normalized collision energy (NCE) of 28%, and fragments were detected in the Orbitrap at a resolution of 60,000 at m/z 200, with first mass fixed at m/z 120. One MS/MS spectrum of a precursor mass was allowed before dynamic exclusion for 20 s with "exclude isotopes" on. Lock-mass internal calibration (m/z 445.12003) was used. The ion source parameters were as follows. Ion spray voltage = 1800 V, no sheath and auxiliary gas flow; and capillary temperature was 275 • C.
Generation of Transgenic Zebrafish Lines with Müller Cell Specific Wild Type DJ-1 and DJ-1c106a Expression in a DJ-1 Null Background
We have previously established a DJ-1-deficient zebrafish line [19]. This line was generated by using the CRISPR-Cas9 method to target exon 1 of the park7 gene to knockout DJ-1 ( Figure 1A). Here, we have reinserted DJ-1 and DJ-1 c106a in glia cells of the knockout line, using ISce1-transgenesis and elements of the glia fibrillary acidic protein (gfap) promotor to enable glial specific expression of DJ-1. In the retina, however, this glial expression is restricted to the Müller cells [20,30], thus making it possible to study the effect of Müller specific DJ-1 expression in a retinal DJ-1 null background. Flag-tagged DJ-1 and mutant are expressed together with GFP, but separated by the viral 2A peptide, which allows stoichiometric unfused expression of the proteins ( Figure 1A-C). Moreover, eGFP expression was prominent around the Müller cell bodies and could also be observed in their processes extending to the photoreceptor layer ( Figure 1B). Less GFP expression was observed extending towards the inner limiting membrane. Eyes from the three zebrafish lines, namely dj1 -/-(DJ-1_KO), dj1 -/-;Tg(gfap:eGFP-2A-dj1) (Müller_DJ-1), and dj1 -/-;Tg(gfap:eGFP-2A-dj1 c106a ) (Müller_DJ-1 c106a ), together with wild-type eyes, were used in this study to evaluate the role of Müller cell expressed DJ-1 in retinal neuronal protection from oxidative stress induced by the loss of DJ-1 ( Figure 1D). Mass-spectrometry-based analysis of isolated retinas showed that Müller cells expressed DJ-1 and mutant DJ-1 levels were 0.10 and 0.18 fold, respectively, when compared to endogenous DJ-1 levels in wild-type whole retina (Supplementary Materials Table S2). method to target a 20 bp region of exon one of the park7 gene [19]. Lines expressing glial specific wild-type DJ-1 or DJ-1 c106a in a DJ-1 null background were constructed by using ISce1 transgenesis and regulatory elements of glial fibrillary acidic protein (GFAP). The viral 2A peptide allows expression of GFP and Flag-DJ1 as uncoupled protein. In the retina, the gfap promotor drives expression only in the Müller glia cells. (B) Glial expression of GFP in the Müller_DJ-1 line (b,d) versus KO_DJ-1 (a,c). Note that GFP expression was determined by using a GFP antibody, followed by a far-red secondary antibody to avoid interference from retinal autofluorescence. Prominent expression is found around the Müller cell bodies. GFP expression also extends through the Müller processes into the photoreceptor layer. Faint expression can be observed in the Müller foot processes on the inner limiting membrane. Bar, 50 μm. (C) A Western blot shows expression of endogenous DJ-1 and Flag-tagged DJ-1 from total brain extracts belonging to animals from which eyes were collected. Ponceau S staining was used as a loading control. Asterisk points to an unspecific band. (D) Sagittal view of the zebrafish eye and workflow employed in this study.
Retinal Degeneration Induced by the Loss of DJ-1 Can Be Inhibited by Müller Cell Expressed DJ-1
An earlier study of retina from DJ-1 knockout mice showed that loss of DJ-1 mainly affected the outer plexiform layer, photoreceptors and retinal pigment epithelium (RPE) [12,31]. Toluidine-blue-stained semithin sections of retina from 1-to-18-month-old DJ-1_KO zebrafish showed an age-dependent degeneration of retina (Suppl. Figure S1. In 5-month-old animals, the loss of DJ-1 expression seemed to have little effect on the structure and morphology of the retina, but in 9-month-old animals a marked change could be observed with vacuole-like structures in the retinal pigment epithelial (RPE) layer and disruption of photoreceptor filament organization ( Figure 2). These changes seemed to be even more manifested in older animals (Suppl. Figure S1). versus KO_DJ-1 (a,c). Note that GFP expression was determined by using a GFP antibody, followed by a far-red secondary antibody to avoid interference from retinal autofluorescence. Prominent expression is found around the Müller cell bodies. GFP expression also extends through the Müller processes into the photoreceptor layer. Faint expression can be observed in the Müller foot processes on the inner limiting membrane. Bar, 50 µm. (C) A Western blot shows expression of endogenous DJ-1 and Flag-tagged DJ-1 from total brain extracts belonging to animals from which eyes were collected. Ponceau S staining was used as a loading control. Asterisk points to an unspecific band. (D) Sagittal view of the zebrafish eye and workflow employed in this study.
Retinal Degeneration Induced by the Loss of DJ-1 Can Be Inhibited by Müller Cell Expressed DJ-1
An earlier study of retina from DJ-1 knockout mice showed that loss of DJ-1 mainly affected the outer plexiform layer, photoreceptors and retinal pigment epithelium (RPE) [12,31]. Toluidine-blue-stained semithin sections of retina from 1-to-18-month-old DJ-1_KO zebrafish showed an age-dependent degeneration of retina (Supplementary Figure S1. In 5-month-old animals, the loss of DJ-1 expression seemed to have little effect on the structure and morphology of the retina, but in 9-month-old animals a marked change could be observed with vacuole-like structures in the retinal pigment epithelial (RPE) layer and disruption of photoreceptor filament organization ( Figure 2). These changes seemed to be even more manifested in older animals (Supplementary Figure S1). We focused on 9-month-old retinas and included the analysis of the transgenic animals with reintroduced Müller-specific wild-type and mutant DJ-1 expression in the DJ-1 null background (Figure 2). Morphological changes observed in the DJ-1_KO included We focused on 9-month-old retinas and included the analysis of the transgenic animals with reintroduced Müller-specific wild-type and mutant DJ-1 expression in the DJ-1 null background (Figure 2). Morphological changes observed in the DJ-1_KO included both a retinal thinning and a reduction of cells in the ganglion cell layer, in addition to the morphological changes observed in the RPE and photoreceptor layers (Figure 2). This retinal degenerative pathology was inhibited by selective Müller cell DJ-1 expression. Reintroducing Müller cell DJ-1 c106a , on the other hand, did only inhibit changes in the photoreceptor layer and to some degree retinal thinning. Thus, distinct Müller cell DJ-1dependent pathways seem to be involved in retinal protection, with at least one being C106-dependent.
Introduction of Müller Cell DJ-1 Expression Inhibits DJ-1 Loss-Induced Ultrastructural Changes in the Retinal Pigment Epithelia Cells
RPE cells have an important role in diurnal phagocytosis of photoreceptors, as distal ends of their outer segments are pinched off and phagocytosed by neighboring RPE cells before they are degraded in a lysosomal-dependent pathway [31]. Different stages of phagosomes could be observed in all samples (Figures 3 and 4, marked arrowhead and p), but in contrast to wild type and Müller_DJ-1, DJ-1_KO and Müller_DJ-1c106a contained a number of large electron-dense structures (Figures 3 and 4, marked *). The morphometric analysis of these structures in the DJ-1_KO showed a mean cut diameter of 3.9 µm; thus, they are far larger than the phagosomes (0.9 µm) ( Figure 4B). Their electron-density and filamentous content resembled the appearance of heterolysosomes, but somehow they seemed to have been arrested in their degradation process. In the RPE cells of DJ-1_KO and Müller_DJ-1c106a, these large electron-dense structures would occupy most of the cytoplasmic space. In both DJ_KO and Müller_DJ-1c106a, large vacuoles in the RPE area were present. These vacuoles most possibly resemble the vacuoles observed at the light microscopic level (Figure 2). both a retinal thinning and a reduction of cells in the ganglion cell layer, in addition to the morphological changes observed in the RPE and photoreceptor layers ( Figure 2). This retinal degenerative pathology was inhibited by selective Müller cell DJ-1 expression. Reintroducing Müller cell DJ-1c106a, on the other hand, did only inhibit changes in the photoreceptor layer and to some degree retinal thinning. Thus, distinct Müller cell DJ-1-dependent pathways seem to be involved in retinal protection, with at least one being C106-dependent.
Introduction of Müller Cell DJ-1 Expression Inhibits DJ-1 Loss-Induced Ultrastructural Changes in the Retinal Pigment Epithelia Cells
RPE cells have an important role in diurnal phagocytosis of photoreceptors, as distal ends of their outer segments are pinched off and phagocytosed by neighboring RPE cells before they are degraded in a lysosomal-dependent pathway [31]. Different stages of phagosomes could be observed in all samples (Figures 3 and 4, marked arrowhead and p), but in contrast to wild type and Müller_DJ-1, DJ-1_KO and Müller_DJ-1c106a contained a number of large electron-dense structures (Figures 3 and 4, marked *). The morphometric analysis of these structures in the DJ-1_KO showed a mean cut diameter of 3.9 μm; thus, they are far larger than the phagosomes (0.9 μm) ( Figure 4B). Their electrondensity and filamentous content resembled the appearance of heterolysosomes, but somehow they seemed to have been arrested in their degradation process. In the RPE cells of DJ-1_KO and Müller_DJ-1c106a, these large electron-dense structures would occupy most of the cytoplasmic space. In both DJ_KO and Müller_DJ-1c106a, large vacuoles in the RPE area were present. These vacuoles most possibly resemble the vacuoles observed at the light microscopic level (Figure 2). A number of electron-lucent vacuoles were observed in the DJ-1 knockout RPE (Figure 4, marked LV). These structures may possibly be lipid vesicles in which lipid content is partly washed out during tissue preparation [32].
Protein Profiling of Resected Retinas
In an attempt to identify proteins associated with the loss of DJ-1 and proteins associated with Müller DJ-1 protection, we performed a label-free quantitative global massspectrometry-based protein analysis. NanoLC-HDMSE analysis was carried out on resected retinas from three animals from each of the groups, namely wild type, DJ-1_KO, Müller_DJ-1 and Müller_DJ-1 c106a , at the age of nine months. A total of 3211 proteins were identified on the basis of one or more peptides with a mass accuracy ≤ 10 ppm and a score ≥ 4. For further quantitative analysis, the protein should be identified with at least two unique peptides and appear in all three samples in at least one group, leaving 1868 proteins. It should be noted that, because the zebrafish proteome has not yet been comprehensively annotated with gene ontology (GO) terms, our data interpretation, in some cases, relies on the knowledge of their mammalian orthologue.
All protein hits qualifying for quantitative analysis are listed in Supplementary Materials Table S2.
Expression of Retinal Cell Markers in Knockout and Transgenic Lines
To establish the level of possible retinal degeneration and/or gliosis after DJ-1 loss, we searched for cell specific markers of Müller cells, retinal epithelial cells (RPE), rod photoreceptors, cone photoreceptors, retinal ganglion cells and microglia/macrophages [20,33,34] (Supplementary Materials Table S1).
No sign of gliosis, as reflected by an increase in Müller cells markers (GFAP and Glutamate synthase), was observed. On the other hand, the ganglion marker Gefiltin and a Rhodopsin variant associated to rod cells were decreased in knockout compared to wild-type retina.
The significant decrease in rhodopsin variant was also observed in Müller_DJ-1c106a.
Loss of DJ-1 Alters Expression of Proteins Belonging to the Respiratory Complex I and Glycolysis Independently of Reinsertion of Müller Cell DJ-1
To get an overview of proteins regulated by loss of DJ-1, we selected proteins with expression levels altered in DJ-1_KO, even though wild-type or mutant DJ-1 were reintroduced in the retinal Müller cells (Table 1). Most probably these identifications reflect protein changes in the neuronal retina or RPE. A majority of these proteins were components of the mitochondrial complex I. All of them were significantly downregulated in DJ-1_KO, Müller_DJ-1 and Müller_DJ-1 c106a , as compared to wild-type retinas. On the contrary, lactate hydrogenase, which converts pyruvate to lactate in glycolysis, was upregulated. Possibly, these changes reflect a shift in metabolism to minimize oxidative stress [35]. Another seeming response to oxidative stress was the upregulation of both glutathione S-transferase and glutathione peroxidase in both knockout and transgenic retinas compared to wild type. A corresponding transcriptional upregulation of Glutathione S-transferase was verified by using in situ hybridization (Supplementary Materials Figure S2). The in situ hybridization showed, in particular, high transcriptional levels of Glutathione S-transferase in the ganglion cell layer and inner nuclear layer in both knockout and Müller_DJ-1c106a as compared to wild type and Müller_DJ-1. LFQ, label-free quantitation; p-values when compared to wild type, * <0.05, ** <0.01 and *** < 0.001. n.d, not determined. NaN: Non Assigned Number (not detected).
Identification of Retinal Proteins Regulated by the Loss of DJ-1, but with Restored Levels after Introducing Müller Cell DJ-1
The morphological analysis showed that structural changes and retinal degeneration in DJ-KO were inhibited by the introduction of wild-type DJ-1 in Müller cells, but not by its mutant form. We therefore searched for retinal proteins dysregulated in both DJ-1_KO and Müller DJ-1 c106 , but with wild-type expression levels in Müller_DJ-1 ( Table 2). Within these criteria, we identified Prosaposin and G-protein-coupled receptor 37a (Gpr37), which are involved in glial-neuron protection [36]. Additionally, we also identified proteins regulating cell structure (Calponin 2), mitochondrial motility (Metaxin), autophagy (SEC23interacting protein) and inflammation (serum Amyloid P component) to be differentially expressed [37][38][39]. Expression levels for a ribosomal protein (Rpl36a) and a nuclear export protein (Exportin 1) were also found to be altered.
Identification of Proteins with Altered Expression in DJ-1 Knockouts Regardless of Introducing Either Müller Cell DJ-1 or Müller Cell DJ-1c106a
Although the cysteine-106 residue of DJ-1 is considered as an oxidative sensor through cysteine oxidation [7], DJ-1-dependent antioxidant function has also proven to be independent of C106 [10,40]. Introducing mutant DJ-1 in Müller cells did not protect from the retinal degeneration induced by DJ-1 loss (Figures 2-4). Intriguingly, it seemed to prevent a general stress response, a response that might be neuroprotective (Table 3). Seven proteins were found to be upregulated only in the DJ-1 knockout: Ependymin, Histone-H1-like, Cathepsin D, Methylmalonyl coA epimerase, Crystallins and Grifin. Crystallin gamma 1 and 2a, and Grifin, known as lens proteins, have all previously been found to be upregulated in retina as a response to stress [41]. Increasing evidence shows that Crystallins may have an important antioxidant function besides being structural lens proteins [42,43]. Methylmalonyl CoA epimerase is involved in lipid catabolism. Cathepsin D is an essential lysosomal protease in RPE cells and Cathepsin D deficiency results in extensive accumulation of lipofuscin [44]. Ependymin, an extracellular lipid-binding protein, is involved in cell adhesion and neuronal regeneration [45].
Discussion
In the present study, we show that loss of DJ-1 in zebrafish induces an age-related retinal degeneration similar to what has been observed in DJ-1-deficient mice [46] and to retinal pathologies associated with neurodegenerative diseases [47]. Here, however, we also demonstrate that this DJ-1 loss-induced degenerative retinal phenotype can be inhibited by reintroducing DJ-1 selectively in the retinal Müller cells.
Müller-specific DJ-1 expression was enabled by expressing DJ-1 under control of elements of the gfap promotor into a DJ-1 knockout line [19]. This promotor drives expression in astrocytes in the brain, but in retina, it is a Müller-specific promotor and does not drive expression in neither retinal astrocytes or microglia ( Figure 1A,B) [20,30]. Both wild-type DJ-1 and DJ-1 c106a were expressed, the latter because the Cysteine 106 is believed to act as an oxidative sensor in DJ-1 and to be essential for at least parts of DJ-1's antioxidant response pathways [10,40].
DJ-1 is a multifunctional protein with its importance in oxidative-stress protection being the most recognized. DJ-1 is expressed in all cell types, but in the brain, astrocytic expression seems highly important, as astrocyte DJ-1 not only protects the astrocytes themselves from oxidative stress, but also neighboring neurons [17,18]. Moreover, elevated DJ-1 expression within activated astrocytes is a pathological feature found in several neurodegenerating diseases, including Parkinson's disease and Alzheimer's disease [48,49]. Müller cells, the predominant glia cells in the retina, exhibit a comparable function in the retina as to astrocytes in the brain [50]. LFQ, label-free quantitation: * p < 0.05, ** p < 0.01 and *** p < 0.001. n.d.: not detected.
The loss of DJ-1 induced an age-dependent degeneration of the ganglion cell layer, prominent morphological changes in the retinal pigment epithelial cells (RPE) and structural changes in the photoreceptor layer (Figures 2-4 and Supplementary Materials Figure S1). The morphological changes in the RPE layer included vesiculation and occurrence of large electron dense structures, with the latter almost occupying the entire cytosol of the RPE cells in the aging DJ-1-deficient retina. All degenerative morphological features were inhibited by reintroducing Müller-selective wild-type DJ-1, but not DJ-1 c106a (Figures 2-4).
No sign of gliosis, as reflected in increased expression of Müller cells markers, was observed in DJ-1-deficient or transgenic retinas (Supplementary Materials Table S1). On the other hand, both DJ-1 knockout retinas and retinas with Müller-specific DJ-1c106a expression showed increased levels of the inflammatory markers Serum Amyloid P component and Prosaposin (Table 2). Both have been associated with Parkinson's disease [51,52].
Our proteomic profiles of whole retinas showed that the loss of DJ-1 affected the central metabolism by downregulation proteins belonging to the mitochondrial complex I and upregulating lactase hydrogenase, which converts pyruvate to lactate (Table 1). This most probably reflects a metabolic shift from oxidative phosphorylation to glycolysis to lower production to reactive oxygen species [53,54]. A similar change in protein profile was also observed in the lines with Müller cell DJ-1 expression, but to a lower degree. A metabolic shift would primarily affect the retinal neurons, and in particular the ganglion cells, which heavily depend on mitochondrial metabolism in contrast to Müller cells [55]. Both DJ-1 knockout and Müller DJ-1-expressing lines showed an upregulation of the oxidative-stress-response proteins glutathione peroxidase and glutathione S-transferase (Table 1), but seemingly this response was not sufficient in order to protect from retinal degeneration. Interestingly, DJ-1 knockout retinas also showed an upregulation of other proteins associated with retinal oxidative-stress response and with antioxidant properties [41][42][43]56,57]: the lens proteins Crystallins and Grifin (Table 3). This upregulation, which might have neuroprotective properties, was not observed in either retina expressing Müller DJ-1 wild type or mutant, thus indicating that both wild-type and mutant form DJ-1 in Müller cells induce an antioxidative response, which avoids initiating this alternative stress response.
Müller cells have an important function in retinal redox homeostasis, as they release glutathione (GSH), the major retinal antioxidant [58]. The tripeptide GSH is synthesized from serine/cysteine, glycine and glutamate, in which the latter is released from surrounding neurons and taken up into Müller cells via EAAT transporters [58]. DJ-1 may have several ways to influence and maintain GSH metabolism in Müller cells. Both glutamine influx and serine metabolism, which provide precursors of GSH synthesis, are reduced in DJ-1-deficient cells [59]. De novo synthesis of serine has shown to be important for Müller cell resistance to oxidative stress [60]. DJ-1 can also increase the uptake of neuronal released glutamate [61], which, in addition to reducing excitotoxicity, indirectly stimulates the Nrf2 pathway [62]. DJ-1 may also influence the availability of GSH by regulating rate-limiting enzymes in GSH synthesis [63]. Thus, reintroducing DJ-1 in Müller cells would not only re-establish redox balance in the Müller cell itself, but also the oxidative-stress-response pathway by which the surrounding DJ-1-deficient retinal neurons and RPE cells depend on.
Müller cells also have an important function in structural organization and assembly of photoreceptor outer segments (POSs), and targeted disruption of Müller cell metabolism affects the assembly of POS [64]. In the DJ-1 knockout retina, POSs appear to be unstructured, whilst both retinas expressing wild-type and C106-mutant DJ-1 in Müller cells appear to maintain proper POS organization (Figure 2).
Our proteomics analysis also suggested a possible role of Müller cell DJ-1 in regulating the neuroprotective prosaposin/GRP37 pathway (Table 2). Prosaposin (PSAP) is a neurotrophic factor mediating its neuroprotective effect through astrocytic GRP37L1 and GRP37 receptors [36]. In both DJ-1 knockout and Müller DJ-1 C106A -expressing retinas prosaposin levels were increased, whereas GRP37a, an ortholog to human GPR37, was only observed in wild-type and Müller DJ-1 retinas ( Table 2). Transcriptional profiling and in situ hybridization of mouse retina have shown that Müller cells are enriched in GRP37 transcripts [65]. Müller DJ-1 may potentially regulate Prosaposin/GPR37 signaling both through its regulation of the C106-dependenten ERK1/2 signaling [66] and through its regulation of PARKIN, which has GPR37 as a substrate [8,67].
Both DJ-1 knockout and retinas only expressing DJ-1 C106A in Müller cells showed age-dependent changes in the RPE cell layer, with accumulation of vesicles and electron dense structures (Figures 2-4). RPE cells phagocytose and digest daily shed photoreceptor outer segments (POSs) though a lysosomal-dependent pathway [31]. We observed different stages of phagosomes in the RPE of all zebrafish lines, but the much larger electron-dense structures were only observed in the knockout and Müller mutant DJ-1-expressing line (Figures 3 and 4). We are unsure of the identity of these structures, but they seemed to include POS-like structures. Thus, indicating that both DJ-1-deficient retinas and Müller DJ-1c106a-expressing retinas, in contrast to Müller wild-type DJ-1-expressing retinas, are dysfunctional in their degradation process of POS. RPE cells in both knockout and Müller cell DJ-1c106a-expressing retinas may be subjected to higher oxidative stress levels and nondegradable components in POS, thus hampering their normal function in POS phagocytosis and degradation [68]. The increase of the lysosomal Cathepsin D and lipid metabolizer Methylmalonyl CoA epimerase in knockout retinas possibly reflects high lysosomal stress in RPE cells (Table 3). Calponin, which plays a role in cell migration and phagocytosis, showed altered expression levels in DJ-1 knockout and Müller cell DJ-1c106a-expressing retinas, as compared to wild-type and Müller DJ-1-expressing retinas ( Table 2). It should be noted that zebrafish and also other vertebrate Müller cells are able to phagocytose cell debris from degenerating photoreceptors [69]. This function may be dysregulated in Müller cell DJ-1-deficient cells as DJ-1 has been proposed to be an activator of phagocytosis [70].
In conclusion, we have shown that loss of retinal DJ-1 induces an inflammatory and antioxidative response. This stress response is not sufficient to avoid severe age-dependent retinal degradation. In contrast, through re-insertion of DJ-1 selectively in Müller cells, the retinal degradation is avoided. This rescue effect is dependent upon the oxidative-stress sensor C106 residue of DJ-1. The Müller cell DJ-1 function seems to involve both regulation of retinal redox homeostasis and possibly also the psap/GPR37 neuroprotective pathway.
Supplementary Materials: The following are available online at [URL]/10 .3390/antiox10121862/s1, Figure S1: Age-dependent retinal degeneration in DJ-1 knockout retina, Figure S2: In situ hybridization of Glutathione S-transferase mRNA in the retina, Table S1: Expression of Retinal cell markers, Table S2: List of the 1868 identified protein used for quantitative analysis. Acknowledgments: Mass spectrometry was carried out at the Proteomics Core Facility and imaging at the Imaging Core Facility at the University of Bergen (UiB). Zebrafish were established and hosted in the zebrafish facility at the Department of Biological Science, UiB.
Conflicts of Interest:
There are no conflicts of interest.
==
Domain: Biology Medicine
|
[{"added": "2021-11-26T16:11:26.103Z", "created": "2021-11-23T00:00:00.000", "id": "244648446", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2076-3921/10/12/1862/pdf", "pdf_hash": "a6ac60eefa89059f7047ce5ec850463521372f00", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:4459", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "58b69d3937eadc798a67527c0647b611b23fa2c7", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Gut microbiota composition reflects disease progression, severity and outcome, and dysfunctional immune responses in patients with hypertensive intracerebral hemorrhage
Objective In this study, we aimed to explore the alterations in gut microbiota composition and cytokine responses related to disease progression, severity, and outcomes in patients with hypertensive intracerebral hemorrhage (ICH). Methods Fecal microbiota communities of 64 patients with ICH, 46 coronary heart disease controls, and 23 healthy controls were measured by sequencing the V3-V4 region of the 16S ribosomal RNA (16S rRNA) gene. Serum concentrations of a broad spectrum of cytokines were examined by liquid chips and ELISA. Relationships between clinical phenotypes, microbiotas, and cytokine responses were analyzed in the group with ICH and stroke-associated pneumonia (SAP), the major complication of ICH. Results In comparison with the control groups, the gut microbiota of the patients with ICH had increased microbial richness and diversity, an expanded spectrum of facultative anaerobes and opportunistic pathogens, and depletion of anaerobes. Enterococcus enrichment and Prevotella depletion were more significant in the ICH group and were associated with the severity and functional outcome of ICH. Furthermore, Enterococcus enrichment and Prevotella depletion were also noted in the SAP group in contrast to the non-SAP group. Enterococci were also promising factors in the prognosis of ICH. The onset of ICH induced massive, rapid activation of the peripheral immune system. There were 12 cytokines (Eotaxin, GM-CSF, IL-8, IL-9, IL-10, IL-12p70, IL-15, IL-23, IL-1RA, IP-10, RANTES, and TNF-α) changed significantly with prolongation of ICH, and the Th2 responses correlated with the 90-day outcomes. Cytokines TNF-α, IP-10, IL-1RA, IL-8, IL-18, and MIP-1β in SAP group significantly differed from non-SAP group. Among these cytokines, only IP-10 levels decreased in the SAP group. Enterococcus was positively associated with IL-1RA and negatively associated with IP-10, while Prevotella was inversely associated in both the ICH and SAP groups. Conclusion This study revealed that gut dysbiosis with enriched Enterococcus and depleted Prevotella increased the risk of ICH and subsequently SAP. The altered gut microbiota composition and serum cytokine profiles are potential biomarkers that reflect the inciting physiologic insult/stress involved with ICH.
Introduction
Intracerebral hemorrhage (ICH) is a devastating disease and a major public health issue worldwide (1). The fatality and longterm mortality rate of ICH have not changed significantly in recent years (2), and effective treatments have not yet been found in the internal medicine and surgery fields (3-7). Thus, alternative treatment options are needed to improve ICH prognosis and lower the risk of mortality.
The gut-brain axis is a bidirectional communication system that involves multiple pathways including neural, hormonal, and immunological signals with the microbiota as the central mediator (8). Acute ischemia rapidly causes severe gastrointestinal paralysis, ischemia and produces excess nitrate leading to intestinal dysbiosis (9,10). In turn, the intestinal flora and their metabolites affect the outcome and prognosis of stroke. The use of broad-spectrum antibiotics before a stroke can reduce the overall diversity of intestinal microbes and reduce cerebral infarction, which involves intestinal immune cell traffic to the meninges (11). Further, germ-free mice with intestinal dysbiosis after stroke had increased lesion volume and functional impairment compared to normal control mice (9). Alterations in gut microbiota composition affect the host immune system with inflammatory cytokine production and immune cell differentiation (12), enhancing the role of neuroinflammation in ICH. Thus, the intestinal microbiota and the interplay with the immune system are intervention strategies in the setting of stroke. Notably, current studies focus on ischemic stroke while the alteration of the microbiota and relative immune system in Frontiers in Immunology frontiersin.org ICH are less studied, particularly in vivo. Here, we characterized the microbiota and peripheral cytokine alterations in ICH patients and analyzed the relationships between changes in fecal microbiota and immune responses.
Materials and methods
The flow chart is shown in Figure 1. A total of 64 ICH patients, 46 coronary heart disease (CHD) controls, and 23 healthy controls were enrolled from April 2018 to December 2020 in the neurology department of Guangdong Provincial Hospital of Traditional Chinese Medicine, Shenyang Second Hospital of Traditional Chinese Medicine, The Fourth Affiliated Hospital of Guangzhou Medical University, and the health examination department of Guangdong Provincial Hospital of Traditional Chinese Medicine. All subjects provided written informed consent to participate in this study.
ICH was diagnosed according to American Heart Association/American Stroke Association guidelines (13). Stroke-associated pneumonia (SAP) following ICH is the most common complication after stroke. In this study, the diagnosis of SAP was based on modified Centers for Disease Control and Prevention (CDC) criteria (14). The inclusion criteria were as follows (1) age >18 years, (2) admission within 7 days of ICH onset, and (3) informed consent obtained and the retention of biological samples completed. The exclusion criteria were as follows: (1) ICH caused by brain tumor, brain trauma, blood diseases, cerebrovascular malformation, or aneurysm, (2) any antibiotics, prebiotics, or probiotics treatment within four weeks before admission, (3) active infection within two weeks before admission, (4) liver and kidney dysfunction, (5) history of gastrointestinal diseases such as gastrointestinal tumor, inflammatory bowel disease, or active gastrointestinal bleeding in the last 3 months, and (6) history of immune-related diseases or receiving immunotherapy. Clinical data such as age, gender, medical history, and neurological deficits were assessed and collected by neurologists.
Sample collection
Stool and serum samples of patients with ICH were collected at T1 (0-3 days after ICH), T2 (4-7 days after ICH), T3 (8-14 days after ICH), and T4 (14-30 days after ICH). In total, 170 stool and 184 serum samples were collected after the onset of symptoms. The number of stool samples for T1 to T4 were 44, 33, 82, and 11, while the number of serum samples for T1 to T4 were 46, 34, 88, and 16. Venous blood (10 mL) was collected at different phases on an empty stomach and centrifuged (3000 rpm, 10 min) within 6 h after collection. After centrifugation, the serum was divided into cryovials and stored in an -80°C refrigerator for cytokine analysis. Fecal Samples were collected in the morning, stored in the Fecal Microbial Collection and Preservation Kit (ML-001A, Shenzhen Dayun Gene Technology Co., Ltd.), and saved in an -80°C refrigerator within 72 h. Stool samples of healthy controls (HC) and CHD group were collected using the same methods. The specimens of all collaborative subcenters were transported through a cold chain and stored uniformly in the Biological Resource Center of Guangdong Provincial Hospital of Traditional Chinese Medicine to avoid repeated freezing and thawing.
DNA extraction, 16S ribosomal RNA gene sequencing
According to the manufacturer's instructions, the DNA was extracted using the magnetic soil and stool genomic DNA extraction kit (Magnetic Soil and Stool DNA Kit, Tiangen Biochemical Technology Co., Ltd.). After extracting total DNA from the stool samples, we used primers, 341F (CCTAYGGG RBGCASCAG) and 806R (GGACTACNNGGGTATCTAAT), to amplify the V3-V4 region of the bacterial 16S rRNA gene. The library was constructed using TruSeq DNA PCR-Free Library Preparation Kit from Illumina. The constructed library was subjected to Qubit quantification and library testing. After it was quantified, the NovaSeq 6000 was used for sequencing.
Statistical analyses
Categorical variables are presented as numbers and percentages, and continuous variables are presented as mean ± standard deviation (SD) or median (interquartile range (IQR)). Comparisons between groups were performed with chi-square tests for categorical variables. Continuous variables that followed the normal distribution were compared with the Student's t test or analysis of variance (ANOVA). Variables inconsistent with the normal distribution and Levene's test were compared with the non-parametric Wilcoxon test or Kruskal-Wallis test. Spearman's rank correlations were calculated between the relative abundance of bacterial communities and environmental variables or cytokine responses. The predictive performance of ICH prognosis was assessed by comparing receiver operating characteristic (ROC) curves. Statistical analysis was performed using SPSS 26.0 (Statistical Package for Social Sciences, Chicago, IL, USA) software. The analysis of intestinal flora was performed using QIIME software (version 1.9.1) and the R language tool (version 3.4.0). Changes in cytokine concentration were plotted using GraphPad Prism 9 (GraphPad Software, Inc.) software. A two-sided, P<0.05 was considered statistically significant.
Clinical characteristics of ICH and control groups
The demographic and clinical information of the 64 ICH patients, 46 CHD controls, and 23 healthy subjects included in this study are shown in Table S1 in the appendix. As Table S1 shows, there was no statistical difference in age and gender between the ICH group and the CHD group. Compared with the CHD group, the ICH group had patients with a higher proportion of hypertension history (76.563% vs 26.087%, P < 0.001). The two groups had similar rates of smoking and alcohol abuse histories. The median triglyceride in the CHD group was higher than that in the ICH group (0.995 vs 1.43, P=0.012), while the levels of total cholesterol, low-density lipoprotein, and highdensity lipoprotein were similar in the two groups. Compared with the HC group, the proportion of men in the ICH group was higher (59.375% vs 26.087%, P = 0.006) and the average age was older (P < 0.001). The proportion of patients with a history of hypertension, active smoking, drinking, and coronary heart disease was higher in the ICH group than in the HC group. Additionally, the median (interquartile range) of the ICH scores, GCS at admission, and NIHSS score at admission of the ICH group were 1 (2), 14 (6), and 10 (9) respectively. There were 27 (42.20%) patients with a neurological recovery defined as NIHSS score improving ≥ 40% after 14 d of standard treatments. In addition, there were 37 patients (57.81%) with functional independence (mRS ≤ 2) at 90 d.
Patients with ICH have altered and more diverse gut microbiota
We then characterized the ICH-associated gut microbiota by high-throughput sequencing of the V3-V4 region of the 16S rRNA gene. The gut microbial composition is shown in Figure S1A. The bacterial diversity and richness in the ICH and the control groups were measured by different methods using the Simpson index, Shannon index, and richness index. ICH patients had more diverse gut microbiota than the controls (Wilcoxon rank-sum test, compared to HC group, P = 0.018 Flow chart and enrolled participants in the current study. for the Simpson index and P < 0.001 for the richness index, Figure 2A; compared to CHD group, P = 0.002 for the Simpson index and Shannon index and P < 0.001 for the richness index, Figure 2B). Specifically, these results remained significant according to longitudinal analyses (Kruskal-Wallis Test, compared to HC group, P = 0.036 for the Shannon index, P < 0.001 for the richness index, Figure S1B; compared to CHD group, P = 0.027 for the Simpson index, P = 0.009 for the Shannon index and P < 0.001 for the richness index, Figure S1C).
To determine whether there were significant differences in the microbiota structure between ICH patients and controls, principal coordinate analysis (PCoA) was used. The microbial A B D C composition of the ICH group was significantly different from that of the HC or CHD group according to Bray-Curtis differences (Permutational multivariate analysis of variance (PERMANOVA) test; ICH vs HC, R2 = 0.028, P < 0.001; ICH vs CHD, R2 = 0.034, P < 0.001) ( Figures 2C, D). Furthermore, the PCoA also revealed that the gut microbiota changed dynamically with the prolongation of ICH ( Figures S1D, E). These results suggested that the richness and diversity of gut microbiota in patients with ICH were significantly different from those of controls.
The gut microbiota profile shows Enterococcus enrichment and Prevotella depletion in the ICH group To identify the most relevant taxa responsible for the observed differences, supervised comparisons of the microbiota between the ICH and control groups were performed by linear discriminant analysis (LDA) effect size (LEfSe) analysis without any adjustment. We used a logarithmic LDA score cutoff of 6.0 to identify important taxonomic differences between the ICH and control groups and found a notable difference in fecal microbiota. We identified, through LEfSe analysis, 19 taxa that were differentially abundant in the HC and ICH groups ( Figure 3A) and 25 taxa in the CHD and ICH groups ( Figure 3B). We observed that the relative abundances of Prevotella and Faecalibacterium were higher in the HC group than those in the ICH group, while the relative abundances of Enterococcus, Parabacteroides, Lachnoclostridium, Acidaminococcus, and Streptococcus were higher in the ICH group than those in the HC group. Moreover, the relative abundances of Prevotella and Roseburia were higher in the CHD group, whereas the relative abundances of Enterococcus, Parabacteroides, and Lachnoclostridium were higher in the ICH group. Notably, the relative abundance of Enterococcus was higher in patients with ICH compared to controls ( Figures 3A, B). Additionally, significant taxa were observed at different times after ICH ( Figure S2A, B). A generalized linear model (GLM) was used to model the microbiota that were significantly different between the ICH and control groups after controlling for possible confounding factors (age, gender, antibiotic use, and comorbidities) (15). As Tables S2A, B shows, Enterococcus, Parabacteroides, Streptococcus, Veillonella, Clostridium _innocuum_group, and Eubacterium_eligens_group differed significantly between the ICH and control groups after adjustment. Further, we found that the ICH score was the most important phenotype that contributed to the flora variation in ICH ( Figure S2D), suggesting that ICH was the major cause of microbiota alteration rather than hypertension or other comorbidities.
To understand microbial community metabolism among the ICH and control groups, MetaCyc was used, which is a database of metabolic pathways and components covering all domains of life (16). It showed that peptidoglycan biosynthesis V (b-lactam resistance) (PWY-6470), the super pathway of b-Dglucuronosides degradation (GLUCUROCAT-PWY), Bifidobacterium shunt (P124-PWY), and hexitol fermentation to lactate, formate, ethanol, and acetate(P461-PWY) were enriched in the ICH group ( Figure S2E), which were also positively correlated with Enterococcus (r = 0.866, r = 0.659, r = 0.697 and r = 0.655, respectively, P <0.001) ( Figure S2F). The functional capacities of the intestinal microbiome were predicted based on 16S data using BugBase (17). At the organism level, three potential phenotypes including anaerobic, facultatively anaerobic, and containing mobile elements were predicted to be significant in the ICH and control groups (P < 0.001) as Figure S3 shows. Among these three phenotypes, the ICH group had more mobile elements and facultative anaerobic bacteria and less anaerobic bacteria as the disease progressed. Meanwhile, the proportion of the facultative anaerobia phenotype was significantly enriched in the ICH group in the T1-T3 phases, with a mild recovery in phase T4. Collectively, these results suggested changes in microbiota profile were closely related to the disrupted intestinal microenvironment.
The gut microbial composition of SAP is shown in Figure S4A. In the PCoA analysis, there was no difference between the patients with and without SAP with respect to gut microbiota (PERMANOVA test, R2 = 0.005, P = 0.572) ( Figure 4A). While the SAP group was different from the non-SAP group using analysis of similarities (ANOSIM) (Kruskal-Wallis test, P < 0.001) ( Figure 4B). The dysbiosis of gut microbiota in patients with and without SAP are shown in Figure S4B. Furthermore, the LEfSe algorithm was used to analyze the flora with significant differences between the two groups. We found that 18 taxa were differentially abundant in the two groups ( Figure 4C). Among them, Enterococcus, Parabacteroides, Blautia, Lachnoclostridium, and Acidaminococcus were significantly enriched, and Prevotella were depleted in patients with SAP compared to the non-SAP group. The relative abundance of Enterococcus was higher in patients with SAP than in non-SAP (P < 0.001) ( Figure 4C); moreover, GLM further confirmed that the Prevotella, Blautia, Ruminococcus_torques_group, Sutterella, and Veillonella were different between the two groups after controlling for age, hematoma volume, NIHSS score, and antibiotic use (Table S2C). Enterococcus, Alistipes, Hungatella, and clostridium_immocuum_group were enriched in the SAP group and were positively correlated with the severity of ICH (admission and discharge NIHSS, ICH score, and hematoma volume), severity of pneumonia (NLR and PSI score (except Alistipes)), and poor outcome of ICH (discharge and 90-day mRS scores ( Figure 4D). Roseburia, Fusobacterium, and Prevotella were enriched in the non-SAP group and were negatively correlated with the severity and poor outcomes of ICH. Enterotypes, clustering human fecal metagenomic samples based on their taxonomic composition, are described as, "densely populated areas in multidimensional space of community composition" (18). Three types of enterotypes are traditionally reported, which are independent of age, gender, cultural background, and geography. We found that ET1 only appears in the SAP group (22%), and ET2 was predominant in the non-SAP group (85%), with a high abundance of Prevotella ( Figures S4C, D). The important genera in different enterotypes are shown in Figure S4E. Moreover, we found that the relative abundances of Enterococcus, Parabacteroides, Lachnospira, UCG_004, and Clostridium_innocuum_group were higher in the patients with ICH after developing pneumonia than those before developing pneumonia ( Figure S4F, G), which indicated that Enterococcus and Parabacteroides could be sensitive biomarkers in the prediction of which patients with ICH develop strokeassociated pneumonia.
Altered cytokine responses in ICH patients with alterations in taxonomic compositions of the gut microbiota
Gut microbial dysbiosis is associated with abnormal immune responses and is often accompanied by abnormal production of inflammatory cytokines (19). Thus, we investigated the dynamic changes of a broad spectrum of cytokines in the process of ICH and assessed the relationship between cytokine responses and clinical characteristics. Levels of different serum cytokines in the process of ICH were shown in Figure S5. There were 12 cytokines changed significantly at the four different time points (Kruskal-Wallis Test, Eotaxin: P = 0.036; GM-CSF: P = 0.006; IL-8: P = 0.027; IL-9: P = 0.011; IL-10: P = 0.030; IL-12p70: P = 0.014; IL-15: P = 0.006; IL-23: P = 0.015; IL-1RA: P = 0.003; IP-10: P < 0.0001; RANTES: P = 0.028; and TNF-a: P = 0.012). Among these cytokines, we found that levels of GM-CSF, IL-12p70, IL-15, IL-1RA, IL-9, IL-23, and TNF-a were gradually increased with the prolonging of time, and levels of these cytokines were positively associated with the 90-day unfavorable outcomes ( Figure 5A). Moreover, levels of IL-10 were gradually decreased with the prolonging of time despite a slight increase at the phase T2. Levels of IP-10 decreased sharply from phase T1 to T2 and increased from phase T2 to T4. Decreased IL-10 levels and increased IP-10 levels were negatively associated with the 90day unfavorable outcomes ( Figure 5A). Further, IL-1RA levels were also positively related with the severity of ICH (admission and discharge NIHSS (r = 0.401 and r = 0.482, P < 0.001), PSI score (r = 0.360, P < 0.001), ICH score (r = 0.476, P < 0.001), hematoma volume(r = 0.309, P < 0.001), length of hospital stay (r = 0.244, P = 0.003) and length of ICU stay (r = 0.459, P < 0.001), white blood cell and neutrophil counts (r = 0.395 and r = 0.303, P <0.001), and poor functional outcomes of ICH (discharge and 90day mRS scores(r = 0.492 and r = 0.285, P < 0.001)) ( Figure 5A). These findings suggested that IL-1RA may be a strong cytokine to predict the severity and poor functional outcomes of ICH. Next, we investigated whether specific cytokine responses correlated with the relative abundance of important genera. As Figure 5B shows, Enterococcus was positively related to IL-1RA (r = 0.229, P = 0.003) and negatively related to IP-10 as well as SDF-1a (r = -0.315, P <0.001 and r = -0.253, P = 0.001). Conversely, Prevotella was negatively corrected with IL-1RA (r = -0.427, P < 0.001) and positively related to IP-10 and SDF-1a (r = 0.248, P = 0.001 and r = 0.182, P = 0.019). Interestingly, Parabacteroides, Anaeroplasma, Lachnospira, Roseburia, Faecalibacterium, and Agathobacter were negatively associated with proinflammatory cytokines such as IL-1a, IL-b, IL-6, TNF-b, and others, which indicated that ICH onset was accompanied by a great alteration in the intestinal microbiota and immune responses.
To further examine the cytokine responses in SAP subjects, we found that there were six cytokines statistically significant between the SAP group and non-SAP group (Wilcoxon rank- Enterococci are promising factors in the prognosis of cerebral hemorrhage To determine signature bacteria that could discriminate the good or poor functional outcomes of short-term and long-term prognosis, we incorporated robust statistical analysis and applied 5fold cross-validation together with random forest to create classification models with consideration of the lowest error rate and standard deviation. The random forest model was used to select important genera. As Figure 6A shows, the combination of Enterococcus, Prevotella, Lachnospiraceae_NK4A136_group, [Clostridium]_innocuum_group, Fusobacterium, Romboutsia, and Sellimonas could distinguish the good or poor outcomes of shortterm prognosis (discharge NIHSS score decrease > 40% for good outcome), with an AUC of 0.8343 (95% CI = 0.7706-0.898). Among these genera, Enterococcus was the major genus in the signature biomarkers' random seed, which indicated that Enterococcus was likely to predict the short-term outcome of ICH. Moreover, 17 genera including Eubacterium, Roseburia, Fusobacterium, Enterococcus, and Prevotella consisted of the random forest of long-term outcomes of ICH (90-day mRS ≤ 2 for good outcome), with an AUC of 0.8364 (95% CI = 0.7764-0.8956) ( Figure 6B).
Discussion
Previous studies have shown that commensal microbiota played a critical role in degenerative and autoimmune diseases of the central nervous system (20, 21). Stroke itself markedly affects the composition of intestinal microbiota and these changes, in turn, can determine stroke outcome (9). Few studies have been conducted to reveal the characteristics of the intestinal microbiota and peripheral immunity associated with ICH in vivo. Here, we reported that ICH induced gut microbiota dysbiosis, which was similar to previous studies on other acute CNS injuries (9,11,22). Additionally, we described the cytokine response after ICH and its relationship to the intestinal microbiota.
Our study revealed the richness and diversity of fecal microbiota were altered in patients with ICH, in contrast to those in the HC group and CHD group. The microbiota structures were different between the ICH and control groups. These observations were consistent with a previous study that suggested the gut microbiota could be altered in ICH (22). In terms of the composition of the gut microbiota, we determined certain specific changes in the composition of the bacterial genera in patients with ICH relative to controls by applying the LEfSe algorithm. We observed a significant increase in putative pathobionts in the ICH group. Enterococcus, a genus of Firmicutes phylum, are considered commensal organisms of the human gastrointestinal tract. However, they can also be pathogenic, usually causing urinary tract infection, bacteremia, endocarditis, burn and surgical wound infections, neonatal sepsis, abdomen and biliary tract infections, and root canal failure (23)(24)(25). In our study, Enterococci levels were higher in the ICH samples, and this genus has been associated with producing bacteriocins, which are linked to mobile elements (24). Additionally, Enterococci are an important clinical cause of bloodstream infection. The incidences of E. faecalis and E. faecium bloodstream infections were 4.5 and 1.6 per 100000, respectively, in a population-based study (26). The researchers showed that E. faecium infections were associated with gastrointestinal illness and affected patients who were invalid and hospitalized, leading to a high mortality rate. Other studies showed that Parabacteroides was also abundant in patients with hypertension (27) and large artery atherosclerotic stroke or transient ischemic attack (28). Similarly, the less studied Acidaminococcus was enriched in hypertension subjects in other cohorts as well (29)(30)(31). A higher abundance of Lachnoclostridium could lower circulating levels of acetate, resulting in increased visceral fat negatively impacting obesity and type 2 diabetes (32). Lachnoclostridium has been found to produce trimethylamine (32). Trimethylamine N-oxide (TMAO) promotes atherosclerosis and is linked to platelet hyperreactivity and inflammation, which in turn participates the development of stroke and its secondary consequences (33). Streptococcus was found to cause neurological damage by producing neurotoxins such as streptomycin, streptodornase, and streptokinase (34). The abnormal increase of these putative pathobionts could produce endotoxins and neurotoxins and were associated with high-risk factors of ICH, which may have contributed to the development of ICH pathogenesis (34). The three main depleted genera in the ICH group, Prevotella, Faecalibacterium, and Roseburia, are major commensal or beneficial microbes. Prevotella are linked to a plant-rich diet composed of carbohydrates and fiber; although, in the gut, they have been linked to inflammatory conditions (35,36). One study A B C FIGURE 5 Correlations among abundances of significant fecal microbiota, clinical indexes, and serum levels of cytokines. (A) Heatmap of Spearman's rank correlation coefficient among cytokines and specific clinical indexes. (B) Heatmap of Spearman's rank correlation coefficient among cytokines and significant genera between ICH and control groups. (C) Heatmap of Spearman's rank correlation coefficient among significant cytokines and significant genera between the SAP and non-SAP groups. *: P < 0.05. **: P < 0.01.
found that subjects with a high Prevotella abundance lost more weight when eating ad libitum whole-grain diets, suggesting Prevotella may control body weight (37). Faecalibacterium and Roseburia have been widely considered critical butyrate acidproducing beneficial bacteria (38, 39), whose populations were depleted in many diseases (29,40,41). Among these genera, our study also showed that the Enterococcus and Parabaceroides populations increased in the ICH group, and had a robust correlation with the severity of disease, inflammatory conditions, and poor outcomes. However, some beneficial microbes, such as Prevotella and Roseburia, correlated inversely with the above-mentioned factors.
Recently, studies on SAP have increased significantly. SAP is the major complication of ICH and has high mortality and morbidity (42). However, there have been few advancements in the prevention and treatment of SAP (43). Increasing evidence has shown that gut microbiota played an essential role in poststroke infection (44,45). Therefore, in this study, we explored the microbiota community of SAP after ICH. We found that more patients with moderate to severe ICH were admitted to the intensive care units of our clinical centers because of high SAP rates. We then found that there were structural differences in the gut microbial communities between the SAP and non-SAP groups. Consistent with the altered gut microbiota in the ICH patients, Enterococcus enrichment and Prevotella depletion were also found in the SAP patients, and Enterococcus was positively associated with the severity of ICH and SAP and poor outcomes of ICH, whereas we found that Prevotella was inversely associated. This suggested that Enterococcus enrichment and Prevotella depletion not only promoted the progression of ICH but also increased the occurrence of SAP. Similar to previous studies, Prevotella was associated with a reduced risk of hospitalacquired pneumonia in adult intensive care unit patients (46) and was reduced in the oropharynx of adults and children with asthma or chronic obstructive pulmonary disease (47). Enterococcus was similarly abundant in SAP following acute ischemic stroke (48) and acquired immune deficiency syndrome (49), which showed that Enterococcus could be related to strokeinduced immunodepression, a leading mechanism of SAP (50)(51)(52)(53). Furthermore, enterotypes could be the potential predictors of SAP, as showed that ET1 was the best indicator in the SAP group driven by Enterococcus, and ET2 was dominant in the non-SAP group with a high abundance of Prevotella. However, our cohort was composed of a small population from the Southern and Northeast regions of China.
In terms of the potential microbial functions, our study showed overgrowth of facultatively anaerobic and mobile element-containing bacteria and the decrease of anaerobic bacteria in patients with ICH, which indicated that the alteration to the gut microbiota may be involved in the development of brain injury. Interestingly, the Enterococcus (facultatively anaerobic) increase and the Prevotella (anaerobic) depletion were in agreement with our findings. Further, we found that metabolic pathways of peptidoglycan biosynthesis and hexitol fermentation to lactate, formate, ethanol, and acetate had a positive association with Enterococcus in the ICH. Peptidoglycan is an essential molecule in the cell wall of both gram-positive and gramnegative bacteria. In a previous study, intraperitoneal injection of 2E7, to neutralize circulating peptidoglycan, suppressed the development of autoimmune arthritis and experimental autoimmune encephalomyelitis in mice (54), which indicated that peptidoglycan could be related to the development of autoimmune disease (55,56). Peptidoglycan has been found in human atherosclerotic lesions (57). Increased baseline levels of peptidoglycan recognition protein-1 (PGLYRP-1), a proinflammatory molecule that binds peptidoglycan, were independently associated with an increased risk of first atherosclerotic cardiovascular disease (ASCVD) in a ten-year cohort, suggesting that PGLYRP-1 may contribute to the development of ASCVD (58). Cerebral ischemia is a contributing mechanism to secondary injury after ICH. Lactate accumulation induced by ischemic damages was observed in the ICH model (59, 60). Lactic acid has been shown to exacerbate ischemic brain injury by activating G protein-coupled receptor 81 (GPR81) and inhibition of GPR81 attenuated the ischemic injury (61). Additionally, early elevated cerebral lactate levels in extracellular fluid were associated with the occurrence of pneumonia in patients with aneurysmal subarachnoid hemorrhage, which may result from systemic hypoxemia or lactatemia with a damaged blood-brain barrier (62). Further, lactate accumulation in the colon could alter gut microbiota composition (63) and modulate immune responses (64). In our study, Lactobacillales (data not shown), Enterococcus, and Streptococcus (lactic acid bacteria) were significantly enriched in the ICH group compared to the HC group. These results suggested that microbiota-derived lactate may participate the secondary injury after ICH and increase the occurrence of SAP.
A growing body of evidence has suggested that intestinal microbes modulated the induction, training, and function of immune system responses, with gut microbiota dysbiosis related to several autoimmune and immune-mediated inflammatory diseases (65-67). Therefore, we investigated the dynamic changes of a broad spectrum of cytokines following ICH and evaluated the relationship between inflammatory cytokine response and long-term outcomes of ICH and signature microbiota. In our study, we found that Eotaxin, GM-CSF, IL-8, IL-9, IL-10, IL-12p70, IL-15, IL-23, IL-1RA, IP-10, RANTES, and TNF-a were changed significantly in the progression of ICH. levels of GM-CSF, IL-12p70, IL-15, IL-1RA, IL-9, IL-23, and TNF-a were increased and levels of IL-10 decreased gradually, which positively correlated with 90-day poor outcomes. Many studies revealed that GM-CSF promoted leptomeningeal collateral growth, decreased the infarct size, and improved long-term functional outcomes in the experimental stroke (68,69). GM-CSF was more than a growth factor and researchers showed that GM-CSF also promoted neuroinflammation by increasing LPS-induced production of proinflammatory mediators (70). In line with previous study, astrocyte-derived IL-15 significantly increased in the ICH patients and experimental ICH and aggravated brain injury following ICH through the proinflammatory response amplification of microglia in the setting of ICH (71). Similarly, astrocytic IL-15 also exacerbates brain damage after ischemic stroke by enhancing cell-mediated immune responses (72). Researchers found that IL-23 signaling could promote Th2 polarization and enhance Th2 expression in allergic inflammation (73). Expression of IL-23 and IL-17 increased in sequence following ICH and IL-23/IL-17 axis promoted secondary brain injury in ICH model mice (74). However, IL-17 levels did not increase in the acute phase in our study, which indicates that IL-23 may influence ICH in a Th17-independent manner. Consistent with previous research, higher IL-1RA, erythrocyte sedimentation rate, and CRP were correlated with dependent stroke outcome (mRS >3) in acute ischemic stroke (75). Moreover, IL-5, IL-6, IL-9, and IL-27 were also positively correlated with long-term functional outcomes. Increased serum levels of IL-6 and IL-10 were detected in intraparenchymal hemorrhage (76), and higher admission IL-6 levels were associated with unfavorable 90-day functional outcomes and hematoma and perihematomal edema volumes (77). Additionally, IL-6 and IL-10 levels were higher in hemorrhage stroke patients with 1-month unfavorable outcomes (78). Hematoma expansion is a major cause of morbidity and mortality after ICH, and inflammation may be associated with its pathogenesis. Higher plasma IL-10 levels were related to the hematoma expansion in spontaneous ICH and worse 30-day outcomes (79). However, a study on IL-10-/-mice showed that the presence of IL-10 was protective against the development of ICH (80). Although IL-10 is regarded as an anti-inflammatory cytokine to prevent inflammatory and autoimmune pathologies by limiting the immune response to pathogens (81), it also exhibits proinflammatory activities. A study showed that IL-10 treatment stimulated lipopolysaccharide (LPS)-induced release of IFN-g and enhanced activation of CTL and NK cells after LPS injection, though IL-10 treatment upon LPS-induced IFN-g release could not be reproduced in whole blood in vitro (82). IL-5, IL-9, IL-10, IL-23, and IL-27 are also related to the Th2 response (T-cell response associated with allergies, progressive systemic sclerosis, and autoimmune disorders) (83). Researchers observed that IL-27 was upregulated centrally and peripherally after ICH, and IL-27 treatment improved ICH outcomes by reducing edema and increasing iron and hematoma clearance (84). However, higher IL-27 levels were correlated with poor 90day outcomes in our results. The findings above suggested that high-dose anti-inflammatory therapy in patients with inflammatory disorders could be associated with undesired proinflammatory effects in vivo.
In contrast, IL-21, CXCL1, IL-4, IL-31, IFN-g, SDF-1a, and IP-10 were negatively associated with 90-day unfavorable outcomes. Recently, IL-4/STAT6 signaling accelerated microglia-and macrophage-mediated hematoma clearance and improved neurofunctional recovery following ICH in blood and collagenase injection models (85). Additionally, in a study about the relationship between ex vivo cytokine synthesis and 3-month outcomes after ischemic stroke, decreased release of IP-10, TNFa, IL-1b, and IL-12; increased release of IL-10 and IL-8; and higher plasma IL-6 levels were associated with poor outcomes (86). Additionally, decreased release of IP-10 and TNF-a after ex vivo blood stimulation with endotoxin was associated with poor outcomes after stroke, suggesting that the inhibition of both the MyD88-dependent and MyD88independent pathways of toll-like receptors (TLR)4 signaling in blood cells was associated with poor prognosis in stroke patients (87). Reduced IFN-g production caused by impaired NK and T cell response was the crucial stroke-induced defect in the antibacterial defense. IFN-g supplementation effectively i n h i b i t e d b a c t e r i a l i n f e c t i o n s a f t e r s t r o k e ( 5 0 ) . Neovascularization after ICH is an important compensatory response that mediates brain repair and improves the clinical outcome. The Tp53 Arg72Pro single-nucleotide polymorphism increased endothelial cell survival and triggered efficient endothelial progenitor cell mobilization via vascular endothelial growth factor and SDF-1a, resulting in neovascularization after experimental ICH (88). In conclusion, the onset of ICH induced massive, rapid activation of the peripheral immune system and Th2 responses were correlated with worse 90-day outcomes.
Furthermore, Enterococcus was positively associated with IL-RA and negatively associated with IP-10 and SDF-1a, while Prevotella showed an inverse association. Peptidoglycan is detected by multiple pattern-recognition receptors and triggers inflammatory responses in immune and nonimmune cells (89). TLR2s are known to be the signaling receptors for peptidoglycan, which induced IL-1RA gene expression by activating the p38 stress-activated protein kinase (90). IP-10 has been shown to have direct antibacterial activity similar to adefensins, like against Escherichia coli and Listeria monocytogenes (91). In addition, IFN-g signaling in enteric glia cells (EGCs) maintains intestinal homeostasis and immunity and improves tissue repair after intestinal damage caused by pathogen infection. Researchers have identified IP-10 as the critical response cytokine in IFN-g signaling, thus the IFNg-EGC-IP-10 axis is essential to the immune response and tissue repair after infectious challenge (92). Collectively, this evidence showed that Enterococcus interacted with cytokines such as IL-1RA, IP-10, and SDF-1; promoted the TLR-2 pathway; inhibited the TLR-4 pathway (87); induced neovascularization; and disturbed the homeostasis of the intestinal microbiota to aggravate the inflammatory response and worsen ICH outcomes.
Stroke-induced immunosuppression (SIIS) was characterized by decreased lymphocyte counts in the spleen, blood, and thymus; impaired early NK and T cell responses, and a shift from Th1 to Th2 (50). This syndrome increased the susceptibility to stroke-associated infections. Among these infections, SAP was the major acute type of ICH and can worsen ICH functional outcomes (93). To elucidate the molecular mechanisms of SAP, the peripheral suppression of the immune system after the occurrence of ICH must be considered. In this study, we found that there were six cytokines that were significantly correlated with SAP, including IP-10, IL-1RA, TNF-a, MIP-1b, IL-18, and IL-8. IP-10 was the only cytokine that was decreased in the SAP group. IFN-g plays a pivotal role in preventing bacterial infections after stroke. Studies have revealed that supplementing with IFN-g by adoptive transfer of IFN-g-producing lymphocytes or recombinant IFN-g treatment inhibited bacteremia and pneumonia (50). However, this did not prove whether the downstream effector of IFN-g was associated with strokeassociated infections or not. IP-10, also called IFN-g-inducible protein 10, is a chemokine secreted from cells stimulated with type I and II IFNs and LPS (94). It is vital in controlling pneumonia by enhancing IFN-g production and reinforcing leukocyte antibacterial responses (95). In a previous study, at the early stage of Klebsiella administration, anti-IP-10 antibody treatment led to 10-to 100-fold increases in the number of Klebsiella pneumoniae CFU isolated from lung homogenates compared to IgG administration. Additionally, adenovirusmediated expression of IP-10 led to 30-to 100-fold reductions in lung and blood CFU in Klebsiella-infected mice in the early stage (95). Therefore, the IFN-g-IP-10 axis may be a candidate pathway for immunotherapy of SAP or severe respiratory tract infection. Decreased secretion of TNF-a and IFN-g has contributed to spontaneous bacterial infections. However, a reduction of endotoxin-induced TNF-a was observed 12 h and 2 d after middle cerebral artery occlusion and returned to control levels on day 5 (50). Our study showed that TNF-a levels were increased in the SAP group, which suggested that the increase of TNF-a present during the late stage of stroke could also be linked to SAP, as the duration of SIIS still remained unknown. Stroke severity was the most important predictor of infection risk, and increased plasma IL-1RA levels were independently associated with infection risk after adjusting for stroke severity. This suggested that IL-1RA was a strong predictor of post-stroke infection (96). Moreover, in a previous study, the A2A2 genotype of the IL-1RA gene was associated with the risk of adverse outcomes of severe community-acquired pneumonia in Indian children (97), though there were some different findings in other studies (98, 99). MIP-1b, an inflammatory chemokine, has an impact on vasculopathy. Researchers have found that MIP-1b inhibition improved endothelial progenitor cell (EPC) function and enhanced EPC homing and ischemia-induced neovasculogenesis (100). Increased IL-18 and IL-8 expressions contributed to the development and severity of stroke (101)(102)(103) and IL-18 also participated in hypoxic-ischemic brain injury (104). Interestingly, a novel innate immunity pathway consisting of lipoteichoid acid, produced by gram-negative bacteria, was sensed by the NLRP6 inflammasome and exacerbated a systemic gram-positive pathogen infection via the production of IL-18 (105). Unexpectedly, IL-18 was not associated with the genus Enterococcus in our study. Further analysis showed an association between the cytokines and genera and revealed that increased bacteria in the SAP group, especially Enterococcus, enhanced the expression of IL-1RA and decreased IP-10 levels to promote SAP. Increased bacteria in the non-SAP group, particularly Prevotella, were inversely related to SAP.
Random forest analysis showed that Enterococci were the critical biomarkers in determining either the good or poor functional outcomes of short-term and long-term prognoses.
Notably, Enterococci were the most important biomarkers in predicting short-term functional outcomes, which was due to its levels gradually increasing throughout the ICH process and peaking at phase T4.
The major findings of our study were that the gut microbiota changed dynamically throughout the duration of ICH, and gut dysbiosis with Enterococcus enrichment and Prevotella depletion not only promoted ICH but also SAP. Moreover, we investigated the dynamic changes of a broad spectrum of cytokines in the process of ICH and confirmed the roles of these cytokines in patients with ICH, and examined the relationship between genera and cytokine responses. Our study does have some limitations. First, the age, gender, and comorbidities of the ICH group and the control groups were not identical, although a generalized linear model was applied to control the possible confounding factors. Additionally, the stool samples in different phases of the ICH process were varied, as ICH could reduce gastrointestinal motility (22). Second, short-chain fatty acid (SCFA) levels, which potentially mediate gut-brain communication, were not tested (106), although a number of studies have revealed that SCFAs played a beneficial and anti-inflammatory role in stroke (107,108). Third, antibiotics are unavoidable, important factors for ICH patients and gut microbiota. ICH patients in this study were recruited from intensive care units and 73.4% of them were diagnosed with pneumonia within 7 days of ICH. Thus, we should consider the effect of antibiotics on intestinal microbiota. However, the impact of different types of antibiotics and their application times on intestinal flora in vivo remains unknown, and it is difficult to control the antibiotics used in unpredictable medical conditions. Finally, for the study scale, we did not analyze the correlations between every taxon and every cytokine tested in this study under the consideration of the limited statistical power of multiple comparisons. Consequently, the related changes in the microbiota and serum cytokines were analyzed under the assumption that the altered microbiota may trigger peripheral inflammatory responses that contributed to ICH or SAP. Additionally, the specific mechanisms underlying the microbiota and ICH process were not explored in this study. Therefore, in a future study, we plan to simulate the intestinal alteration by enriching Enterococcus or depleting Prevotella in experimental ICH to verify the potential targets and elucidate their causal relationship in the gut-brain axis. Moreover, specific immune responses stimulated by a particular species, or a group of gut microorganisms, need to be investigated. We will analyze the dynamic changes in cytokine responses in patients with SAP in our subsequent studies to identify the changes in inflammatory responses. More study patients will be enrolled in the future to support our findings. As discussed, SIIS was a key mechanism of ICH and SAP. More information about SIIS including the duration, the cytokine storm, and its activation is needed.
Conclusion
In summary, to the best of our knowledge, this is the first study to show that patients with Enterococcus enrichment and Prevotella depletion in the gut microbiota had increased risk of ICH and SAP in vivo. Changes in a broad spectrum of cytokines associated with the signature microbiota proved that microbiota alterations with aberrant host immune responses were related to ICH pathogenesis. Elucidation of the interaction between intestinal microbiota and the peripheral immune response would help to understand ICH pathogenesis. The altered gut microbiota composition and serum cytokine profiles are potential biomarkers that reflect the inciting physiologic insult/stress involved with ICH. Gut microbiota modulation may help to the development of intervention strategies targeting microbiota dysbiosis for ICH.
Data availability statement
The original contributions presented in the study are publicly available. This data can be found here: [URL]. ncbi.nlm.nih.gov/bioproject/PRJNA805052.
Ethics statement
The studies involving human participants were reviewed and approved by all participating centers. The patients or volunteers provided their written informed consent to participate in this study.
Author contributions
JG, LZ, JZ, JL and YC designed the study and provided guidance on the data analysis and interpretation/presentation of the data. JG is the subject primarily responsible for the clinical trial and provided a critical review of the manuscript. JL, YC, and XS performed and analyzed all the data. JL drafted most sections of the manuscript. GT, JZ, ZL, XY, TH, GH, LW and YH conceived the study, supervised the participants, and revised the manuscript. CW and BW participated in specimen processing and quality control. YZ, QG, BG, ML, RW, JY and WC organized and managed the study including samples and data collection, and quality assurance. All authors contributed to the article and approved the submitted version. JL and YC have an equal contribution to this manuscript.
Funding
The present study was funded by the National Natural Science
==
Domain: Biology Medicine
|
[{"added": "2022-10-14T13:30:56.167Z", "created": "2022-10-14T00:00:00.000", "id": "252878287", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.frontiersin.org/articles/10.3389/fimmu.2022.869846/pdf", "pdf_hash": "5def024db25cc66ab7884bbb105a5344af873f0b", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:1581", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "fb75c8ab3e40b215e0b5076f18194f5893ebf05d", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Genetic and epigenetic factors fine-tune TGFB1 expression within the osteoarthritic articular joint
Objective : Osteoarthritis (OA) is an age-related disease characterised by articular cartilage degeneration. It has a large heritability and genetic screens have identified single nucleotide polymorphisms (SNPs) marking genomic risk loci. One such locus is marked by G>A SNP rs75621460, downstream of TGFB1 . This gene encodes TGF- β 1, the correct expression of which is essential for cartilage maintenance. We have used a combination of human patient samples (n=319) and a chondrocyte model to characterise the impact of rs75621460 in multiple articular joint tissues. Methods : Patient samples were genotyped and DNA methylation (DNAm) levels quantified by pyrosequencing. Gene reporter and electrophoretic mobility shift assays were used to determine differential nuclear protein binding to the region. The functional impact of DNAm upon TGFB1 expression was tested using targeted epigenome editing. Results : We identified that rs75621460 is located within a TGFB1 enhancer, and that the OA risk A-allele alters transcription factor binding, decreasing enhancer activity. Protein complexes binding to A (but not G) induced DNAm at flanking CG-dinucleotides. Strong correlations were observed between patient DNAm levels and TGFB1 expression, the direction of which was opposing between cartilage and synovium. This demonstrated biological pleiotropy in the impact of the SNP within different tissues of the articulating joint. Conclusion : The OA risk SNP rs75621460 impacts TGFB1 expression by modulating the function of a gene enhancer. We propose a mechanism by which the SNP impacts enhancer function, providing novel biological insight into one mechanism of osteoarthritis genetic risk, which may facilitate the development of future pharmacological therapies. risk loci 37 . The very small range of DNAm values over which correlations with TGFB1 expression occur potentially suggest the effects operate in a sub-population of chondrocytes within the tissue. In vitro methylation of the enhancer reduced the activity of both alleles in a reporter assay, and EMSA analysis indicated that DNAm at CpG2 impacted upon protein binding. The SP1 antibody bound to both methylated and unmethylated probes, consistent with previous reports into binding of this transcription factor 38,39 . We further identified that a targeted increase of DNAm at CpGs3-6 could reduce TGFB1 expression in the absence of the rs75621460 A-allele.
OA is an age-related, degenerative disease of the articulating joints, affecting over 40 million Europeans 15 . The disease hallmark is the thinning and loss of articular cartilage, often accompanied by a low-grade synovial inflammation within the affected joint 16 . This leads to chronic impairment of joint function, with a resultant increased risk of premature death due to secondary co-morbidities 17,18 . A typical clinical end-point is surgical replacement of the affected joint. Currently, there are no disease-modifying OA drugs and novel treatments are urgently required.
The causes of primary OA are complex. Yet, with an estimated heritability of ~50%, genetic influences contribute highly to disease susceptibility 19 . Genome-wide association studies (GWAS) have revealed the highly polygenic nature of OA and, over 90 significant association signals have been reported. Risk variants are often intergenic and thought to operate by mediating differential expression of their target genes. This places OA in the "enhanceropathy" category of common diseases, in which subtle but detrimental changes in gene expression through aberrant activity of DNA regulatory elements, or "enhancers", contribute to disease progression 20 .
In 2019, an OA risk signal was reported at chr19q13.2, marked by intergenic SNP rs75621460 (G>A; minor allele frequency (MAF), 0.03) 14 . The SNP lies 2.4kb downstream of TGFB1 and has >99% probability of being the single causal variant at this locus 14 . In this study we investigate rs75621460 and the encompassing region of DNA for regulatory activity.
Accepted Article
This article is protected by copyright. All rights reserved Furthermore, we quantify genetic variation and epigenetic modifications within the region and measure the impact upon expression of TGFB1 in multiple human joint tissues.
In silico analysis of the locus
An in-silico analysis of the locus was performed using ROADMAP chromatin state data 21 , RNAsequencing (RNA-seq) data from hip OA and neck-of-femur (NOF) fracture cartilage 22 , and ATAC-sequencing data from knee OA cartilage 23 . P-values for RNA-seq data were calculated using a Wald test within the DESeq2 package. The ROADMAP 18-state model utilises 6 histone post-translational modifications to assign one of eighteen chromatin states to cell-specific epigenomes and was used here to identify potential regulatory function in two cell types: E006, embryonic stem cell-derived mesenchymal stem cells (MSCs); and E049, bone marrow-derived cultured chondrocytes. Analysed knee articular cartilage ATAC-seq data was downloaded directly from GEO (accession GSE108301) 23 . Population allele frequencies of rs75621460 were taken from LDlink.
Luciferase reporter analysis
A 553bp region encompassing rs75621460 was amplified from pooled blood DNA, cloned into the pGL3-Basic firefly reporter vector (Promega), and sequenced to identify clones with the ancestral G-or derived A-allele at rs75621460. Tc28a2 immortalised chondrocytes were seeded into a 96-well plate 24h prior to transfection with the relevant pGL3-promoter luciferase vector construct (100ng) and pRL-TK Renilla vector (1.5ng) using Fugene HD transfection reagent (Promega). After 24h, cells were lysed and luciferase activity was measured by GloMax Navigator (Promega). For each well, luciferase activity was normalised to that of Renilla as previously described 24 .
Electrophoretic Mobility Shift Assay (EMSA)
Nuclear protein was extracted from Tc28a2 cells as previously described 25 . For each allele of rs75621460, forward and reverse single stranded DY682-labeled oligonucleotides (Eurofins), spanning 15bp each side of the SNP, and encompassing CpG2 were annealed to generate doublestranded probes (Table S1). Four probe combinations were generated containing either the G allele or A allele at rs75621460 that were unmethylated or methylated at CpG2. Reactions were carried
Accepted Article
This article is protected by copyright. All rights reserved out as previously described 25,26 . For supershift assays, 2µg of the indicated antibody was added to the binding reaction (Table S2).
CRISPR-Cas9
The CHOPCHOP CRISPR Design Tool 27 was used to design guide RNA (gRNA) sequences which were predicted to have low off-target effects, a GC content between 40 and 70%, and with a high targeting efficiency immediately upstream (gRNA1) and downstream (gRNA2) of rs75621460. The selected gRNAs created an 84bp deletion encompassing rs75621460 (Table S3).
Gene expression analysis
cDNA was reverse transcribed from total RNA using the Superscript IV standard protocol (Invitrogen) after an initial 15-minute treatment with 1 unit of amplification grade DNaseI (Invitrogen). Gene expression was measured by quantitative reverse transcription-polymerase chain reaction (qRT-PCR) using pre-designed TaqMan assays (Integrated DNA Technologies).
Gene expression was quantified using TaqMan chemistry, normalised to housekeeping genes 18S, HPRT1 and GAPDH and expressed as 2 -Δct as described previously 29 .
Patient samples and extraction of nucleic acids
Human tissue samples were obtained from patients undergoing hip or knee joint replacement surgery due to end-stage OA or NOF fracture. Arthroplasty was conducted at the Newcastle upon Tyne NHS Foundation Trust hospitals. The Newcastle and North Tyneside Research Ethics
Accepted Article
This article is protected by copyright. All rights reserved Committee granted ethical approval for the collection, with each donor providing verbal and written informed consent (REC reference number 14/NE/1212). Further details of the patient samples used in this project are provided in Table S4. RNA was extracted from cartilage by TRIzol-chloroform (Life Technologies) separation, following which the RNA was purified from the aqueous phase using the RNeasy Mini Kit (Qiagen). Both DNA and RNA were extracted from whole blood and synovium using the EZNA DNA/RNA Isolation kit (Omega Bio-Tek). For genotyping DNA was used directly. For methylation analysis, 500ng DNA was bisulphite converted using the EZ DNA methylation kit (Zymo Research).
Pyrosequencing
PyroMark Q24 Advanced (Qiagen) was used to genotype all patient DNA samples as previously described 24 . Pyrosequencing was also used to quantify DNAm at six CpGs flanking rs75612460 following bisulphite conversion of DNA (EZ DNA Methylation Kit, Zymo). Each sample was amplified in duplicate. Samples were excluded from the analysis if the replicates differed by >5%.
Assays were designed using PyroMark assay design software 2.0 and primer sequences are listed in Table S5.
Lucia Reporter Assay
A 546bp region containing either the G or the A allele of rs75621460 was amplified and cloned into the pCpG-free-promoter-Lucia vector (Invivogen). Primer sequences are listed in Table S6.
Clones were transformed into competent GT115 cells (Invivogen) according to the manufacturer's protocol. Plasmids were methylated or mock-methylated in vitro using M. SssI (NEB). Successful methylation was determined by digest with methylation-sensitive SmaI (New England Biolabs).
Tc28a2 cells were transfected with 100ng of pCpG-free-promoter construct, along with 10ng of the pGL3-promoter vector (Promega) and luminescence measurements were made as described above.
Accepted Article
This article is protected by copyright. All rights reserved and expanded each time to 90% confluency. At each passage, cells were isolated for extraction of nucleic acids (Purelink, Thermo Fisher).
Statistical analyses
Genotype and methylation correlations were calculated using Kruskal-Wallis testing. For Lucia reporter assays we corrected for multiple comparisons using the methods of Holm-Sidak or Dunn, as specified in the figure legends. Changes in gene expression following Cas9 modulation were calculated using paired t-tests. AEI and DNAm relationships were determined using linear regression analysis. The exact details of all statistical tests is provided in the relevant figure legend. All tests were performed in GraphPad Prism 8.3.1.
Results
The region encompassing rs75621460 is a gene enhancer OA risk SNP rs756215460 is an intergenic variant at chromosome 19q12, positioned between CCDC97 and TGFB1 (Fig.1a). ChIP-seq data from mesenchymal stem cells (MSCs) and differentiated chondrocytes, along with ATAC-seq from OA knee chondrocytes indicate that the SNP resides within a chromatin-accessible region with post-translational histone modifications H3K27ac (yellow) and H3K27me3 (red), indicating that this region possesses regulatory function ( Fig.1a). TGFB1 expression was significantly (P<0.01) upregulated in OA hip cartilage. No significant change (P>0.05) in CCDC97 expression was observed (Fig.1b).
We cloned the 550bp accessible chromatin region into a luciferase reporter vector. The ancestral G-allele construct conferred a 2.7-fold increase in luciferase activity (Fig.1c). The derived A-allele (OA-risk) also demonstrated regulatory activity (1.6-fold), which was significantly lower (P<0.05) than that of G.
A multiple sequence alignment revealed that the G-allele is highly conserved in mammals ( Fig.1d). Within human populations, the A-allele emerged at a frequency >1% only in Europeans.
Differential allelic protein binding occurs at rs75621460
We used electrophoretic mobility shift assays (EMSAs) to characterise proteins binding to rs75621460. This revealed several complexes with a greater binding affinity to the G-allele than A
Accepted Article
This article is protected by copyright. All rights reserved ( Fig.2a, arrows 1 and 4). Furthermore, proteins were identified which exclusively bound to one of the two alleles ( Fig.2a arrows 2-3 and 5-6). Unlabelled probes were added to the reaction at increasing concentrations ( Fig.2b and c). The unlabelled A-probe was unable to strongly compete for binding for the higher molecular weight complexes bound to the G-probe (Fig.2b). However, some lower molecular weight complexes were outcompeted by increasing concentrations of the unlabelled A-probe, indicated by two arrows on Figure 2b. The unlabelled G-probe was able to compete for binding of all protein complexes to the labelled A-probe, with only one exception, indicated with an arrow on Figure 2c. The TRANSFAC database predicted four transcription factors which differentially bind to the alleles of rs75621460: SP1, MAZ, KLF17, and ETF ( Fig.2d). All four were predicted to bind exclusively to the G-allele. EMSA was performed using antibodies raised against the four proteins (Fig.2e). A supershifted band was observed in the presence of the SP1 antibody. This complex (indicated by an arrow) was bound to both alleles, however with a greater abundance at the G probe (Fig.2e). These combined EMSA results indicate that the G-allele binds proteins with greater affinity than the A-allele, and that distinct protein complexes bind to the region in chondrocytes, determined by the allele carried at rs75621460.
TGFB1 is the gene target of the rs75621460 enhancer
We deleted an 84bp region of the enhancer encompassing rs75621460 from the genome of Tc28a2 immortalised chondrocytes using CRISPR-Cas9 and a pair of gRNAs (gRNA 1 and 2) (Fig.3a).
No change in CCDC97 expression was measured (P=0.12) following deletion of the region ( Fig.3b). However, a significant decrease in TGFB1 expression was observed in Tc28a2-Δ84, in which mean expression was 0.48 of that measured in wild-type cells (P=0.003).
Methylation quantitative trait locus (mQTL) analysis of rs75621460
The deletion introduced in Tc28a2-Δ84 cells encompassed six CG-dinucleotides (CpGs), positions at which eukaryotic DNA can be methylated. This included a single upstream CpG (CpG1), and five downstream CpGs (CpG2-6) (Fig.4a). We investigated whether DNAm at these CpGs was modulated by SNP genotype. Due to the low MAF at rs75621460, we screened 206 human hip and knee cartilage samples to identify sufficient individuals carrying the A allele for analysis and identified 190 major allele homozygotes (GG), and 16 heterozygotes (GA).
We quantified cartilage DNAm at the 6 CpGs and stratified values by SNP genotype. All homozygous (GG) individuals investigated (n=93-101 across the six CpGs) were hypomethylated
In samples for which both DNA and RNA were available, we tested for correlations between DNAm and TGFB1 expression (methylation and expression QTLs, meQTLs). In cartilage, data were analysed together (n=31), and also by joint site (hip, n=14; knee, n=17).
Across both tissues, homozygous patient samples (GG) showed no significant meQTLs (P>0.09, Fig.4f). Conversely, very strong correlations were observed amongst the heterozygote samples. In cartilage, this was dependent upon the joint site from which cartilage was taken, with stronger meQTLs measured in knee (r 2 =0.47 -0.99) than in hip (r 2 =0.01 -0.65) (Fig. S2a). In knee cartilage and synovium, the strongest effect was observed for CpG1 (upstream of the SNP), where r 2 values were 0.99 and 0.90, respectively (Fig.4g). In both knee tissues, increasing DNAm at Correlations between DNAm and TGFB1 expression were observed at the five downstream CpGs (Fig.4g). In knee cartilage, a very strong meQTL (r 2 =0.92) operated at CpG2
Accepted Article
This article is protected by copyright. All rights reserved (Fig. S2a). Strikingly, the impact of rs75621460 upon DNAm at the downstream CpGs was paradoxical in the two distinct knee joint tissues. In cartilage, increasing DNAm correlated with increased TGFB1, whereas in synovium the opposite effect was seen (Fig.4f).
Heterozygous DNAm at CpG2 was stratified by DNAm at CpG1 (to identify correlations between CpGs upstream and downstream of the SNP) and at CpG3 (to identify correlations between CpGs downstream of the SNP). In synovium, positive correlations were observed between CpG1 and CpG2 (r 2 =0.84) and between CpG2 and CpG3 (r 2 =0.97) (Fig. S2b). However, in cartilage, correlations were observed at the downstream CpGs (r 2 =0.92-0.94), but not between CpG1 and CpG2 (r 2 =0.04-0.31), which are physically separated by rs75621460 (Fig. S2c). This validates the observations made in the meQTL analysis and suggests that in cartilage, distinct mechanisms regulate DNAm upstream and downstream of the SNP.
The detected meQTLs were the strongest in cartilage, the central tissue in OA pathogenesis. We therefore continued to use a chondrocyte model for subsequent downstream analyses.
DNA methylation in the enhancer attenuates activity
We next investigated whether DNAm at the CpGs flanking rs75621460 have a functional impact upon enhancer activity. The enhancer was cloned into a CpG-free reporter vector and expressed in Tc28a2 cells in either an unmethylated or methylated state. Methylation of the cloned region resulted in a significant reduction in enhancer activity in constructs containing both the G-(P=0.004) and A-allele (P=0.019) (Fig.5a), demonstrating that DNAm influences chondrocyte enhancer activity independently of rs75621460 genotype.
We repeated the EMSA, this time including probes that were methylated at CpG2, the sole CpG contained within the probe sequence. We compared nuclear protein binding to both alleles in the unmethylated or methylated state. The six bands of interest that were previously identified ( Fig.2a) are highlighted (Fig.5b). All of these protein complexes were able to bind to methylated probes (Fig.5b). Interestingly, for bands 1 and 4, methylation of the A-probe appeared to recover protein binding (Fig.5b).
We conducted a supershift assay using the methylated probes and also included an antibody for DNMT3a with the aim of detecting recruitment of a de novo DNA methylating enzyme by proteins bound to the A-allele. However, the only visible shift identified using this panel of antibodies was in the lanes containing anti-SP1, which was able to bind to both the
Accepted Article
This article is protected by copyright. All rights reserved unmethylated and methylated probes (Fig.5c and d). These EMSA data, along with our reporter assay data indicate that methylation of the region attenuates activity of the enhancer. However, DNAm at the single most proximal CpG to rs75621460 (CpG2) does not prevent the binding of proteins adjacent to the SNP, including SP1.
Modulation of the epigenome using DNMT3a-dCas9
Finally, we investigated whether DNAm flanking the SNP could functionally impact TGFB1 expression in the absence of the derived A-allele. We used a DNMT3a-dCas9 fusion protein for targeted editing of DNAm at the six CpGs in Tc28a2 cells, which are homozygous (GG) at rs75621460. Five gRNAs (gRNA3-7) were designed to target the region (Fig. S3a). DNMT3a-dCas9 was expressed alone (non-targeting control), or along with one of the five gRNAs, and DNAm was quantified over three cell passages (Fig. S3b). Four of the five guides (gRNAs4-7) successfully increased DNAm at one or more CpGs, an effect which was lost passively (Fig. S3b).
As gRNA3 did not modulate DNAm at any of the targeted six CpGs, it was not included in subsequent experiments.
Discussion
TGF-1 has a well-established role in OA pathophysiology, however this is the first study to identify an interplay between genetic and epigenetic regulation of TGFB1 expression in the context of disease risk. We have characterised an intergenic TGFB1 enhancer within the
Accepted Article
This article is protected by copyright. All rights reserved articulating joint, at which the alleles of an OA risk SNP impact upon DNAm and regulate TGFB1 in vivo.
We confirmed the SNP region as an in vitro enhancer at which the rs75621460 OA-risk Aallele reduces enhancer activity compared to the highly-conserved ancestral G-allele. The conservation of the G-allele amongst distinct human populations and throughout mammalian evolution illustrates the importance of the G for protein binding and enhancer function. EMSA analysis supported this data, showing that different alleles at rs75621460 could bind distinct proteins. The emergence of the A-allele in European populations implicates a selection advantage resulting from population-specific pressure, yet that this selection also simultaneously confers a detriment to cartilage health in older age, a phenomenon known as antagonistic pleiotropy.
Additionally, we identified that the transcription factor SP1, which has previously been shown to play a role in TGFB1 regulation 31 , binds to complexes at both alleles. Deletion of the region in chondrocytes confirmed TGFB1 as the enhancer gene target.
The absence of eQTLs in patient samples was perhaps unsurprising due to our modest sample size. Interindividual variability in gene expression often necessitates sample sizes involving hundreds of patients for the detection of significant genotype-expression correlations 32,33 . A complementary approach for eQTL analysis, which greatly increases sensitivity, involves measuring allelic imbalance between the expression of gene transcripts, and has been widely applied to investigations of OA risk loci 24,34-46 . We were unable to utilise this approach here due to the low MAF and the absence of a suitable TGFB1 transcript SNP.
However, we have demonstrated how the use of a secondary endophenotype, DNAm, can provide a more sensitive approach to investigate the impact of genotype upon gene expression within an individual.
We identified mQTLs at six CpGs in two tissues of the articulating joint, indicating that genetic and epigenetic interplay at the locus contributes to disease aetiology as observed at other OA risk loci 37 . The very small range of DNAm values over which correlations with TGFB1 expression occur potentially suggest the effects operate in a sub-population of chondrocytes within the tissue. In vitro methylation of the enhancer reduced the activity of both alleles in a reporter assay, and EMSA analysis indicated that DNAm at CpG2 impacted upon protein binding. The SP1 antibody bound to both methylated and unmethylated probes, consistent with previous reports into binding of this transcription factor 38,39 . We further identified that a targeted increase of DNAm at CpGs3-6 could reduce TGFB1 expression in the absence of the rs75621460 A-allele.
Accepted Article
This article is protected by copyright. All rights reserved It has previously been documented that regulatory SNPs can confer tissue-specific effects upon genes, resulting in biological pleiotropy 32,40 . At this locus, the directly opposing effects in cartilage and synovium are the result of a shared effect of a single variant rather than the colocalisation of two distinct effects 40 . This emphasises that whilst integration of epigenetic data is a useful post-GWAS tool 14,41 , functional analyses in appropriate disease models are imperative to elucidate tissue-specific pathological mechanisms.
We propose a molecular mechanism of TGFB1 regulation in cartilage, as follows (Fig. S4).
Substitution of the highly-conserved G-allele at rs75621460 alters the consensus sequence for protein binding. In the presence of the G-allele, a protein complex with strong transcriptional activity binds to the sequence (Fig. S4a). This complex does not modulate DNAm, hence there is We speculate that in synovium, where a paradoxical correlation was observed, tissuespecific proteins which have a repressive effect upon TGFB1 could bind to the A-allele (Fig. S4c).
In both tissues, the OA risk A-allele results in attenuated enhancer activity, and decreased TGFB1 expression.
Elucidating the mechanism of TGFB1 expression in synovium was not within the scope of this study and requires further investigation. Furthermore, the impact of the SNP upon downstream TGF-β signalling remains unknown. The SNP resides within a region of open chromatin in fibroblast-like synoviocytes 42 . This knowledge, along with our data, indicates that the region is also utilised to regulate TGFB1 in synovium. The TGFB1 enhancer is an interesting focus for future studies, especially in the context of inflammatory joint diseases such as rheumatoid arthritis. Additionally, the use of single-cell technologies could identify subpopulations of cells within joint tissues in which these mechanisms operate. Furthermore,
Accepted Article
This article is protected by copyright. All rights reserved novel techniques for targeted subnuclear proteomics profiling provide a promising tool to identify the exact proteins modulating TGFB1 expression in distinct tissue types 43 .
TGF-β has been well-studied in the context of OA pathophysiology 44,45 . In healthy cartilage, TGF-β acts as an anabolic factor to stimulate the synthesis of ECM proteins, conveying a chondroprotective effect against mechanical loading in a healthy joint 46
Accepted Article
This article is protected by copyright. All rights reserved
Accepted Article
This article is protected by copyright. All rights reserved (D) Analysis of differential transcription factor binding to the G and A-allele at rs75621460 using the transcription factor database, TRANSFAC.
(E) Supershift experiment with antibodies targeting SP1, MAZ, KLF17, and ETF compared to no antibody (control) to the EMSA reaction containing the G or A-allele probe. The arrow indicates the position of supershifted complexes. (C) Labelled probes containing rs75621460 G-allele with CpG2 unmethylated or methylated were incubated with antibodies raised against SP1, MAZ, KLF17, ETF, or DNMT3a (all 2μg) and Tc28a2 nuclear protein extract. The arrow indicates the SP1-supershifted complex.
(D) Labelled probes containing rs75621460 A-allele with CpG2 unmethylated or methylated were incubated with antibodies raised against SP1, MAZ, KLF17, ETF, or DNMT3a (all 2μg) and Tc28a2 nuclear protein extract. The arrow indicates the SP1-supershifted complex. (B) Tc28a2 percentage methylation levels at the 6 CpGs surrounding the SNP in no guide controls (black, dashed line) or using a gRNA targeting the region (colours as described above, solid line) (n=6 biological replicates, each with two technical repeats). Combinations of gRNAs were also used: gRNA 4+6 and gRNA 5+7 (light blue, solid line). Subsequent changes in TGFB1 expression were also measured following targeted editing of methylation, and gene expression was normalised to that in no-guide controls (n=6 biological repeats, each with 3 technical repeats). P values were calculated using paired t-tests following testing of control values for normality (D'Agostino and Pearson, P=0.37).
==
Domain: Biology Medicine
|
[{"added": "2021-06-04T01:26:53.218Z", "created": "2021-01-01T00:00:00.000", "id": "232338246", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1002/art.41736", "pdf_hash": "c380b28fa8a46efaf629caf06997586d7d45ea6c", "pdf_src": "ScienceParsePlus", "provenance": "peS2o-0028.json.gz:5105", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "69576b8a93e08d3fdc8a1216a379771f99622476", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Bmi-1-induced miR-27a and miR-155 promote tumor metastasis and chemoresistance by targeting RKIP in gastric cancer
We previously reported an inverse relationship between B cell-specific Moloney murine leukemia virus integration site 1 (Bmi-1) and Raf kinase inhibitory protein (RKIP), which is associated with the prognosis of gastric cancer (GC). In this study, we further explored the microRNA (miRNA) regulatory mechanism between Bmi-1 and RKIP. Microarray analysis was first carried out to identify miRNA profiles that were differentially expressed in cells overexpressing Bmi-1. Then, miRNAs that could regulate RKIP were identified. Quantitative real-time PCR (qRT-PCR) and Western blotting were performed to measure the expression of Bmi-1, miR-155, miR-27a and RKIP. RKIP was confirmed as a target of miR-27a and miR-155 through luciferase reporter assays, qRT-PCR and Western blotting. The effects of the Bmi-1/miR-27a/RKIP and Bmi-1/miR-155/RKIP axes on tumor growth, proliferation, migration, invasion, colony-formation ability, metastasis and chemoresistance were investigated both in vitro and in vivo. The downregulation of RKIP by Bmi-1 occurred at the protein but not mRNA level. This indicates probable posttranscriptional regulation. miRNA expression profiles of cells with ectopic expression of Bmi-1 were analyzed and compared to those of control cells by microarray analysis. A total of 51 upregulated and 72 downregulated miRNAs were identified. Based on publicly available algorithms, miR-27a and miR-155 were predicted, selected and demonstrated to target RKIP. Bmi-1, miR-27a and miR-155 are elevated in human GC and associated with poor prognosis of GC, while RKIP is expressed at lower levels in GC and correlated with good prognosis. Then, in vitro tests shown that in addition to regulating RKIP expression via miR-27a and miR-155, Bmi-1 was also able to regulate the migration, invasion, proliferation, colony-formation ability and chemosensitivity of GC cells through the same pathway. Finally, the in vivo test showed similar results, whereby the knockdown of the Bmi-1 gene led to the inhibition of tumor growth, metastasis and chemoresistance through miR-27a and miR-155. Bmi-1 was proven to induce the expression of miR-27a and miR-155 and thus promote tumor metastasis and chemoresistance by targeting RKIP in GC. Overall, miR-27a and miR-155 might be promising targets for the screening, diagnosis, prognosis, treatment and disease monitoring of GC.
Background
Gastric cancer (GC) is one of the leading causes of cancer morbidity and mortality worldwide, especially in Eastern Asian and Eastern European countries [1][2][3]. Patients in advanced stages benefit less than expected from palliative chemotherapies, mainly due to tumor metastasis and chemoresistance [4,5], although the underlying molecular mechanism remains largely unknown. In our previous studies, the polycomb-group protein B cellspecific Moloney murine leukemia virus integration site 1 (Bmi-1) was reported to be associated with tumor size, clinical stage and prognosis of GC [6,7]. Furthermore, we demonstrated that Bmi-1 is inversely associated with Raf kinase inhibitory protein (RKIP), a regulator of apoptosis induced by chemotherapeutic agents and a clinically relevant cancer metastasis suppressor gene [8,9]. Based on previously reported results, the negative correlation between Bmi-1 and RKIP was deemed valuable in predicting patient survival and therapeutic response in GC [10]. The inverse expression pattern of Bmi-1 and RKIP was confirmed in GC cell lines by utilizing in vitro gene overexpression and silencing methods, suggesting the likelihood of RKIP being regulated by Bmi-1. However, the regulatory mechanism between Bmi-1 and RKIP needs further elucidation.
A variety of factors may be involved in the mechanism regulating the expression of Bmi-1 and RKIP. Among all potential regulatory mechanisms, microRNAs (miRNAs) emerged as a top research target due to their significance in various biological activities. miRNAs are an endogenous group of small noncoding RNAs that regulate up to 60% of human protein-coding genes at the posttranscriptional level by binding to the 3′ untranslated region (3'UTR) of a target mRNA [11][12][13]. miRNAs are able to affect tumor cell proliferation, invasion, metastasis and chemoresistance by regulating gene expression. For instance, Bmi-1 upregulates miR-21 and miR-34a in addition to regulating GC stem cell-like properties via the activation of the AKT-NF-κB pathway [14]. RKIP is also reported to be directly regulated by miR-543, which controls cell proliferation and metastasis in human prostate cancer cells [15]. Given the importance of miRNAs, the dysregulation of miRNAs is thought to be closely related to metastasis and chemoresistance in GC. We hypothesized that miRNAs may account for the negative correlation between Bmi-1 and RKIP. Therefore, this study aimed to explore Bmi-1-induced miRNAs that regulate RKIP. The candidate miRNAs were identified and predicted and were then validated by in vitro and in vivo experiments.
Overall, a novel microRNA regulatory mechanism of the Bmi-1-RKIP signaling axis was found. The inhibition of the RKIP tumor suppressor and a mechanism underlying tumor metastasis and resistance to chemotherapy in GC were also elucidated in this study. These findings may be significant in terms of screening, diagnosis, prognosis, disease monitoring and therapeutic value in GC.
Tissue specimens
Fifteen pathologically confirmed GC specimens with tumor and adjacent normal paired fresh-frozen tissues were obtained at Sun Yat-sen memorial hospital, Sun Yat-sen University (Guangdong, China). No preoperative treatments were received. Total RNA and protein of frozen tissues were extracted for quantitative real-time PCR (qRT-PCR) and Western blotting assays. The clinical specimens were obtained with patients' informed consent and permission from the Institutional Ethics Committee of Sun Yat-sen memorial hospital. RNA-seq data, clinicopathological information and follow-up data of GC were downloaded from The Cancer Genome Atlas (TCGA) database ( [URL]). Patients with censored overall survival time, gene expression profiles or certain pathological classification were excluded. The cut-off values of Bmi-1, miR-27a-3p (miR-27a), miR-155-5p (miR-155) and RKIP expression in Kaplan-Meier analysis were determined by Xtile software [16,17].
Cell lines
The human GC cell lines (BGC823 and SGC7901) were obtained from the Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China). The human gastric epithelial immortalized GES-1 cell line was purchased from Beijing Institute for Cancer Research. The human leukemia monocytic U937 cell line was purchased from American Type Culture Collection (ATCC, Washington, USA). All these cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM, Gibco, MA, USA) supplemented with 10% fetal bovine serum (FBS, Biological Industries, Beit Haemek, Israel) and incubated at 37°C in a humidified atmosphere of 5% CO 2 .
Plasmids, siRNAs and stable cell lines pLNCX2-Bmi-1 was constructed as previously described [7]. Stable cells overexpressing Bmi-1, SGC7901-Bmi-1, BGC823-Bmi-1, GES-1-Bmi-1#1 and GES-1-Bmi-1#2 were generated by retroviral transfection as described in our previous study [7,10]. The Bmi-1 overexpressing cells were also named Vector-Bmi-1, and the negative control was named Vector-Ctrl. Lentiviral shRNA virus specific to human Bmi-1 and Bmi-1-specific siRNAs were purchased from GenePharma (Shanghai, China). The shRNA sequences (the same as the specific siRNAs) for targeting Bmi-1 are shown in Supplementary Table S1. To construct cell lines for constitutive miRNA expression, lentiviruses containing GFP-miR-27a, GFP-miR-155, or GFP-negative control miRNA vector were purchased from GenePharma, Inc. The lentiviral vectors were used to infect targeted cells. BGC823 and SGC7901 cells were pre-seeded in a 6-well plate overnight and infected with 10 μl of virus. Infected cells were selected by adding 400 ng/ml puromycin for 5 days and then transferred to cultured flasks for proliferation. Stable cell lines were verified by qRT-PCR.
Microarray data processing
The Bmi-1-overexpressing GES-1 cell line GES-1-Bmi-1#1 and its vector control cells were harvested with TRIzol, and miRNAs were extracted using a miRNeasy mini kit (QIAGEN, Beijing, China). Then, samples were sent to KangChen Bio-tech Inc. (Shanghai, China), quantified and analyzed the differential expression of miRNAs using a miRNA microarray. Microarray hybridization, data generation, and normalization were performed following standard protocols. Finally, differentially expressed miRNAs were identified through fold-change filtering. The cut-off fold change was 2.0.
Luciferase assay
The RKIP 3'UTR was amplified by PCR separately using the primers described in Supplementary Table S1 from cDNA of SGC7901 cells. The PCR product was ligated into the multiple cloning region of the pGL3 luciferase reporter plasmid (Promega, Wisconsin, USA) according to the manufacturer's recommendations. 293T cells plated in 96-well plates at a density of 4 × 10 4 cells per well were cotransfected with 100 ng of the constructed luciferase plasmid or the control luciferase plasmid and 15 ng of the pRL-TK Renilla plasmid (Promega) using the Lipofectamine 2000 reagent (Thermo Fisher Scientific, MA, USA). miR-27a mimic/mut and miR-155 mimic/mut (50 nmol/L) were then cotransfected with the luciferase plasmid containing the RKIP 3'UTR for microRNA detection. After 24 h, cells were lysed and detected for Renilla and firefly luciferase activity using the Dual Luciferase Reporter Assay Kit (Promega). Three independent cotransfection experiments were carried out in triplicate.
Animal experiments
All mouse experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals by the National Institutes of Health and were approved by the Animal Care and Use Committee of Sun Yat-sen University. To study the effect of the Bmi-1/miR-27a/RKIP and Bmi-1/miR-155/RKIP axes on GC tumorigenesis, chemoresistance and metastasis, we conducted tumor xenograft and tail vein injection studies in mice.
For the tumorigenesis experiments, BGC823-shcon, BGC823-shBmi-1, BGC823-NC mimic, BGC823-miR-27a mimic, BGC823-miR-155 mimic, BGC823-shBmi-1 + NC mimic, BGC823-shBmi-1 + miR-27a mimic and BGC823-shBmi-1 + miR-155 mimic cells (1.5 × 10 6 cells in 0.15 ml of PBS) were subcutaneously inoculated into the right flanks of female BALB/c nude mice (5 weeks old) to establish tumor xenografts. Tumor size was monitored and calculated every 3 days. The tumor volume was estimated using the equation: tumor volume (mm 3 ) = (length in mm) × (width in mm) 2 /2. In the chemoresistance experiments, tumor-bearing xenografts were established the same as in the tumorigenesis experiment. Once the subcutaneous tumors grew big enough to be palpable, intraperitoneal (IP) injection of either 5-Fu (15 mg/kg) or vehicle control (PBS) was performed every other day for 14 days. On the 21st day after implantation, mice in the tumorigenesis and chemoresistance groups were sacrificed. For the tumor metastasis experiment, 1 × 10 6 cells were injected into mice via the tail vein. The mice were sacrificed 30 days after cell administration. The tumors, lungs or livers dissected in all animal experiments were fixed and paraffin-embedded for histopathological analysis.
Immunohistochemistry
Paraffin-embedded tissue sections that were 4 μm thick were used in immunohistochemistry. Following deparaffinization, hydration, antigen retrieval and blocking, the slides were successively incubated with primary antibodies at 4°C overnight and secondary antibodies conjugated with horseradish peroxidase (HRP) at 37°C for 30 min. The slides were then incubated in DAB solution for 10 min and counterstained with hematoxylin for 1 min. Images of the stained slides were obtained with a microscope (Nikon, Tokyo, Japan).
Statistical analysis
All statistical analyses were performed with Statistical Product and Service Solutions version 22.0 (SPSS, Illinois, USA) and GraphPad Prism 5.0 software (GraphPad, Inc., CA, USA). Data are presented as the mean ± standard error ( − x ± SE). The statistical significance of the differences was determined by t-test and one-way ANOVA test according to the homogeneity of variances. P < 0.05 was considered to be statistically significant.
The detailed methodology of qRT-PCR, Western bloting and functional experiments can be found in the Supplementary Materials.
Results microRNAs induced by Bmi-1 that regulate RKIP expression were screened and selected miRNAs have been confirmed to function in disease pathogenesis by regulating the expression of genes mainly in posttranscriptional gene silencing. To determine whether miRNAs are involved in Bmi-1 repression of RKIP, we compared the miRNA expression profiles of cells with ectopic expression of Bmi-1 (GES-1-Bmi-1) with the control cells (GES-1). Microarray analysis identified 51 upregulated and 72 downregulated miRNAs in GES-1-Bmi-1 cells ( Fig. 1a and Supplementary Table S2), including miR-27a, miR-155, and miR-516, which were upregulated and had been previously found to be associated with gastric adenocarcinoma or other gastrointestinal tumor invasiveness. Among all the differentially upregulated miRNAs, miR-27a, miR-761, miR-155, miR-4725, miR-4784, miR-4329, miR-4765 and miR-516b were predicted to target RKIP based on analysis with publicly available algorithms (Fig. 1b). miR-27a and miR-155 were selected and proven to be upregulated in GES-Bmi-1 cells by qRT-PCR (Fig. 1c). As shown in Fig. 1d, miR-27a and miR-155 were downregulated when the Bmi-1 gene was knocked down in BGC823 and SGC7901 GC cell lines.
The clinical significance of Bmi-1, miR-27a, miR-155 and RKIP The association between Bmi-1, RKIP and the clinical outcome of gastric cancer has been reported in our previous study [10]. To further explore the clinical significance of Bmi-1, miR-27a, miR-155 and RKIP in GC, we first analyzed their RNA expression levels by qRT-PCR. The results demonstrated higher Bmi-1, miR-27a and miR-155 expression and lower RKIP expression in GC tissues ( Fig. 1e and Fig. S1A). Furthermore, Western blotting examination revealed that 93.33% (14/15) of patients with GC had an upregulated protein level of Bmi-1, and 73.33% (11/15) had a downregulated protein level of RKIP (Fig. S1B). The inverse association between Bmi-1 and RKIP is consistent with our previous research results. Moreover, we analyzed the RNA expression levels of Bmi-1, miR-27a, miR-155 and RKIP in human GC samples from TCGA database using StarBase ( [URL]) [20]. As shown in Fig. S1D, Bmi-1, miR-27a, and miR-155 were significantly upregulated, while RKIP was downregulated in GC, which is consistent with Fig. 1e and Fig. S1A. We divided the patients with available histological information from TCGA into two groups by Lauren classification (intestinal-type and diffuse-type) [21]. Tumors exhibiting signet ring type were classified as diffuse. Patients with high expression of Bmi-1, miR-27a, and miR-155 and low expression of RKIP had a shorter 3-year overall survival in intestinal-type GC. Moreover, there was no significant correlation between the above four indicators and the 3-year overall survival in diffuse-type GC ( Fig. 1f and Fig. S1C). The above results suggest that these four indicators may have different clinical significance in different pathological classifications of GC. Taken together, our results reveal that Bmi-1, miR-27a and miR-155 are elevated in human GC and associated with poor prognosis of GC, while RKIP is expressed at lower levels in GC and correlated with good prognosis.
miR-27a and miR-155 regulate RKIP directly
As shown in Fig. 2a and b, miR-27a and miR-155 significantly decreased the luciferase activity of the constructed plasmid containing the 3'UTR of RKIP. In addition, RNA EMSAs were performed to demonstrate the direct interaction between miR-27a and its cognate RKIP target sequence, but such interaction was not observed in miR-155. As shown in Fig. 2c, miR-27a and its cognate RKIP mRNA oligonucleotide could bind together to form an electrophoretically stable miRNA/mRNA complex.
However, miR-155 displayed two bands at different migration rates by PAGE, indicating that it did not bind with the oligonucleotide of RKIP mRNA to form a stable complex (Fig. 2d). Furthermore, the miR-27a/RKIP mRNA complex was able to interact with SGC7901 cytoplasmic extracts to form new complexes (Fig. 2e, lane 1). To address the role of miR-27a and miR-155 in RKIP expression, Western blot analysis and corresponding densitometry analysis were performed in BGC823 and SGC7901 GC cell lines after transfection with either miR-27a mimic/inhibitor or miR-155 mimic/inhibitor. The results showed that RKIP protein expression was reduced in the miR-27a and miR-155 mimic groups but increased in the miR-27a and miR-155 inhibitor groups compared to the control groups ( Fig. 2f and Fig. S3A). However, the RNA expression of RKIP did not change after transfection with the miRNAs (Fig. 2g).
Bmi-1 regulates RKIP expression through miR-27a and miR-155 in GC cells
To confirm that the regulatory effects of Bmi-1 on RKIP are mediated through miR-27a and miR-155, we introduced miR-27a inhibitor and miR-155 inhibitor to GC cells with ectopic expression of Bmi-1 and miR-27a mimic and miR-155 mimic to GC cells with Bmi-1 gene knockdown. Although the protein expression of RKIP in BGC823 and SGC7901 cells decreased considerably after transfection with Bmi-1, both the miR-27a and miR-155 inhibitors were able to restore RKIP expression. In contrast, the protein expression of RKIP in BGC823 and The predicted binding site on the RKIP 3'UTR for miR-155 and miR-27a and the mutant sequence. b The relative luciferase activity of 293 T cells cotransfected with miRNAs and the indicated luciferase plasmid. *P < 0.05 vs. NC mimic, **P < 0.01 vs. NC mimic. c EMSA showed the formation of a stable dye-miR-27a/RKIP-Target complex, which was inhibited by the competitive interaction between excessive amounts of cold-miR-27a (unlabeled miR-27a) and RKIP-Target. d EMSA demonstrated no interaction between miR-155 and RKIP-Target. e The formation of a miR-27a/RKIP-Target protein complex by adding a cell cytoplasmic extract. f Western blot analysis showed the expression of RKIP and EMT-related marker genes after treatment with miR-155 mimic/inhibitor or miR-27a mimic/inhibitor. g Gene expression of RKIP and EMT-related marker genes modulated by miR-155 mimic/inhibitor or miR-27a mimic/inhibitor. *P < 0.05 vs. NC mimic, # P < 0.05 vs. NC inhibitor SGC7901 cells substantially increased after transient transfection with siBmi-1, and the miR-27a mimic and miR-155 mimic repressed RKIP expression ( Fig. 3a and Fig. S3B). Besides, the RNA expression of RKIP did not change after transfection with the miRNAs or siBmi-1 (Fig. 3b). Altogether, these findings suggest that RKIP expression is negatively regulated by Bmi-1 through miR-27a and miR-155 in GC cell lines.
The migratory and invasive behaviors of BGC823 and SGC7901 cells were detected using Transwell assays. We found that Bmi-1 gene knockdown reduced the number of migrating and invading cells compared to the mock control, whereas cotransfection of miR-27a or miR-155 increased the number of such cells ( Fig. 4a and Fig. S3C).
The MTS experiments indicated that the overexpression of miR-27a and miR-155 led to an increase in cell proliferation, while the downregulation of Bmi-1 inhibited cell proliferation, which could be counteracted by cotransfection of miR-27a or miR-155 (Fig. 4b). Colony formation assays both in soft agar and culture plates demonstrated that compared to the negative control, fewer colonies developed when Bmi-1 was inhibited. However, cotransfection of miR-27a or miR-155 reversed the reduced colony-formation abilities (Fig. 4c).
Chemotherapy sensitivity of the tumor cells was then assessed using the MTS assay, and it was observed that Bmi-1-knockdown cells displayed less chemoresistance than control cells when treated with two conventional chemotherapeutic agents (5-Fu and oxaliplatin). Moreover, to determine whether Bmi-1 plays an important role in apoptosis via miR-27a or miR-155 during drug treatment, the miR-27a mimic and miR-155 mimic were also transfected into shBmi-1 cells. We found that the miR-27a mimic and miR-155 mimic significantly The results of qRT-PCR were in accordance with the Western blotting assays. *P < 0.05 vs. Vector-Ctrl/siNC attenuated the impact of Bmi-1 silencing on druginduced apoptosis (Fig. 4d). Moreover, the miR-27a inhibitor and miR-155 inhibitor could also weaken the effects of Bmi-1 overexpression on migration, invasion and chemoresistance in gastric cancer cells (Fig. S4 and Fig. S3C). Therefore, it can be deduced that Bmi-1 partly regulates chemosensitivity via a microRNA-dependent mechanism.
For the indicated time after subcutaneous injection, an obvious attenuation in tumor growth was observed in mice that were injected with shBmi-1, whereas mice that were injected with the miR-155 mimic or miR-27a mimic showed the opposite effect and had increased tumor growth (Fig. 5a-c). This was also confirmed by the immunohistochemical results (Fig. 5d). As expected, we observed that the BGC823-shBmi-1 group had fewer lung and liver metastases than the controls. We also noted that the BGC823-miR-27a mimic and BGC823-miR-155 mimic groups exhibited a significant increase in the number of metastatic foci in both the lungs and livers. Moreover, a dramatic upregulation was found in the number of foci in the lungs and livers of mice injected with BGC823-shBmi-1 + miR-155 mimic or BGC823-shBmi-1 + miR-27a mimic cells relative to BGC823-shBmi-1 (Fig. 6a-c).
To assess the chemosensitivity of tumor cells, we established subcutaneous xenografts and then evaluated their response to 5-Fu treatment. After 14 days of treatment, the mice were sacrificed, and their (Fig. 6d-f). However, shBmi-1 displayed a significantly stronger effect in reducing tumor growth and increasing the sensitivity towards 5-Fu compared to the control. Notably, tumors in the groups cotransfected with shBmi-1 + miR-155 mimic or shBmi-1 + miR-27a mimic grew much faster than tumors in the shBmi-1 group, indicating the ability of the miR-155 mimic and miR-27a mimic to decrease the chemosensitivity of BGC823 cells to 5-Fu (Fig. 6g-i). Overall, the effect of shBmi-1 could be attenuated by reducing RKIP expression through the action of the miR-27a mimic or miR-155 mimic. Additionally, the results of immunohistochemical staining of Bmi-1, RKIP, Vimentin, BCL2 Associated X, Apoptosis Regulator (Bax) and B-cell lymphoma-2 (Bcl-2) detected in isolated tumors (Fig. S5A and B) were in accordance with the results of the Western blot assay. Moreover, the negative correlation between the expression of Bmi-1 and RKIP was further proven by microscopic observation of tumors derived from the shcon and shBmi-1 xenograft models, which provided further evidence to support our hypothesis. Therefore, based on these indications from correlation studies, it can be concluded that the Bmi-1/miR-27a/RKIP and Bmi-1/miR-155/RKIP axes are major determinants of cell growth, metastasis and resistance to chemotherapy in GC.
Discussion
Recently, therapies targeting metastasis and chemoresistance of GC have attracted the attention of researchers [22,23]. However, insufficient understanding of the underlying molecular mechanisms involved creates a barrier for the development of effective strategies in clinical practice. We previously reported the distinctive biological activity of Bmi-1/RKIP and their inverse relationship in GC, but the detailed mechanisms needed further investigation. In this study, we screened for Bmi-1-induced miRNAs and successfully identified the Bmi-1/miR-27a/RKIP and Bmi-1/miR-155/RKIP signaling axes. For the first time, we elucidated the molecular mechanism of how Bmi-1 regulates RKIP in GC, which ultimately affects the metastasis and chemoresistance of GC. Our results suggest that therapies targeting the Bmi-1/miR-27a/RKIP and Bmi-1/miR-155/RKIP signaling axes would be effective in GC patients, especially those with a high risk of metastasis and chemoresistance.
RKIP is a proven metastasis suppressor protein in various human cancer types and is linked to an invasive phenotype [24,25]. Low levels or the absence of RKIP expression in GC is associated with poor patient prognosis [10,26]. However, the molecular mechanisms of how RKIP expression is downregulated in GC remain unknown. Polycomb group (PcG) proteins are epigenetic gene-silencing proteins that have been shown to play important roles in human cancer occurrence and progression [27,28]. Bmi-1, the first functionally identified PcG member, is frequently dysregulated in various cancers and strongly correlates with tumor aggressiveness; thus, its presence predicts a poor prognosis [10,29,30], but little attention has been paid to the downstream regulatory mechanism of Bmi-1 in GC promotion. The opposite biological functions and negative correlation between Bmi-1 and RKIP suggested a potential regulatory mechanism in GC. The results from our previous study led to the likely hypothesis that RKIP is regulated by Bmi-1 [10].
To directly examine the role of Bmi-1 in RKIP expression, we introduced a Bmi-1 expression plasmid by retroviral infection into the nonmalignant human gastric epithelial GES-1 cell line. RKIP protein expression was suppressed by ectopic expression of Bmi-1 in GES-1 cells [10]. However, forced expression of Bmi-1 barely changed the mRNA level of RKIP (Fig. S2A). Therefore, RKIP protein expression is likely to be regulated at the posttranscriptional level in GC cells. This result is in line with previous reports that have suggested the critical role of posttranscriptional mechanisms in the regulation of RKIP expression in human hepatocellular carcinoma [31]. Mechanisms such as protein degradation and microRNA-mediated suppression may also play a role in the posttranscriptional regulation of the expression of specific genes. Moreover, protein half-life analysis demonstrated that forced expression of Bmi-1 did not induce RKIP protein degradation (Fig. S2B). On the other hand, microRNAs are RNAs with sizes between 20 and 25 nucleotides, and they contribute to gastric carcinogenesis by regulating the expression of oncogenes and tumor suppressors at the posttranscriptional level to affect cell proliferation, apoptosis, motility and invasion [32][33][34]. microRNAs have been shown to be involved in regulating RKIP expression in human cancer [24,35]. To our knowledge, this study is the first to report a difference in miRNA expression based on the overexpression of Bmi-1 in GES-1 cells. Potential miRNAs targeting RKIP were selected using miRNA microarray and publicly available algorithms (TargetScan). Upon combining the outcomes of the literature review, miR-27a and miR-155 were identified as candidate microRNAs that are involved in the Bmi-1 repression of RKIP.
In this study, we demonstrated that Bmi-1 functions as a repressor of RKIP, which is mediated by miR-27a and miR-155. miR-27a and miR-155 were also able to directly repress RKIP expression, which consequently affected the expression of genes necessary for regulating metastasis and chemoresistance, including Vimentin, Cadherin 1 (E-cadherin), Bax and Bcl-2. Through in vitro and in vivo experiments, it was observed that Bmi-1-downregulated cells showed slower growth, poorer migration and invasion capacities, and a clear sensitivity to in vitro drug treatment. Further in vivo experiments in mice demonstrated that miR-27a and miR-155 downregulated RKIP and promoted metastasis to the lungs and livers. The level of inhibition to oxaliplatin and 5-Fu sensitivity was also greater than that of the control. Furthermore, the miR-27a mimic and miR-155 mimic could significantly counteract the metastasis and chemoresistance effects in Bmi-1-downregulated cells. Similarly, the miR-27a inhibitor and miR-155 inhibitor weakened the effects of Bmi-1 overexpression on migration, invasion and drug resistance in gastric cancer cells. The results were consistent with previous results that highlight the significant role of RKIP in altering tumor cell metastasis and chemoresistance both in vitro and in vivo. In addition, we observed differences in miR-27a mimic-and miR-155 mimic-induced cell proliferation and the size of the tumors when compared to control cells. This implies that there may be other pathways influencing cancer growth simultaneously. Collectively, these findings strongly indicate the pivotal role of RKIP in Bmi-1-mediated promotion of metastasis and suppression of chemosensitivity.
Some researchers have reported that miR-27a is highly expressed in GC tissues and that the overexpression of miR-27a promotes the tumorigenicity, metastasis and chemoresistance of GC [36][37][38], which is in accordance with this study. In addition, previous studies illustrated that increased expression of miR-155 was closely related to tumor invasion and metastasis in advanced GC, implying poor prognosis [39]. Qu Y et al. showed that miR-155 downregulated the expression of TGFβR2 to promote the proliferation and migration of GC cells, which is consistent with our study [40]. Moreover, it was reported that Helicobacter pylori infection was closely linked to miR-155, possibly upregulating miR-155 expression to inhibit the DNA mismatch repair (MMR) gene and induce a mutant phenotype that is conducive to error-prone translation synthesis and thus promotes GC progression [41][42][43]. However, some studies have reported that miR-155 is expressed at low levels in GC tissue and acts as a tumor suppressor gene [44][45][46]. Combined with the prognostic analysis of miR-155 shown in Fig. 1f, we propose that miR-155 plays different roles in GC of different histological types. Therefore, the occurrence and development of GC is complex. Although the carcinogenic role of miR-27a and miR-155 in GC has been reported, our study demonstrates that miRNA as a key junction plays a posttranscriptional regulatory role in the Bmi-1/RKIP pathway, further revealing the specific molecular mechanism of GC metastasis and chemoresistance.
Previous published literature illustrates that GC is histologically complex and can be characterized by the expression profile of microRNAs. It was reported that miR-105, miR-145, and miR-133a were upregulated in diffuse-type lesions, while miR-498 and miR-494 were upregulated in intestinal-type GC [47,48]. We analyzed the clinical significance of miR-27a and miR-155 from TCGA and found that these two indicators were not identical in different histological types, suggesting that these two indicators could be signatures linked to the tumorigenesis and development of GC. Therefore, we need to include a larger patient population and collect follow-up information to clarify the correlation between miR-27a, miR-155 and clinical prognosis in further studies. Moreover, we will verify the expression of miR-27a and miR-155 and its clinical significance in different histological types.
Conclusions
In conclusion, the present study indicates that Bmi-1 negatively regulates the metastasis suppressor gene RKIP via microRNA-mediated posttranscriptional mechanisms in human GC. Bmi-1-induced miR-27a and miR-155 were candidate microRNAs identified by microarray analysis and were verified to regulate RKIP. Furthermore, the Bmi-1/miR-27a/RKIP and Bmi-1/miR-155/RKIP signaling axes might be potent targets for novel therapeutic approaches against human GC due to their demonstrated roles in tumor metastasis and drug resistance. Future studies should focus on these aspects.
Additional file 1: Fig. S1 The association between clinical data and Bmi-1 and RKIP. A. qRT-PCR analysis of Bmi-1 and RKIP RNA expression in 15 paired GC tissues (T) and adjacent normal tissue samples (N). B. Western blotting analysis of Bmi-1 and RKIP in 15 paired GC tissues. The definitions of T and N were the same as mentioned in A. C. Kaplan-Meier analysis of the 3-year overall survival of patients with intestinal-type or diffuse-type GC from TCGA. D. Bmi-1, miR-27a and miR-155 were upregulated, while RKIP was downregulated significantly in GC tissues from the TCGA database. *P < 0.05, **P < 0.01. Fig. S2 Bmi-1 does not upregulate RKIP at the mRNA level nor induce RKIP protein degradation. A. Bmi-1 and RKIP mRNA expression in GES-1 cells overexpressing Bmi-1. *P < 0.05 vs. GES-1-Vector. B. GES-1-Bmi-1 cells and GES-1-Vector cells were subjected to the protein synthesis inhibitor cycloheximide for the indicated period of time. The half-life of RKIP protein in Bmi-1-transduced cells was comparable to that in the control cells, which indicated that Bmi-1 did not induce RKIP protein degradation. Fig. S3 Quantification of Western blotting assays as well as invasion and migration assays. A. The densitometry analysis of bands from the Western blotting assays in Fig. 2f. * P < 0.05 vs. NC mimic/NC inhibitor. B. The densitometry analysis of bands from the Western blotting assays in Fig. 3a. * P < 0.05 vs. Vector-Ctrl/siNC. C. Analysis of the quantities of invading cells in migration and invasion assays. *P < 0.05 vs. shcon, **P < 0.01 vs. shcon/Vector-Ctrl, ## P < 0.01 vs. NC mimic/NC inhibitor. Fig. S4 miR-27a inhibitor and miR-155 inhibitor weakened the effects of Bmi-1 overexpression in functional experiments. A. Bmi-1 upregulation induced gastric cancer cell migration and invasion, which were decreased by the miR-155 inhibitor or miR-27a inhibitor (100 × magnification). B. The reduced ability of cell proliferation due to the transient transfection of the miR-155 inhibitor or miR-27a inhibitor was improved by Bmi-1 overexpression. C. Colony formation assays either in soft agar or on plates showed that the Bmi-1 overexpression group generated more colonies than any other group, and the effect could be reversed by miR-155 inhibitor or miR-27a inhibitor. D. The IC 50 values of cells treated with 5-Fu or oxaliplatin were detected by CCK8 reagent. The increase in Bmi-1 reduced chemosensitivity, while the miR-155 inhibitor and miR-27a inhibitor lowered the IC 50 . * P < 0.05 vs. Vector-Ctrl, # P < 0.05 vs. NC inhibitor. Fig. S5 Immunohistochemistry of tumors for the detection of Bmi-1, RKIP, Vimentin, Bax and Bcl-2. A. Image from immunohistochemistry of isolated tumors from animals. The animals were subcutaneously implanted with cells stably overexpressing miR-155 or miR-27a and then subjected to intraperitoneal injection of 5-Fu. B. Immunostained sections of different tumors from animals that were implanted with cells stably transfected with shRNA or cotransfected with shRNA and miRNA. Magnification: 200 × . Supplementary Table S1. Sequences of primers. Supplementary Table S2. All differentially expressed miRNAs.
==
Domain: Biology Medicine
|
[{"added": "2023-01-17T15:01:58.172Z", "created": "2020-06-24T00:00:00.000", "id": "255855418", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://molecular-cancer.biomedcentral.com/track/pdf/10.1186/s12943-020-01229-y", "pdf_hash": "0aa275c9218bddcf0495cb869698fb3b5f7a5b85", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:9864", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "0aa275c9218bddcf0495cb869698fb3b5f7a5b85", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Phytoecdysteroids Accelerate Recovery of Skeletal Muscle Function Following in vivo Eccentric Contraction-Induced Injury in Adult and Old Mice
Background: Eccentric muscle contractions are commonly used in exercise regimens, as well as in rehabilitation as a treatment against muscle atrophy and weakness. If repeated multiple times, eccentric contractions may result in skeletal muscle injury and loss of function. Skeletal muscle possesses the remarkable ability to repair and regenerate after an injury or damage; however, this ability is impaired with aging. Phytoecdysteroids are natural plant steroids that possess medicinal, pharmacological, and biological properties, with no adverse side effects in mammals. Previous research has demonstrated that administration of phytoecdysteroids, such as 20-hydroxyecdysone (20E), leads to an increase in protein synthesis signaling and skeletal muscle strength. Methods: To investigate whether 20E enhances skeletal muscle recovery from eccentric contraction-induced damage, adult (7–8 mo) and old (26–27 mo) mice were subjected to injurious eccentric contractions (EC), followed by 20E or placebo (PLA) supplementation for 7 days. Contractile function via torque-frequency relationships (TF) was measured three times in each mouse: pre- and post-EC, as well as after the 7-day recovery period. Mice were anesthetized with isoflurane and then electrically-stimulated isometric contractions were performed to obtain in vivo muscle function of the anterior crural muscle group before injury (pre), followed by 150 EC, and then again post-injury (post). Following recovery from anesthesia, mice received either 20E (50 mg•kg−1 BW) or PLA by oral gavage. Mice were gavaged daily for 6 days and on day 7, the TF relationship was reassessed (7-day). Results: EC resulted in significant reductions of muscle function post-injury, regardless of age or treatment condition (p < 0.001). 20E supplementation completely recovered muscle function after 7 days in both adult and old mice (pre vs. 7-day; p > 0.05), while PLA muscle function remained reduced (pre vs. 7-day; p < 0.01). In addition, histological markers of muscle damage appear lower in damaged muscle from 20E-treated mice after the 7-day recovery period, compared to PLA. Conclusions: Taken together, these findings demonstrate that 20E fully recovers skeletal muscle function in both adult and old mice just 7 days after eccentric contraction-induced damage. However, the underlying mechanics by which 20E contributes to the accelerated recovery from muscle damage warrant further investigation.
INTRODUCTION
Lengthening (eccentric) muscle contractions elicit higher force production, with lower energy cost, than shortening (concentric) contractions. While eccentric muscle contractions are performed on a daily basis (e.g., descending stairs), eccentric exercise has also been utilized in rehabilitative settings as an effective countermeasure against muscle atrophy and weakness, as well as a modality to treat tendinopathies (1). If repeated multiple times, eccentric contractions are a useful tool to induce a physiologically-relevant injury to skeletal muscle, resulting in damage to the contractile apparatus and loss of function (2,3). Skeletal muscle possesses the remarkable ability to repair and regenerate after an injury or damage; however, this ability is impaired with aging (4-7). The repair/regeneration of skeletal muscle tissue after damage relies on a series of highlycoordinated and time-dependent processes (8).
While the exact mechanisms responsible for the impaired regenerative response in aged skeletal muscle continue to be explored, it appears that changes to inflammatory processes (9), protein metabolism (10), and endogenous hormones (11) all play significant roles. Evidence suggests that age-related alterations in immune cell (i.e., macrophage) phenotype may contribute to the impaired regenerative capacity of aged skeletal muscle. Greater increases in M2-like macrophages were observed in aged skeletal muscle, compared to young, in both humans (12) and mice (13) at the same time point during regeneration following eccentric damage. Interestingly, no age-related differences in pro-inflammatory cytokine expression (IL-1B, TNFα, IFNγ) have been observed in regenerating mouse muscle tissue (14). Additionally, the healthy maintenance of skeletal muscle mass is achieved by an intricate balance between protein synthesis and protein breakdown. It has been demonstrated extensively that aged skeletal muscle displays a reduced ability for anabolic stimuli [e.g., resistance exercise (15,16) and dietary protein ingestion (17,18)] to stimulate protein synthesis, resulting in a negative protein balance; a phenomenon aptly named "anabolic resistance." Lastly, correlations exist between declining gonadal hormones and diminished skeletal muscle health in elderly men (11) and women (19). Some previous studies have reported efficacy of hormone therapy (HT) on improving or maintaining skeletal muscle health in certain aged populations (20)(21)(22)(23); however, the long-term potential for adverse side effects with HT [e.g., increased risk for the development of certain types of cancers and cardiovascular events (24,25)] may outweigh the possible benefits. Clearly, the underlying mechanisms responsible for impaired regeneration in aged skeletal muscle are multifaceted. While treatments that target exercise, dietary protein intake, and endogenous hormones have shown marginal success, there is still a great need for developing effective, natural interventions to enhance muscle regeneration with aging.
Phytoecdysteroids (PEs) are natural plant steroids found in a variety of hardy plant species, such as Ajuga and Leuzea, as well as commonly consumed spinach (Spinacia oleracea). PEs possess a plethora of medicinal, pharmacological, and biological properties, with no adverse side effects in mammals (26,27). Characterized as polyhydroxylated basic carbon ring structures of 27-29 carbons, PEs elicit immunoprotective, antioxidant, anabolic, hepatoprotective, hypoglycemic, and physical performance enhancing effects (28). While over 250 different PEs have been identified, 20-hydroxyecdysone (20E) is the most widely investigated. It has been suggested that the anabolic effects of 20E are mediated via a G-protein coupled cell surface receptor (29), as opposed to an intracellular androgen receptor. Thus, 20E is considered anabolic, but non-androgenic since it does not increase prostate or seminal vesicle mass in young castrated rats after 10 days of treatment (30), nor does it alter organ or testes mass in aging mice with 28 days of treatment (31). Regarding skeletal muscle, 20E has been reported to increase grip strength in young rats and stimulate protein synthesis via the PI3K/Akt pathway in C2C12 myotubes in vitro (32). Further, Toth et al. (33) reported that 7 days of 20E treatment increases fiber cross sectional area in healthy soleus, but not the extensor digitorum longus muscle, as well as enhances muscle growth in regenerating (myotoxin-injected) soleus muscles of young rats. Conversely, we recently reported that 28 days of 20E treatment does not alter muscle mass or fiber size, nor does a single acute treatment of 20E stimulate anabolic signaling in skeletal muscle tissue from sedentary aging C57BL/6 mice (31). From these findings, we concluded that a concurrent stress (e.g., recovery from damage) may be required for 20E to elicit beneficial effects on skeletal muscle.
Taken together, it appears that 20E may have the potential to modulate muscle regeneration. With a few exceptions [e.g., (33)], most studies to date have only examined the effect of phytoecdysteroids on muscle size or mass in sedentary, healthy skeletal muscle. No studies have assessed the functional characteristics of skeletal muscle after injury/damage with phytoecdysteroid treatment. Therefore, the purpose of this study was to investigate if 20E accelerates the functional recovery of skeletal muscle after eccentric damage in adult and old mice.
Experimental Design
Male C57BL/6 adult [7.4 ± 0.1 and 7.8 ± 0.4 months of age in the PLA (n = 4) and 20E (n = 7) treatment groups, respectively] and old [26.4 ± 0.4 and 26.5 ± 0.4 months of age in the PLA (n = 7) and 20E (n = 7) treatment groups, respectively] were used in this study. The ages of the adult mice in this study were the equivalent of 30-35 year-old humans, while the old mice were the equivalent of 70-75 year-old humans (9). First, mice were anesthetized (4% isoflurane and maintained with 2% isoflurane) and pre-eccentric damage in vivo contractile function of the anterior crural muscle group [tibialis anterior (TA), extensor digitorum longus (EDL), and extensor hallucis longus] was measured. Mice were immediately subjected to the eccentric contraction-induced muscle damage protocol, and then in vivo contractile function was reassessed. Upon completion of the post-eccentric damage contractile function test, mice were allowed to recover from anesthesia and then returned to their cage. Once fully recovered from anesthesia, mice were randomly assigned to either the placebo control treatment group (PLA) or the 20-hydroxyecdysone treatment group (20E) and received the first treatment via oral gavage. Mice were administered daily treatment doses, at approximately the same time of day, for 6 days. Twenty-four hours after the seventh and final treatment dose, mice were anesthetized again and 7-day recovery in vivo contractile function was measured, as described above (repeated measures design). Finally, mice were sacrificed under anesthesia and the TA and EDL muscles were harvested, weighed for mass, and mounted for histological analysis.
Phytoecdysteroid Treatment
Mice assigned to the 20E treatment groups received daily doses of 50 mg • kg −1 body mass (BM) 20E (E6425-HE; Bosche Scientific, New Brunswick, NJ, USA) dissolved in phosphate-buffered saline (PBS) via oral gavage for 7 days. Mice assigned to the vehicle control treatment groups (placebo; PLA) received daily doses of equivalent volume PBS for 7 days. BM was recorded each day. The dose of 50 mg • kg −1 BM is based on the previous studies by Gorelick-Feldman et al. (32) and Lawrence et al. (31).
In vivo Contractile Function Testing and Eccentric Contraction-Induced Damage Protocol
Contractile function of the anterior crural muscles was measured three times in each mouse: pre-and post-eccentric damage, as well as after the 7-day treatment period, via the torque-frequency relationship, as previously described (34,35). Under anesthesia, hair on the left hindlimb was removed with dilapidation cream and thoroughly rinsed with water. Mice were then placed on a heated (37 • C) platform and the left foot was secured with an aluminum foot-cover and tape to the footplate affixed to the shaft of a dual-mode servomotor (300B-LR, Aurora Scientific, Aurora, ON, Canada). A clamp secured to a micro-manipulator (World Precision Instruments, Sarasota, FL) was used to position and hold the left knee in place during the procedure. The ankle joint was held at 90 • of passive dorsiflexion with respect to the tibia and the tibia was positioned at 90 • with respect to the femur. Sterilized 30-gauge needle electrodes (Grass Instruments, Warwick, RI) were inserted through the skin for stimulation of the left common peroneal nerve, each was positioned and held in place with a micro-manipulator. Single isometric contractions (1 Hz) were used to obtain initial needle electrode placement; optimal stimulation voltage (5-10 volts) and needle electrode placement was confirmed by 5-10 isometric contractions (200ms train duration, 0.1-ms pulse width at 300 Hz). Following needle electrode placement, a torque-frequency curve measured peak isometric torque produced by the anterior crural muscle group at 10 ascending stimulation frequencies, from 20 to 300 Hz, with 2 min rest between each contraction.
After completion of the initial pre-eccentric damage in vivo contractile function testing, the anterior crural muscle group was immediately subjected to the eccentric contraction-induced muscle damage protocol using 150 eccentric contractions (300 Hz, 120-ms train of 0.1-ms pulses, with 38 • of angular movement at 2,000 • •s −1 starting in 19 • of dorsiflexion, moving to 19 • of plantarflexion) described by Corona et al. (35). Muscle function was also assessed every 10th eccentric contraction via individual 300 Hz isometric contractions (Figure 1). Warren et al. previously reported that decreases in isometric torque during this eccentric contraction-induced muscle damage protocol are the result of muscle injury, not fatigue (36). After completion of the eccentric contraction-induced damage protocol and a 5min delay, post-eccentric damage in vivo contractile function was measured. Finally, after the 7-day treatment period, mice were anesthetized and 7-day recovery in vivo contractile function was measured as described above.
Additionally, separate groups (n = 10/each) of sham-treated adult and old mice were utilized as experimental controls. Mice in the sham-treated groups performed all of the same experimental procedures described above (including the 7-day 20E and PLA treatments), except they were not subjected to the eccentric damage protocol. During the time required to complete the eccentric damage protocol (∼30 min), mice remained anesthetized and resting before continuing with the post-and eventually the 7-day sham-recovery contractile function tests.
Contractile Function Data Acquisition and Analysis
The muscle lever system (Aurora Scientific 1300A), stimulator and force transducer were connected to a signal interface (Aurora Scientific, Model 610A) that sends the analog signal to an analog-to-digital converter card (National Instruments, Austin, TX) on a computer with Dynamic Muscle Control software (Aurora Scientific, DMC 610A). The force output data were analyzed utilizing Dynamic Muscle Analysis software (Aurora Scientific, DMA 610A). Raw (group mean) torque-frequency relationships displayed in Figure 2 and Supplementary Figure 1 were analyzed for statistical significance prior to modeling. To generate the EC 50 , torque-frequency relationship data were modeled with the following four-parameter logistic fit equation using GraphPad Prism version 8.4.3, GraphPad Software, San Diego, California USA: where x is the stimulation frequency; min and max are the smallest (i.e., twitch) and largest (i.e., peak tetanic) torques estimated from the best-fit torque-frequency relationship curve, respectively; EC 50 is the stimulation frequency required to generate 50% of maximal estimated torque (max-min); and n is the Hill Slope Coefficient indicating the slope of the steepest portion in the estimated torque-frequency relationship curve (37). The EC 50 provides further insight on contractile function from the estimated torque-frequency relationships (38).
Histological Analyses
The TA muscle was harvested from the left hindlimb and mounted for histological analysis following measurement of 7-day recovery in vivo contractile function. Harvested tissues were mounted on cork using a mixture of tragacanth gum and optimal cutting temperature medium (Fisher Scientific, Houston, TX), frozen in liquid nitrogen-cooled isopentane, and stored at −80 • C until sectioning. TA muscle samples were cut into 10 µm cross-sections using a cryostat (CryoStar Model HM 505; ThermoFisher Scientific Inc.) and mounted on positively charged microscope slides. Muscle section quality, tissue integrity, and markers of muscle damage were assessed using common histological techniques for cytosolic and nuclear components using Mayer's hematoxylin and eosin (H&E) solutions (Millipore Sigma, St. Louis, MO). H&E-stained muscle sections were imaged with 4× and 10× objectives using an Olympus IX81 light microscope and cellSens Imaging Software (Olympus, Waltham, MA). Markers of muscle damage (e.g., edema, overt fiber damage, presence of infiltrating inflammatory cells, and centrally-located myonuclei) (39)(40)(41) were assessed in each image using qualitative indices on a scale of 0-3 to provide a muscle damage score: with 0 = no apparent muscle damage; 1 = minimal muscle damage; 2 = moderate muscle damage; and 3 = severe muscle damage.
Skeletal muscle repair and regeneration after damage is often accompanied by remodeling of connective tissue. Gomori trichrome staining (Millipore Sigma, St. Louis, MO) was performed on TA muscle sections (10 µm) as this technique differentiates skeletal muscle from connective tissue. Gomori trichrome-stained muscle sections were imaged with 4× and 10× objectives using an Olympus IX81 light microscope and cellSens Imaging Software. No analyses were performed on Gomori trichrome-stained sections.
Statistical Analyses
All data are expressed as mean ± SEM. Torque-frequency relationship data were analyzed using a repeated measures three-way factorial ANOVA (treatment, two levels; time, three levels; stimulation frequency, 10 levels). Measures of contractile performance data (eccentric damage isometric contractions, twitch torque, maximal tetanic torque, twitch:tetanic ratio, and EC 50 ) and animal body masses were analyzed using a repeated measures three-way factorial ANOVA (treatment, two levels; time, two or three levels; age, two levels). Individual muscle masses were analyzed using a repeated measures two-way factorial ANOVA (treatment, two levels; age, two levels). The a priori level of significance was set at p < 0.05. Following a significant F-ratio, Fisher's LSD pairwise post-hoc comparisons were made. Data were analyzed using SPSS (IBM Corp., Armonk, NY). No statistical analyses were performed for the qualitative analysis for muscle damage scores.
Animal Subjects and Skeletal Muscle Mass
Animal subject characteristics are shown in Table 1. Body mass did not differ between any treatment group (time x treatment x age interaction, p = 0.875); however, all groups lost body mass between initial and the 7-day recovery time point (main effect of time, p = 0.008). Furthermore, TA and EDL muscle mass (normalized to body mass; mg•g −1 ) were significantly lower in the old mice (main effect of age, p = 0.003 for both TA and EDL), regardless of treatment (treatment x age interaction, p = 0.453 for TA and p = 0.326 for EDL). Animal and muscle characteristics for sham-treated groups followed similar patterns of significance as the eccentric damage groups stated above (Supplementary Table 1).
Eccentric Contraction-Induced Muscle Damage
The eccentric contraction-induced muscle damage protocol results in a ∼40-50% decline (main effect of time, p < 0.001) in anterior crural muscle group contractile function, measured by maximal isometric torque (Figure 1). Additionally, there were no differences in the loss of contractile function in response to eccentric damage between ages of mice or treatment groups (time x treatment x age interaction, p = 0.740).
In vivo Skeletal Muscle Contractile Function
To investigate if 20E accelerates the recovery of skeletal muscle after eccentric damage in adult and old mice, in vivo skeletal muscle contractile function was measured before, immediately after, and at 7 days of recovery from eccentric damage. Indeed, the eccentric contraction-induced damage protocol significantly reduced isometric torque immediately post-damage in both adult and old mice (main effect of time, p < 0.001; Figures 2B,E, respectively; time × treatment × age interaction, p = 0.753). Most remarkable, however, was the significant time × treatment interaction (p < 0.001) demonstrating that 20E treatment resulted in full recovery of isometric torque at 7 days-post eccentric damage in both adult and old mice, while neither PLA-treated group (both adult and old) recovered by 7 days (Figures 2C,F, respectively). Additionally, we performed experiments on adult and old sham-treated mice that had not performed the eccentric damage protocol. In vivo skeletal muscle contractile function was not altered over time or treatment in any of the adult or old sham-treated groups (time × treatment × age interaction, p = 0.766), thus demonstrating the repeatability of our experimental procedures (Supplementary Figure 1).
Further analysis of contractile function revealed that there was only a significant main effect of time (p < 0.001) in FIGURE 1 | Percent of initial isometric torque during the 150 eccentric contraction-induced muscle damage protocol. Maximal tetanic isometric torque of the anterior crural muscles was measured after every 10th eccentric contraction of the damage protocol in adult and old mice. There was a significant 40-50% decline in isometric torque after the 150-eccentric contraction muscle damage protocol, regardless of age or treatment condition. Values represent mean ± SEM. PLA, placebo; 20E, 20-hydroxyecdysone. * significantly different than initial; p < 0.001. twitch torque (20 Hz), as twitch was significantly lower in all treatment groups and ages immediately after eccentric damage (p < 0.001) and remained lower at 7 days (p = 0.032; Figures 3A,E, respectively), compared to pre-damage. While no time x treatment x age interaction (p = 0.677) existed in maximal tetanic torque (250 Hz), there was a significant time x treatment interaction (p < 0.001). Maximal tetanic torque was significantly lower immediately after eccentric damage in all treatment groups and ages (p < 0.001); however, only the 20E-treated groups in both adult ( Figure 3B) and old mice ( Figure 3F) fully recovered maximal tetanic torque by returning to pre-damage levels at 7 days of recovery. Similar to twitch torque, there was only a significant main effect of time in the twitch to tetanic ratio (p < 0.001), as the ratio was significantly lower in all groups immediately after eccentric damage (Figures 3C,G), but recovered with 7 days of recovery (time × treatment × age interaction, p = 0.630). Furthermore, there were no age-or treatment-related differences regarding how eccentric damage caused muscle contractile dysfunction or recovery from muscle damage (all treatment x age interactions, p > 0.500). In other words, adult and old mice displayed similar responses to eccentric muscle damage and recovery with 20E treatment.
There was no time x treatment x age interaction (p = 0.586) for EC 50 ; however, there was a significant main effect of time (p < 0.001), such that EC 50 was significantly higher in all age and treatment groups immediately post-damage (p < 0.001), before beginning to return back to pre-damage levels by 7 days of recovery, i.e., there was still a significant difference between 7-d vs. Pre (p = 0.009), but also a difference between 7-d vs. Post (p = 0.002) time points (Figures 3D,H). These data signify that a greater stimulation frequency is required to elicit 50% of maximal tetanic torque immediately after eccentric damage.
Histology
H&E staining was performed to observe morphological muscle damage (i.e., edema, fiber damage, presence of infiltrating inflammatory cells, and centrally-located myonuclei). Markers of muscle damage were minimally observed (muscle damage scores ≤ 1.0 on a scale of 0-3) at 7 days post-injury in TA muscles subjected to eccentric damage (Figures 4Av-viii), compared to sham-treated muscles (Figures 4Ai-iv). Muscle damage scores obtained via qualitative/semi-quantitative analyses revealed that muscles in the PLA-treated groups appear to have higher muscle damage scores, compared to 20E-treated groups (39)(40)(41). Furthermore, the old PLA-treated mice appeared to have the greatest muscle damage score at 7 days post-injury, compared to any other group (Figure 4B).
Gomori trichrome staining of the TA muscles revealed that connective tissue staining appears to be higher in the muscles subjected to eccentric damage (Figures 5E-H)" compared to sham-treated muscles (Figures 5A-D). It does not appear that 20E treatment has any effect on connective tissue staining in regenerating muscle tissue at 7 days post-injury.
DISCUSSION
When skeletal muscle is subjected to contractile stress and/or strain that supersedes the normal capabilities of the muscle, an injury or damage occurs and functional capacity (i.e., muscle torque/force or strength) is diminished. Here in this study, as with many others (34,35,(42)(43)(44), we reaffirm that functional capacity is significantly impaired immediately, and for several days after, 150 eccentric contractions. This eccentric contraction-induced damage protocol results in ∼40-50% reductions in isometric torque of the anterior crural muscles (Figure 1). Consistent with previous studies in rodents (45,46), these declines in functional capacity with eccentric contractions were similar between the adult and old mice and between treatments. In other words, all mice in this study sustained similar degrees of muscle damage, regardless of age or treatment. It is important to note that 20E (or placebo) treatments did not begin until the mouse had recovered from anesthesia after FIGURE 3 | Isometric twitch torque, maximal tetanic torque, twitch:tetanic ratio, and EC 50 of the anterior crural muscles in Adult and Old mice at the pre-injury, immediately post-injury, and 7-day recovery time points. Analysis revealed significant reductions in twitch torque (20 Hz; A,E) in all groups immediately after (Post) and at 7 days (7-d) following the eccentric contraction-induced muscle damage protocol, regardless of treatment condition. Maximal tetanic torque (250 Hz) was also significantly reduced in both adult (B) and old mice (F) Post eccentric damage, compared to Pre, regardless of treatment condition. Only the 20E-treated adult and old mice fully recovered maximal tetanic torque to pre-injury levels by 7-d. Twitch:Tetanic ratio was significantly reduced at Post in both adult (C) and old mice (G), regardless of treatment condition, compared to Pre, but Twitch:Tetanic recovered by 7-d. EC 50 was significantly higher in all age and treatment groups at Post, but EC 50 at 7-d was still significantly different than Post and Pre (D,H). Values represent mean ± SEM. PLA, placebo; 20E, 20-hydroxyecdysone; EC 50 , stimulation frequency required to elicit 50% of maximal tetanic torque. * significantly different than Pre within the same group (p < 0.05); & significantly different than Post within the same group (p < 0.05). the initial muscle function testing session, which included the eccentric damage protocol. Therefore, it is not possible for 20E to have provided any prophylactic or protective effects on muscles suffering eccentric damage. Arguably, functional loss of strength is the most important and reliable indicator of skeletal muscle injury, not to mention the clinical implications that loss of muscle function imparts (3). The force-producing capabilities of the muscle usually return to baseline (pre-injury) levels between 2 and 4 weeks post-eccentric contraction-induced injury (34,42,44,46,47). But this process may take longer depending on the severity of damage incurred (48). The regenerative capacity of skeletal muscle is impaired during aging, and this can lead to prolonged or incomplete recovery from eccentric damage, as well as further loss of muscle mass and strength. Seminal work by Brooks and Faulkner (46) demonstrated that while young and aged mice respond similarly to eccentric damage (i.e., loss of strength and markers of muscle damage), young mice had recovered muscle function by 4 weeks post-damage, whereas aged mice had not. The most intriguing and novel finding of the current study was that 20-hydroxyecdysone (20E) completely recovered skeletal muscle functional capacity by 7 days postinjury, not only in the adult mice, but also in the old mice. Based on only measuring a single time point and the relatively short recovery period examined in the current study (7 days), it is difficult to forecast just how long it would have taken for the placebo-treated mice in our study to recover muscle function back to pre-injury levels. It seems plausible that some, but not full recovery of muscle function is possible in just 7 days after eccentric muscle damage (35,49,50). This study demonstrates, for the first time, that oral supplementation with phytoecdysteroids accelerates the functional recovery of skeletal muscle in both adult and old mice after eccentric contractioninduced injury.
A major tenet of muscle contraction and functional capacity is the ability for a neural impulse to stimulate the release of calcium (Ca +2 ) from the sarcoplasmic reticulum (SR) leading to the development of force at the sarcomere, a process widely known as excitation-contraction (EC) coupling. Traditionally, it was thought that disruption of the force-generating and force-transferring structures of the sarcomere was responsible for the loss of muscle function after eccentric contractioninduced muscle damage (51). However, there is a dissociation between muscle function and histological markers of muscle damage that disputes the latter causal relationship. Recall that the greatest declines in muscle function occur immediately following eccentric contraction-induced injury, yet markers of muscle damage are not fully apparent until 1-2 days after injury (34,47,52). Studies from the past few decades have established that the early functional loss of strength from eccentric contractions (0-5 days post-injury) is primarily the result of EC coupling dysfunction, not contractile structure disruption. Warren et al. (53) were the first to describe eccentric contraction-induced EC coupling dysfunction in mouse soleus muscle in vitro. Later, Ingalls et al. (47), exhibited that the site for EC dysfunction in response to eccentric damage lies at the level of the t-tubule and the SR Ca +2 release channel (i.e., the "triad"), specifically via disruption of the interface between dihydropyridine receptors (DHPR) and ryanodine receptors (RyR1). The decreased twitch to tetanic ratio observed immediately after injury provides an indirect indication of EC coupling dysfunction with eccentric damage; however, this impairment was recovered by 7 days, regardless of treatment or age (50,54). While 20E elicits a rapid, but transient (30-180 s) increase in intracellular [Ca +2 ] in C2C12 myotubes in vitro (29), it is highly unlikely that 20E is producing any long-term effects on Ca +2 kinetics or repair of the dysfunctional EC coupling mechanism resulting in the observed rescue of muscle function at 7 days in this study.
It has been estimated that EC coupling dysfunction is responsible for 57-75% of the functional loss of muscle strength acutely after eccentric contractions (51). However, in addition to reversing EC coupling dysfunction, other mechanisms also contribute to the recovery of muscle function observed in the days to weeks following eccentric damage. As the normal progression of damage and regeneration processes occur, disruptions to sarcomeric (contractile apparatus) and sarcolemmal (membrane) structures have been observed within days of the eccentric exercise bout (35,55,56). Furthermore, there is sufficient evidence that alterations in skeletal muscle protein metabolism may be responsible for the remaining strength deficits in the 1-2 weeks after eccentric damage (34,42). Therefore, a need to restore protein balance may account for the prolonged recovery time. Skeletal muscle protein synthesis signaling, via activation of the mTORC1 pathway, is stimulated in the range of 1-7 days post-damage in response to various eccentric or lengthening contraction protocols in rodents and humans (57)(58)(59)(60)(61). Phytoecdysteroids, particularly 20E, promote anabolic responses in many tissues, including skeletal muscle. Gorelick-Feldman et al. demonstrated that 20E stimulates protein synthesis (32), primarily through the activation of PI3K/Akt signaling (29), in a dose-dependent manner in C2C12 myotubes in vitro. Furthermore, pre-treatment of myotubes with the G protein inhibitor, Bordetella pertussis toxin (PTX), abolishes the 20E effect on Ca +2 influx and Akt activation, suggesting that 20E functions through a G proteincoupled cell surface receptor (29). We recently reported that 20E does not provide an anabolic stimulus, via activation of the PI3K/Akt/mTORC1 pathway, in skeletal muscle from aging sedentary mice, despite using the same dose of 20E as in the current study (50 mg•kg −1 BM) (31). From this study, we concluded that an additional stimulus, for example exercise or injury, may be required for 20E to elicit anabolic effects on skeletal muscle. The novel observation that 20E treatment fully recovers skeletal muscle function in just 7 days after eccentric damage is much earlier than other studies using similar eccentric damage protocols in mice (34,35,42,47). Therefore, we conclude that 20E accelerates the functional recovery of muscle after eccentric damage. While we cannot definitively state that the recovery of eccentric contraction-induced skeletal muscle dysfunction after just 7 days with 20E supplementation is due solely to the anabolic effects of 20E (via PI3K/Akt/mTORC1 signaling), based on previous literature it seems plausible that 20E may provide an anabolic stimulus to damaged muscle leading to accelerated recovery of muscle function (62). Moreover, since aged skeletal muscle displays anabolic resistance, 20E may be able to stimulate alternative anabolic pathways to those traditionally responsible for the anabolic resistance in aged muscle. Thus, daily 20E treatments, starting immediately after eccentric damage, could be providing an anabolic stimulus required to accelerate muscle recovery after eccentric damage, compared to placebotreated mice. Whether 20E functions via the canonical protein synthesis pathway leading to recovery of skeletal muscle function needs to be investigated in future studies.
Skeletal muscle damage triggers a widespread series of events that can essentially be divided into two main stages: tissue degeneration and tissue regeneration. Tissue degeneration is necessary for removing damaged and dysfunctional structures, whereas tissue regeneration works to repair or rebuild muscle structures and regain function. Both stages rely heavily on mechanisms of inflammatory and myogenic pathways, not to mention many others. As discussed previously, inflammation is not responsible for the immediate decline of muscle function with eccentric contractions. However, inflammatory processes are essential for the successful regeneration of skeletal muscle and recovery of muscle function after injury (9). One of the hallmarks of histological muscle damage is the ordered infiltration of immune cells; first to respond are the neutrophils that appear in the first 24 h after injury, followed by M1-like macrophages around day 2, and finally M2-like macrophages by day 4 post-injury (8). In the current study we did not assess immune cells or inflammatory mechanisms, but phytoecdysteroids, including 20E, have been suggested to possess anti-inflammatory properties (27). While the anti-inflammatory effects of 20E directly on skeletal muscle are not known, previous reports suggest that the PI3K/Akt/mTORC1 pathway regulates immune (macrophage) cell function (63,64). Theoretically, if 20E treatment is activating the PI3K/Akt/mTORC1 pathway during the 7-day recovery period, this could be potentially beneficial in managing the immune cell/inflammatory response to eccentric muscle damage and accelerating muscle recovery. However, we have no direct evidence that phytoecdysteroids influence inflammatory mechanisms during recovery from eccentric muscle damage, but this area certainly warrants further investigation.
Regarding our histological findings, it appears that markers of muscle damage are still apparent, albeit minimal, after 7 days of recovery from eccentric damage. It is interesting that despite a full recovery of muscle function (i.e., isometric torque) at 7 days in our 20E-treated groups, markers of muscle damage remain evident. This is not surprising as similar findings have been described previously wherein muscle function had recovered, but markers of muscle damage and regeneration, particularly centralized nuclei, are still visible weeks after eccentric contraction-induced muscle damage (35,44). As described above, much of the inflammatory/immune response contributing to muscle regeneration has run its course by 7 days post-eccentric injury and what remains is tissue remodeling and growth processes. While the muscle tissue has been the main focus of this study, it is also important to recognize the importance of the association between the muscle fiber and connective tissue, or extracellular matrix (ECM), that surrounds the muscle fibers. Whether the ECM is damaged in response to eccentric contractions is still unclear (65). Generally speaking, aged skeletal muscle tissue contains more ECM than young muscle. One of the major complications with the age-related impairment in muscle regeneration after injury is not only the decreased myogenic potential, but also increased fibrogenesis (growth of connective tissue) (66). During regeneration of aged skeletal muscle, muscle tissue is replaced with connective tissue and, consequently muscle function and muscle quality decline. To our knowledge, no studies have investigated the effect of phytoecdysteroids on connective tissue or components of the ECM during muscle regeneration. If 20E is influencing connective tissue remodeling, it cannot be determined from our histological observations. Therefore, we cannot conclude whether 20E is providing any positive benefit to muscle damage markers or tissue remodeling in the 7-day recovery period after eccentric damage. However, future studies should investigate the potential role of 20E with regards to markers of muscle damage and tissue morphology with extended time points after eccentric muscle damage.
CONCLUSION
In conclusion, eccentric contraction-induced damage results in significant declines in skeletal muscle function. Normal repair/regeneration of skeletal muscle tissue after damage relies on a series of highly-coordinated and time-dependent processes, including, but not limited to inflammatory, myogenic, and protein balance mechanisms. However, the regeneration process is impaired with aging. 20-hydroxyecdysone (20E) possesses anabolic, anti-inflammatory, and antioxidant properties. Here, for the first time, we demonstrate that daily treatment with 20E fully recovers skeletal muscle function in both adult and old mice within just 7 days after eccentric damage. The underlying mechanics by which 20E contributes to the accelerated recovery from muscle damage warrant further investigation. Taken together, it is reasonable to suggest that 20E has potential to be utilized as a supplementary intervention for muscle recovery after damage and in aging.
DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.
ETHICS STATEMENT
The animal study was reviewed and approved by the Appalachian State University Institutional Animal Care and Use Committee.
AUTHOR CONTRIBUTIONS
KZ and RS were responsible for the conceptualization of the project, development of methodology, project administration, monitoring of data collection, statistical analysis, and writing of this manuscript. JG and CH were responsible for data collection, processing, and analysis and editing of this manuscript. All authors significantly contributed to the article and approved the submitted version.
==
Domain: Biology Medicine
|
[{"added": "2021-11-08T14:22:58.267Z", "created": "2021-11-08T00:00:00.000", "id": "243834527", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.frontiersin.org/articles/10.3389/fresc.2021.757789/pdf", "pdf_hash": "9e0a1d944fe5d57e3ffb78fb28216c5037d34189", "pdf_src": "Frontier", "provenance": "peS2o-0028.json.gz:4505", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "9e0a1d944fe5d57e3ffb78fb28216c5037d34189", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Effect of Tiaojingzhixue Fang on the expression of sex hormone and endometrial tissue mRNA in perimenopausal patients with abnormal uterine bleeding
This study aimed to investigate the effect of Tiaojingzhixue decoction on the expression of sex hormone and endometrial mRNA in perimenopausal patients with abnormal uterine bleeding. For this purpose, 84 patients with perimenopausal abnormal uterine bleeding who were treated in our hospital from January 2018 to January 2019 were divided into study group and control group, the control group was treated with mifepristone for six months, the clinical efficacy, time of symptom relief and disappearance, endometrial thickness, menstrual volume, adverse events, expression of sex hormone (ER), Progesterone receptor (PR), Lutropin (LH), Follicle-stimulating hormone (FSH), Estradiol (e2) and Vascular endothelial growth factor in endometrial tissue were compared between the two groups. The results showed that the clinical efficacy of the study group (97.61%) was significantly higher than the control group (80.95%) (P<0.05). There was no significant difference in endometrial thickness between the two groups before treatment, but after treatment for 1 month, 3 months and 6 months, endometrial thickness was thinner in the study group (P<0.05). There was no significant difference in menstrual volume between the two groups before treatment (P <0.05), but it was lower in the study group (P <0.05). The incidence of adverse reactions in the study group was 11.90% lower than in the control group (26.19%). The expression levels of ER, PR, LH, FSH and E2 were almost the same before treatment and there was no significant difference between the two groups (P>0.05), but they were lower in the study group after treatment (P<0.05). Before treatment, there was no significant difference in the level of VEGF between the two groups (P>0.05); after treatment, the level of Vegf in the study group was lower than the control group (P<0.05). After treatment, the level of Vegf in both groups was significantly higher than before treatment (P<0.05). In general, Tiaojingzhixue decoction can decrease the level of sex hormone and increase the expression of VEGF in patients with perimenopausal abnormal uterine bleeding.
Introduction
Abnormal uterine bleeding (AUB) is a common disease in gynecology. Patients present with irregular menstrual cycles, including frequency, regularity, duration, and flow outside pregnancy. Studies have shown that as many as one-third of women will experience abnormal uterine bleeding during their lifetime, most often during menarche and perimenopause (1). The normal menstrual cycle is 24 to 38 days, lasts 7 to 9 days, and blood loss is 5 to 80 ml. AUB may be a manifestation of the hormonal environment or a clinical manifestation of benign and malignant lesions of the reproductive tract in perimenopausal women. Postmenopausal bleeding needs special attention because it is a clinical indication of endometrial cancer in developed countries and cervical cancer in China (2). Abnormal uterine bleeding can also be classified as acute or chronic. Acute AUB is excessive bleeding and requires immediate intervention to prevent further bleeding. Acute AUB can occur alone or in combination with chronic AUB, which is irregular menstrual bleeding for most of the past 6 months.
Perimenopause includes the last two to eight years before menopause and one year after the onset. Menorrhagia represents large amounts of regular bleeding for a long time and may be associated with fibroma, adenomyosis, or endometrial polyps.
Menorrhagia and uterine bleeding represent other types of AUB, whose root cause is other endometrial changes (3).
Abnormal uterine bleeding is related to cytokines that regulate cell proliferation in vivo, such as the Basic fibroblast Factor (bGGF) vascular endothelial growth factor, etc. (4,5). The reduced expression of bGGF will slow down the repair of the endometrium and aggravate the symptoms of abnormal uterine bleeding (6). These effects on the uterus are also related to the expression of progesterone receptors and estrogen receptors in the body (7, 8), and the level of expression is closely related to the severity of the disease. Low levels of estrogen can have negative feedback to the hypothalamus, and the endometrium is disturbed by a single estrogen, leading to polyp hyperplasia, retreat bleeding, and perimenopausal abnormal uterine bleeding.
This study mainly studied the effects of Tiaojing Zhixue Prescription on the expression of sex hormones and endometrial tissue mRNA in perimenopausal patients with abnormal uterine bleeding. The report is as follows:
Materials and methods General Information
A total of 84 patients with perimenopausal abnormal uterine bleeding treated in our hospital from January 2018 to January 2019 were selected as the research objects, and they were divided into study groups and control groups, with 42 patients in each group. The study group was 45-54 years old, with an average age of (49.11±4.01) years. The course of the disease was 2-7 months, with an average course of (4.95±1.23) months. Pregnancy 1-4 times, there were 19 cases of mild anemia, 18 cases of moderate anemia and 5 cases of severe anemia. The control group was 43-53 years old, with an average age of (47.32±4.98) years. The course of the disease was 1-7 months, with an average course of (4.22±1.45) months. Pregnancy 1-4 times, there were 20 cases of mild anemia, 17 cases of moderate anemia and 5 cases of severe anemia. There was no statistically significant difference in age and course of disease between the two groups (P<0.05), indicating comparability. This study has been approved by the Ethics Association of our hospital.
Inclusion Criteria
(i) All patients were diagnosed with abnormal perimenopausal uterine bleeding; (ii) Patients have no contraindication to the drugs used; (iii) Patients were not treated with other hormone drugs before treatment; (iv) None of the patients had fertility requirements; (v) Patients can cooperate with researchers; (vi) Patients and their families were informed of the study and signed informed consent.
Exclusion Criteria
(i) Patients with malignant tumors; (ii) patients with cervical lesions; (iii) patients with diseases of the endocrine system; (iv) patients with severe dysfunction of liver, kidney and other organs; (v) patients with combined mental diseases.
Research Methods
Immunohistochemistry was used to detect the expression of the estrogen-progesterone receptor in tissue samples from 84 patients. Dewaxing and hydration 4μm thick paraffin tissue slices were washed with buffer solution 3 times, each time for 3min, and then put into the microwave oven for continuous heating at 92°C ~ 98°C for 12min. After that, goat serum was dropped to seal the non-specific antigen, and then mouse anti-human ER monoclonal antibody was added (Company name: Abcam, Type: Rabbit anti-human polyclonal antibody) was placed in a refrigerator at 4°C overnight and rewarmed for 30min. The indirect method of Streptomyces antibiotin protein with horseradish peroxidase was added to detect the receptor expression, and chromogenic agent was used for chromogenic process (9). After chromogenic process, hemoxylin was used for redyeing and neutral adhesive was used for sealing. According to the improved Sinicrope method, PR and ER staining intensity ratings of uterine tissue cells were as follows: 0 for colorless or almost no staining, 1 for light brown-yellow staining, 2 for light brown-yellow staining and 3 for dark brown-yellow staining; Scores of PR and ER staining cells in sections: 0 scores for positive cells less than 5%, 1 score for 5%-25%, 2 scores for 25%-50%, 3 scores for 50%-75% and 4 scores for 75%-100%. The final PR ER staining score was obtained by adding the staining intensity to the staining cell number score. After adding staining intensity score and staining cell number score, 0 ~ 2 was divided into grade 1, 3 ~ 4 was divided into grade 2, and 5 ~ 6 was divided into grade 3. All stained sections were examined by 2 physicians. If the results were consistent, the results would be final.
Intravenous plasma was collected on fasting in the morning before and after treatment in all patients and centrifuged in a centrifuge at 1700r/min at low temperature. The supernatant was removed and stored in a refrigerator at 4℃ for later detection.
Patients in the study group were treated with meridian regulating hemostasis formula, which was prescribed as follows: Astragalus membranaceus 18 g 12 g, dangshen 30 g, Fried atractylodes, motherwort 30 g, cattail pollen carbon 12 g (Fried), 15 g, 12 g field thistle madder, 30 g purslane, herba schizonepetae, sanguisorba 30 g, 6 g ink dry lotus 30 g, 6 g cohosh, notoginseng powder (a blunt), glue 9 g, 3 g anemarrhena asphodeloides bge 9 g, phellodendri 6 g, burnt hawthorn, burnt divine qu, burnt malt 9 g, decocted in water, one dose daily, morning and evening. The drug was administered for 5 to 10 days, starting on the fifth day of rebleeding, and was observed for 2 consecutive cycles of three menstrual cycles.
The control group was given didrogesterone tablets (Chinese drug approval Batch No. H20050395, drug specification: 5mg) once a day, 12.5mg/ time
Observation Indicators
(i) Compare the clinical efficacy of the two groups after treatment. Significant effect: Abnormal uterine bleeding disappeared effectively: abnormal uterine bleeding was significantly reduced than before; Ineffective: Abnormal uterine bleeding symptoms, menstrual time, cycle and so on have no significant difference with before treatment, or even more serious; (ii) The symptom relief time and complete disappearance time were compared between the two groups after treatment. (iii) Color vaginal ultrasound was used to examine the endometrial thickness of the two groups before and after treatment for six months. (iv) The menstrual volume of the two groups was detected and compared by menorrhagia. According to the area of blood-stained with tampons in the two groups of patients, the scoring rules are as follows: 1 point for blood-stained area less than 1/3; The area between 1/3 and 2/3 is 5 points; More than 2/3 of the area is 20 points, the higher the score means more menstruation.(5) The incidence of adverse reactions between the two groups was compared; (6) The expression levels of ER, PR, LH, FSH and E2 were compared between the two groups. Serum LH, FSH and E2 levels were analyzed by ELISA. The contents of ER and PR were determined by adding staining intensity score and staining cell number score. (7) The expression level of VEGF in endometrial tissue was compared between the two groups. The expression of VEGF was detected by immunohistochemistry.
Statistical methods
SPSS 22.0 software was used to analyze the obtained data. Mean ± standard deviation was used to represent the counting data. T-test was used for comparison between the two groups. [N (%)] was used to represent the count data, and χ2 was used to test the difference between groups. Rank sum test was used for grade data and P<0.05 was considered statistically significant.
Results and discussion Comparison of clinical efficacy between the two groups after treatment
The experimental results showed that the clinical efficacy of the study group (97.61%) was higher than that of the control group (80.95%), and the comparison between the two groups was statistically significant (P<0.05), as shown in Table 1.
Comparison of symptom relief time and complete disappearance time between the two groups after treatment
After the experiment can be obtained, the symptom relief time and symptom disappear time of the study group are less than the control group. There was significant statistical significance between the two groups (P<0.05), as shown in Table 2. Comparison of endometrial thickness between the two groups before and after treatment After the experiment, the endometrial thickness of the two groups before treatment was almost the same, and there was no significant statistical significance between the two groups (P<0.05), which was comparable. The endometrial thickness of the study group was thinner than that of the control group after treatment for 1 month, 3 months and 6 months, and the comparison between the two groups was statistically significant (P<0.05). The endometrial thickness of the two groups after treatment was thinner than that before treatment, and there was statistically significant difference between the two groups before and after treatment (P<0.05), as shown in Table 3.
Comparison of menstrual volume between the two groups before and after treatment
After the test, the menstrual volume of the two groups was almost the same before treatment, and there was no significant statistical significance before treatment (P<0.05), which was comparable. After treatment, the menstrual volume of the study group was lower than that of the control group, and there was significant statistical significance between the two groups (P<0.05). After treatment, menstrual volume in both groups was lower than before, and the comparison between the two groups before and after treatment was statistically significant (P<0.05), as shown in Table 4.
Comparison of adverse reactions between the two groups after treatment
The results showed that the incidence of adverse reactions in the study group was 11.90% lower than that in the control group (26.19%), and the comparison between the two groups was statistically significant (P<0.05), as shown in Table 5.
Comparison of ER, PR, LH, FSH and E2 expression levels between the two groups before and after treatment
The results showed that the expression levels of ER, PR, LH, FSH and E2 were similar before treatment, and there was no significant statistical significance between the two groups (P<0.05), indicating comparability. After treatment, the expression levels of ER, PR, LH, FSH and E2 in the study group were lower than those in the control group. After treatment, the expression levels of ER, PR, LH, FSH and E2 in the two groups were lower than before treatment, with significant statistical significance (P<0.05), as shown in Table 6.
Comparison of VEGF expression levels between the two groups before and after treatment
After the test, VEGF levels in the two groups were almost the same before treatment, and there was no significant statistical significance before treatment (P>0.05). After treatment, the level of VEGF in the study group was lower than that in the control group, and there was statistical significance between the two groups after treatment (P<0.05). VEGF levels in both groups were higher after treatment than before, with significant statistical significance (P<0.05), as shown in Table 7. ABU is a very common female disease, accounting for 1/3 of gynecological diseases, more than 70% of which are perimenopausal, and its main clinical manifestations are frequent and regular cycles, blood loss and abnormal period time (10). The principle of its pathogenesis is that with the increase of women's age (9), ovarian function declines and the mechanism of regulating sex hormones in the body is disturbed, which eventually leads to breakthrough or withdrawal bleeding (11). The endometrial estrogen receptor volume increases, resulting in changes in the endometrium (12)(13)(14). The most common cause of abnormal uterine bleeding before and after menopause is endometrial polyps, which are more common in postmenopausal women with atypia and malignant tumor proliferation. Endometrial polyps may occur in women of childbearing age and postmenopause, and they are one of the most common causes of abnormal uterine bleeding (15). According to studies, endometrial polyps were detected in about 26% of AUB patients (16)(17)(18). In most cases, polyps are benign, but between 0.5% and 13% can grow or become malignant. Postmenopausal women with hypertension and obesity are more likely to develop cancer (19)(20)(21). In perimenopausal patients with abnormal uterine bleeding, mRNA expression is also abnormal, in which the level of VEFG changes greatly, which is the most specific and effective factor in regulating abnormal uterine bleeding. VEGF expression can regulate itself or other secretory mechanisms to stimulate endothelial vascular proliferation and structure formation, which can promote vascular growth and improve vascular permeability. Repair the vascular endothelium in the bleeding uterus, accelerate the formation of vascular formation factors in the uterus, endothelial cells to form new blood vessels. And the endometrium after repair of capillaries, reduce uterine bleeding. The level of VEGF in perimenopausal patients with abnormal uterine bleeding is lower than that of normal people, so when bleeding, it cannot be regulated, forming a vicious cycle.
In this study, many of the herbs of tiaojing Hemostasis prescription have the function of supplementing qi, nourishing blood and stopping bleeding. For example, Sanqi powder has the effect of stopping bleeding, dispersing stasis and calming pain. Ejiao has the effect of nourishing blood (22). Tuckahoe has the effect of improving water permeability, invigorating the spleen and calming heart. Scutellaria charcoal has the effects of clearing heat, dampness, relieving fire and detoxifying (23). Perimenopausal AUB patients are often accompanied by varying degrees of anemia, so this prescription can repair the patient's uterus and replenish blood and qi.
The clinical treatment effect of the study group was better than that of the control group, and the adverse reactions were less than that of the control group, indicating that tiaojing Zhixue Prescription had a good effect on the repair of patients' uterus, and patients almost had no adverse reactions. Among them, the Traditional Chinese medicine of supplementing qi and nourishing blood plays a great role, possibly because it can reduce the generation of sex hormones in the uterus, which can reduce the stimulation of sex hormones to the uterus so that the patient's uterine bleeding volume can be significantly reduced, and the expression of VEGF in the uterus can be increased to repair it. The decrease in adverse reactions is due to the fact that drugs regulate the body during treatment. When the disease occurs, the patient's blood loss increases, and the patient's qi and blood will be lost, and the drugs in the prescription will coordinate with each other to regulate the qi and blood in the body. The level of sex hormone in the study group was also significantly lower than that before the treatment and lower than that in the control group, indicating that the regulation of jingjing hemostasis was obvious in the regulation of sex hormone in the patients, which may be to regulate and protect the secretion of sex hormone through the regulation and protection of the decline of the ovary to restore the normal level of sex hormone in the body. The expression level of VEGF in uterine tissues was rise less than the control group, because the experiment formula of patients with uterine bleeding after, to regulate the expression of VEGF, which makes the body of the expression of VEGF levels increase, and then to repair of the uterus, menstrual quantity gradually returned to normal, the thickness of the lining of the uterus to normal levels. The excessive increase of VEGF may lead to endometrial hyperplasia in patients, resulting in aggravation of the disease, so the treatment effect of the study group is better.
In the study of Martins et al. (24), it was detailed that abnormal uterine bleeding has many causes. Uterine polyps, adenomyosis, uterine malignant tumors and hyperplasia, ovulation dysfunction, are related to the disorder of sex hormones and abnormal mRNA expression in patients. Therefore, after understanding it, we can prevent the disease or carry out the symptomatic treatment. In the study of Jewson et al. (25), progesterone was involved in the regulation of abnormal uterine bleeding, and the regulation of PRA and PRB receptors played A role, and their increase in the body would lead to endometrial hyperplasia, which may deteriorate into endometrial cancer. Hormones and sex issues are very important in married life (26,27). All of these are consistent with the study of this paper and have theoretical help for the treatment of abnormal uterine bleeding in the perimenopausal period.
In conclusion, Tiaojing Zhixue Prescription can regulate the expression of sex hormones and mRNA in perimenopausal patients with abnormal uterine bleeding, and play a therapeutic effect. Downregulation of basic fibroblast growth factor increases cisplatin sensitivity in A549 non-small cell lung cancer cells. J Cancer Res Ther 2018; 14 (7)
==
Domain: Biology Medicine
|
[{"added": "2022-02-06T16:28:53.369Z", "created": "2022-02-04T00:00:00.000", "id": "246566785", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "http://www.cellmolbiol.org/index.php/CMB/article/download/4127/1947", "pdf_hash": "240db699319bb09e536874802810a948a6db06b0", "pdf_src": "Anansi", "provenance": "peS2o-0028.json.gz:2851", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "67b3a67cf3c00ac76d83d0c530d0b2a1ca1d5668", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
A Sequalae of Lineage Divergence in Staphylococcus aureus from Community-Acquired Patterns in Youth to Hospital-Associated Profiles in Seniors Implied Age-Specific Host-Selection from a Common Ancestor
The rapidly changing epidemiology of Staphylococcus aureus and evolution of strains with enhanced virulence is a significant issue in global healthcare. Hospital-associated methicillin-resistant S. aureus (HA-MRSA) lineages are being completely replaced by community-associated S. aureus (CA-MRSA) in many regions. Surveillance programs tracing the reservoirs and sources of infections are needed. Using molecular diagnostics, antibiograms, and patient demographics, we have examined the distributions of S. aureus in Ha’il hospitals. Out of 274 S. aureus isolates recovered from clinical specimens, 181 (66%, n = 181) were MRSA, some with HA-MRSA patterns across 26 antimicrobials with almost full resistances to all beta-lactams, while the majority were highly susceptible to all non-beta-lactams, indicating the CA-MRSA type. The rest of isolates (34%, n = 93) were methicillin-susceptible, penicillin-resistant MSSA lineages (90%). The MRSA in men was over 56% among total MRSA (n = 181) isolates and 37% of overall isolates (n = 102 of 274) compared to MSSA in total isolates (17.5%, n = 48), respectively. However, these were 28.4% (n = 78) and 12.4% (n = 34) for MRSA and MSSA infections in women, respectively. MRSA rates per age groups of 0–20, 21–50, and >50 years of age were 15% (n = 42), 17% (n = 48), and 32% (n = 89), respectively. However, MSSA in the same age groups were 13% (n = 35), 9% (n = 25), and 8% (n = 22). Interestingly, MRSA increased proportional to age, while MSSA concomitantly decreased, implying dominance of the latter ancestors early in life and then gradual replacement by MRSA. The dominance and seriousness of MRSA despite enormous efforts in place is potentially for the increased use of beta-lactams known to enhance virulence. The Intriguing prevalence of the CA-MRSA patterns in young otherwise healthy individuals replaced by MRSA later in seniors and the dominance of penicillin-resistant MSSA phenotypes imply three types of host- and age-specific evolutionary lineages. Thus, the decreasing MSSA trend by age with concomitant increase and sub-clonal differentiation into HA-MRSA in seniors and CA-MRSA in young and otherwise healthy patients strongly support the notion of subclinal emergences from a resident penicillin-resistant MSSA ancestor. Future vertical studies should focus on the surveillance of invasive CA-MRSA rates and phenotypes.
Introduction
Staphylococcus aureus is one of the leading causes of skin and soft tissue infections that are either uncomplicated, severe, or invasive in nature [1][2][3]. It is also a leading bacterial agent in septic pneumonia and other respiratory tract, surgical site, prosthetic joint, and cardiovascular infections, all leading to severe nosocomial bacteremia [4]. As such, S. aureus is the most dominant Gram-positive species that is specifically adapted to humans as commensals that rapidly turns into deadly pathogens in various clinical diseases. It precisely inhabits the human skin and anterior nares such as the axillae and groin with around~20% of nasal colonization in different populations [1]. S. aureus typically causes a wide range of skin infections, including impetigo, skin abscess, furuncles, wound infections, and septic shock syndrome [5]. This wide range of diseases is due to the large and variable arsenals of virulence factors such as surface proteins, degradative enzymes, cytotoxins, biofilms formations, and antibiotic resistance genes. This is in addition to an array of differential expression profiles of intrinsic chromosomal genes that are turned on by host microenvironmental conditions [6]. Thus, surveillance of infections becomes imperative for the distribution and frequency of S. aureus infections in hospitals.
One of the major reasons for the seriousness of S. aureus infections is the widely spread methicillin-resistant S. aureus (MRSA) lineages causing increased mortality, morbidity, and hospital stays, as compared to methicillin-sensitive S. aureus (MSSA) lineages [7]. Balkhy et al. (2016) [8] reported on the strategic planning of the Gulf Cooperation Council Center for Infection Control (GCC-IC) that has placed the emergence of antimicrobial resistance (AMR) on the top of its agenda since 2014. This was soon followed by the second round table discussion on the "One Health" concept from the Gulf Cooperation Council Countries ("Part Two: A Focus on Human Health") [9]. Nevertheless, despite enormous efforts, and as is the case globally, calls for regional MRSA surveillance programs were made especially with the emergence of strains that require no underlying risk factors to cause illness, as well as the propagation of chimeric resistance elements in both HA-MRSA and CA-MRSA [10]. . This preceded several reports on the rise of S. aureus resistant strains in the region. Most of the MRSA isolates were predominantly from skin, soft tissue, wounds, and nasal swabs [6,[11][12][13][14][15]. The antibiotics most commonly used for MRSA infections (skin and soft tissue infection) included fusidic acid, mupirocin, vancomycin, and clindamycin. Most resistances were found in the eastern region of Saudi Arabia compared to Riyadh and other cities [6,16,17], while resistance has increased by three times the usual in the UAE and Gulf countries, which suffered higher rates of resistance [17][18][19][20][21].
The origin of most deadly infections of S. aureus is a simple skin inhabitant variant of this species. The success of this highly contagious pathogen on skin and soft tissue infections depends on the elaborate adaptive mechanisms of virulence and immune evasion of host defenses. These include but are not limited to the local adaptive response to treatment and control options: for instance, production of specific toxins that destroy phagocytes such as leucocidins, which trigger phagocyte apoptosis, inhibition of complement factors, and inhibition of agglutination and the formation of thrombi. This way the resident skin S. aureus turns into a deadly pathogen in a four-stage mechanism bypassing initial immune response and neutrophils and surviving in blood, followed by abscess formation in few days, and finally launching a severe attack where variants mutate to persist [22,23]. The result of this is the emergence of highly successful strain variant from a skin reservoir of existing ones. For instance, MRSA genotypes with unique virulence, including a high prevalence of PVL and fusidic acid resistance in Kuwait hospitals was reported [24][25][26][27]. To avoid host Diagnostics 2023, 13, 819 3 of 15 immunity, S. aureus occupies intracellular shelter within host cells such as the phagocytes neutrophils and monocytes [28,29], as well as a series of non-phagocytic cells including epithelial and endothelial cells, keratinocytes, and osteoblasts [30]. Persistent in S. aureus in neutrophils is a virulence mechanism that serves as a reservoir leading to chronic as well as acute infections [31]. The rapid adaptational changes and strain differentiation into virulence come from genome plasticity, transfer of resistance, and an array of gene subsets in an accessory genome making this species one of the most contagious human-associated pathogens in history. This makes it imperative for multipoint local surveillance programs to establish rigid strategic control plans.
Methicillin-resistant Staphylococcus aureus is subdivided into lineages based on molecular typing (sequence type STs), genomic, and other data. There are a large number of new lineages that replaced old ones. Of particular concern is the progressive global replacement of hospital-associated MRSA (HA-MRSA) by community-associated MRSA (CA-MRSA) lineages. It has been widely known that MRSA was initially confined to hospitals and resistant to almost all types of beta-lactams; whereas, CA-MRSA is known to carry Panton-Valentine leucocidin and is susceptible to non-beta-lactams [32]. The CA-MRSA lineages are well known for their rapid community transmission, aggressive skin and soft tissue infections, and severe community-acquired necrotizing pneumonia. These lineages were top listed as significant threats since the previous MRSA pandemic that caused mortality rates similar to that of AIDS, tuberculosis, and viral hepatitis combined [2,[33][34][35][36]. At present, the evolution and emergence of invasive lineages of CA-MRSA, HA-MRSA, as well as MSSA causing bloodstream infections are being increasingly reported, leading to changes in global clonal profiles [7,37] with complete replacement of HA-MRSA by CA-MRSA occasionally in some cases. In China, since 2013, CA-MRSA strains have included ST59 largely replacing HA-MRSA ST239 [38]. This has also been the case in South Asia [39], in Africa [40], Australia, USA [41], India [42], and Canada [43].
The evolutionary lines of MRSA lineages have not been clear in the Middle Eastern countries. It has been shown that the European CC1-MRSA-IV appeared around 1995 and was widespread throughout Europe and into the Middle East [44]. Since then, the evolutionary lines of these lineages have not been clear and have been further complicated by the appearance of livestock lineages such as bovine mastitis lineage poultry lineage, and food-associated lineages in Arab countries [45]. The World Health Organization (WHO) labeled MRSA as one of the indicators for antimicrobial resistance in the Sustainable Development Goals connected to the health target 3.d [46]. Furthermore, the emerging MRSA lineages have become critically important in pediatrics targeting otherwise healthy young individuals. However, the rates, frequencies, and distribution of S. aureus lineages in hospitals are not adequately addressed owing to the paucity in high quality data on local strain profiles and infection patterns. Thus, this study aims to investigate the molecular profiles, antibiograms, and age-and gender-specific distributions of S. aureus lineages isolated from cutaneous infections and to establish a precise understanding of skin carriage as a source of transmission dynamics.
Microbiological Diagnosis, Antimicrobial Susceptibility Testing, and Patients' Demographics
Microbiological Diagnosis, antimicrobial susceptibility, and patient demographic data were all obtained from laboratory records, hospital medical records, and other sources in the hospitals. All records of clinical specimens from different hospital departments were processed for selection of non-duplicate isolates of Staphylococcus aureus from hospitals in Ha'il from September to December 2021. These were then subjected to subsequent diagnostic characterizations into methicillin-resistant S. aureus (MRSA) and methicillin-susceptible S. aureus (MSSA) isolates. For this, standard microbiological analysis and antimicrobial sensitivity testing, followed by molecular profiling of lineages, were conducted as follows.
For non-automated routine protocols, specimens were aseptically and professionally collected in suitable transport media, swabs, and/or media, send to the lab, and processed immediately or cultured for primary identifications using standard conditions and media incubated at 37 • C incubations for at least 18 h. Stock cultures of bacterial isolates were immediately kept in broth media at −80 • C for future reference and vertical studies. For automated protocols, specimens or cultures were concomitantly prepared and used for identification by automated testing and ID susceptibility testing using automated systems. Most the this phase was performed on the BD Phoenix system (BD Biosciences, Franklin Lakes, NJ, USA) and MicroScan plus (Beckman Coulter, Brea, CA, USA). When required, sensitivities were confirmed by in vitro cultures in agar diffusions interpreted by zone interpretive standards for this region. The susceptibility testing and breakpoint interpretive standards were carried out in accordance with the recommendations of the Clinical and Laboratory Standard Institute (CLSI document M100S-26) [47].
Resistance Classifications of MRSA Lineages Based on Standard Definitions for Classification: As Multi-Drug Resistant Bacteria (MDR)
Staphylococcus acquired resistance classifications are based on standard definitions for classifications that considers MRSA isolates as multi drug-resistant (MDR) for their methicillin resistance and resistance to beta-lactams except for the community-acquired lineages (CA-MRSA), since they are susceptible to beta-lactams. These definitions are according to the recommendations of the CDC and European Centre for Disease Control. The following definitions are usually accepted as standard according to Magiorakos et al., [48]: MDR= non-susceptibility to at least one agent in three or more antimicrobial categories; XDR = non-susceptibility to at least one agent in all but two or fewer antimicrobial categories (i.e., bacterial isolates remain susceptible to only one or two categories); PDR = non-susceptibility to all agents in all antimicrobial categories as reported.
We did not include known intrinsic resistances to particular drugs. Thus, MRSA criteria for defining S. aureus MDR classifications must include one or more of the following to apply: 1. hospital-acquired MRSA is always considered MDR by virtue of being an MRSA; 2. non-susceptible to ≥1 agent in >3 antimicrobial categories.
Molecular Detection and Characterization of S. aureus Lineages by Multi-Gene GeneXpert System
Simultaneous confirmation, identification, and molecular profiling of S. aureus directly from the specimen is carried out using the latest versions of the Cepheid GeneXpert ® Dx system with specific all-in-one cartilages of the SA Complete and MRSA assay kits, following manufacturers' recommendations and names and codes included in each kit. The system consists of an instrument, personal computer, and preloaded software for running tests and viewing the results. Depending on the kit used, assay definition files are imported into the software, such as the Blood Culture assay definition file. To start, when the computer is turned on the GeneXpert lunches automatically or is clicked to start, then log-on, create test, or orders. In the following steps sample patient IDs are scanned, and kit barcodes will re-populate boxes with Assay, Reagent Lot ID, Cartridge SN, and Expiration Date. Although some specimens like sputum require a few simple steps such as homogenizations and mixing before loading into cartridges, the general protocol for all kits is concise, automated, user-friendly, and highly robust. For example, Xpert MRSA/SA SSTI (skin and soft-tissue infections, as well as wounds, surgical infections) swabs from deep tissue, surgical site and wound infections were inserted into the sample reagent vial (break-in). Then, vortex well, particularly for samples with mucus contents or debris, dispense the sample into Port-S, insert the cartridge, and start the test. Multi-gene primers, probes, and reagents in kits allow for robust automated direct confirmation of S. aureus at species level with subsequent differentiation into S. aureus lineages directly from specimens. This is accomplished by the built-in primers for nuc spa, mecA and the mec (SCCmec) gene direct detections from specimens utilizing automated real-time polymerase chain reaction (PCR) in a single-use, disposable, self-contained cartridge with PCR reagents inserted and inoculated directly with swabs/samples. This method allows for minimizing laboratory media influence on S. aureus that is known to trigger sensing genes, leading to adaptive genome expressions and the emergence of different types. In addition, it reduces cross-contamination between specimens as well as cross-sequence contaminations in molecular tests. These are all remote since the cartridge is a disposable, closed, and self-contained kit. Furthermore, a sample processing control (SPC) and a probe check control (PCC) are also included. The SPC is present to control for adequate processing of the target bacteria and to monitor the presence of inhibitor(s) in the PCR reaction. The PCC verifies reagent rehydration, PCR tube filling in the cartridge, probe integrity, and dye stability.
Statistical Analysis
Collected data were analyzed using Statistical Package for Social Sciences software (IBM SPSS; Version 24 SPSS version 23.0 for Windows (SPSS, Inc., Chicago, IL, USA). Descriptive and stratified analysis were conducted; we present absolute numbers, proportions, and graphical distributions. We conducted exact statistical tests for proportions and show pvalues (based on Chi square test values) where appropriate (a p-value < 0.05 was considered statistically significant).
Discussion
The rapidly changing epidemiology of S. aureus leading to increased evolution of strains with enhanced virulence has been one the most significant public health and healthcare issues in this era. With increased human dynamics, new lineages constantly emerge crossing host barriers and making transmission dynamics between human-human, zoonotic and anthroponotic, and livestock transmissions such as poultry-and bovine mastitis-lineages. Consequently, wide-spread evolutionary changes are taking place
Discussion
The rapidly changing epidemiology of S. aureus leading to increased evolution of strains with enhanced virulence has been one the most significant public health and healthcare issues in this era. With increased human dynamics, new lineages constantly emerge crossing host barriers and making transmission dynamics between human-human, zoonotic and anthroponotic, and livestock transmissions such as poultry-and bovine mastitis-lineages. Consequently, wide-spread evolutionary changes are taking place in hospital strain profiles for the introduction and replacement by community-associated lineages. Middle Eastern countries, in particular Saudi Arabia, have been making significant developments as a global economic hub. This makes it imperative for understanding and monitoring local and global strain profiles associated with humans and livestock products.
In the Arabian Peninsula in general and Saudi Arabia specifically, MRSA became evident as a severe cause of serious health issues, with numerous studies reporting prevalence rates up to nearly 50% [49] and over [50]. In some instances, the prevalence and incidence rates are much higher often with serious consequences and poor patient outcomes. Despite enormous efforts in primary MRSA screening and containment protocols in place, surveillance reports on the profiles of these lineages in the Middle East are limited. In addition, under most common protocols and practices, the prevalence rate of MRSA shows high variation in regional distribution like in western (42%), central (32%), and eastern (27%) areas, respectively [51,52]. Thus, there is a serious paucity of high quality data on S. aureus across the region. In this study, of all 276 staphylococcal isolates, 183 (66%) were MRSA, which is quite high but falls well within prediction rates for the increasing frequencies since the previous report of 38% [53]. Usually, the rates are not constant, and they do differ with geographical distribution in different countries in the region. As the Kingdom of Saudi Arabia is a major economic country in the region and the world in addition to its holy Islamic sites annually visited by millions of people from all over the world, increasing rates of potential global lineages is well justified.
In the current study, the MRSA isolates showed an extreme resistance against imipenem (IMI), amoxicillin/clavulanic acid (AUG), cefotaxime (CTX), cefoxitin (FOX), and oxacillin (OX) at the rate of >82%. In addition, nearly all isolates were fully resistant to penicillin (P) at 99% and ampicillin (AMP) at 98%. This scenario is well known to be substantially associated with high morbidity and mortality due to severe hospital-associated methicillin-resistant S. aureus (HA-MRSA) in bloodstream and soft tissue infections. Mono-therapeutics with first-line antimicrobials as the glycopeptide vancomycin or the lipopeptide daptomycin are also associated with reduced susceptibilities and therapeutic failure, unless combined with first-line agents to improve β-lactam susceptibility in a seesaw effect [54]. Nevertheless, solid evidence exists against the use of beta-lactams either alone or in combination as they have been found to enhance MRSA virulence by the powerful universal gene-expression hub, the SarA gene family that code for proteins involved in quorum-sensing [55]. By virtue of being MRSA, the resistance reported here are not surprising as much as to the pattern of CA-MRSA resistance and the number of resistant isolates circulating.
We report on a high number of MRSA isolates susceptible to several non-beta-lactams often reaching to 100% susceptibility consistent with CA-MRSA pattern. These were tigecycline (TGC) 100%, linezolid (LNZ) 98.3% rifampicin (RD) 97.2%, teicoplanin (TEC) 97.2%, vancomycin (VA) 96%, levofloxacin (LE) 91%, tetracycline (TE) 89.3%, trimethoprim/sulfamethoxazole (SXT) 84.5%, and inducible macrolide (iMLS) 81%. This is in agreement with the previous studies that all MRSA isolates identified were susceptible to non-beta-lactams [56,57]. For this, vancomycin should be kept as the first choice for empiric treatment of MRSA with continued use of linezolid to be considered as the last resort. In line with previously conducted studies, the same trends have been reported by other authors [58,59]. However, in this study, high susceptibility was reported towards TGC, TE, SXT, and LE which is an improvement from previously reported resistances from the Middle East [60][61][62]. The changes reported in the spectrum of antimicrobial susceptibility of previously resistant patterns is potentially the outcome of stricter MRSA screening protocols in-pace and antimicrobial stewardship. This promising MRSA susceptibility indicated that rifampicin and sulfamethoxazole/trimethoprim could be a better empirical option in these regions. As this study was from Ha'il hospitals in the remote northern regions of Saudi Arabia, the risk of resistance transfer among Gram-positives is low and there is no report yet on the two globally emerging rifampicin resistant S. epidermidis lineages found in 24 countries that concomitantly reduce susceptibility to vancomycin and teicoplanin [63]. Thus, the full resistance to beta-lactam and susceptibility to non-beta-lactam antibiotics reported here is the typical pattern of CA-MRSA lineages [32]. This is further supported by the high number of these isolates from young, otherwise healthy patients, which is one property of this lineage. This makes it imperative for future large-scale surveillance of all outpatients and inpatients in a downstream molecular surveillance to identify sequence-type, clonal complex, pvl gene, and resistance cassette types [64]. Moreover, our isolates have shown more than 99% susceptibility towards linezolid (LNZ), nitrofuran (F), cefotaxime (CTX), cefoxitin (FOX), oxacillin (OX), teicoplanin (TEC), rifampicin (RD), vancomycin (VA), gentamicin (CN), tetracycline (TE), and sulfamethoxazole/trimethoprim (SXT).
About 98.9% of methicillin-susceptible Staphylococcus aureus in this study were penicillinresistant. However, they were 100% susceptible in vitro to the potentiated amoxicillin/clavulanic acid (AUG) and other non-beta-lactams such as daptomycin (DAP), mupirocin (MUP), imipenem (IMI), tigecycline (TGC). The increased use of clavulanate potentiated amoxicillin may have enriched penicillin resistance in methicillin-susceptible phenotypes in the present study. This observation has been previously reported in increased childhood nasal colonization of penicillinase producing methicillin-susceptible S. aureus [65]. Similarly, an epidemic outbreak of Panton-Valentine leukocidin (PVL) positive methicillin-susceptible S. aureus (MSSA) was reported in a maternity hospital that was initiated by postpartum mastitis and neonatal skin infections [66]. Due to biased sequencing of MRSA lineages, MSSA strains frequently receive less attention albeit they are associated with serious infections in humans. Genome sequencing of a highly virulent yet pan-susceptible MSSA isolate from a fatal case of sepsis and bacteraemia in a dengue patient revealed a novel combined genotype (t091/ST2990). A β-lactamase plasmid, staphylococcal enterotoxin, and enterotoxin-like genes were identified in addition to phylogenetic evidence of common ancestry with the European MRSA clone [67]. In the current study, higher susceptibility was seen for MSSA than MRSA to both beta-lactam and non-beta lactam antibiotics indicating the usefulness of continued surveillance in identifying susceptibility profiles. A 15-year retrospective surveillance at two tertiary care institutions in Boston, MA with 31,753 adult inpatients revealed S. aureus infection declined from 2000 to 2014 by 4.2%, due to an annual decline in MRSA of 10.9%. Consequently, penicillin-susceptible S. aureus (PSSA) increased by 6.1% annually, while the rates of methicillin-susceptible penicillin-resistant S. aureus (MSSA) did not change (10% to 11%; p value 0.43). Furthermore, 3/14 MSSA and 2/21 PSSA isolates arose from the loss of resistance-conferring genes. The decline in S. aureus infections has been accompanied by a shift toward increased antibiotic susceptibility [68]. Thus, constant surveillance and resistance programs are critical for the evolution of susceptible strains, empiric therapy, and combating invasive S. aureus.
Our result depicts high prevalence of MRSA among elderly patients with underlying risks aged >50 years (32%). Past studies presented considerably variable data regarding the distribution of MRSA in varied age groups. There are ample studies from Saudi Arabia and other Middle Eastern countries that showed MRSA common occurrence in elderly patients [56,69], and this is attributed to the common risk factors including age, co-morbidities, and long hospital stays. However, the frequency of infection reported in this study among otherwise healthy and mostly young patient groups, 0-20 years (15%) and 21-48 years (17%) with no underlying risk of comorbidity or hospitalization, is consistent with established host properties of CA-MRSA [70]. These were mostly male patients harboring 37% (n = 102 of 276) of total isolates followed by female patients at 28.4% (n = 78 of 276) with the ratio of 1.3:1. This may be attributed to the local differences in more male outdoor socializations than females in addition to the lack of hygiene practice [69,[71][72][73].
Conclusions
Despite enormous efforts and strict MRSA screening in pre-hospital admissions, the rates of the S. aureus lineages, particularly in men and senior patients with underlying risk, are still the highest. This is potentially due to increased use of beta-lactams known to enhance S. aureus virulence. Intriguingly, most of the isolates had CA-MRSA patterns with high susceptibility to non-beta-lactams and increased prevalence in young and otherwise healthy individuals. However, almost all MSSA phenotypes identified in this study were only penicillin-resistant. Taken together, we report on three S. aureus lineages, each with unique evolutionary dynamics and host-specificity, i.e., the decreasing trend of MSSA by age with the concomitant increase and sub-clonal differentiation into HA-MRSA in seniors and CA-MRSA in young and otherwise healthy patients. These profiles strongly imply age-specific evolutionary selection of these strains from a resident MSSA ancestor. Future vertical studies should focus on the surveillance of invasive CA-MRSA rates and phenotypes.
==
Domain: Biology Medicine
|
[{"added": "2023-02-24T17:04:00.924Z", "created": "2023-02-21T00:00:00.000", "id": "257138634", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2075-4418/13/5/819/pdf?version=1676984729", "pdf_hash": "b56ec2872fcc8ab7584808e316b21b8cdf16b761", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:2049", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "b42967571c2694a09323300fd9877d9cfcbcabf9", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Targeting CD47/SIRPα as a therapeutic strategy, where we are and where we are headed
Immunotherapy using PD-1 and CTLA4 inhibitors to stimulate T cell immunity has achieved significant clinical success. However, only a portion of patients benefit from T cell-based immunotherapy. Macrophages, the most abundant type of innate immune cells in the body, play an important role in eliminating tumor cells and infectious microbes. The phagocytic check point protein CD47 inhibits the phagocytic activity of macrophages through binding to SIRPα expressed on macrophages. Blockade of the interaction between CD47 and SIRPα could restore phagocytic activity and eliminate tumor cells in vitro and in vivo. In this manuscript, we review the mechanism of action and development status of agents (antibodies targeting CD47 and SIRPα, SIRPα-Fc fusion proteins, and bi-specific antibodies) that block CD47/SIRPα interaction in preclinical studies and in the clinical setting. In addition, small molecules, mRNA, and CAR-T/M that target the CD47/SIRPα axis are also reviewed in this article.
Introduction
Tumor cells evade immune destruction by transmitting inhibitory signals to lymphocytes and myeloid cells [1]. Blockade of these inhibitory molecules, which include CTLA4, PD-1, and PD-ligand 1 (PD-L1), could restore T cell function and promote elimination of tumor cells. Immune checkpoint inhibitors (ICIs) have improved outcomes for patients with multiple types of cancers. However, many patients do not respond to this type of immunotherapy [2,3]. In some cases, these therapeutic agents have been associated with disease progression [4,5], the cause of which is currently being investigated. Therefore, drugs that act on a novel class of targets to immobilize a broader immune cell population are needed to improve upon current therapeutic options.
Macrophages [6] are typically the first dedicated innate immune cells to detect the presence of infectious pathogens or tumor cells. Macrophages are derived from monocyte precursors that circulate in the blood and migrate into tissues, after which they differentiate into tissue macrophages such as Kupffer cells in the liver, alveolar macrophages in the lung, and microglia in the brain. Circulating monocytes and resident tissue macrophages can directly kill tumor cells via phagocytosis (innate immune response) and can activate the adaptive immune response. However, these immune responses can be inhibited by ligand binding to inhibitory receptors expressed on the macrophage cell surface [7]. Signal regulatory protein alpha (SIRPα) is a transmembrane protein expressed on all myeloid cells, including monocytes, macrophages, and neutrophils. SIRPα contains immunoreceptor tyrosine-based inhibition motifs (ITIMs) that can be phosphorylated, resulting in recruitment of inhibitory molecules such as Src homology 2 (SH2) domain-containing protein tyrosine phosphatase (SHP)-1 and SHP-2 [8]. Binding of CD47 to SIRPα triggers coupling of SIRPα to these phosphatases, resulting in inhibition of phagocytic activity [9,10]. CD47 is ubiquitously expressed on many types of cells to prevent phagocytosis by phagocytes. However, tumor cells overexpress CD47 to evade the immune system through inhibition of myeloid cell-mediated elimination [11]. Inhibition of CD47-SIRPα interaction restored the phagocytic activity of phagocytes in vitro and in vivo [12][13][14][15]. Targeting the CD47/SIRPα axis has become a promising strategy to promote tumor elimination through innate immunity. This review focuses on development, safety, and efficacy of agents that target the CD47/SIRPα axis in preclinical and clinical studies.
CD47/SIRPα: the molecules and biology
SIRPα [16,17], also named SHPS-1 or CD172a, is a transmembrane glycoprotein mainly expressed on neurons and myeloid cells that is particular enriched on macrophages. Human SIRPα is coded by the SHPS-1 gene located at human chromosome 20p13. The open reading frame region is composed of eight exons, including a signal peptide, extracellular domain, a transmembrane segment, and three parts of one cytoplasmic domain. The extracellular domain consists of three Ig-like regions, an NH2-terminal immunoglobulin (Ig) variable (V) region (domain 1, D1), and two Ig constant (C) regions (domain 2 and 3). The cytoplasmic region contains two immunoreceptor tyrosine based inhibitory motifs (ITIMs) and a proline-rich region (YYYY), which bind to Src homology (SH2) domain-containing molecules.
CD47 is a 52 kD transmembrane glycoprotein belonging to the immunoglobulin superfamily. Human CD47 is encoded by the CD47 gene located at the q13.12 region of chromosome 3. Human CD47 contains an NH 2 -terminal Ig variable-like extracellular domain (ECD), a 5-transmembrane spanning helical bundle domain, and a short intracellular COOH-terminal domain (CTD) [18]. CD47 is an essential component of the innate immune system, and binding of its extracellular domain with its ligands αVβ3, SIRPα, and thrombospondin-1 (Tsp-1) activates different signaling pathways that control cell proliferation and differentiation, angiogenesis, and immune regulation. The CTD is alternatively spliced and can exist as four isoforms, ranging from 4 to 36 residues. A schematic diagram of the compositions of CD47 and SIRPα proteins are shown in Fig. 1.
Therapeutic strategy
The critical role of the CD47/SIRPα axis in the innate immune response suggests that these two proteins may be attractive therapeutic targets. Antagonists targeting the innate immune checkpoint CD47/SIRPα pathway are currently in clinical development. These antagonists include 1) monoclonal antibodies targeting CD47 or SIRPα, 2) SIRPα-Fc fusion proteins, 3) bispecific antibodies (BsAb), 4) small molecules to down-regulate CD47 on tumor cells, 5) RNAi and, 6) CD47-chimeric antigen receptor-T cell/Macrophages.
Monoclonal antibodies and fc fusion proteins
Three types of agents targeted to the CD47/SIRPα axis were developed: antibodies, SIRPα-Fc fusion proteins targeted to CD47, and antibodies targeted to SIRPα. The mechanisms of CD47-SIRPα blocking agents are summarized in Fig. 3. Agents targeted to CD47 should block the CD47-SIRPα interaction to remove the anti-phagocytic signal and restore the phagocytic activity of macrophages [27]. In addition, engagement of FcRs to limit activity is considered to be necessary for agents targeted to CD47 [28]. In addition, anti-SIRPα antibodies using inert Fc to prevent toxicity resulting from SIRPα expressed on myeloid immune cell perhaps have therapeutic potential.
The second therapeutic mechanism is bridging innate and acquired immunity [29]. Tumor cells are recognized, taken up by antigen presenting cells (APCs, such as dendritic cells and macrophages), and presented to naive T cells, resulting in T cell activation. Antibodies targeted to CD47 could induce direct killing of tumor cells by inhibiting protein kinase A via Giα, resulting in clustering of CD47 in the membrane, and caspase-independent programmed cell death [30].
Anti-CD47 monoclonal antibodies
More than ten anti-CD47 antibodies are in different phases of clinical development ( Table 1). All of these antibodies are based on human IgG4-Fc, except AO-176, which is based on human IgG2-Fc [31]. Clinical study results of these antibodies were as follows (ordered by clinical trial developmental state).
Magrolimab (Hu5F9-G4) Magrolimab, previously known as Hu5F9-G4, was the first anti-CD47 antibody to enter clinical trials, and is currently in Phase III development. The clinical trials for magrolimab are listed in the Table 1. In a preclinical study [32], combined treatment with magrolimab and rituximab showed significant clearance of Raji cells in vitro and elimination of AML tumor cells in vivo. Magrolimab blocks the CD47/ Fig. 2 Biology of CD47/SIRPα interaction. CD47 binds to SIRPα to transmit inhibitory signals to macrophages and to inhibit or lessen phagocytic activity through uncoupling of receptor binding and signal transduction. Therapeutic agents block the interaction between CD47 and SIRPα to remove the inhibitory signal and restore the phagocytic activity SIRPα interaction and the anti-phagocytic signal on macrophages, and binding of rituximab to FcRs initiates pro-phagocytic signaling. Activation of pro-phagocytic signaling is a beneficial effect of anti-CD47 antibody treatment. Magrolimab caused significant hemagglutination and phagocytosis of RBCs in vitro, which indicated potential toxicity. Non-human primate pharmacokinetic and toxicology studies showed dose-dependent anemia. To overcome treatment-related anemia and thrombocytopenia, a low priming dose was given to stimulate production of new RBCs and to facilitate tolerance of subsequent higher maintenance doses. The same strategy (1 mg/kg priming dose on day 1) effectively controlled anemia during subsequent infusion of magrolimab in a clinical trial. The saturation concentration (receptor occupancy) on circulating white and red cells was 30 mg/ kg [33]. In a phase Ib [34] (NCT02953509) study evaluating relapsed or refractory non-Hodgkin's lymphoma, patients with diffuse large B-cell lymphoma (DLBCL) or follicular lymphoma were treated with magrolimab in combination with rituximab. The results showed that 50% of the patients had an objective (i.e., complete or partial) response, with 36% having a complete response. Specifically, the rates of objective response and complete response were 40 and 33%, respectively, among patients with DLBCL, and 71 and 43%, respectively, among those with follicular lymphoma. Calreticulin (CRT) is a member of the endoplasmic reticulum lectin chaperone family of proteins that plays important biological roles in Ca 2+ homeostasis (in the endoplasmic reticulum, ER), integrin-dependent cell adhesion (in the cytoplasm), and immune response activation (on the plasma membrane) [35]. CRT translocates from the ER to the cell surface during immunogenic cell death in response to various stress factors such as chemotherapy, irradiation, photodynamic therapy, and cytokines. CRT on the surface of stressed or dying cells acts as an "eat me" signal by binding to LRP1 on macrophages [36].
Azacitidine is a chemotherapeutic drug that could induce ICD through translocation of CRT to the cell surface. Interaction of CRT with LRP1 transmits "eat me" signals to macrophages and promotes phagocytosis of dying cells [37]. Combined treatment with magrolimab and azacitidine in a preclinical study resulted in significantly increased phagocytic activity in vitro, and elimination of HL60 in vivo [38]. Preliminary data from a trial (NCT04778397) of magrolimab combined with azacitidine for treatment of AML resulted in a 65% OR and a 40% CR in 34 patients. Transient on-target anemia was observed, and 56% of patients with AML became red blood cell transfusion independent in response to this Lemzoparlimab (TJC4) Lemzoparlimab (TJC4), from I-MAB, is a human IgG4 antibody targeted to CD47 that was screened using a phage display system. The crystal structure of the TJC4/CD47 complex (straighter headto-head orientation) showed that TJC4 binds to a different epitope than does magrolimab (tilted head-to-head orientation) [42]. In addition, TJC4 showed minimal binding to RBC and no hemagglutination was observed at 100 μg/ml in vitro. No significant erythrocyte toxicity was observed in cynomolgus monkeys dosed with 10-100 mg/kg QW. A phase I trial (NCT03934814) [43] evaluating treatment of R/R advanced solid tumors and lymphoma with TJC4 alone, or in combination with pembrolizumab or rituximab, is ongoing. No DLT or SAE (severe adverse events) were observed, and treatmentassociated anemia occurred in 30% of patients (6 of 20, 2 receiving 3 mg/kg, 2 receiving 10 mg/kg, 1 receiving 20 mg/kg, and 1 receiving 30 mg/kg). Anti-drug antibody (ADA) events occurred in 25% of patients, but there were no concerns regarding safety or pharmacokinetics (PK). Maximal saturation of peripheral T cells was achieved at a dose of 20 mg/kg administered weekly.
AO176 AO176, developed by Arch Oncology, is a humanized IgG2 subclass anti-CD47 antibody [31]. AO176 was shown to bind to integrin-β1 expressed on tumor cells, but not on RBC. Interestingly, AO176 blocked the CD47/SIRPα interaction to stimulate phagocytosis of tumor cells, and also directly killed tumor cells (non-ADCC). A tolerability and hematologic study on Cynomolgus monkeys showed no anemia. In phase I/ II clinical trials [44], Grade 3 TRAE (treatment-related adverse events) were observed in > 10% of patients. In addition, DLT was observed at 20 mg/kg. Studies evaluating AO176 monotherapy and combination therapy with paclitaxel are ongoing.
CC-90002 CC-90002 is a humanized monoclonal IgG4 CD47 antibody. An early clinical trial (NCT02641002) that evaluated treatment of Acute Myeloid Leukemia (AML) and high-risk myelodysplastic syndrome (MDS) was terminated due to poor activity and safety profiles [45,46]. Another phase I trial (NCT02367196) evaluated treatment of advanced solid (alone) and hematologic malignancies (in combination with rituximab). In a combination trial with patients with R/R NHL [47], the ORR (overall response rate) was 13% with 25% achieving stable disease. However, 50% had anemia (of any grade), 33% had thrombocytopenia, and DLTs were observed in 3 subjects (1 subject infusion-related reaction at 15 mg/ kg Q2W and 2 subjects had grade 3 thrombocytopenia requiring platelet transfusion occurring at 30 mg/kg).
SGN-CD47M
SGN-CD47M, developed by Seagen (Seattle Genetics), is a CD47 targeting probody drug conjugate (PDCs). Probody therapeutics [48] are antibody prodrugs designed to remain inactive until proteolytically cleaved and activated in the tumor microenvironment. Probody drug conjugates [49] can be activated by multiple proteases in the tumor microenvironment, but remain inactive in the circulation and in normal tissues.
The efficacy of PDCs depends on multiple factors including binding affinity and specificity for the antigen, efficiency of cleavage in the tumor microenvironment, lack of cleavage in normal tissues, and internalization efficiency. Studies focused on PDCs are in the early clinical stage, and safety and efficacy have yet to be determined. Clinical trial (NCT03957096) of SGN-CD47M for treatment of advanced solid tumors was terminated based on portfolio prioritization.
Other drugs in early clinical development for safety and dosing evaluation are listed in Table 1.
SIRPα-fc Among the 10 allelic variants of SIRPα, SIRPα V1 and V2 are the most prevalent variants [50]. The affinity of wild type SIRPα binding to CD47 is in the micromolar range, which is 1000-fold weaker than that of anti-CD47 antibodies. Soluble SIRPα binding to CD47 on tumor cells could block inhibitory signals and enhance phagocytic activity. Six SIRPα-fusion proteins are currently in phase I or phase II clinical trials (Table 2, ordered by clinical trial developmental state).
Evorpacept (ALX148)
Evorpacept is comprised of SIRPα variant 1 domain 1 (V1D1) and inactive human IgG1-Fc [51]. Evorpacept (CV1) was selected from mutant libraries and includes 9 mutations (V6I, A271, I31F, E47V, K53R, E54S, H56P, L66T, V92I), which resulted in a 50,000-fold increase in affinity compared with that of wild type SIRPα. Preclinical data showed that evorpacept augmented macrophage antitumor activity in vitro and in vivo in combination with tumor-opsonizing antibodies (trastuzumab, obinutuzumab and cetuximab). However, no phagocytosis or antitumor activity was observed following treatment with evorpacept alone. A trial [52] in which patients with NHL received evorpacept alone at 10 mg/kg or at 15 mg/kg in combination with rituximab resulted in ORRs of 40.9 and 63.6%, respectively. The CD47 receptor occupancy on RBC and CD4 T cells was approximately 90% at 10-15 mg/ kg. A phase I study [53] of evorpacept in combination with pembrolizumab, trastuzumab, or zanidatamab, and/or chemotherapeutic agents, evaluating treatment of advanced solid malignancy is ongoing. Preliminary results showed anti-cancer activity of evorpacept in combination with pembrolizumab (AP) and/or chemotherapy (5FU + platinum) in patients with second line or greater HNSCC (head and neck squamous cell carcinoma) with prior platinum therapy. The ORR in patients with checkpoint inhibitor-naïve HNSCC (n = 10) treated with AP was 40%, but 0% in patients with HNSCC who had previously received checkpoint inhibitors (n = 10). A phase II study of evorpacept in combination with pembrolizumab for treatment of HNSCC was recently initiated. Treatment with evorpacept in combination
Antibodies targeting SIRPα on myeloid cells
Anti-SIRPα antibodies induce weak or no phagocytic activity alone, but induce significantly increased phagocytic activity when combined with opsonizing antibodies (rituximab, cetuximab) [58][59][60]. Several issues with anti-SIRPα antibodies are important to consider. First, since SIRPα is expressed on myeloid cells, anti-SIRPα antibodies using inactive human IgG-Fc to avoid Fc effector-mediated toxicity on these immune cells may be advantageous [60]. Second, SIRPγ expressed on T and NK cell shares 74.37% amino acid similarity with the extracellular domain with SIRPα [61]. SIRPγ on T cells binds to CD47 on APCs to mediate cell-cell adhesion and enhances antigen presentation, resulting in T cell proliferation and cytokine secretion [62]. Development of a SIRPα targeting antibody with specificity toward SIRPα to avoid interference with the interaction between CD47 and SIRPγ may preserve T cell activity. Third, antibody internalization could lead to rapid clearance of antibody in vivo. Higher doses or multiple dosing is required to ensure that levels remain therapeutically relevant [63,64]. Internalization of SIRPα decreases the inhibitory signal and may enhance the ability of antibodies to restore phagocytic activity [59,65]. Finally, Although SIRPα has ten known variants, V1, V2, and V8 are the most prominent (over 90%) haplotypes in the human population. Antibodies targeting all three of these variants may be more potent than those that target a single variant [66,67]. Antibodies targeting SIRPα currently being evaluated are listed in the Table 4. Trials evaluating treatment with SIRPα antibodies in combination with immunotherapies are in the early clinical stage of development, and include OSE-172 [68] from OSE Immunotherapeutics (co-developed with Boehringer Ingelheim), CC-95251 from Celgene, and FSI-189 from Gilead [59].
Bi-specific molecules
Bi-specific molecules bind two targets or two distinct epitopes of one target. The antigen binding sites of bispecific molecules could consist of two antibodies or proteins (ligand or receptor), or could consist of one antibody and one protein. Bi-specific antibodies can bind two target antigens in-cis and in-trans. Rational design of bispecific molecules based on biological activity may result in distinct effects or improved efficacy when compared to combination treatments. Four bispecific antibodies, catumaxomab [69] (withdrawn in 2017), blinatumomab [70], emicizumab [71], and amivantamab-vmjw [72] have been approved by EMA or FDA. Bi-specific molecules can be constructed from CD47 targeting antibodies or SIRPα and other antigen-targeting molecules (Fig. 4). Target antigens could include: A) tumor associated cell surface antigens (PD-L1, CD20, CD19, MSLN (Mesothelin), Claudin18.2, and Her2), B) immune checkpoint proteins (PD-1, CD40, 41BB), and C) cytokines or receptors (CSF-2 receptor, VEGF). Reduced affinity for CD47 and increased affinity for the second target may reduce toxicity and enhance efficacy. For type A antibodies, IgG1-Fc was selected to enhance antibodymediated killing of tumor cells (ADCC, ADCP, and CDC). However, inactivated Fc is preferred when used in type B bi-specific molecules. Use of CD47-targeting biologics has shown significant clinical efficacy for treatment of R/R AML, NHL, and MDS. The combination of Hu5F9 and rituximab showed particular promise as a treatment approach for R/R NHL [34]. Several CD47related BsAb are in early-stage clinical trials. NI1701 [73], which targets CD47 and CD19, is an IgG-like BsAb constructed using modified knobs-into-hole technology [74], and contains an IgG1-Fc. Preclinical data showed that NI1701 [75] selectively binds to CD47 and CD19 co-expressing cells, but interacts poorly with normal healthy cells (CD47 + CD19 − ), resulting in avoidance of normal cells acting as sinks for binding of the antibodies, and reduced toxicity. In vitro and in vivo studies showed that NI1701 more potently killed tumor cells than did anti-CD47 and anti-CD19 antibodies alone, or in combination. A BsAb named NI1801 [73] targeted to CD47 and MSLN showed similar preclinical activity as NI1701. IMM0306, a bispecific antibody fusion protein targeted to CD20 (rituximab) and CD47 (SIRPα) with wild type IgG1-Fc [76], is in a phase I trial evaluating treatment of R/R CD20-positive B-cell non-Hodgkin's lymphoma. SL-172154 [77] is a fusion protein targeted to CD47 with SIRPα and CD40 with CD40L, and is in a phase I trial for treatment of solid tumors. A BsAb, HX009, comprised of the extracellular region of SIRPα V2D fused with an anti-PD1 antibody (HX008) was evaluated in a phase I trial. a antibody in clinical trial The study showed that HX009 blocked both CD47/SIRPα and PD-1/PDL-1 interactions, and interacted with CD47 on tumor cells and PD1 on T cells to help present tumor antigens to T cells, resulting in activation of the innate and acquired immune responses. Several BsAbs have been developed to inhibit CD47/SIRPα and PD-1/PD-L1 interactions. One such BsAb, IBI322 [78], was designed as a selective CD47 binding CD47/PD-L1 bispecific antibody. IBI322 showed no negative hematological effects in cynomolgus monkeys, but the binding affinity of IBI322 to cynomolgus CD47 was not disclosed, and IBI322 showed dose-dependent binding to RBC. Other BsAb antibodies listed in Table 2 are in the preclinical proof-ofconcept stage.
Engineered T cells and macrophages
T cells with chimeric antigen receptors (CARs) showed promising therapeutic efficacy against hematologic malignancies, and several CAR-T therapeutics have been approved [79][80][81][82][83]. However, low treatment response rates of solid tumors to CAR-T treatment were observed. Golubovskaya et al. [84] showed that CD47-CAR-T cells effectively killed ovarian, pancreatic, and other cancer cells, and induced production of high levels of IL-2, which correlated with expression of CD47 antigens. Treatment with CD47-CAR-T cells may be a novel strategy for treating different types of cancers. Huyen [85] designed a third generation of CD47-CAR-T cell that could effectively kill lung cancer cells (A549) and inhibit lung cancer cell metastasis. A dual CAR-T targeting CD47 and TAG-72 (tumor-associated glycoprotein 72) generated by Shu [86] showed promising results against ovarian cancer in preclinical experiments. An Anti-PD-L1 (A12) CAR-T with the ability to secrete anti-CD47 VHH (variable heavy domain of heavy chain antibodies or nanobodies) (A4), developed by Xie [87], represented a novel strategy for cancer treatment. An A12-A4 CAR-T showed better anti-cancer activity than an A12 CAR-T plus soluble A4 in C57BL/6 PD-L1-KO mice bearing B16F10 cells. Each of these CD47-CAR-T cells are in the preclinical stage, and efficacy and safety should be further investigated. Development of CAR-M1 (M1: classically activated macrophages) is an emerging therapeutic strategy [88], and several studies of engineered CAR-M cells [89][90][91] showed tumor cell elimination activity in vitro and in vivo. These studies showed that CAR-M induced phagocytosis and induced M2 to M1 polarization through secretion of pro-inflammatory factors and chemokines [90]. CD47 is ubiquitously expressed on the surfaces of multiple hematopoietic and solid tumor cells. However, adverse events due to cytokine secretion by macrophages during immune checkpoint activation [92,93] and CAR-T [94,95] treatment was common. Technological improvements for preparation and production of CAR-M are needed to offer scalable and reproducible manufacturing processes.
Small molecules, peptides, and microRNA
RRx-001 [96] is an anticancer agent designed to induce M2 to M1 polarization and to promote recovery of phagocytic activity of macrophages toward tumor cells.
The anti-phagocytic inhibitory signal was removed or reduced through downregulation of both CD47 and SIRPα gene expression on tumor cells and macrophages, respectively. Elimination of tumor cells was shown in in vitro and in vivo. Phase III clinical trials (NCT03699956, NCT02489903) against small cell lung cancer [97,98] are ongoing.
D4-2 [99], a macrocyclic peptide targeted to mouse SIRPα was designed to inhibit the interaction between CD47 and SIRPα and promote macrophage-mediated phagocytosis of tumor cells when combined with rituximab. PKHB1 [100], a TSP-1-derived CD47 agonist peptide, induced cell death (CRT exposure and DAMP release) in chronic lymphocytic leukemia cells.
MicroRNAs (miRNAs), which are 20-22 nucleotides in length, play important roles in cancer pathogenesis and progression since they can repress the target gene at the translational level by directly binding to the 3'untranslated regions (3'UTRs) [101].
Overexpression miR-378a [102] in mice peritoneal macrophages downregulates SIRPα mRNA expression. Phagocytosis of Ishikawa cells by macrophages-miR-378a and macrophages was carried out in vitro. Phagocytic index in macrophages-miR-378a group is 3 times than that in macrophages group.
Zhao [103] reported that miR-200a inhibited the expression of CD47 by directly targeting the 3'UTR of the CD47 mRNA. MicroRNA 200a suppressed nasopharyngeal carcinoma (NPC) cell proliferation, migration, and invasion, and promoted phagocytosis of NPC cells by Fig. 4 Types of bis-specific molecules targeted CD47 and other molecules. Bispecific Ab-type 1: KIH format, targeted to CD47 (yellow) and other tumor associated antigens (CD19, CD20, and MSLN, blue); Bi-specific Ab-type 2: KIH format, targeted to CD47 (yellow) and immune checkpoint molecules (PD-1, CD40, and 41BB, green); Bi-specific Ab-type 3: SIRPα (cyan) fusion to N-terminus of H-chain of IgG (targeting tumor associated antigen, blue); Bi-specific Ab-type 4: ligand/receptor to modulate TME (GM-CSF, and VEGFR2, pink) fusion to the C-terminus of the H-chain of IgG (yellow); Bi-specific Ab-type 4: SIRPα fusion to the N-terminus of IgG-Fc and ligand/receptor to modulate TME (GM-CSF, VEGFR2, pink) fusion to the C-terminus macrophages through down-regulation of CD47 expression on NPC cells.
MicroRNA 708 [104] was directly targeted CD47 and resulted in downregulation of CD47 on T cell acute lymphoblastic leukemia cell line. MicroRNA 708 expression in the T-ALL cell line was sufficient to promote phagocytosis by macrophages in vitro, and inhibited tumor engraftment in vivo.
Conclusions and future perspectives
Following the clinical success of therapeutic antibodies targeting T cell checkpoint molecules, combination therapies using checkpoint inhibitors with other agents have been a major theme of clinical oncology studies. Specifically, inhibitors of the CD47/SIRPα pathway have emerged as promising therapeutic candidates. Overexpression on tumor cells makes CD47 an ideal target for cancer therapy. Antibodies targeting CD47 showed promising results against MDS and AML [105][106][107]. However, side effects such as anemia, hyperbilirubinemia, thrombocytopenia, and lymphopenia induced by CD47-targeting molecules are of specific concern and need to be addressed with development of new therapeutic agents.
In addition, the therapeutic effects of agents targeting the CD47/SIRPα axis on solid tumors are limited. Additional therapeutic strategies, including combination therapy and bi-specific antibodies, may be promising. Combination therapy with opsonizing antibodies, immune checkpoint inhibitors, chemotherapeutic agents to activate FcR on macrophages, and T cell sensitizers that induce immunogenic cell death to stimulate a more potent immunological effect all have potential. Agents targeted to the CD47-SIRPα axis should not only block the CD47/SIRPα interaction, but also activate signaling on macrophages (FcγR, CRT expression). Other agents including small molecules, mRNA, and CAR-T/M that block the CD47/SIRPα interaction are also in development. Many promising strategies targeting the CD47-SIRPα axis are in development and offer a great deal of hope to patients with cancer.
==
Domain: Biology Medicine
|
[{"added": "2023-01-30T15:21:45.616Z", "created": "2022-04-13T00:00:00.000", "id": "256365631", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://biomarkerres.biomedcentral.com/track/pdf/10.1186/s40364-022-00373-5", "pdf_hash": "99f6734d484f4d8a1d261563e60bdf25f05efc14", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:121", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "99f6734d484f4d8a1d261563e60bdf25f05efc14", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Predictive Value of The G1–G6 Transcriptomic Molecular Classication of Hepatocellular Carcinoma for its Biological Behavior and Clinicopathological Features in China
Background: Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide and a large number of genetic alterations are involved in the carcinogenetic process. A G1–G6 transcriptomic classication was previously proposed in a French study, and Korean and Singaporean groups indicated its potential application in Asian HCC patients. However, the genomic proles of Chinese patients are distinct from patients of other regions, and therefore the suitability of this method in Chinese HCC patients has remained unknown. Materials/Methods: In this study, we tested the transcriptomic group classication from the French cohort on a cohort of HCC patients from China. a total of 107 HCC cases from China were selected for the G1– G6 transcriptomic molecular classication. The correlation between the G1–G6 molecular classication and clinicopathological features were analyzed. RNA sequencing and bioinformatics analysis were performed to screen related targets and molecular signaling pathways. Results: We investigated the G1–G6 signatures in 107 Chinese HCC patients. HCC cases from China (n=107) were distributed as follows: G1 (17.76%), G2 (1.87%), G3 (18.69%), G4 (9.35%), G5 (23.36%), and G6 (28.97%) groups. We observed concordance between the genetic proles and clinical features of Chinese HCC patients and French HCC patients. We found that the G1–G3 subgroups were associated with high serum alpha-fetoprotein (AFP) level, high copy number of hepatitis B virus (HBV) DNA, complex histopathological structure, macrovascular invasion, negative or weak Hep-Par1 expression, programmed death-ligand 1 expression, and liver cancer stemness. The G1 subgroup was mainly related to liver cancer stemness, and G3 subgroup showed the worst prognosis. The G5 and G6 subgroups were associated with activation of the Wnt/β-catenin pathway. Compared with the G1–G4 group, the G1–G3 group showed signicantly higher expression levels of regenerating family member 1 beta (REG1B), regenerating family member 3 gamma (REG3G), and inositol 1,4,5-trisphosphate receptor type 1 (ITPR1), and enriched calcium signaling pathway. Conclusions: Our results clarify the correlation between G1–G6 molecular classication and molecular markers and molecular signaling pathways in the Chinese HCC population and initially established a link between the phenotype and molecular characteristics. This study enhances our understanding of the heterogenicity of China HCC and indicates that the G1–G6 signatures can be used to identify potential therapeutic biomarkers against HCC patients in China. was used to analyze the correlation between the G1–G6 molecular classication and tumor biological behavior. After checking data for normal distribution and variance homogeneity, comparisons between two groups were evaluated using independent-sample t-tests or Mann–Whitney U test. For three or more groups, differences were statistically analyzed by one-way ANOVA or the non-parametric test (Kruskal– Wallis test). Kaplan–Meier and Cox proportional hazards survival regression analysis were used to evaluate the prognostic signicance of the G1–G6 molecular classications within 36 months. All p values were two-tailed, and P < 0.05 was considered statistically signicant.
Background
Hepatocellular carcinoma (HCC) is the most common malignancy worldwide, with more than half of the new HCC cases and deaths every year occurring in China [1]. The main causes of HCC include chronic hepatitis B virus (HBV) or hepatitis C virus (HCV) infection, alcoholic or non-alcoholic steatohepatitis, autoimmune hepatitis, and several metabolic diseases. However, the etiologies of HCC between the Euro-American area and Asia-Paci c region vary widely. HBV infection was endemic in China, where the prevalence rate of HBsAg is 7.2% [2], while HCV infection is common in western countries. In China, HCC only has a 5-year survival rate of 14.1% and a recurrence rate of about 70% [3]. Approximately 70-80% of HCC patients are diagnosed at an advanced stage and can receive only palliative care [4]. Although several promising drugs were developed over the last decades, these drugs have failed to meet clinical endpoints in phase III trials [5]. The failure of these trials is, at least in part, because of the lack of effective molecular markers or the minimal validation of known molecular markers in different populations.
Recently, researchers have been attempting to establish a molecular classi cation for HCC, which is prognostically informative, to accurately identify patients after curative resection who will bene t from additional early therapeutic interventions to prevent a recurrence [6,7]. Several studies from China focused on the correlation between molecular markers and the differentiation degree and invasion ability of HCC.
However, few studies have investigated the correlation between HCC molecular classi cation and pathological characteristics. A HCC classi cation method based on the expression of 16 genes was proposed by Boyault et al [8]. The authors classi ed tumors into six groups, G1-G6, using a minimal subset of 16 genes. This transcriptomic group classi cation was tested on cohorts from Singapore [9] and South Korea [10] with comparisons to European HCC patients. The results from these groups were generally in line with the results of Boyault et al, but there were still many inconsistencies. The Singapore study reported that the G1 subgroup had a higher proportion of patients with HBV in Singapore compared with European samples. HBV infection was found in the G2 subgroup of European HCC patients but not found in the G2 subgroup of Singapore HCC patients. In addition, the G6 subgroup was closely associated with satellite nodules in the European HCC population but not in the Singapore HCC population. The Singapore study also found no associations of clinical features with G4-G6 subgroups in the Southeast Asia HCC population. The probability of microvascular invasion (MVI) in the G3 subgroup was 5 times higher than other subgroups in Singapore samples. These differences suggested that the G1-G6 transcriptomic classi cation is not completely applicable to HCC patients in all regions of Asia. Furthermore, the genomic pro les of HCC patients are distinct in populations from China and other countries. For example, the TP53 mutation rate in Chinese populations is signi cantly lower than that in Korean and Singapore populations [11,12]. Therefore, it is valuable to validate the clinical relevance of the 16 gene HCC classi cation method in a Chinese population with HCC.
In this study, to determine whether the HCC molecular typing method established by the European team applies to Chinese HCC patients, a total of 107 HCC cases from China were selected for the G1-G6 transcriptomic molecular classi cation. The correlation between the G1-G6 molecular classi cation and pathological features, biological behaviors, serum marker levels, 3-year overall survival (OS) rate, and other indicators were analyzed in combination with the clinicopathological data of the HCC patients. We also detected and analyzed molecular markers and molecular signal pathways related to the G1-G6 molecular classi cation. RNA sequencing and bioinformatics analysis were performed to screen related targets and molecular signaling pathways.
HCC samples and clinical data
The use of human clinical samples was approved by the Medical Ethics Committee of Mengchao Hepatobiliary Hospital of Fujian Medical University. A total of 120 HCC specimens and paired adjacent noncancerous liver tissues were randomly selected from patients undergoing hepatectomy between January 2014 to December 2017. The patient follow-up information and clinicopathological characteristics were obtained from the biological sample bank and original pathology report. We also collected 12 fresh HCC tissues and paired adjacent noncancerous liver tissues for further analysis by RNA-seq. Data from The Cancer Genome Atlas (TCGA) database were obtained from TCGA portal sites ( [URL] real-time PCR (qPCR)
The qPCR procedures were performed as previously described [13]
Western blot
Tissues were harvested and lysed in cold extraction buffer (RIPA, Beyotime Biotechnology, Shanghai, China) as previously described [14]. Samples were centrifuged at 13,000 × g at 4°C for 15 min, and the protein concentration in the supernatants was measured using the Pierce BCA Protein Assay kit (Thermo Fisher Scienti c Inc., Waltham, IL, USA) according to the manufacturer's protocol. Equal amounts of protein samples were subjected to 10 % sodium dodecyl sulfate-polyacrylamide gel electrophoresis and subsequently electro-transferred to polyvinylidene di uoride membranes. The membranes were blocked with 5 % BSA diluted in Tris-buffered saline (20 mM Tris, 150 mM NaCl, pH 7.4) containing 0.05 % Tween-20 for 2 h and then incubated with primary antibodies at 4°C overnight. The primary antibodies are listed in Table 2. The membranes were washed ve times and then incubated with peroxidase-conjugated secondary antibodies (1:3,000, ab6721, Abcam, Cambridge, MA, USA) in blocking buffer for 1 h at RT.
Band intensities were quantitated by an enhanced chemiluminescence detection system using the SuperSignal™ West Pico Plus Kit (Thermo Fisher Scienti c Inc.). The protein density was quanti ed with Bio-Rad Image Lab software and ImageJ software. Tissue microarray (TMA) construction A total of 58 para n-embedded tissue samples from HCC patients (G1: 14 cases, G2: 1 case, G3: 14 cases, G4: 9 cases, G5: 10 cases, and G6: 10 cases) were randomly selected under a strati ed sampling procedure to generate TMAs. The donor wax blocks were made of the para n sections and stained by hematoxylin-eosin (HE) staining. To exclude tissues with necrotic or bleeding areas, the corresponding positions of cancer tissues and adjacent liver tissues were observed and marked under a microscope.
The recipient block was cast by melting conventional para n wax in molds for making blank blocks. The donor tissue blocks were transferred into the recipient wax wells and prepared with a 1.5 mm perforated needle. Next, we placed the wax blocks in the oven at 65°C for approximately 7-9 min and removed them at the semi-melted state; the blocks were cooled slightly at room temperature and moved to a refrigerator at 4°C. The freeze-thaw process was repeated so that the tissue core in the blocks and the wall of the pore were integrated.
The sections were treated with a 3 % peroxidase solution to block endogenous peroxidase. The sections were then incubated with 5 % BSA blocking solution to reduce the non-speci c background signal and false positives. Next, the sections were incubated overnight with the primary antibodies at 4°C. A section was incubated in antibody diluent alone without the primary antibody as a control. The sections were further incubated with ImmPress horseradish peroxidase anti-rabbit IgG antibody (Maixin Inc., Fujian, China) secondary antibodies. The immunoreactions were visualized using a DAB Kit (Maixin Inc.). Following counterstaining and mounting, digital images from the sections were created using a wholeslide scanner (KF-PRO-005-EX, KFBIO, Ningbo, China) and images were captured and analyzed using K-Viewer version 1.5.3.1 software (KFBIO).
RNA-seq and bioinformatics analysis
Total RNA was extracted from tissues using Trizol reagent (TransGen Biotech) according to the manufacturer's instructions. RNA quality was con rmed using an Agilent 2100 Bioanalyzer. Libraries for sequencing were created with the Illumina NEBNext® Ultra™ RNA Library Prep Kit or NEBNext® Ultra™ Directional RNA Library Prep Kit. Brie y, PCR ampli cation was performed to obtain the nal DNA library.
After the library was constructed, a Qubit2.0 Fluorometer was used for the preliminary quanti cation. RNA-seq was performed on the Illumina sequencing platform by Shanghai Jikai Company (Shanghai, China). Quality control indicated that the sequencing error rates and data ltering reads of the 24 samples were controlled within the acceptable range. RNA-seq, bioinformatics analysis, and TCGA database analysis were used to explore the target genes and molecular signaling pathways between the proliferation class (G1-G3, group H) and non-proliferation class (G4-G6, group L). We compared two or multiple gene expressions under different conditions using statistical methods, identi ed the speci c genes that correlated with the conditions, and analyzed the biological signi cance (quality control, matching, quantitative analysis process, signi cant difference analysis, and function of enrichment) of these speci c genes.
Statistical analysis
Statistical analysis was conducted by GraphPad Prism version 8 software (GraphPad Software, San Diego, CA, USA) and IBM SPSS Version 25 software. The chi-square test (Fisher exact probability method) was used to analyze the correlation between the G1-G6 molecular classi cation and tumor biological behavior. After checking data for normal distribution and variance homogeneity, comparisons between two groups were evaluated using independent-sample t-tests or Mann-Whitney U test. For three or more groups, differences were statistically analyzed by one-way ANOVA or the non-parametric test (Kruskal-Wallis test). Kaplan-Meier and Cox proportional hazards survival regression analysis were used to evaluate the prognostic signi cance of the G1-G6 molecular classi cations within 36 months. All p values were two-tailed, and P < 0.05 was considered statistically signi cant.
Results
Distribution of G1-G6 subgroups in HCC patients from Europe and China studies Gene expression analysis was performed on surgically resected HCC samples from Chinese patients who were grouped into G1-G6 transcriptomic categories according to the expression of 16 predictor genes. First, we isolated RNA samples from 120 HCC tumor samples for HCC classi cation, and 107 RNA samples with su cient tissue quantity and good quality were analyzed by qPCR. Data from these samples were analyzed according to the standard method proposed by the Europe group [8]. We found that the HCC cases from China (n=107) were distributed as follows: G1: n=19 (17.76%), G2: n=2 (1.87%), G3: n=20 (18.69%), G4: n=10 (9.35%), G5: n=25 (23.36%), and G6: n=31 (28.97%) subgroups.
Several studies have shown that no matter what molecular typing method is used, HCC can be de ned into two major groups, including a proliferation group and a non-proliferation group. Proliferation tumors are associated with aberrant activation of signaling pathways, while non-proliferative tumors display a well-differentiated phenotype [15,16]. Correspondingly, G1-G3 (combined with G1, G2, and G3) of the G1-G6 molecular typing can be classi ed as a proliferation group, while G4-G6 (combined with G4, G5, and G6) was classi ed as a non-proliferative group. Next, we pooled classes into the G1-G3 group and G4-G6 group. We found that the proportions of the G1-G3 group and G4-G6 group in the China cohort were 38.32% and 61.68%, respectively. In the European cohort, 35 I-II 4 2 3 0 6 9 III-IV 15 0 14 10 19 22 TNM stage 107 I-II 8 2 10 6 18 19 III-IV 11 0 To analyze the correlation between proliferation and clinicopathological characteristics of HCC patients based on G1-G6 molecular classi cation, we integrated the proliferative group (G1-G3) and nonproliferative group (G4-G6) by clinicopathological information. We found that the proliferative group (G1-G3) was correlated with high serum AFP level (P=0.002), high copy number of HBV DNA (P=0.047), complex histological subtype (P=0.010), macrovascular invasion (P=0.012), and negative or weak positive Hep-Par1 (P=0.012) ( Table 4). Using follow-up data of patients after hepatectomy, we also analyzed the correlation between HCC prognosis and G1-G6 molecular classi cation. As shown in Fig. 1C, the G1-G3 HCC patients showed a shorter 3-year OS than G4-G6 HCC patients (P=0.010). The recurrence-free survival of the G1-G3 group tended to be shorter than the G4-G6 group (P=0.072) ( Figure 1D). Further analysis showed that G1 patients tended to have a shorter 3-year OS than non-G1 patients (P=0.1867) ( Figure 1E) and G3 patients had the worst prognosis (P=0.0125) ( Figure 1F). Cox regression univariate analysis showed that the TNM stage (P=0.0002), BCLC stage (P=0.001), and G1-G6 molecular classi cation (P=0.012) were important factors that affected the overall prognosis of HCC patients. All factors with a p-value < 0.1 detected in univariate analyses were included in multivariate analyses, and the subsequent results showed that the G1-G6 molecular classi cation (P=0.035) is an independent prognostic risk factor for poor OS of HCC patients (Table 5). Predictor of HCC classi cation using qPCR A total of 19 para-cancer liver tissue samples were randomly selected to investigate the correlation between expression of the 16 genes in the proliferation group (G1-G3) compared with the nonproliferation group (G4-G6). Our results showed that the expression levels of AFP (P=0.0200), CDH2 (P=0.0022), HN1 (P<0.0001), NRAS (P=0.0096), PAK2 (P=0.0024), RAB1A (P=0.0020), and SAE1 (P<0.0001) were signi cantly higher in the proliferation group (G1-G3) than in the non-proliferation group (G4-G6) ( Fig. 2A). Conversely, the expression levels of LAMA3 (P=0.0059) and PAP (P=0.0495) were markedly higher in the non-proliferation group (G4-G6) than in the proliferation group (G1-G3) (Fig. 2B).
There were no signi cant differences in the expressions of the following genes between the proliferation group (G1-G3) and non-proliferation group ( PD-L1 is highly expressed in G1-G3 subgroups Several molecular targeted drugs have entered clinical trials as palliative and complementary treatments for HCC [17]. For example, the clinically approved antibodies targeting PD-1 or its ligand PD-L1 were shown to cause lasting responses in up to 25% of advanced HCC patients in two early trials. In HCC, the proportion of tumor cells expressing PD-L1 is approximately 15% [18]. Several studies showed in amed and membrane expression of PD-L1 in ancestral liver cancer HCC, while lymph epithelioma-like HCC is in amed by in ammatory cells that express PD-L1 in large numbers [16,19,20]. The expression pattern of PD-L1 has an important effect on HCC prognosis, which is poor if both tumor cells and macrophages express PD-L1 [21]. Screening of patients who may bene t from PD-1/PD-L1 inhibitor therapy is the most important clinical concern. The correlation between G1-G6 classi cation and tumor cell PD-L1 expression is not clear. Therefore, we examined the expression of the PD-L1 by IHC in a TMA. The TMA comprised 58 resected HCC specimens and 58 specimens of adjacent non-malignant and premalignant para-cancer liver tissue from the same patients. Staining was compared on consecutive tissue sections of the TMA to enable automated (unbiased) image analysis and direct comparison of the samples. The staining was quanti ed according to the percentage of area stained for each protein in comparison with isotype-matched control IgG (Fig. 3A). Representative images for the positive PD-L1 antibody staining are shown in Fig. 3B, along with histograms showing the pooled quanti ed data for the antigen (Fig. 3C). We found that the positive staining of PD-L1 in G1-G3 subgroups was signi cantly higher than in G4-G6 subgroups (P=0.028).
G5 and G6 HCC subgroups are associated with activation of the Wnt/β-catenin pathway signaling Aberrant activation of Wnt/β-catenin signaling plays a key role in HCC progression, with approximately half of HCC cases acquiring mutations in either CTNNB1 or AXIN1 [22]. Glutamine synthetase (GS) is classically overexpressed when the Wnt/β-catenin signaling pathway is activated due to CTNNB1 mutations and could be used as surrogate marker [23]. In our study, we found that the protein levels of βcatenin were substantially higher in G5 and G6 HCC subgroups than in G1-G4 subgroups ( Fig. 4A and B). Additionally, the mRNA expression level of GS in HCC liver tissues was signi cantly increased compared with levels in para-cancer liver tissues, and the G5 and G6 HCC subgroups showed signi cantly higher expression levels of GS than the G1-G4 subgroups (Fig. 4C). In line with these results, western blot analysis showed that protein levels of β-catenin and GS were higher in G5 and G6 subgroups than other subgroups (Fig. 4D). Further statistical analysis found that the ratio of GS expression to β-catenin expression in the G5 and G6 groups was approximately three-fold higher compared with levels in the G1-G4 subgroups, which suggests that the Wnt/β-catenin pathway is highly activated in G5 and G6 HCC. The activation level was highest in G6. These results suggested that G5 and G6 HCC subgroups were closely associated with the robust activation of Wnt/β-catenin signaling.
The G1 subgroup is associated with maintenance of tumor cell stemness Increasing studies have reported the presence of a small number of cells in tumor tissue with strong stemness [24,25]. We found that the G1-G3 HCC subgroups were closely associated with poor differentiation and high invasion of tumor cells. We further checked whether the proliferation group (G1-G3) was correlated with the stemness characteristics of HCC. Representative images for IHC of stemnessrelated molecular markers, EpCAM and Sox9, are shown in Fig. 5A and 5C. EpCAM staining was located at the cytomembrane and cytoplasm while SOX9 expression was identi ed in the nucleus. The positive expressions of EpCAM and Sox9 in the G1-G3 group were signi cantly higher than in the G4-G6 group ( Fig. 5B and D). The G1 subgroup showed the highest positive percentages of EpCAM and Sox9 compared with other subgroups (Fig. 5E). The G3 subgroup did not show a signi cant difference compared with other subgroups (Fig. 5F). qPCR analysis also showed a similar pattern as the IHC results ( Fig. 5G and H). These results suggested that the G1 subgroup is associated with the maintenance of stemness in tumor cells.
Transcriptome sequencing analysis for G1-G6 molecular classi cation of Chinese HCC patients The results of our study suggest that G1-G3 group can be used as an independent risk factor for evaluating HCC patients' prognosis and is signi cantly correlated with poor prognosis after hepatectomy.
To clearly understand the poor prognosis group based on G1-G6 molecular classi cation of Chinese HCC patients, 12 samples of HCC tissues (2 cases of each subgroup) and the corresponding adjacent liver tissues were randomly selected. The expression levels of genes related to differentiation, proliferation, and function of tumor cells were examined in these samples and analyzed by RNA-seq and bioinformatics. The error rates of the 24 samples were all within the acceptable range ( Supplementary Fig. 1). The proportions of ltered reads of the 24 samples were all within the acceptable range, and the proportion of clean reads of each sample accounted for more than 90%, which suggested that the data were suitable for subsequent analysis (Supplementary Fig. 2).
The all HCC group (CA group) included 12 samples of HCC tissues, and the para-HCC group (pCA group) included 12 corresponding samples of adjacent liver tissues. High proliferative group (group H) includes 6 samples of G1-G3 cancer tissues, and low proliferative group (group L) includes 6 samples of G4-G6 cancer tissues. Compared with the pCA group, there were 3233 up-regulated genes and 1654 downregulated genes in the CA group, while 33,299 genes showed no changes between the two groups (Fig. 6A). Compared with group L (G4-G6 subgroup), there were 1,269 up-regulated and 1,445 down-regulated genes in group H (G1-G3 subgroup), and 33,692 genes showed no differential expression between the two groups (Fig. 6B). We also found that 108 genes overlapped between the H group (compared with group L) and CA (compared with group pCA) (Fig. 6C), and 27 genes were commonly down-regulated ( Fig. 6D). Among the 27 genes, four genes (REG1B, REG3G, C19orf18, and ITPR1) were signi cantly upregulated or downregulated in group H compared with group L ( Table 6). Analysis using TCGA database also con rmed that the expression levels of REG1B, REG3G, and ITPR1 were signi cantly increased in HCC liver tissues compared with normal liver tissues (Fig. 6E). Using the follow-up information database of TCGA patients, Kaplan-Meier analysis was conducted to analyze the prognosis of HCC patients. As shown in Fig. 6F, HCC patients with a high expression of ITPR1 showed a signi cantly shorter 3-year OS than HCC patients with low or medium expression of ITPR1. However, there were no statistical differences between HCC patients with high expressions of REG1B or REG3G and the corresponding control groups. These results suggested that ITPR1 may promote the occurrence and development of HCC and affect the prognosis of HCC patients. We next performed KEGG pathway analysis using the KEGG database and Clusterpro le software. Fisher exact analysis was used to test and calculate p-values, and the pathways with a p-value less than 0.05 were retained. The p-value represented the signi cance of enrichment of differential genes; a lower the pvalue indicates a greater correlation between the pathway and the differential genes. A Padj value less than 0.05 was taken as the threshold of signi cant enrichment. The KEGG pathway enrichment analysis revealed that the calcium signaling pathway, cGMP-PKG signaling pathway, renin secretion, neuroactive ligand-receptor interaction, and cell metabolism-related signaling pathways were signi cantly enriched in the G1-G3 subgroups (Fig. 7A-B).
Discussions
In this study, our results indicate that the G1-G6 signatures were generally in concordance between the genetic pro les of Chinese HCC patients and HCC patients in the original French study. Additionally, our data suggest that the G1-G3 group can be used as an independent risk factor for evaluating the prognosis of HCC patients, since the G1-G3 group was signi cantly correlated with poor prognosis after hepatectomy and the G3 group showed the worst prognosis among all subgroups. Furthermore, we demonstrated that the G5-G6 group was associated with activation of the Wnt/β-catenin pathway.
Moreover, we discovered that the G1-G3 group was associated with PD-L1 expression and the G1 subgroup is mainly related to liver cancer stemness. Importantly, we identi ed four genes (REG1B, REG3G, C19orf18, and ITPR1) that were signi cantly up-regulated or down-regulated in the proliferation group (G1-G3) compared with the non-proliferation group (G4-G6). Additionally, we found that the calcium signaling pathway, cGMP-PKG signaling pathway, renin secretion, neuroactive ligand-receptor interaction, and cell metabolism-related signaling pathways were signi cantly enriched in the G1-G3 subgroup.
Although our results were generally consistent with the studies from Europe and Singapore, there were still many inconsistencies. First, the distribution of the HCC subgroups in the Chinese cohort was signi cantly different from that in the original European cohort. In the European study, the G4 and G1 subgroups represented the largest and smallest subgroup, respectively. In our study, the G6 and G2 subgroups were the largest and smallest groups, respectively. Second, the Singapore research reported that the incidence of MVI in the G3 subgroup was markedly higher than in other subgroups. However, we did not nd a similar characteristic of the G3 subgroup of China HCC patients. Instead, we found that the incidence of MVI was signi cantly lower in the G5 subgroup than the other subgroups. Third, inconsistent with the European study but in line with the Singapore study, we did not nd any subgroups characterized with satellite nodules. Finally, our data showed that the G1-G3 subgroups had a shorter 3-year OS rate and the G3 subgroup has the worst prognosis, which was different from the studies in Europe and Singapore. These differences suggest that some of the clinical characteristics of HCC patients are distinct in the populations from China and other countries.
In this study, we found a high serum AFP level in the G1 group and a correlation between the G5-G6 group and Wnt/β-catenin pathway activation based on previous studies. Several studies showed that HCC with β-catenin mutations have slow proliferation, well-differentiated characteristics, and trabecular type false adenoid type histology, with cholestasis and intratumoral in ltrating immune de ciency [16,26].
β-catenin is a key molecular marker in the epithelial-mesenchymal transition. In physiological conditions, β-catenin binds cadherin and promotes the release of cadherin when β-catenin translocates to the nucleus [27]. In HCC, E-cadherin is reduced on the cell surface and N-cadherin is expressed once tumor cells undergo epithelial-mesenchymal transition [28]. It implicates that G5 and G6 subgroups may have some features as mentioned above.
We found several additional correlations in this study that had not been reported previously. For example, the G1-G3 group was associated with PD-L1 expression and liver cancer stemness. These ndings provide a new direction for the diagnosis and treatment of HCC. In the future, if G1-G3 patients were treated with targeted therapy for PD-1/PD-L1 and liver cancer stem cells, this may improve the survival and prognosis of a patient with highly malignant HCC with an extremely poor prognosis.
The pathogenesis of HCC is complex and involves multiple signaling pathways [29,30]. Signaling pathways involved in growth factors activated by intracellular calcium ions play a crucial role in various physiological and biochemical processes [31]. HCC encodes a variety of tyrosine kinases, including epidermal growth factor receptor (EGFR), whose functions are mainly in glucose and lipid metabolism, liver brosis, regeneration, and carcinogenesis [32,33]. Modica et al. [34] reported that EGF-dependent HCC proliferation may be affected by calcium ion, and intracellular dissociative calcium is a key regulator for EGFR signal propagation and apoptosis of tumor cells. Calcium not only promotes the proliferation of proliferating HCC cells but also induces the apoptosis of non-proliferating HCC cells, which might contribute to the reason why the proliferation group (G1-G3) has a poor prognosis. Considering the high hepatic activity of calcium and the overexpression of calcium channels in G1-G3 group, manipulation of calcium in ux in HCC cells may inhibit tumor metastasis.
ITPR1 encodes an intracellular receptor for type 1 inositol 1,4,5-triphosphate and mediates the release of calcium from the endoplasmic reticulum. A few recent studies indicated that ITPR1 affects tumor progression by regulating autophagy. For example, ITPR1 can protect renal cancer cells against natural killer cells by inducing autophagy [35,36]. In our study, we found that HCC patients with a high expression level of ITPR1 have a signi cantly lower 3-year OS rate than HCC patients with low or medium expression levels of ITPR1. Therefore, we speculate that ITPR1 may cooperate with calcium ions to promote the proliferation of proliferating HCC cells, thus leading to the increase of postoperative recurrence risk of HCC patients.
As a second messenger of cell information transmission, cGMP promotes cell division and inhibits cell differentiation. Our RNA-seq analysis showed that the cGMP-PKG signaling pathway was enriched in G1-G3 type HCC, which might contribute to the proliferative characteristics of the G1-G3 subgroups. In this study, we found that G1-G3 type HCC patients showed a high serum AFP level. A differential analysis based on the HCC transcriptome data from TCGA showed that the genes expressed in HCC tissues with high AFP expression are involved in neuroactive ligand-receptor interaction pathways [37]. Together, these results suggest that the proliferative characteristics of G1-G3 subgroups might be from the enrichment of neuroactive ligand-receptor interaction pathways.
Conclusions
In this study, we found that the G1-G6 transcriptomic molecular classi cation was applicable in the China HCC cohort regardless of the ethnic origin of patients. Through the results of these cohorts, we may be able to assertively establish the clinical and pathological relevance of the 16 gene score and use the classi cation system to develop therapeutic strategies for HCC patients worldwide in the future. Our results help clarify our understanding of the correlation between G1-G6 molecular classi cation and molecular markers and molecular signaling pathways in the Chinese HCC population and initially established a link between the phenotype and molecular characteristics. However, we mostly performed correlation analysis of G1-G6 molecular classi cation, focusing on the postoperative prognosis monitoring of HCC patients, and more studies are needed to fully clarify the molecular mechanism of HCC occurrence and development.
Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.
==
Domain: Biology Medicine
|
[{"added": "2021-10-15T15:22:22.680Z", "created": "2021-10-12T00:00:00.000", "id": "241114317", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "https://www.researchsquare.com/article/rs-956345/latest.pdf", "pdf_hash": "2f3c1db8ac8cc7fc534e63398302f0da0df83dc0", "pdf_src": "Adhoc", "provenance": "peS2o-0028.json.gz:365", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "59c5f450942da303d39703429f7fce2831bbfedc", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Histone Modifications in NAFLD: Mechanisms and Potential Therapy
Nonalcoholic fatty liver disease (NAFLD) is a progressive condition that encompasses a spectrum of liver disorders, beginning with the simple steatosis, progressing to nonalcoholic steatohepatitis (NASH), and possibly leading to more severe diseases, including liver cirrhosis and hepatocellular carcinoma (HCC). In recent years, the prevalence of NAFLD has increased due to a shift towards energy-dense dietary patterns and a sedentary lifestyle. NAFLD is also strongly associated with metabolic disorders such as obesity and hyperlipidemia. The progression of NAFLD could be influenced by a variety of factors, such as diet, genetic factors, and even epigenetic factors. In contrast to genetic factors, epigenetic factors, including histone modifications, exhibit dynamic and reversible features. Therefore, the epigenetic regulation of the initiation and progression of NAFLD is one of the directions under intensive investigation in terms of pathogenic mechanisms and possible therapeutic interventions. This review aims to discuss the possible mechanisms and the crucial role of histone modifications in the framework of epigenetic regulation in NAFLD, which may provide potential therapeutic targets and a scientific basis for the treatment of NAFLD.
Introduction 1. Pathogenesis of Nonalcoholic Fatty Liver Disease
Nonalcoholic fatty liver disease (NAFLD) is an increasingly common condition [1], which is strongly associated with metabolic diseases such as obesity, type 2 diabetes mellitus (T2DM), hyperlipidemia, and atherosclerosis [2,3]. NAFLD patients with metabolic syndrome or T2DM typically have an increased risk of death [2]. The prevalence of NAFLD ranges from 13.5% in Africa to 31.8% in the Middle East, and NAFLD is now the leading cause of chronic liver disease worldwide. In Western countries, the prevalence of NAFLD is 20-40% in adults, and 10-30% of the NAFLD patients eventually develop nonalcoholic steatohepatitis (NASH) [4]. Researchers have indicated that NASH patients who progress to the stage of cirrhosis have an increased risk of hepatocellular carcinoma (HCC) [5]. However, there is no approved pharmacotherapy for the treatment of NASH or liver cirrhosis. Therefore, NAFLD is a prevalent disease with a high social burden and unmet medical need.
From simple steatosis to NASH, cirrhosis, and hepatocellular cancer, NAFLD is a progressive disease [6] (Figure 1). Simple steatosis, or nonalcoholic fatty liver (NAFL), is characterized by the accumulation of triglycerides (TG) in hepatocytes. Hepatic steatosis can occur due to various factors. Excess dietary fat is certainly one. In addition, dietary carbohydrates raise blood glucose and insulin levels, leading to the activation of carbohydrate responsive element-binding protein (ChREBP) and sterol regulatory element-binding protein-1c (SREBP-1c) in the liver. These are two major transcription factors upregulating genes in fatty acid synthesis, thereby promoting de novo lipogenesis (DNL) [7][8][9], which produces free fatty acids (FFAs) from acetyl-coenzyme A (CoA) and leads to the accumulation of lipid droplets in hepatocytes upon activation. Furthermore, insulin resistance (IR), which is a common condition in patients with obesity or T2DM, could decrease the capacity of adipose tissue to store lipids, thereby increasing the amount of FFAs in the blood and, eventually, in the liver. In addition to producing energy through oxidation and storage in hepatocytes as TG, hepatic FFAs can be coupled to apolipoproteins and secreted in very-low-density lipoproteins (VLDL). Hepatic steatosis occurs when the amount of hepatocyte triglyceride production increases more than the amount of VLDL triglyceride release [10]. Therefore, an imbalance in lipid homeostasis causes the accumulation of lipids in the liver.
Unlike simple steatosis, NASH is a state in which inflammation and fibrosis occur in addition to the lipid accumulation in the liver. Therefore, NASH is characterized by the accumulation of excessive FFAs, increased oxidative/endoplasmic reticulum (ER)stress, fibrosis, and inflammation in the liver [11,12]. It is defined clinically by the presence of steatosis, hepatocellular ballooning, and lobular inflammation, accompanied by variable degrees of fibrosis upon a liver biopsy [13]. When there is an increased uptake/synthesis or hampered removal of fatty acids, they can be used as substrates for the production of lipotoxic species that induce cellular stress, hepatocyte apoptosis, and liver injury [14]. Indeed, the occurrence of NASH is associated with a number of cellular stresses, such as endoplasmic reticulum (ER) stress, mitochondrial damage, and oxidative stress, which is related to the production of reactive oxygen species (ROS) (Figure 1). It is noteworthy that NASH is progressive, and more severe forms may occur, such as liver cirrhosis and HCC [11]. HCC is the fifth-most prevalent form of cancer and the third leading cause of cancerrelated death [15], highlighting the importance of preventing the progression of hepatic steatosis to HCC. Unlike simple steatosis, NASH is a state in which inflammation and fibrosis occur in addition to the lipid accumulation in the liver. Therefore, NASH is characterized by the accumulation of excessive FFAs, increased oxidative/endoplasmic reticulum (ER)stress, fibrosis, and inflammation in the liver [11,12]. It is defined clinically by the presence of steatosis, hepatocellular ballooning, and lobular inflammation, accompanied by variable degrees of fibrosis upon a liver biopsy [13]. When there is an increased uptake/synthesis or hampered removal of fatty acids, they can be used as substrates for the production of lipotoxic species that induce cellular stress, hepatocyte apoptosis, and liver injury [14]. Indeed, the occurrence of NASH is associated with a number of cellular stresses, such as endoplasmic reticulum (ER) stress, mitochondrial damage, and oxidative stress, which is related to the production of reactive oxygen species (ROS) (Figure 1). It is noteworthy that NASH is progressive, and more severe forms may occur, such as liver cirrhosis and HCC [11]. HCC is the fifth-most prevalent form of cancer and the third leading cause of cancer-related death [15], highlighting the importance of preventing the progression of hepatic steatosis to HCC.
Factors Influencing NAFLD
NAFLD is a complex disease resulting from multiple aspects of lifestyle, diet, genetic variants, and epigenetic factors (Figure 2). The lack of physical exercise has been associated with increased body weight, insulin resistance, and an increased risk of metabolic syndrome and NAFLD [16]. Unhealthy diets may lead to overnutrition characterized by the excessive intake of fats or fructose, which are metabolized in the liver through DNL and converted to TGs. Consequently, this metabolic pathway contributes to the accumulation of fat in the liver. Insulin resistance has long been acknowledged as a crucial factor in the development of NAFLD [17]. There is a growing body of evidence supporting that the microbiome also plays a critical role in NAFLD [18][19][20]. The gut-liver axis has been connected to a number of disorders associated with obesity, including NAFLD, and these two organs are interdependent at many levels [21].
Factors Influencing NAFLD
NAFLD is a complex disease resulting from multiple aspects of lifestyle, diet, genetic variants, and epigenetic factors (Figure 2). The lack of physical exercise has been associated with increased body weight, insulin resistance, and an increased risk of metabolic syndrome and NAFLD [16]. Unhealthy diets may lead to overnutrition characterized by the excessive intake of fats or fructose, which are metabolized in the liver through DNL and converted to TGs. Consequently, this metabolic pathway contributes to the accumulation of fat in the liver. Insulin resistance has long been acknowledged as a crucial factor in the development of NAFLD [17]. There is a growing body of evidence supporting that the microbiome also plays a critical role in NAFLD [18][19][20]. The gut-liver axis has been connected to a number of disorders associated with obesity, including NAFLD, and these two organs are interdependent at many levels [21]. and basal conditions such as metabolic syndrome, genetic variants, and epigenetic factors. In addition, non-epigenetic factors may influence NAFLD through epigenetic mechanisms, such as DNA methylation, histone modifications, and ncRNA. PNPLA3, patatin-like phospholipase domain-containing 3; HMT, histone methyltransferase; HDM, histone demethylase; HAT, histone acetyltransferase; HDAC, histone deacetylase; ncRNA, non-coding RNA.
In addition to external or environmental cues, genetic variants significantly influence the progression of NAFLD [22]. One solid example is the genetic variant in the PNPLA3 gene (patatin-like phospholipase domain-containing 3, rs738409), a substitution of cytosine with guanine resulting in a change of codon 148 from isoleucine to methionine. PNPLA3 is involved in the lipolysis of lipid droplets in hepatocytes. The I148M variant escapes proteasome degradation, accumulates on lipid droplets, and blocks the function of adipose triglyceride lipase and lipolysis [23]. PNPLA3-I148M has exhibited a robust correlation with increased hepatic TG accumulation and hepatic inflammation in human genome-wide association studies [24]. PNPLA3-I148M expression in liver resulted in an increase in the liver triglyceride content in a transgenic mouse model [25]. Thus, it is an important genetic risk factor and a valid therapeutic target for NAFLD [26]. Similarly, a few other genetic risk factors have been identified, such as a splice variant (rs72613567 T>A) in HSD17B13 and an E to K substitution (rs58542926 C>T) in TM6SF2, which have been well summarized recently [27,28]. In addition to external or environmental cues, genetic variants significantly influence the progression of NAFLD [22]. One solid example is the genetic variant in the PNPLA3 gene (patatin-like phospholipase domain-containing 3, rs738409), a substitution of cytosine with guanine resulting in a change of codon 148 from isoleucine to methionine. PNPLA3 is involved in the lipolysis of lipid droplets in hepatocytes. The I148M variant escapes proteasome degradation, accumulates on lipid droplets, and blocks the function of adipose triglyceride lipase and lipolysis [23]. PNPLA3-I148M has exhibited a robust correlation with increased hepatic TG accumulation and hepatic inflammation in human genome-wide association studies [24]. PNPLA3-I148M expression in liver resulted in an increase in the liver triglyceride content in a transgenic mouse model [25]. Thus, it is an important genetic risk factor and a valid therapeutic target for NAFLD [26]. Similarly, a few other genetic risk factors have been identified, such as a splice variant (rs72613567 T>A) in HSD17B13 and an E to K substitution (rs58542926 C>T) in TM6SF2, which have been well summarized recently [27,28].
Due to the fact that epigenetic alterations can be dynamic and reversible, epigenetic mechanisms have been widely implicated in the initiation and progression of NAFLD, even in collaboration with other environmental cues (Figure 2). Epigenetic inheritance studies the changes in heritable gene expression or cellular expression caused by specific mechanisms without altering the DNA sequence [29]. Alterations in DNA methylation, histone variants and modifications, chromatin remodeling, and non-coding RNA-based mechanisms may all result in epigenetic changes. Several epigenetic mechanisms are of significant importance in the NAFLD spectrum of diseases [30]. For example, the cytosine methylation (5mC) of mitochondrial DNA (mtDNA) had been found in the liver biopsies of patients with NAFLD, and patients with NASH had higher levels of DNA methyltransferase 1 [31]. In another study, AAV-miR-20b administration induced hepatic steatosis and reduced FA oxidation in HFD-fed mice, possibly by decreasing the level of PPARα [32]. More broadly, histone modification changes may lead to the dysregulation of multiple biological processes associated with NAFLD, such as hepatic lipid accumulation, ER stress, oxidative stress, mitochondrial damage, and inflammation, which may be used as a single mechanism or work in synergy with the environmental factors on the development of NAFLD [22]. Due to length limitations, in this review, we focus on the regulatory mechanism of histone modifications, their pathological implications, and the potential therapeutic applications in the treatment of NAFLD.
Regulation of NAFLD by Histone Modifications
The histone core is an octamer, comprising two H2A, H2B, H3, and H4 molecules. The histone core and the DNA coiled on it constitute the nucleosome, serving as the basic structural unit of chromatin. The N-terminal tails of histone H2A, H2B, H3, and H4 may extensively modified post-translationally and serve as a hub receiving diverse regulatory signals from diet, lifestyle, and other environmental cues. With the technical improvements and scientific advancements, the list of modification types has expanded, encompassing not only classical acetylation, methylation, ubiquitination, and phosphorylation but also newly discovered ones, including lactylation and dopaminylation [33,34]. Among the classical modifications, acetylation, methylation, and ubiquitination primarily happen on lysine or arginine, while phosphorylation happens on serine or threonine. Distinct modification types may alter the chromatin structure and gene expression differentially. For example, histone acetylation generally correlates with transcriptional activation, whereas deacetylation tends to exert a transcriptionally repressive role. From the perspectives of their relevance in NAFLD, the classical histone modifications have been extensively investigated, and therefore, more details are summarized below.
Histone Methylation in NAFLD and its Therapeutic Implications
Histone methylation is catalyzed by histone methyltransferases (HMTs), and the histone demethylases (HDMs) remove the methylation marks on lysine or arginine residues. Notably, distinct lysine or arginine residuals require specific HMT and HDM enzymes, which also provide a strong specificity toward individual methyltransferase or demethylase, thereby reenforcing their distinct functional impacts.
Early evidence suggested histone H3 lysine 4 (H3K4) methyltransferase MLL2 could influence the metabolism via random ENU mutagenesis. A study revealed that a germline Mll2 mutation led to insulin resistance and impaired glucose tolerance in mice [35]. In another study, HFD feeding led to the activation of ABL1 kinase, which phosphorylated PPARγ2 and enhanced the MLL4-PPARγ2 interaction. Consequently, overnutrition enhanced the recruitment of MLL4 to the promoter of PPARγ2-regulated steatosis target genes. This, in turn, increased the H3K4 methylation and transcriptional activation. Thus, the interaction between MLL4 and PPARγ2 proteins played a role in the development of fatty liver in HFD-fed mice [36,37] (Figures 3 and 4). Moreover, the activation of hepatic stellate cells (HSCs), a key event in the transition from NAFLD to NASH, is also regulated by histone methylation. In activated HSCs, ASH1, another HMT for H3K4 methylation, directly bound to the regulatory genomic regions of alpha smooth muscle actin (α-SMA), collagen I, tissue inhibitor of metalloproteinase-1 (TIMP1), and transforming growth factor beta 1 (TGF-β1) to facilitate H3K4 methylation and transcriptional activation. Conversely, inhibiting ASH1 led to the downregulation of these fibrogenic gene expressions [38,39].
Protein arginine methyltransferase 5 (PRMT5) affects gene expression by methylating the arginine residues on histones, including H4R3, H3R8, and H2AR3. A previous study showed PRMT5 promoted the development of hepatic steatosis under a high-fat diet by facilitating the suppression of transcription regulators in mitochondrial biogenesis such as PPARα [40] (Figure 3A). PPARα functions as a pivotal transcription factor governing processes like fatty acid uptake, mitochondrial and peroxisomal fatty acid oxidation, and ketogenesis in the liver [41,42]. Mechanistically, PRMT5 was upregulated in the liver upon HFD. Conversely, the silencing or deletion of PRMT5 led to diminished AKT phosphorylation while increasing the expression of PPARα and PGC-1. This, in turn, elevated mitochondrial and peroxisomal fatty acid oxidation, demonstrating a propensity to slow down the development of fatty liver, although the mechanism with which PRMT5 regulates AKT phosphorylation remains unclear [40,43].
Methylations on K27 or 9 of histone H3 (H3K27 or H3K9 methylation) are typically linked to transcriptional repression. Polycomb repressive complex 2 (PRC2), a multicomponent protein complex, is the only methyltransferase for H3K27 and conserved from fungi to mammals. EZH2, the enzymatic subunit of PRC2, catalyzes the mono-, di-, and trimethylation of H3K27, which plays an important role in cell proliferation and differentiation. It has been reported that the EZH2 protein level was downregulated in the rat liver of a diet-induced NAFLD model and the fatty acid-induced insulin-resistant HepG2 model, although lacking a mechanism [44]. On the other hand, many studies have pointed toward out that EZH2/PRC2 activity or H3K27me3 may facilitate the disease progression to HCC via the repression of multiple tumor-suppressive microRNAs [45]. For example, it was reported that miR-200c was repressed by chromatin H3K27me3, and EZH2 depletion upregulated miR-200c and inhibited the growth of Huh7 in vitro and in vivo [46]. Methylations on K27 or 9 of histone H3 (H3K27 or H3K9 methylation) are typically linked to transcriptional repression. Polycomb repressive complex 2 (PRC2), a multicomponent protein complex, is the only methyltransferase for H3K27 and conserved from fungi to mammals. EZH2, the enzymatic subunit of PRC2, catalyzes the mono-, di-, and trimethylation of H3K27, which plays an important role in cell proliferation and differentiation. It has been reported that the EZH2 protein level was downregulated in the rat liver of a diet-induced NAFLD model and the fatty acid-induced insulin-resistant HepG2 model, although lacking a mechanism [44]. On the other hand, many studies have pointed toward out that EZH2/PRC2 activity or H3K27me3 may facilitate the disease progression to HCC via the repression of multiple tumor-suppressive microRNAs [45]. For example, it was reported that miR-200c was repressed by chromatin H3K27me3, and EZH2 depletion upregulated miR-200c and inhibited the growth of Huh7 in vitro and in vivo [46].
Unlike H3K27, there are multiple H3K9 methyltransferases, such as SUV39H2/KMT1B and the dimeric G9a, and they exhibit differential catalytic activities and target genes [47]. In methionine-and choline-deficient (MCD) diet-fed mice, SUV39H2 promoted hepatic steatosis by downregulating SIRT1, a NAD + -dependent histone deacetylase executing a protective role in the liver (more details in Section 2.4) [48]. Conversely, Suv39h2 deletion alleviated diet-induced NASH in mice [49]. As H3K9 methylation is usually mechanistically associated with transcriptional repression, these effects may or may not be direct modulations and require further investigation. Nonetheless, the presented evidence underscores the role of histone methylation in the pathogenesis of NAFLD. Unlike H3K27, there are multiple H3K9 methyltransferases, such as SUV39H2/KMT1B and the dimeric G9a, and they exhibit differential catalytic activities and target genes [47]. In methionine-and choline-deficient (MCD) diet-fed mice, SUV39H2 promoted hepatic steatosis by downregulating SIRT1, a NAD + -dependent histone deacetylase executing a protective role in the liver (more details in Section 2.4) [48]. Conversely, Suv39h2 deletion alleviated diet-induced NASH in mice [49]. As H3K9 methylation is usually mechanistically associated with transcriptional repression, these effects may or may not be direct modulations and require further investigation. Nonetheless, the presented evidence underscores the role of histone methylation in the pathogenesis of NAFLD.
Potential Targets and Compounds Modulating Histone Methyltransferases in NAFLD
As NAFLD is a multifactorial chronic disease, treatment methods and approaches are still scarce despite the increased attention it has received. Since epigenetic modifications and NAFLD are closely related, altering histone modifications holds the potential to offer new opportunities in the treatment of NAFLD.
Targeting EZH2 has been intensively studied for its potential therapeutic implications in the treatment of NAFLD [50]. A fundamental event in the pathogenesis of hepatic fibrosis is the activation of quiescent HSC and their subsequent transformation into myofibroblasts.
EZH2 has been reported to promote this transformation by suppressing the expression of PPARγ [51]. In addition, EZH2 was found to be inhibited by SIRT1, the protective NAD + -dependent histone deacetylase, and EZH2 inhibition is required for the protective effect of SIRT1 activation in myofibroblasts [52]. Furthermore, the inhibition of EZH2 decreased fibrogenic gene transcription in the TGF-β1-treated HSCs [53]. Indeed, DZNep, an HMT inhibitor, and GSK-503, a specific EZH2 inhibitor, prevent the progression of liver fibrosis in vivo by decreasing H3K27 methylation [44,50,53,54]. In addition, the herbal prescription Yang-Gan-Wan (YGW) and its active ingredients, rosmarinic acid (RA) and baicalin (BC), showed the potential to treat liver fibrosis by de-repressing Pparγ in an epigenetic-dependent way, which suppressed the expression of EZH2 and reduced H3K27 di-methylation [55].
In addition, the expression of histone H3K9 methyltransferase G9a and the DNA methyltransferase DNMT1 was found to be upregulated in human cirrhotic liver and during mouse HSC activation. Using a dual chemical inhibitor of G9a and DNMT1 CM272, the authors showed that the inhibition of them simultaneously disrupted the profibrogenic metabolic reprogramming of HSC induced by TGF-β1 and inhibited liver fibrogenesis in vivo. Thus, the dual targeting of G9a and DNMT1 may provide a potential therapeutic approach for the treatment of liver fibrosis [56]. Targeting the complex epigenetic mechanisms involved in fibrogenesis with innovative molecules like CM272 may pave the way for better therapies.
Histone Demethylation in NAFLD and its Therapeutic Implications
In addition to HMTs, HDMs have been shown to be involved in NAFLD development. The histone demethylase Plant Homeodomain Finger 2 (PHF2) can erase the H3K9me2 mark. Mice with adenovirus overexpressing Phf2 in the liver showed that increased levels of DAG and TG were protected from insulin resistance and inflammation [57] (Figure 3A). The reason was that the overexpression of Phf2 could increase the level of SCD1, which catalyzes the desaturation of saturated fatty acids (SFA) to monounsaturated fatty acids (MuFA), and MuFA can prevent lipotoxicity. Consequently, PHF2 could prevent the progression to NASH with inflammation and fibrosis [57]. In addition, the lysine (K)-specific demethylase KDM7A belongs to the PHF2/PHF8 family of the Jumonji C (JmjC) domain-containing demethylase (JMJD demethylase) and has two identifiable domains: a PHD and a JmjC domain [58]. KDM7A overexpression could erase the H3K9me2 and H3K27me2 repressive markers on the DGAT2 promoter, thereby increasing the expression of DGAT2 and TG accumulation, which, finally, induced hepatic steatosis [59]. As SCD1 and DGAT2 enzymes are potential targets for the treatment of NAFLD and clinical trials are ongoing, PHF2 and KDM7A could provide potential therapeutic targets in treating NAFLD.
The KDM4 family is made up of four isoforms, KDM4A to D (also called JMJD2A to D). KDM4A, B, and C encode their respective proteins containing one JmjC, one JmjN, two PHD domains, and two Tudor domains. KDM4D is different from the other three isoforms, because it lacks both the PHD and Tudor domains [60,61]. KDM4D catalyzed the di-demethylation and tri-demethylation of H3K9, which stimulated TLR4 expression and triggered hepatic fibrogenesis by activating the NF-κB pathway. Meanwhile, KDM4D was significantly upregulated during HSC activation [60,62,63]. Unlike KDM4D, the other three isoforms were downregulated in HSC activation and facilitated the transcription of miR-29 together with SREBP2 to antagonize liver fibrosis [64]. In addition, KDM4B/JMJD2B could upregulate PPARγ2 and its target genes related to lipid droplet formation and fatty acid uptake by removing H3K9 methylation to promote hepatic steatosis [65]. A later study demonstrated that KDM4B plays a pivotal role in liver X receptor alpha (LXRα)mediated lipogenesis [66], which provided another mechanism for KDM4B in hepatic steatosis (Figure 3A).
In addition to KDM4s, the role of the KDM3 subfamily in the liver has also been studied. KDM3A, KDM3B, and KDM3C (JMJD1A-C) are roughly 50% identical at the amino acid level and are all capable of removing dimethyl and monomethyl marks from H3K9 and H4R3 and nonhistone proteins, to a lesser extent. PPARγ expression was epigenetically regulated by KDM3A/JMJD1A during HSC activation (Figure 3C). KDM3A knockdown led to the elevated expression of fibrosis markers in HSCs and a mouse liver fibrosis model [67]. KDM3C/JMJD1C facilitated nutrient signaling to elevate the triglyceride levels in the liver and plasma by promoting the expression of lipogenesis genes [68]. Mechanistically, USF-1 recruited JMJD1C into multiple lipogenic genes in the fed state to demethylate H3K9me2 and increase chromatin accessibility (Figure 3A). JMJD1C was phosphorylated at T505 by the mTOR complex, which enhanced the direct interaction between USF-1 and JMJD1C and transduced the nutrient signal [68]. Therefore, a potential treatment approach for hepatosteatosis and IR is to target JMJD1C phosphorylation by mTOR, a critical lipogenic insulin signaling cascade [68]. KDM6B/JMJD3, a H3K27 demethylase, is another notable HDM in liver metabolism and NAFLD [69]. Under fasting conditions, JMJD3 and SIRT1 worked synergistically to activate fatty acid oxidation genes, such as Fgf21, Cpt1a, and Mcad. The liver-specific downregulation of JMJD3 reduced fatty acid oxidation and led to hepatic steatosis [70] (Figure 3B).
Potential Targets and Compounds Modulating Histone Demethylase in NAFLD
Small molecules that activate JMJD3 or promote the interaction of JMJD3 with SIRT1 specifically decreased the lipid levels, which may provide a therapeutic approach to treat obesity and hepatosteatosis [70]. Furthermore, fasting conditions also induced Fibroblast Growth Factor-21 (FGF21) signaling, which required JMJD3 to activate hepatic autophagy and lipid degradation through upregulating the global autophagy network genes [71] (Figure 4). This study also demonstrated that FGF21 administration to alleviate fatty liver in HFD mice was mediated by JMJD3 [71]. Thus, targeting the histone demethylase JMJD3 could be a potential treatment for NAFLD.
GSK2879552 is an LSD1 inhibitor that showed beneficial effect to inhibit FASN expression and ameliorate hepatic steatosis in mice. Mechanistically, the transcription factor Slug was upregulated in hepatocytes by insulin in fed state, which recruited the histone K3K9 demethylase LSD1 to Fasn promoter and promote FASN expression. So GSK2879552 blocked lipogenesis activated by Slug-LSD1 pathway and may be a useful therapeutic rationale for the treatment of NAFLD [72]. In addition, Gomisin N (GN) is a phytochemical from Schisandra chinensis, exhibiting hepatoprotective, anti-cancer, and anti-inflammatory properties [73]. In one study, the administration of GN was found to downregulate the expression of PPARγ2 and JMJD2B in the liver of HFD-induced obese mice [65], which may contribute to the alleviation of HFD-induced hepatic steatosis. GN was also found to reduce the tunicamycin-induced hepatic ER stress and TG accumulation in mice [74]. Thus, GN might be helpful for the treatment of NAFLD, although it may not work mainly through histone demethylation.
Histone Acetylation in NAFLD and Its Therapeutic Implications
Histone acetylation is a post-translational modification and mostly occurs at specific lysine residues in the N-terminal tails of the histone H3 and H4. Histone acetylation is always associated with chromatin opening and transcriptional activation. Histone acetylation is usually quite dynamic and jointly determined by histone acetyltransferases (HATs) and histone deacetylases (HDACs), which add or remove the acetyl group on particular histone lysine residues.
Histone acetylation may affect the expression of individual critical genes. In HepG2, H3 and H4 acetylation at fatty acid synthase gene FASN promoter was transiently increased upon insulin stimulation in a manner of cross-regulation with ChREBP, although the HAT was not identified [75]. In addition, it has been demonstrated that liver-specific knockdown of nuclear receptor subfamily 2, group F, member 6 (NR2F6) alleviated obesity-associated hepatosteatosis and MCD diet-induced NASH through downregulating CD36 expression in mouse models [76]. NR2F6 bound directly to CD36 promoter in hepatocytes, recruited nuclear receptor coactivator 1 (SRC-1), a component of p300/CBP HAT complex, and promoted H3 acetylation on CD36 promoter [76]. Interestingly, NR2F6 expression was increased in the livers of NAFLD patients and reduced by metformin treatment in obese mice [76]. Therefore, NR2F6 antagonists might offer a therapeutic approach for treating NAFLD through histone acetylation.
In addition, histone acetylation may be involved in the regulation of multiple genes/ pathways simultaneously. Homozygous knock-in of a serine-to-alanine mutation at Ser196 (S196A) in LXRα to abolish the phosphorylation could affect the hepatic H3K27 acetylome and transcriptome during the progression of NAFLD. For example, the H3K27Ac at the Ces1f gene locus and the expression of Ces1f were high in the liver of LXRα-S196A mice comparing with WT mice when fed high-fat-high-cholesterol (HFHC) diet. Ces1f is a member of the carboxylesterase 1 family that controls hepatic lipid mobilization. Meanwhile, the H3K27Ac and expression of inflammation and fibrosis related genes including Spp1 and Col1e1 were reduced in S196A. So, LXRα-S196A could induce liver steatosis but prevent cholesterol accumulation, inflammation and fibrosis, thereby slowing the development from simple hepatic steatosis to NASH [77].
Potential Targets and Compounds Modulating Histone Acetylase in NAFLD
A few studies further suggested that histone acetylation can be a potential target for NAFLD. The active phosphorylated form of FTY720/fingolimod, a prodrug treating multiple sclerosis, could reduce FASN expression by histone acetylation alteration, inhibit ceramide production and hepatic steatosis in diet-induced NAFLD mice [78]. Tannic acid (TA), a HAT inhibitor, inhibited lipid accumulation in vivo and reduced the mRNA expression of genes associated with lipogenesis. Mechanistically, TA eliminated the occupancy of p300 on the sterol regulatory elements (SREs) in the promoters of FASN and ATP-citrate lyase (ACLY) genes, thereby decreasing acetylation of H3K9 and H3K36 [79].
The biguanide medicine metformin is the most popular anti-diabetic medication for the treatment of type 2 diabetes (T2D), which relieves hyperglycemia by reducing hepatic gluconeogenesis and improving insulin sensitivity [80][81][82]. It was reported that metformin activated AMPK, which directly phosphorylated and activated HAT1, promoted histone acetylation, and upregulated genes in mitochondria biogenesis [83]. Another study demonstrated that metformin promoted the phosphorylation of CBP at Ser436, which resulted in the dissociation of the CREB-CBP-TORC2 complex and downregulated the expression of the genes encoding gluconeogenic enzymes. Thus, metformin dramatically reduced the blood glucose level [84]. In addition, metformin increased hepatic protein levels of SIRT1 and GCN5 to inhibit hepatic gluconeogenesis [85].
Histone Deacetylation in NAFLD and Its Therapeutic Implications
The histone deacetylases (HDACs) family includes four distinct classes, namely class I, II, III and IV. HDAC1-3 and HDAC8 constitute Class I HDACs. Class II includes HDAC 4, 5, 6, 7, 9, and 10. On the other hand, class III HDAC enzymes, also known as sirtuins or silent information regulators (SIRT1-7), rely on NAD + as a cofactor. Class IV HDAC exclusively consist of HDAC11 [86].
HDACs promote a dense chromatin structure and inhibit transcription by deacetylating lysine residues. In a diet-induced NAFLD and HCC mouse model, SREBP1 directly upregulated the expression of HDAC8, which worked with EZH2 concordantly through H4 deacetylation and H3K27 trimethylation to repress Wnt antagonists, thereby activating the Wnt pathway [87,88]. Snail1, a zinc-finger transcription factor, was reported to repress the expression of FASN through recruiting HDAC1/2 to deacetylate H3K9 and H3K27 at FASN promoter. In HFD-fed mice, Snail1 overexpression in the liver decreased the insulin-stimulated lipogenesis in hepatocytes and attenuated the fatty liver. Conversely, disruption of the insulin-snail1 pathway may lead to NAFLD (Figure 3B) [89]. HDAC6 was also reported to deacetylate the transcription factor FOXO1. S100 calcium binding protein A11 (S100A11) was upregulated in NAFLD liver, and then blocked the interaction between HDAC6 and FOXO1 to stimulate lipogenesis and liver steatosis [90]. Here the role of HDAC6 is beneficial along the line of NAFLD prevention.
SIRT1 is a NAD + coenzyme-dependent histone deacetylase. Various studies have shown that SIRT1 can influence NAFLD through a variety of pathways. On one hand, SIRT1 is a unique upstream regulator of LKB1/AMPK sensing energy signaling [91]. Under fasting condition, the SREBP-1c acetylation level in mouse liver was consistently reduced and its interaction with SIRT1 was increased. The SREBP-1c-SIRT1 interaction was decreased after feeding, while SREBP-1c acetylation went up and better promoted lipogenesis [92]. In HFD model, liver-specific Sirt1 knockout impaired PPARα/PGC-1α signaling and reduced fatty acid oxidation, thereby resulting in increased hepatic steatosis. This provided compelling evidence for a significant association between SIRT1 and hepatic fatty acid metabolism [93]. In diet-induced and genetically obese mice, pharmacological SIRT1 activators suppressed the hepatic lipid and cholesterol levels as well as liver steatosis [94]. In addition, SIRT1 could also inhibit NF-κB activity to reduce the inflammatory response, which is a powerful defender against pathologic conditions like fatty liver [95,96]. Thus, interventions stimulating SIRT1 activity could potentially offer therapeutic benefits for the management of hepatic diseases and metabolic syndrome associated with obesity [93].
Like SIRT1, SIRT6 has been implicated in the negative regulation of lipid metabolism and inflammation. Liver-specific Sirt6 knockout mice exhibited a tendency to increase hepatic steatosis, inflammation and insulin resistance under high-fat and high-fructose (HFHF) diet through upregulated BTB domain and CNC homolog 1 (Bach1), a nuclear repressor of Nrf2 [97]. In a similar study, liver-specific Sirt6 deletion led to fatty liver through reduced β-oxidation and enhanced glycolysis, lipogenesis and TG synthesis [98]. Mechanistically, SIRT1 interacts with FOXO3a and NRF1 on SIRT6 promoter to positively regulate SIRT6 expression [98]. Indeed, multiple studies support the idea that SIRT6 promotes β-oxidation in a fasting state. In one study, SIRT6 regulated hepatic PPARα activity in vivo via deacetylation of the cofactor NCOA2 at K780 [99]. In parallel, SIRT6 could increase the activity of long-chain acyl-CoA synthase 5 (ACSL5) by deacetylating K98, K361 and K367 and promote fatty acid β-oxidation, thereby increasing the cellular lipid utilization and ultimately resisting the NAFLD process [100]. In addition, SIRT6 engaged in an interaction with acetyltransferase GCN5 and increased its activity, thereby suppressing hepatic gluconeogenesis. Therefore, hepatic SIRT6 activation may be therapeutically useful in the prevention of IR and NAFLD [101]. Another study demonstrated that SIRT6 overexpression in liver reduced steatosis, inflammation, and fibrosis caused by a HFHF diet, indicating the SIRT6 activator may be a promising therapeutic direction for treating NASH by reducing oxidative stress and inflammation [97]. Together, it is valuable to further explore the therapeutic agonists of SIRT1 and SIRT6 for the treatment of NAFLD.
Potential Targets and Compounds Modulating Histone Deacetylase in NAFLD HDAC chemical inhibitors have been developed to treat cancer, and many of them have also been tested in mice models of NAFLD or obesity. The treatment of mice with valproic acid (VPA), a class I and II HDAC inhibitor, could decrease collagen deposition and HSC activation in the CCl 4 model [102]. VPA is expected to have potential in preventing the further progression of liver fibrosis. Suberoylanilide hydroxamic acid (SAHA), another HDAC inhibitor, was shown to reduce liver fibrosis in rats through the suppression of TGF-β1 signaling [103]. In addition, sodium butyrate (NaB) increased H3K9Ac on the PPARα promoter, enhanced fatty acid oxidation, and inhibited the NF-κB inflammatory pathway, thereby alleviating NAFLD in the rat HFD model [104]. Thus, HDAC inhibitors may hold promise for the treatment of NAFLD.
Resveratrol (RSV), a natural polyphenol, not only exhibits anti-inflammatory and antioxidative characteristics but also activates SIRT1 [105]. RSV alleviated HFD-induced hepatic steatosis in mice liver and reduced lipid droplet accumulation in a SIRT1/ATF6dependent manner [106]. In addition to the studies on mice models, there have been clinical studies demonstrating resveratrol has the potential to improve NAFLD in patients [107,108].
Patients receiving a daily dose of 300 mg resveratrol for 3 months had lower ALT and aspartate transaminase (AST) levels, higher lipid metabolism, and less inflammation [108]. On the other hand, it was also reported that patient treated with 3000 mg of resveratrol daily for 8 weeks did not show an improvement in their ALT and AST levels, hepatic steatosis, and insulin resistance [109]. Therefore, the dose and long-term effects of RSV require more study. Furthermore, SRT1720 is a specific SIRT1 activator that prevents diet-induced obesity and insulin resistance through enhancing fatty acid oxidation in the liver, muscle, and brown adipose tissue in mice [110]. SRT1720 has a significant protective effect against NAFLD in monosodium glutamate (MSG) mice. Treatment with SRT1720 reduced the expression of markers of oxidative stress, as well as inflammatory cytokines [111].
Interestingly, Olaparib, a poly (ADP-ribose) polymerase (PARP) inhibitor approved for cancer treatment, could increase hepatic SIRT1 activity/expression in mice [112] and showed the potential to treat fatty liver disorders. Although upregulating SIRT1 is only one aspect of the multifaceted mechanism of Olaparib, mice receiving Olaparib for 5 weeks showed significantly attenuated liver injury, inflammation, and fibrosis in a MCD diet model [112]. The monoterpenic phenol carvacrol (CVL), which exists in a variety of essential oils from the Labiatae family plants, also upregulates SIRT1. It has prospective hepatoprotective and neuroprotective effects [113]. Combined CVL and rosiglitazone treatment in HFD-fed mice improves the symptoms of diabetes mellitus, such as the reduction in the ALT/AST, plasma glucose, and insulin levels [114].
In addition to SIRT1 and SIRT6, SIRT3 is a mitochondrial sirtuin that has been investigated widely, which is essential for the maintenance of mitochondrial functions [115]. Protocatechuic acid (PCA), also known as 3,4-dihydroxybenzoic acid, is a natural phenolic compound in various food plants. Although it also has antioxidant and anti-inflammatory properties [116,117], PCA bound and upregulated the SIRT3 protein to prevent liver damage and steatosis in HFD-fed mice, likely by regulating the acetylation and degradation of Acyl-CoA synthetase family member 3 (ACSF3) and fatty acid metabolism [118].
Histone Ubiquitination and Phosphorylation in NAFLD
Histone ubiquitination is the process by which the ubiquitin molecule is specifically conjugated to histones on lysine residues in the presence of a family of enzymes, such as when activating, binding, and degrading ubiquitin. The ubiquitination of histones plays a role in changing the conformation of chromatin and recruiting and activating downstream readers or chromatin regulator proteins. For example, the overexpression of RNF20 effectively inhibited IL-6, TNFα, and VEGFA to prevent TGF-β-induced hepatic fibrosis via the ubiquitination of H2BK120 (H2BK120ub) [119]. Unlike acetylation and methylation, histone H3 serine 10 (H3S10) phosphorylation is a marker for mitotic chromatin and affected by the counteractions between kinases and phosphatases. Histone phosphorylation often crosstalks with other histone modifications and may regulate the chromatin status upon DNA damage and other stresses. For example, it was reported that ChREBP bound to ChoRE to upregulate the expression of FASN through the histone acetylation, methylation, and phosphorylation of histone H3 serine 10 (H3S10) [120], promoting hepatic steatosis. There are relatively scarce reports on histone ubiquitination and phosphorylation in NAFLD, and more studies may be in the queue.
Conclusions and Future Perspectives
NAFLD is a chronic and progressive hepatic disorder characterized by the increase of an excessive amount of fat in the liver. This accumulation of fat induces stress and damage to hepatocytes, resulting in inflammation and fibrosis. If left untreated, these pathological processes can ultimately lead to the development of liver injury, cirrhosis, hepatocellular carcinoma, and, ultimately, mortality [121]. Epigenetic mechanisms, specifically histone modifications here, are dynamic and can be reversibly regulated by a variety of external cues, such as nutrient signals. The role and mechanisms of histone modifications and the related enzymes in NAFLD have been investigated, and some of the interesting findings are summarized here (briefly listed in Table 1). Some of the useful compounds affecting these histone modifications are briefly listed in Table 2. Although they may not be directly useful in the treatment of patients, they provide specific tools for target validation in cell and animal models. One of the noticeable features is that multiple histone modifications can be concordantly regulated and coordinately promote or inhibit specific gene networks. A few key cases are summarized in Figure 3 and may happen in different types of cells critical in NAFLD, such as hepatocytes and HSCs. Another feature worth mentioning is that many histone modifications are regulated by nutrient availability and status. Two cases are shown in Figure 4 to support this notion. More broadly, cofactors for histone modification enzymes include acetyl-CoA for acetyltransferases, S-adenosyl methionine for methyltransferases, and NAD + for Class III HDACs. The concentrations of these cofactors are comprehensively affected by energy metabolism and other physiological/pathological conditions. In this way, histone modifications and other epigenetic marks may capture and reflect the integrated state of external inputs. Gomisin N JMJD2B Alleviating HFD-induced hepatic steatosis and reducing hepatic ER stress in mice [65,74] Tannic acid HAT Inhibiting lipid accumulation in mice [79] Metformin HATs (p300/CBP/ GCN5) Reducing the blood glucose level and inhibiting hepatic gluconeogenesis in mice [83][84][85] Valproic acid HDAC Inhibiting HSC activation in mice [102] SAHA HDAC Reducing liver fibrosis in rats [103] NaB HDAC Enhancing fatty acid oxidation and inhibiting NF-κB inflammatory pathway in rats [104] Resveratrol SIRT1 Alleviating hepatic steatosis in mice; reducing ALT/AST and improving insulin resistance in human with NAFLD [106][107][108] SRT1720 SIRT1 Decreasing the expressions of marker genes for oxidative stress and inflammatory cytokines and ameliorating fatty liver in mice [110,111] Olaparib SIRT1 Decreasing hepatic triglyceride accumulation, inflammation and fibrosis in NASH mice [112] Carvacrol SIRT1 Decreasing hepatic marker enzymes (ALT/AST) activities in combination with rosiglitazone in mice [114] Protocatechuic acid SIRT3 Preventing liver damage and steatosis in mice and rats [118] As histone modification plays an important role in the development of NAFLD and may be reversely regulated, there is increasing interest in the development of novel therapies focusing on modulating epigenetic variations [57]. Histone-modifying enzymes may provide targets for NAFLD therapy. However, NAFLD is a complex disease associated with multiple metabolic disorders, and its potential side effects still need to be considered. Therefore, although there is interest in the development of histone modification enzyme inhibitors as the treatment for NAFLD, more research needs to be done to examine the possible side effects, to discover target-selective inhibitors, and carefully assess their effectiveness in patients. Targeting the complex epigenetic mechanisms in NAFLD with dual-inhibitory molecules may also be tried. The prospective unexplored potential in histone modifications remains to be investigated and released in the future.
Figure 1 .
Figure 1. The pathological spectrum and pathogenesis of NAFLD. NAFLD is a progressive disease, including a broad spectrum of liver conditions, from simple steatosis to NASH, with the potential to progress to more severe stages such as cirrhosis and HCC. High-fat or -fructose diets could increase the free fatty acids in the liver. Free fatty acids have a key role in the development of NAFLD, and this proceeds three ways in the liver.(1) FFAs enter the mitochondria and are oxidized to produce energy and ketone bodies.(2) Esterified to TG and stored in lipid droplets.(3) Secreted and excreted as VLDL. NASH is characterized by steatosis, inflammation, and fibrosis. Cellular stresses due to lipotoxicity cause cell death and liver injury, which ultimately leads to hepatic fibrosis after the activation of HSCs. NASH, nonalcoholic steatohepatitis; HCC, hepatocellular carcinoma; TG, triglyceride; VLDL, very-low-density lipoprotein; β-ox, β-oxidization; DNL, de novo lipogenesis; FFA, free fatty acid; HSC, hepatic stellate cell.
Figure 1 .
Figure 1. The pathological spectrum and pathogenesis of NAFLD. NAFLD is a progressive disease, including a broad spectrum of liver conditions, from simple steatosis to NASH, with the potential to progress to more severe stages such as cirrhosis and HCC. High-fat or -fructose diets could increase the free fatty acids in the liver. Free fatty acids have a key role in the development of NAFLD, and this proceeds three ways in the liver.(1) FFAs enter the mitochondria and are oxidized to produce energy and ketone bodies.(2) Esterified to TG and stored in lipid droplets.(3) Secreted and excreted as VLDL. NASH is characterized by steatosis, inflammation, and fibrosis. Cellular stresses due to lipotoxicity cause cell death and liver injury, which ultimately leads to hepatic fibrosis after the activation of HSCs. NASH, nonalcoholic steatohepatitis; HCC, hepatocellular carcinoma; TG, triglyceride; VLDL, very-low-density lipoprotein; β-ox, β-oxidization; DNL, de novo lipogenesis; FFA, free fatty acid; HSC, hepatic stellate cell.
Figure 4 .
Figure 4. Representative signaling mechanisms of NAFLD regulation via histone-modifying enzymes.(1)Under HFD feeding, insulin and glucose signaling activates ABL1 kinase, which then phosphorylates PPARγ2, and the PPARγ2-MLL4 complex forms to promote hepatic steatotic target genes.(2) Fasting initiates cAMP/PKA signaling, resulting in the phosphorylation of SIRT1 and the formation of a JMJD3-SIRT1-PPARα complex in hepatocytes to increase the expression of its own gene and SIRT1-targeted β-oxidation via H3K27me3 demethylation.
Figure 4 .
Figure 4. Representative signaling mechanisms of NAFLD regulation via histone-modifying enzymes.(1)Under HFD feeding, insulin and glucose signaling activates ABL1 kinase, which then phosphorylates PPARγ2, and the PPARγ2-MLL4 complex forms to promote hepatic steatotic target genes.(2) Fasting initiates cAMP/PKA signaling, resulting in the phosphorylation of SIRT1 and the formation of a JMJD3-SIRT1-PPARα complex in hepatocytes to increase the expression of its own gene and SIRT1-targeted β-oxidation via H3K27me3 demethylation.
Author
Contributions: Conceptualization, W. Q.; writing-original draft preparation, Y. S.; writingreview and editing, W. Q.; visualization, Y. S.; funding acquisition, W. Q. All authors have read and agreed to the published version of the manuscript.
Table 1 .
Summary of histone modifications and the corresponding mechanisms involved in NAFLD.
Table 2 .
Summary of histone modification-modulating compounds tested in NAFLD models.
==
Domain: Biology Medicine
|
[{"added": "2023-09-30T15:15:33.074Z", "created": "2023-09-27T00:00:00.000", "id": "263805214", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1422-0067/24/19/14653/pdf?version=1695830156", "pdf_hash": "24189757adece8d6e7ddd0616a785b7bdbbc8697", "pdf_src": "ScienceParsePlus", "provenance": "peS2o-0028.json.gz:7006", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "8499e8c8185b379dfe8256fe76d4ab25fdfadd4c", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Ameliorative Effects of Oyster Protein Hydrolysates on Cadmium-Induced Hepatic Injury in Mice
Cadmium (Cd) is a widespread environmental toxicant that can cause severe hepatic injury. Oyster protein hydrolysates (OPs) have potential effects on preventing liver disease. In this study, thirty mice were randomly divided into five groups: the control, Cd, Cd + ethylenediaminetetraacetic acid (EDTA, 100 mg/kg), and low/high dose of OPs-treatment groups (100 mg/kg or 300 mg/kg). After continuous administration for 7 days, the ameliorative effect of OPs on Cd-induced acute hepatic injury in Cd-exposed mice was assessed. The results showed that OPs significantly improved the liver function profiles (serum ALT, AST, LDH, and ALP) in Cd-exposed mice. Histopathological analysis showed that OPs decreased apoptotic bodies, hemorrhage, lymphocyte accumulation, and inflammatory cell infiltration around central veins. OPs significantly retained the activities of SOD, CAT, and GSH-Px, and decreased the elevated hepatic MDA content in Cd-exposed mice. In addition, OPs exhibited a reductive effect on the inflammatory responses (IL-1β, IL-6, and TNF-α) and inhibitory effects on the expression of inflammation-related proteins (MIP-2 and COX-2) and the ERK/NF-κB signaling pathway. OPs suppressed the development of hepatocyte apoptosis (Bax, caspase-3, and Blc-2) and the activation of the PI3K/AKT signaling pathway in Cd-exposed mice. In conclusion, OPs ameliorated the Cd-induced hepatic injury by inhibiting oxidative damage and inflammatory responses, as well as the development of hepatocyte apoptosis via regulating the ERK/NF-κB and PI3K/AKT-related signaling pathways.
Introduction
Cadmium (Cd) is a widespread environmental toxicant that poses a serious threat to human health [1]. Due to its high solubility in water, Cd can easily enter the human body through the food chain from polluted soils and water [2]. Cd exposure will cause metabolic dysfunction and eventually lead to irreversible damage to multiple organs [3,4]. About 50-70% of the absorbed heavy metal accumulates in the kidney and liver [5]. Acute Cd exposure primarily results in liver accumulation and hepatic injury [6]. The liver has been considered one of the main target organs of Cd [7]. Recent research has shown that acute Cd exposure leads to severe hepatic injury, accompanied by oxidative damage, inflammation, and apoptosis [8]. Therefore, reducing oxidative damage, ameliorating inflammatory response, and preventing the development of hepatocyte apoptosis may be practical strategies for the treatment of Cd-induced hepatic injury.
Protein hydrolysates from oysters (Crassostrea hongkongensis) have multiple health benefits, including anti-oxidation [9], anti-inflammatory [10], anti-apoptosis [11], anticancer, and other properties [12]. In previous studies, oyster-derived hydrolysates have been shown to be protective against D-galactosamine(D-GalN)-induced hepatic injury [13]. The peptides (SCAP1, SCAP3, and SCAP7) produced from oyster protein hydrolysis (OPs) present strong antioxidant and anti-cancer properties [14]. In addition, OPs are considered to be a safe and effective dietetic treatment for alcoholic liver disease by declining ethanolinduced oxidative stress and inflammation [15]. In addition, oyster ferritin was found to efficiently reduce the damage of heavy metals in mice [16]. The evidence suggests that OPs might be a potential candidate for ameliorating Cd-induced hepatic injury. Therefore, this study aimed to investigate the ameliorative effect of OPs on hepatic oxidative damage, inflammation, and apoptosis in Cd-exposed mice.
Sequence Analysis of the Main Peptides of OPs
As shown in Figure 1, the peaks of OPs were mostly in the range of 300 to 900 m/z. Overall, 177 peptides, with molecular weights ranging from 550.250 to 1387.697 Da, and an intensity ranging from 3,857,400 to 171,200,000, were identified from the OPs. The peptide fingerprinting of 40 characteristic peptides in OPs was analyzed using liquid chromatography-tandem mass spectrometry (LC-MS/MS). The scores for evaluating the matches between the theoretical and experimental mass spectrums were obtained by comparing the UniProt database; 20 peptide sequences with higher scores are listed in Table 1. Of interest, they contain a higher percentage of hydrophobic amino acids, such as proline (P, 37/180 residues in 20 peptides), valine (V, 14/180), and alanine (A, 8/180). Specific amino acid motifs, such as PVX, was repeated six times, and PxxP was repeated eight times, X being either a glycine, serine, or a proline residue, x being either an alanine, glycine, valine, threonine, asparagine, leucine, glutamic acid, aspartic acid, or an arginine, can be recognized. Hydrophobic amino acids proline or proline-rich peptides were reported to possess good anti-Cd, anti-oxidation, and anti-inflammatory properties [17][18][19][20][21][22]. Protein hydrolysates from oysters (Crassostrea hongkongensis) have multiple health benefits, including anti-oxidation [9], anti-inflammatory [10], anti-apoptosis [11], anti-cancer, and other properties [12]. In previous studies, oyster-derived hydrolysates have been shown to be protective against D-galactosamine(D-GalN)-induced hepatic injury [13]. The peptides (SCAP1, SCAP3, and SCAP7) produced from oyster protein hydrolysis (OPs) present strong antioxidant and anti-cancer properties [14]. In addition, OPs are considered to be a safe and effective dietetic treatment for alcoholic liver disease by declining ethanolinduced oxidative stress and inflammation [15]. In addition, oyster ferritin was found to efficiently reduce the damage of heavy metals in mice [16]. The evidence suggests that OPs might be a potential candidate for ameliorating Cd-induced hepatic injury. Therefore, this study aimed to investigate the ameliorative effect of OPs on hepatic oxidative damage, inflammation, and apoptosis in Cd-exposed mice.
Sequence Analysis of the Main Peptides of OPs
As shown in Figure 1, the peaks of OPs were mostly in the range of 300 to 900 m/z. Overall, 177 peptides, with molecular weights ranging from 550.250 to 1387.697 Da, and an intensity ranging from 3,857,400 to 171,200,000, were identified from the OPs. The peptide fingerprinting of 40 characteristic peptides in OPs was analyzed using liquid chromatography-tandem mass spectrometry (LC-MS/MS). The scores for evaluating the matches between the theoretical and experimental mass spectrums were obtained by comparing the UniProt database; 20 peptide sequences with higher scores are listed in Table 1. Of interest, they contain a higher percentage of hydrophobic amino acids, such as proline (P, 37/180 residues in 20 peptides), valine (V, 14/180), and alanine (A, 8/180). Specific amino acid motifs, such as PVX, was repeated six times, and PxxP was repeated eight times, X being either a glycine, serine, or a proline residue, x being either an alanine, glycine, valine, threonine, asparagine, leucine, glutamic acid, aspartic acid, or an arginine, can be recognized. Hydrophobic amino acids proline or proline-rich peptides were reported to possess good anti-Cd, anti-oxidation, and anti-inflammatory properties [17][18][19][20][21][22].
Composition of Amino Acid in OPs
According to the data from the automatic analyzer, the content of total amino acids in OPs was 33.73 g/100 g ( Table 2). The content of essential amino acids in OPs was 11.82 g/100 g and accounted for 35.04% of the total amino acids. The content of hydrophobic amino acids was 12.76 g/100 g and accounted for 37.83% of the total amino acids. OPs were rich in Glutamic acid (Glu, 4.33 g/100 g), Aspartic acid (Asp, 3.36 g/100 g), Alanine (Ala, 2.63 g/100 g), Proline (Pro, 2.61 g/100 g), and Lysine (Lys, 2.61 g/100 g). Table 3 shows the free amino acid content and composition of OPs. The contents of total free amino acids in OPs were 15.80 g/100 g, indicating OPs contained abundant free amino acids. Among these free amino acids, Pro (2.49 g/100 g), Glu (2.12 g/100 g), Tyr (1.51 g/100 g), Lys (1.43 g/100 g), Gly (1.38 g/100 g), and Val (0.91 g/100 g) were highly detected in OPs. The essential free amino acids accounted for 33.73% of the free total amino acids, and free hydrophobic amino acids accounted for 38.86%.
Effects of OPs on Hepatic Dysfunction in Cd-Exposed Mice
Compared with the control group, the Cd-exposed mice group showed the highest levels of serum alanine transaminase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), and lactate dehydrogenase (LDH) ( Figure 2). EDTA therapy is the most widely used for treating patients with acute or chronic Cd disease [23]. Thus, it was used as a positive control in this test. The supplement of OPs significantly decreased the levels of serum ALT, AST, ALP, and LDH (p < 0.01). The effects of OPs were met even better than that of the positive agent EDTA treatment. OPs exhibited a good ameliorative effect on hepatic dysfunction in Cd-exposed mice. amino acids. Among these free amino acids, Pro (2.49 g/100 g), Glu (2.12 g/100g), Tyr (1.51 g/100g), Lys (1.43 g/100 g), Gly (1.38 g/100 g), and Val (0.91 g/100 g) were highly detected in OPs. The essential free amino acids accounted for 33.73% of the free total amino acids, and free hydrophobic amino acids accounted for 38.86%.
Effects of OPs on Hepatic Dysfunction in Cd-exposed Mice
Compared with the control group, the Cd-exposed mice group showed the highest levels of serum alanine transaminase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), and lactate dehydrogenase (LDH) ( Figure 2). EDTA therapy is the most widely used for treating patients with acute or chronic Cd disease [23]. Thus, it was used as a positive control in this test. The supplement of OPs significantly decreased the levels of serum ALT, AST, ALP, and LDH (p < 0.01). The effects of OPs were met even better than that of the positive agent EDTA treatment. OPs exhibited a good ameliorative effect on hepatic dysfunction in Cd-exposed mice. Control: Intraperitoneal injection of 0.9% NaCl (saline) once daily; Cd: Mice were injected intraperitoneally with CdCl 2 5 mg/kg daily; EDTA: Mice were injected with CdCl 2 (5 mg/kg) intraperitoneally after 1 hour of oral administration with EDTA (100 mg/kg) daily; Cd+L-OPs: Mice were injected with CdCl 2 (5 mg/kg) intra-peritoneally after 1 hour of oral administration with a low dose of OPs (100 mg/kg) daily. Mice were injected with CdCl 2 (5 mg/kg) intra-peritoneally after 1 h of oral administration with a high dose of OPs (300 mg/kg) daily. The data were expressed as mean ± SEM, n = 6 in each group. Compared with the control group, ** p < 0.01; compared with the Cd group, ## p < 0.01.
Effect of OPs on Hepatic Injury in Cd-Exposed Mice
The organ weight coefficient is commonly used to evaluate the toxic effect [24]. The liver weight coefficient of the mice in the Cd-exposed group was significantly higher than that of the mice in the control group (p < 0.01). The OPs markedly lowered the liver coefficients in Cd-exposed mice (p < 0.01). Histopathological sections of the liver stained with H&E are shown in Figure 3B. Compared with the control group, the mice in the Cd group showed obvious pathological changes in liver tissue, including apoptotic bodies, hemorrhage, lymphocyte accumulation, and inflammatory cell infiltration around the central vein. In the OPs and EDTA-treated groups, liver tissue retained its normal appearance and had fewer apoptotic bodies. OPs showed a protective effect on liver tissue against Cd.
traperitoneally with CdCl2 5 mg/kg daily; EDTA: Mice were injected with CdCl2 (5 mg/kg) intraperitoneally after 1 hour of oral administration with EDTA (100 mg/kg) daily; Cd+L-OPs: Mice were injected with CdCl2 (5 mg/kg) intra-peritoneally after 1 hour of oral administration with a low dose of OPs (100 mg/kg) daily. Mice were injected with CdCl2 (5 mg/kg) intra-peritoneally after 1 hour of oral administration with a high dose of OPs (300 mg/kg) daily. The data were expressed as mean ± SEM, n = 6 in each group. Compared with the control group, ** p < 0.01; compared with the Cd group, # # p < 0.01.
Effect of OPs on Hepatic Injury in Cd-exposed Mice
The organ weight coefficient is commonly used to evaluate the toxic effect [24]. The liver weight coefficient of the mice in the Cd-exposed group was significantly higher than that of the mice in the control group (p < 0.01). The OPs markedly lowered the liver coefficients in Cd-exposed mice (p < 0.01). Histopathological sections of the liver stained with H&E are shown in Figure 3B. Compared with the control group, the mice in the Cd group showed obvious pathological changes in liver tissue, including apoptotic bodies, hemorrhage, lymphocyte accumulation, and inflammatory cell infiltration around the central vein. In the OPs and EDTA-treated groups, liver tissue retained its normal appearance and had fewer apoptotic bodies. OPs showed a protective effect on liver tissue against Cd. Figure 3. Effects of OPs on liver coefficient and hepatic injury in Cd-exposed mice. (A) Liver coefficient = liver weight(g)/mouse weight(g); (B) Histopathology with H&E staining (200×) of the liver in mice after treatment for 7 days; CV: Central veins; Arrow: lymphocyte accumulation in the parenchyma; asterisk (*): hemorrhage. Bar = 100 μm. The data were expressed as mean ± SEM, n = 6 in each group. Compared with the control group, ** p < 0.01; compared with the Cd group, # # p < 0.01.
Effect of OPs on Hepatic Oxidative Indexes in Cd-exposed Mice
An increased MDA level is an important indicator of oxidative stress. The Cd-exposed mice group showed the highest levels of hepatic MDA, while the OPs treatment significantly lowered the level of MDA (p < 0.01, Figure 4A). More importantly, the effects of OPs were better than that of EDTA. OPs markedly inhibited lipid peroxidation (MDA as an indicator) and MDA production in Cd-exposed mice. As shown in Figure 4B-D, antioxidant markers such as SOD, CAT, and GSH-Px were significantly reduced in the Cd-exposed mice group compared to the control group. OPs retained higher activity of Figure 3. Effects of OPs on liver coefficient and hepatic injury in Cd-exposed mice. (A) Liver coefficient = liver weight(g)/mouse weight(g); (B) Histopathology with H&E staining (200×) of the liver in mice after treatment for 7 days; CV: Central veins; Arrow: lymphocyte accumulation in the parenchyma; asterisk (*): hemorrhage. Bar = 100 µm. The data were expressed as mean ± SEM, n = 6 in each group. Compared with the control group, ** p < 0.01; compared with the Cd group, ## p < 0.01.
Effect of OPs on Hepatic Oxidative Indexes in Cd-Exposed Mice
An increased MDA level is an important indicator of oxidative stress. The Cd-exposed mice group showed the highest levels of hepatic MDA, while the OPs treatment significantly lowered the level of MDA (p < 0.01, Figure 4A). More importantly, the effects of OPs were better than that of EDTA. OPs markedly inhibited lipid peroxidation (MDA as an indicator) and MDA production in Cd-exposed mice. As shown in Figure 4B-D, antioxidant markers such as SOD, CAT, and GSH-Px were significantly reduced in the Cd-exposed mice group compared to the control group. OPs retained higher activity of antioxidant enzymes in Cd-exposed mice. OPs exhibited a strong reductive effect on Cd-induced oxidative stress in the liver.
Mar. Drugs 2022, 20, 758 6 of 20 antioxidant enzymes in Cd-exposed mice. OPs exhibited a strong reductive effect on Cdinduced oxidative stress in the liver. . The data were expressed as mean ± SEM, n = 6 in each group. Compared with the control group, ** p < 0.01; compared with the Cd group, # p < 0.05, # # p < 0.01.
Effect of OPs on The Hepatic Inflammatory Response (IL-1β, IL-6, TNF-α) in Cd-exposed Mice
Interleukin IL-1β and IL-6 are two stimulators of the hepatic synthesis of acute-phase proteins in the inflammatory response to stress and important biological markers of hepatic inflammation [25,26]. TNF-α is an important biological marker in substantial hepatic tissue damage [27]. As shown in Figure 5A-C, the mice in the Cd-exposed group have the highest level of hepatic inflammatory cytokines (IL-1β, IL-6, and TNF-α) (p < 0.01). OPs significantly attenuated Cd-induced the high level of hepatic IL-1β, IL-6, and TNF-α (p < 0.01 and p < 0.05, respectively). The results from quantitative reverse-transcription PCR analysis (qRT-PCR) manifested that OPs inhibited the expression of hepatic IL-1β, IL-6, and TNF-α in Cd-exposed mice (p < 0.01). These results revealed that the protection of OPs against Cd was associated with its attenuation of the Cd-induced hepatic inflammation in Cd-exposed mice. . The data were expressed as mean ± SEM, n = 6 in each group. Compared with the control group, ** p < 0.01; compared with the Cd group, # p < 0.05, ## p < 0.01.
Effect of OPs on The Hepatic Inflammatory Response (IL-1β, IL-6, TNF-α) in Cd-Exposed Mice
Interleukin IL-1β and IL-6 are two stimulators of the hepatic synthesis of acute-phase proteins in the inflammatory response to stress and important biological markers of hepatic inflammation [25,26]. TNF-α is an important biological marker in substantial hepatic tissue damage [27]. As shown in Figure 5A-C, the mice in the Cd-exposed group have the highest level of hepatic inflammatory cytokines (IL-1β, IL-6, and TNF-α) (p < 0.01). OPs significantly attenuated Cd-induced the high level of hepatic IL-1β, IL-6, and TNF-α (p < 0.01 and p < 0.05, respectively). The results from quantitative reverse-transcription PCR analysis (qRT-PCR) manifested that OPs inhibited the expression of hepatic IL-1β, IL-6, and TNF-α in Cd-exposed mice (p < 0.01). These results revealed that the protection of OPs against Cd was associated with its attenuation of the Cd-induced hepatic inflammation in Cd-exposed mice. , and TNF-α in different groups. These data are expressed as the mean ± SEM, n = 6 in each group. Compared with the control group, ** p < 0.01; compared with the Cd group, # p < 0.05, # # p < 0.01.
Effect of OPs on The Expression of Hepatic COX-2, MIP-2, NF-κB, and p-ERK in Cdexposed Mice
COX-2 is a key enzyme in initiating hepatic inflammatory reactions [28]. Meanwhile, macrophage inflammatory protein (MIP)-2 is a potent neutrophil attractant and activator, contributing to the pathogenesis of inflammatory diseases [29]. MIP-2 and COX-2 would be elevated in Cd-induced inflammation [30]. As shown in Figure 5A, enhanced COX-2 and MIP-2 staining were observed around the central vein of hepatocytes in the Cd-exposed group. OPs treatment noticeably reduced the hepatic COX-2 and MIP-2 levels.
In the process of Cd-induced inflammation, the extracellular signal-regulated kinase (ERK) signal pathway would be activated, and the nuclear factor-κB (NF-κB) subsequently was up-regulated [31]. Western blotting assays illustrated that the expression of NF-κB and p-ERK were highly induced in the Cd-exposed group. However, the OPs treatment effectively dampened the expression of NF-κB and p-ERK (p < 0.01; Figure 6B,C). The above results implied that OPs might alleviate hepatic inflammation by inhibiting the expression of inflammatory activators (COX-2 and MIP-2) and related inflammatory pathways (NF-κB and ERK). , and TNF-α in different groups. These data are expressed as the mean ± SEM, n = 6 in each group. Compared with the control group, ** p < 0.01; compared with the Cd group, # p < 0.05, ## p < 0.01.
Effect of OPs on
The Expression of Hepatic COX-2, MIP-2, NF-κB, and p-ERK in Cd-Exposed Mice COX-2 is a key enzyme in initiating hepatic inflammatory reactions [28]. Meanwhile, macrophage inflammatory protein (MIP)-2 is a potent neutrophil attractant and activator, contributing to the pathogenesis of inflammatory diseases [29]. MIP-2 and COX-2 would be elevated in Cd-induced inflammation [30]. As shown in Figure 6A, enhanced COX-2 and MIP-2 staining were observed around the central vein of hepatocytes in the Cd-exposed group. OPs treatment noticeably reduced the hepatic COX-2 and MIP-2 levels.
In the process of Cd-induced inflammation, the extracellular signal-regulated kinase (ERK) signal pathway would be activated, and the nuclear factor-κB (NF-κB) subsequently was up-regulated [31]. Western blotting assays illustrated that the expression of NF-κB and p-ERK were highly induced in the Cd-exposed group. However, the OPs treatment effectively dampened the expression of NF-κB and p-ERK (p < 0.01; Figure 6B,C). The above results implied that OPs might alleviate hepatic inflammation by inhibiting the expression of inflammatory activators (COX-2 and MIP-2) and related inflammatory pathways (NF-κB and ERK).
Effect of OPs on Hepatic Apoptosis in Cd-exposed Mice
Apoptosis-related mitochondrial Bcl2-associated X protein (Bax) and Caspase-3 are two important pro-apoptotic factors. Under the stimulation of oxidative stress caused by Cd, the hepatic Bax increased, then the downstream Caspase-3 was up-regulated, and eventually, apoptosis occurred [32]. Anti-apoptotic Bcl-2 plays a central regulatory role in apoptosis [33]. Accordingly, we examined the effect of OPs on Cd-induced hepatic apoptosis by measuring the levels of pro-apoptotic factors (Bax and caspase-3) and anti-apoptotic factor Bcl-2 in Cd-exposed mice. As shown in Figure 7A-C, Cd significantly decreased the levels of anti-apoptotic Bcl-2 but increased the levels of pro-apoptotic factors (Bax and caspase-3) (p < 0.01). On the contrary, OPs treatment significantly increased Bcl-2 while decreasing Bax and caspase-3 levels in Cd-exposed mice (p < 0.01). The qRT-PCR results showed that OPs significantly induced the expression of Bcl-2 while suppressing the expression of Bax and caspase-3 in Cd-exposed mice (p < 0.01). OPs exhibited a strong anti-apoptotic effect on Cd-induced apoptosis in mice.
Cd can regulate the PI3K/AKT signaling pathway to induce apoptosis [34,35]. Additionally, PI3K/AKT signaling pathway also plays a crucial role in the regulation of inflammatory protein expressions (COX-2 and MIP-2) [36]. Western blotting assays demonstrated that Cd exposure led to the elevation of the expression of PI3K and p-AKT, accompanied by the imbalance of pro-/anti-apoptotic proteins (Bax, caspase-3 and Bcl-2) (p < 0.05). By contrast, the OPs treatment inhibited the activation of the PI3K/AKT signaling pathway and restored the balance of pro-/anti-apoptotic proteins in Cd-exposed mice. The results implied that OPs might alleviate hepatic apoptosis by restoring the balance of pro-
Effect of OPs on Hepatic Apoptosis in Cd-Exposed Mice
Apoptosis-related mitochondrial Bcl2-associated X protein (Bax) and Caspase-3 are two important pro-apoptotic factors. Under the stimulation of oxidative stress caused by Cd, the hepatic Bax increased, then the downstream Caspase-3 was up-regulated, and eventually, apoptosis occurred [32]. Anti-apoptotic Bcl-2 plays a central regulatory role in apoptosis [33]. Accordingly, we examined the effect of OPs on Cd-induced hepatic apoptosis by measuring the levels of pro-apoptotic factors (Bax and caspase-3) and antiapoptotic factor Bcl-2 in Cd-exposed mice. As shown in Figure 7A-C, Cd significantly decreased the levels of anti-apoptotic Bcl-2 but increased the levels of pro-apoptotic factors (Bax and caspase-3) (p < 0.01). On the contrary, OPs treatment significantly increased Bcl-2 while decreasing Bax and caspase-3 levels in Cd-exposed mice (p < 0.01). The qRT-PCR results showed that OPs significantly induced the expression of Bcl-2 while suppressing the expression of Bax and caspase-3 in Cd-exposed mice (p < 0.01). OPs exhibited a strong anti-apoptotic effect on Cd-induced apoptosis in mice.
Cd can regulate the PI3K/AKT signaling pathway to induce apoptosis [34,35]. Additionally, PI3K/AKT signaling pathway also plays a crucial role in the regulation of inflammatory protein expressions (COX-2 and MIP-2) [36]. Western blotting assays demonstrated that Cd exposure led to the elevation of the expression of PI3K and p-AKT, accompanied by the imbalance of pro-/anti-apoptotic proteins (Bax, caspase-3 and Bcl-2) (p < 0.05). By contrast, the OPs treatment inhibited the activation of the PI3K/AKT signaling pathway and restored the balance of pro-/anti-apoptotic proteins in Cd-exposed mice. The results implied that OPs might alleviate hepatic apoptosis by restoring the balance of pro-/anti-apoptotic proteins via inhibiting the PI3K/AKT signaling pathway in Cd-exposed mice.
Discussion
As one of the main target organs of Cd, acute hepatic injury was observed in Cdexposed mice. Fortunately, the OPs treatment clearly ameliorated the Cd-induced hepatic injury in this study. In particular, oxidative damage, inflammation, and cell apoptosis, as crucial triggers and contributors to the development of Cd-induced hepatic injury [30,34,35,37], were improved after OPs application in Cd-exposed mice.
Extensive literature indicates that the health benefits of protein hydrolysates may be partly attributed to their rich in free amino acids and peptides [38,39]. Extracts rich in free amino acids can be used in pharmaceutical applications [40]. A recent study found that free amino acids were related to the antioxidant property of protein hydrolysates of mackerel [38]. The protein hydrolysates with higher contents of free amino acids exhibited better antioxidant properties [41] and metal-chelating ability [42]. In the present study, a high level of free amino acids (15.8%) was detected in OPs. According to a previous report, the royal jelly hydrolysates with 8.389% of free amino acids had a stronger antioxidant activity than those of royal jelly with 0.572% of free amino acids [43]. A similar study also found the anchovy sprat hydrolysates with higher contents of free amino acids (16.28%- (F) The quantitative densitometric analysis of PI3k, p-AKT, Bax, caspase-3 and Bcl-2. These Data are expressed as the mean ± SEM, n = 6 in each group. Compared with the control group, * p < 0.05, ** p < 0.01; compared with the Cd group, # p < 0.05, ## p < 0.01.
Discussion
As one of the main target organs of Cd, acute hepatic injury was observed in Cdexposed mice. Fortunately, the OPs treatment clearly ameliorated the Cd-induced hepatic injury in this study. In particular, oxidative damage, inflammation, and cell apoptosis, as crucial triggers and contributors to the development of Cd-induced hepatic injury [30,34,35,37], were improved after OPs application in Cd-exposed mice.
Extensive literature indicates that the health benefits of protein hydrolysates may be partly attributed to their rich in free amino acids and peptides [38,39]. Extracts rich in free amino acids can be used in pharmaceutical applications [40]. A recent study found that free amino acids were related to the antioxidant property of protein hydrolysates of mackerel [38]. The protein hydrolysates with higher contents of free amino acids exhibited better antioxidant properties [41] and metal-chelating ability [42]. In the present study, a high level of free amino acids (15.8%) was detected in OPs. According to a previous report, the royal jelly hydrolysates with 8.389% of free amino acids had a stronger antioxidant activity than those of royal jelly with 0.572% of free amino acids [43]. A similar study also found the anchovy sprat hydrolysates with higher contents of free amino acids (16.28-27.53%) exhibited stronger ferrous-chelating activity and radical-scavenging activity compared to those with lower contents of free amino acids (9.05%) [44]. These data indicated that OPs were rich in free amino acids, which may contribute to the potential health benefits of OPs against Cd toxicity.
Serum ALT and AST are leaked from damaged hepatocytes [45,46]. Cd intoxication led to a significant elevation in the levels of ALT and AST [47]. In the present study, significant improvements were observed in the hepatic injury and dysfunction biomarkers (serum AST, ALT, ALP, and LDH) in Cd-exposed mice after the OPs treatment. Compared to the Cd group, OPs significantly decreased hepatic dysfunction biomarkers in a dosedependent manner. This result is in line with an earlier report, in which oyster protein hydrolysate could reduce hepatic dysfunction biomarkers (serum AST, ALT, and ALP) and inflammatory response in alcoholic liver disease mice [48]. Likewise, Shi, Sun [49] reported that ganoderma lucidum peptides have an alleviative effect on D-GalN-induced hepatocellular injury via reversing AST and ALT levels in the liver. Moreover, Mumtaz, Ali [50] found that elevated level of LDH, AST, and ALT in the Cd-exposed batch was improved by ascorbic acid. Early evidence indicates that hepatic injury and cirrhosis usually lead to metabolic disturbances of amino acids [51]. The bioactive properties of protein hydrolysates mainly depend on free amino acids and peptides [38,39,52]. The present study showed that OPs are rich in hydrophobic free amino acids (i.e., Pro) and proline-rich peptides. Among these amino acids, Pro plays a beneficial role in plants under changing environments, including Cd stress [53]. Exogenous Pro could increase antioxidant enzyme activities and confer tolerance to cadmium stress in cultured tobacco cells [22]. Pro has shown tissue-protective effects against D-galactosamine-induced hepatic injury [54]. Dietary Pro could effectively decrease AST and ALT levels of shrimp under NH3 stress [55]. The derivatives of Pro, N-acetyl-seryl-aspartyl-lysyl-proline, were found to attenuate bile duct ligation-induced liver fibrosis by restoring hepatic dysfunction (serum AST and ALT) in mice [56]. Pro and proline-rich proteins are often implicated in stress tolerance in plants [57][58][59]. Salivary proline-rich proteins possess good antioxidant properties [60]. Hypothalamic proline-rich polypeptides were found to protect brain neurons in aluminum neurotoxicosis [61]. These data support the idea that OPs could ameliorate hepatic injury and improve hepatic dysfunction in Cd-exposed mice.
Oxidative stress is often implicated in the induction of multi-organ injury under Cd exposure [62]. Lipid peroxidation is a major consequence of Cd-induced oxidative stress [63]. The consequences of the peroxidative of membrane lipids have been considered in relation to the tissue aspects of liver injury, and these peroxidative reactions play a critical role in the pathogenesis of acute liver necrosis [64]. According to a previous report, the liver, kidneys, and heart were most susceptible to Cd-induced oxidative stress in mice [65]. Some amino acid derivatives, such as N-Acetylcysteine, showed ameliorative effects on cisplatininduced multiple organ toxicity in rats [66][67][68]. Betulinic acid was found to alleviate the kidney and liver damage induced by Cd [69]. In this study, Cd exposure induced serious hepatic toxicity and oxidative stress, which were significantly improved after the OPs supplement. These data suggest that amelioration of hepatic oxidative injury may be the key to the treatment of Cd toxicity by OPs in mice.
Oxidative stress plays a crucial role in Cd-induced hepatic toxicity [70]. The development of liver injury usually involves the lipid peroxidation of hepatic cell membranes in Cd-exposed mice [71]. According to a recent report, Cd-induced hepatic injury is tightly coupled with enhanced lipid peroxidation (MDA) and the significant depletion of antioxidants (CAT and SOD) [72]. In this study, the OPs supplement clearly reduced the formation of MDA and significantly restored the activity of antioxidant enzymes (SOD, CAT, and GPH-Px) in the liver of the Cd-exposed mice. OPs displayed a strong antioxidant activity, which might also be attributed to their abundance of hydrophobic amino acids. Commonly, protein hydrolysates with higher content of hydrophobic amino acids possess better antioxidant properties due to their more effective interaction with lipid-soluble free radicals and the prevention of lipid peroxidation [17,[73][74][75]. Thus, this evidence clearly indicated that OPs possess good antioxidant properties to delay hepatic oxidative injury via retaining hepatic antioxidant enzymes and preventing MDA production in Cd-exposed mice.
Hepatic histopathological damage in Cd-exposed mice is characterized by apoptotic bodies, hemorrhage, lymphocyte accumulation, and inflammatory cell infiltration in liver tissue. Increasing evidence demonstrates that hepatic injury and fibrosis are accompanied by the elevation of the inflammatory response [76]. As well known, TNF-α and IL-6 are two key inflammatory mediators of tissue injury-induced inflammatory response [77]. IL-1β and IL-6 are two stimulators of the hepatic synthesis of acute-phase proteins in the inflammatory response to stress [25,26]. Cd exposure will trigger an acute inflammatory response in mice [78]. The present study showed that hepatic apoptotic cells in Cd-exposed mice were significantly minimized, and histopathological appearance was obviously improved after OPs treatment. In addition, the Cd-triggered inflammatory responses (IL-1β, IL-6, and TNF-α) were significantly inhibited, as expected. This result is consistent with the findings in an earlier report, in which peptides from oyster soft tissue hydrolysates selectively repressed TNF-α, IL-1β, and IL-6 [79]. To go even further, we found that as important activators and regulators of inflammatory responses, MIP-2 [29], COX-2 [28], NF-κB, and the ERK signal pathway [80], were significantly stimulated in Cd-exposed mice. The results are in agreement with a previous study, in which Cd activated the ERK signal pathway, then subsequently up-regulated TNF-α, COX-2, IL-1β, IL-6, and NF-κB in swine [31]. Likewise, Huang, Xia [30] reported that Cd exposure led to an increase in MIP-2 and COX-2. Actually, ROS production could activate MAPK signaling to induce inflammation and skin aging by promoting the phosphorylation of ERK [32]. Peng, Chen [81] also found that the upregulated ERK phosphorylation in ultraviolet B-exposed mice was significantly inhibited by the application of oyster protein hydrolysates. A recent study found that seahorse protein hydrolysates could significantly inhibit p-ERK levels in ethanol-exposed cells. [82]. In this study, Western blotting assays showed that OPs significantly decreased the levels of p-ERK and NF-κB proteins, as well as the MIP-2 and COX-2 in Cd-exposed mice. Therefore, the ameliorative effect of OPs in Cd-caused liver injury may be related to its anti-inflammation properties via suppressing the production of inflammatory mediators and inhibiting the inflammatory response associated with NF-κB and the ERK signal pathway.
In addition to inflammatory responses, hepatic injury is also accompanied by the development of apoptosis in Cd-exposed mice. Reducing Cd-induced apoptosis is also considered to be one of the feasible ways to prevent Cd-induced hepatic injury [69]. In the process, the NF-κB inflammation pathway indirectly activated the apoptosis-related factors Bcl-2, Bax, and Caspase-3 [31]. The present study revealed that the OPs supplement strongly up-regulated the expression of the anti-apoptotic factor Bcl-2 while significantly down-regulated the expression of the pro-apoptotic factors (caspase-3 and Bax), eventually restoring the balance of pro-/anti-apoptotic proteins in Cd-exposed mice. Moreover, the PI3K/AKT pathway is an important signaling pathway associated with apoptosis [83]. Actually, Cd selectively induces MIP-2 and COX-2 through the activation of the PI3K/AKT [30]. A previous study showed that curcumin alleviated lipopolysaccharideinduced hepatic injury and apoptosis via inhibiting the PI3K/AKT and NF-κB pathways [84]. MiR-130a alleviated neuronal apoptosis and changes in the expression of Bcl-2/Bax and caspase-3 in cerebral infarction rats through the PI3K/AKT signaling pathway [85]. In the present study, we found that the OPs supplement significantly inhibited the expression of PI3K and p-AKT proteins. These results are in agreement with an earlier report, in which Selenomethionine ameliorated Cd-induced hepatocyte apoptosis by suppressing the PI3K/AKT pathway [86]. Therefore, we may conclude that OPs possess the ability to ameliorate hepatocyte apoptosis, possibly by restoring the balance of pro-/anti-apoptotic proteins via suppressing the PI3K/ AKT pathway in Cd-exposed mice.
In conclusion, from our study, we found that OPs could effectively ameliorate Cdinduced hepatic injury through their antioxidative and anti-inflammatory properties. In addition, OPs displayed an important role in restoring the balance between pro-apoptotic and anti-apoptotic proteins by suppressing the activation of the PI3K/AKT pathway, contributing to the development of hepatocyte apoptosis in Cd-exposed mice (Figure 8). These results may provide a new insight for a better understanding of the ameliorative function of OPs to Cd toxicity and provide a theoretical basis for the use of OPs to prevent or treat Cd-induced hepatic injury.
Mar . Drugs 2022, 20, 758 12 of 20 addition, OPs displayed an important role in restoring the balance between pro-apoptotic and anti-apoptotic proteins by suppressing the activation of the PI3K/AKT pathway, contributing to the development of hepatocyte apoptosis in Cd-exposed mice (Figure 8). These results may provide a new insight for a better understanding of the ameliorative function of OPs to Cd toxicity and provide a theoretical basis for the use of OPs to prevent or treat Cd-induced hepatic injury.
Animals and Experimental Design
Thirty Specific-Pathogen-Free (SPF) mice (Kunming mice, 25-35 g) were obtained from Changsha Tianqin Biotechnology Co., Ltd. (Changsha, China), and the animals were maintained at the Guangdong Ocean University Animal Centre under light (12 h of light and dark) and temperature (~ 25 °C). The animals were given a standard laboratory diet and water. The experiment was approved by the Animal Ethics Committee of Guangdong Ocean University (No.: GDOU-LAE-2020-009). The animals were randomly divided into 5 groups (n = 6). Control group: Intraperitoneal injection of 0.9% NaCl (saline) once daily. Cd-exposed group: the mice were injected intraperitoneally with CdCl2 5 mg/kg daily [87].
Animals and Experimental Design
Thirty Specific-Pathogen-Free (SPF) mice (Kunming mice, 25-35 g) were obtained from Changsha Tianqin Biotechnology Co., Ltd. (Changsha, China), and the animals were maintained at the Guangdong Ocean University Animal Centre under light (12 h of light and dark) and temperature (~25 • C). The animals were given a standard laboratory diet and water. The experiment was approved by the Animal Ethics Committee of Guangdong Ocean University (No.: GDOU-LAE-2020-009). The animals were randomly divided into 5 groups (n = 6). Control group: Intraperitoneal injection of 0.9% NaCl (saline) once daily. Cd-exposed group: the mice were injected intraperitoneally with CdCl 2 5 mg/kg daily [87]. EDTA-treated group (positive control): the mice were injected with CdCl 2 (5 mg/kg) intraperitoneally after 1 h of oral administration with EDTA (100 mg/kg) daily. The low dose of OPs(L-OPs)-treated group: the mice were injected with CdCl 2 (5 mg/kg) intraperitoneally after 1 h of oral administration with OPs (100 mg/kg) daily. High dose of OPs (H-OPs)-treated group: the mice were injected with CdCl 2 (5 mg/kg) intraperitoneally after 1 h of oral administration with OPs (300 mg/kg) daily. It has been shown that ethylenediaminetetraacetic acid (EDTA) can alleviate cadmium toxicity by enhancing antioxidant enzyme activity and inhibiting inflammatory responses [88]. Therefore, it can be used as a positive control. The doses of EDTA and OPs were determined based on previous studies [89,90]. After 7 days, the mice were executed by cervical dislocation, and the liver was stored at −80 • C for further analysis.
Preparation of OPs
OPs were prepared by enzymatic hydrolysis from the oyster (Crassostrea hongkongensis) meat as described previously [81,90]. Hong Kong oyster meat (~3 kg) was homogenized in distilled water (1:3 (w/w) at 8000 rpm for 5 min. Homogenized oysters were hydrolysed using neutral protease (2 × 10 5 U/g, Pangbo Biotech, Nanning, China) at a protease/substrate ratio of 2.0% (w/v) (pH 7.0). The neutral protease was incubated for 4 h at 50 • C and then inactivated at 100 • C for 15 min. The hydrolysate was centrifuged at 15,000 rpm for 20 min at 4 • C to obtain the supernatant. The supernatant was collected and ultrafiltered using a membrane bioreactor system with a molecular mass cut-off value of 3 kDa to collect the fractions (<3 kDa). The samples were collected and freeze-dried for further analysis.
Peptide Sequence Analysis Based on LC-MS/MS
The peptide sequence analysis used an Easy-nLC 1200 system coupled to a Q-Exactive quadrupole-Orbitrap mass spectrometry (Thermo Fisher Scientific, San Jose, CA, USA). One µL of the samples was injected with an autosampler into an Acclaim Pep Map RPLC C18 column (150 µm i.d. × 150 mm, C18, particle size: 1.9 µm, pore size: 100 Å, Dr. Maisch GmbH, Darmstadt, Hessen, Germany) with mobile phase A: 0.1% formic acid in water; B: 0.1% formic acid in the water, 80% acetonitrile. The flow rate was 600 nL/min, and the LC linear gradient ranged from 4% to 40% for 55 minutes and 10 minutes at 95% mobile phase B. Finally, the molecular mass, sequence, peak area (with respect to base peak), and relative peak area (peak area/total peak area) of the peptides were identified and calculated as previously described [81,91]. The conditions of the mass spectrometer were as follows: Resolution: 70,000, AGC target: 3e6, NCE/stepped NCE: 28. The samples were analyzed with a full-scan MS mode in the range of 100-1500 m/z to obtain the total ion chromatogram. The raw MS files were analyzed and searched against the target protein database based on the species of the samples using Byonic.
LC-MS/MS Analysis of Free Amino Acids
The amino acid composition and content of the OPs were measured, as previously described, with little modification [92]. The OPs samples were accurately weighed to 50 mg and mixed with 600 µL of a water-methanol solution (1:1, v/v) with 10% formic acid in a 2 mL tube. Then, 100 mg of glass beads were added to the mixed samples and vortexed for 30 s. The samples were transferred to a high-throughput tissue grinding machine (MB-96, Meibi, Jiaxing, Zhejiang, China) and vibrated at 60 Hz for 2 min. The tube was centrifuged at 12,000 rpm for 5 min at 4 • C. Ten µL of supernatant was transferred to a new tube containing 490 µL of the water-methanol solution (1:1, v/v) with 10% formic acid and then vortexed for 30 s. Then, 100 µL of the diluted samples were mixed with 100 µL of 100 µg/L double isotope internal standard (Trp-d3, D87103, Medical Isotopes, USA) and vortexed for 30 s. The mixed samples were filtered through a 0.22 µm hydrophilic PTFE filter and transferred into a labeled vial, and subsequently analyzed via LC-MS/MS.
Mass spectrometric analysis was performed using an AB SCIEX AB4000 Mass Spectrometer (AB SCIEX, Framingham, MD, USA) equipped with an electrospray ionization (ESI) source using the following parameters: capillary voltage: 5500 V, temperature of the turbo heaters: 500 • C, nebulizer gas (GS1): 50 psi, auxiliary gas (GS2): 50 psi, and curtain gas (CUR): 30 psi, Collision Gas: 6 psi. All of the amino acids were detected in the multiple reaction monitoring mode (MRM).
Analysis of Amino Acid Composition
According to the previous method with a slight modification [81]. Approximately 30 mg of the sample and 10 mL of 6 mol/L HCl were added to a hydrolysis tube containing phenol. After the tube was vacuumed, the mixture was washed with nitrogen and hydrolyzed at 110 • C for 22 h. After cooling to room temperature, the filtrate is filtered and spun dry under reduced pressure in a centrifuge tube. The 0.02 mol/L HCl solution was added to a dried centrifuge tube and dissolved, and the solution was transferred to the upper sample bottle and determined using an amino acid analyzer (L-8900, Hitachi, Tokyo, Japan). Then, the contents of amino acids in the sample could be determined according to the peak area in comparison with the standard. The content of tryptophan in OP was determined after hydrolysis with 6 mol/L of NaOH instead of HCl.
Histopathology Examination
The liver samples were fixed in 4% paraformaldehyde for 24 hours and embedded in paraffin. The embedded liver tissue was sectioned into 5 µm sections and fixed on slides, stained with hematoxylin and eosin (H&E) and observed under a BX 53 Olympus microscope according to the method described [93].
Analysis of Liver Function
The blood samples were collected as previously described [94,95]. Blood samples were gained by removing the eyeballs of mice. Blood was then centrifuged at 4000 rpm for 30 min at 4 • C. The samples were incubated in an electro-thermostatic water bath at 37 • C for 30 min. The serum was collected and subjected to the examination of the activities of alanine transaminase (ALT), aspartate aminotransferase (AST) activities, the activities of lactate dehydrogenase (LDH), and alkaline phosphatase (ALP) with respective commercial kits. The determination of AST, ALT, LDH, and ALP was performed by the instruction of the kits (Nanjing Jianchen Bioengineering Institute, Nanjing, China).
Measurement of MDA, SOD, CAT and GSH-Px Activities
The changes in hepatic oxidative stress were monitored as previously described [65]. The liver homogenate was centrifuged to obtain the supernatant at 3500 rpm for 10 min at 4 • C, and the superoxide dismutase (SOD), catalase (CAT), glutathione peroxidase (GSH-Px) activity, and malondialdehyde (MDA) levels were measured according to the manufacturer's instructions.
Immunohistochemistry Analysis
MIP-2 and COX-2 expressions in hepatic tissue were evaluated by immunohistochemistry staining as previously described [96]. The prepared hepatic sections were recovered, and the endogenous peroxidase in tissues was inactivated with 0.1% hydrogen peroxide containing methanol for 15 min. Then, the sections were incubated with a rabbit polyclonal MIP-2 and COX-2 antibody at 4 • C overnight. Subsequently, the sections were washed with PBS and incubated with rabbit anti-mouse (1:1000) secondary antibody at room temperature for 30 min. The sections were rinsed with PBS 3 times and stained with diaminobenzidine (DAB). Additionally, they were evaluated under an optical microscope (Olympus Optical Co., Ltd., Tokyo, Japan).
Quantitative Reverse-Transcription PCR (qRT-PCR) Analyses
The total RNA from each sample was isolated using the Trizol reagent (Sango Shanghai, China), and the first strand cDNA was synthesized using the StarScript II First-strand cDNA Synthesis Mix With gDNA Remover (Genstar) according to the manufacturer's instructions. Quantitative reverse-transcription PCR (qRT-PCR) was conducted to determine the mRNA levels of the IL-1β, IL-6, TNF-α, Bax, Caspase-3, and Bcl-2 (the primer sequences are shown in Table 4), the GAPDH gene was used as an internal control [8,30,97]. Real-time PCR reactions were performed on a CFX Real-time system (CFX96, Bio-Rad, Hercules, CA, USA). All of the samples were analyzed in triplicate, and the 2 −∆∆Ct method was used to analyze gene expression levels.
Western Blotting Analyses
The hepatic tissues were lysed with RIPA lysis buffer (Servicebio technology, Wuhan, China), supplemented with protease inhibitor (Servicebio), and homogenized with an ultrasonic processor. According to the manufacturer's instructions, the total protein of the liver tissue was extracted with a commercial kit (Servicebio technology, Wuhan, China). Then, the concentration of the protein was measured with a BCA kit (Beyotime technology, Shanghai, China). Then, the proteins were transferred to a polyvinylidene fluoride (PVDF) membrane, followed by blocking with 5% skim milk (0.5% TBST) and sealed for 1 h. Additionally, then, PDVF membranes were incubated with primary antibodies against NF-κB (1:1000), p-ERK (1:1000), PI3K (1:1000), p-AKT (1:1000), Caspase-3 (1:1000), Bcl-2 (1:1000), Bax (1:1000), and GAPDH (1:3000) were incubated overnight at 4 • C. They were washed with TBST at room temperature on a decolorizing shaker three times. After washing, PVDF membranes were incubated with secondary antibodies (1:3000) at room temperature for 2 h. The antibodies were purchased from Proteintech Group, USA. Finally, they were developed and fixed with developing and fixing reagents, and the Alpha software processing system analyzes the optical density values of the target band.
Statistical Analysis
The data are expressed as the mean ± SE. Data analyses were carried out using JMP Pro 13. The data were analyzed using general descriptive statistics. One-way analysis of variance (ANOVA) at 95%. p < 0.05 was considered statistically significant.
==
Domain: Biology Medicine
|
[{"added": "2022-12-01T11:23:54.087Z", "created": "2022-11-30T00:00:00.000", "id": "254103012", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1660-3397/20/12/758/pdf?version=1670218633", "pdf_hash": "0b9e3dd9761327e83dbc0ee094c6ba63471dbc15", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:779", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "5b94b8540cd3dae9f6ad00df766f305f123b641e", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Advancing Vaccine Strategies against Candida Infections: Exploring New Frontiers
Candida albicans, along with several non-albicans Candida species, comprise a prominent fungal pathogen in humans, leading to candidiasis in various organs. The global impact of candidiasis in terms of disease burden, suffering, and fatalities is alarmingly high, making it a pressing global healthcare concern. Current treatment options rely on antifungal drugs such as azoles, polyenes, and echinocandins but are delimited due to the emergence of drug-resistant strains and associated adverse effects. The current review highlights the striking absence of a licensed antifungal vaccine for human use and the urgent need to shift our focus toward developing an anti-Candida vaccine. A number of factors affect the development of vaccines against fungal infections, including the host, intraspecies and interspecies antigenic variations, and hence, a lack of commercial interest. In addition, individuals with a high risk of fungal infection tend to be immunocompromised, so they are less likely to respond to inactivated or subunit whole organisms. Therefore, it is pertinent to discover newer and novel alternative strategies to develop safe and effective vaccines against fungal infections. This review article provides an overview of current vaccination strategies (live attenuated, whole-cell killed, subunit, conjugate, and oral vaccine), including their preclinical and clinical data on efficacy and safety. We also discuss the mechanisms of immune protection against candidiasis, including the role of innate and adaptive immunity and potential biomarkers of protection. Challenges, solutions, and future directions in vaccine development, namely, exploring novel adjuvants, harnessing the trained immunity, and utilizing immunoinformatics approaches for vaccine design and development, are also discussed. This review concludes with a summary of key findings, their implications for clinical practice and public health, and a call to action for continued investment in candidiasis vaccine research.
Introduction 1. Brief Overview of Candidiasis and Its Global Burden
Throughout the annals of medical history, the Candida species has undergone a metamorphic evolution, transcending from inconspicuous pathogens deemed as mere nuisances to potent and pervasive adversaries responsible for severe human ailments.
Candidiasis, a well-recognized mycosis, encompasses a spectrum of cutaneous, mucosal, and organ infections that emerge at different life stages and are often associated with identifiable predisposing factors [1]. Central to this condition is the colonization of Candida spp., indispensable commensal yeasts found on human skin, in the gut, and in the genitourinary tract in up to 60% of healthy individuals, as well as at sites other than the ones enlisted [1]. This infectious colonization poses a grave health risk to immunocompromised responsible for candidiasis; however, other species collectively na species-C.glabrata, C. parapsilosis, C. tropicalis, and C. krusei-a medical attention (Figure 1) [2,3]. Taxonomic revisions have led to th pathogenic Candida species, but the term candidiasis still encompa by these agents [4]. Morphologically, Candida presents as small, ov like fungi, propagating through budding or fission [5]. An essential is their non-photosynthetic, eukaryotic nature, possessing an extern the plasma membrane, and abundant sterols, notably ergosterol, in t [6]. and C. krusei account for more than 90% of invasive infections [5].
Evidently, there is an escalating global burden of candidiasis health crisis, with the number of cases on the rise due to the gr immunocompromised individuals [7,8]. Invasive candidiasis, includ and deep tissue-seated candidiasis, is the prevailing fungal ailmen hospitals in high-income nations, with a global prevalence spanning 250,000 to 700,000 individuals annually. Its incidence rate falls with cases per 100,000 persons, while mortality rates vary from 40% to 5 to note that global annual incidence rates cannot be reliably determi availability of comprehensive studies. This increasing burden is m than tuberculosis; however, the domain of anti-Candida intervention [8,10]. The United States incurs medical expenses exceeding USD 7. While superficial infections primarily acquired within the community, result in significant morbidity, deep-seated and systemic Candida infections more commonly originate in healthcare settings. Notably, C. albicans, C. glabrata, C. parapsilosis, C. tropicalis, and C. krusei account for more than 90% of invasive infections [5].
Evidently, there is an escalating global burden of candidiasis which presents a dire health crisis, with the number of cases on the rise due to the growing population of immunocompromised individuals [7,8]. Invasive candidiasis, including both bloodstream and deep tissue-seated candidiasis, is the prevailing fungal ailment encountered within hospitals in high-income nations, with a global prevalence spanning from approximately 250,000 to 700,000 individuals annually. Its incidence rate falls within the range of 2 to 14 cases per 100,000 persons, while mortality rates vary from 40% to 55% [9]. It is important to note that global annual incidence rates cannot be reliably determined due to the limited availability of comprehensive studies. This increasing burden is matched by none other than tuberculosis; however, the domain of anti-Candida interventions is still in its infancy [8,10]. The United States incurs medical expenses exceeding USD 7.2 billion annually due to fungal diseases [7]. Moreover, the emergence of Candida auris, a new and highly virulent Candida species, has added to the complexity of the epidemiology [11]. The recent emergence of drug-resistant C. auris has exacerbated the burden on healthcare infrastructure. On 25 October 2022, WHO published a list of 19 fungal priority pathogens wherein Candida auris and Candida albicans are categorized under critical groups, whereas Candida glabrata, Candida tropicalis, and Candida parapsilosis are categorized under high-risk groups [12].
Vaccines 2023, 11, 1658 3 of 28 The earliest known Candida auris isolate was first identified in South Korea, dating as far back as 1996 [13], and has since rapidly spread across the globe. It has a robust ability to persist on surfaces, and coupled with that, asymptomatic carriers and international travel have contributed to its geographical dissemination and emergence as a major cause of nosocomial outbreaks. However, at that time, it was initially misidentified as a different fungal species, highlighting the challenges in recognizing and characterizing this emerging pathogen. These outbreaks have extended their presence beyond their initial identification in Asia, encompassing countries such as Japan, India, and Pakistan. Additionally, C. auris has emerged as a formidable concern in Europe, with reported cases in the United Kingdom and Spain. Across the Americas, the fungus has gained a foothold, affecting countries like Colombia, Venezuela, Panama, and the United States [14]. Perhaps most concerning is its development of multidrug resistance, displaying significant levels of both inherent and acquired resistance to azoles, echinocandins, and amphotericin B, limiting treatment options and increasing healthcare costs. Candida auris infections are associated with high mortality rates, particularly among immunocompromised patients, further impacting the global burden of disease. It is essential to highlight that the list of countries grappling with C. auris continues to grow, underscoring the pathogen's ability to traverse borders and establish itself as a global health challenge.
C. glabrata is the second most prevalent species in the USA, north-western Europe, and Canada, especially among elderly patients and organ transplant patients. In contrast, C. parapsilosis and C. tropicalis are more frequently encountered in southern Europe, South America, India, and Pakistan. C. krusei, which is the least common among the five main species, tends to occur more often in patients with severe immunodeficiency disorders.
Another less common species, C. dubliniensis, is more prevalent among HIV-infected patients. It is important to note that the current epidemiology of invasive candidiasis is significantly influenced by the selection pressure imposed by the use of antifungal medications, both in terms of prophylaxis and treatment. The widespread use of antifungal agents has led to a shift towards non-albicans species, some of which are increasingly resistant to treatment.
A glaring fact to be highlighted here is that around 50% of invasive Candida infections go undetected, leading to potential underestimation of the true incidence of these infections. Currently, there is a conspicuous lack of approved diagnostic methods that are both sensitive and precise. Detecting deep-seated organ involvement poses a significant challenge, with blood cultures yielding positive results in fewer than 40% of patients who do not have concurrent candidemia. This underscores a significant gap in our ability to promptly identify and manage these infections within healthcare settings. Undiagnosed Candida infections can result in delayed or inadequate treatment, leading to adverse patient outcomes, prolonged hospitalizations, and heightened healthcare expenditures. This places an additional burden on healthcare systems and also endangers patient well-being. In recent years, several non-culture-based techniques have emerged as valuable diagnostic aids, particularly in ruling out invasive candidiasis due to their high negative predictive value. Early initiation of antifungal therapy and effective source control are pivotal factors for survival in patients with invasive candidiasis. Nevertheless, the delay in definitive treatment often occurs due to the limited sensitivity of microbiologic cultures, which are currently considered the gold standard for diagnosis. Hence, there is an urgent need for future research and the development of diagnostic tools that offer both sensitivity and precision. The challenges posed in the diagnosis of Candida spp.infections are reviewed elsewhere in detail [15,16].
Challenges in Management of Candida Infections
While superficial mycoses, such as skin and nail infections, are generally manageable, invasive fungal infections present a severe challenge to healthcare systems globally. These infections can range from mild mucosal candidiasis to life-threatening bloodstream infections. The risk factors contributing to the surge in candidiasis cases include HIV infection, immunosuppressive therapy, and prolonged hospital stays. Candida-related bloodstream infections are alarmingly common in healthcare-associated settings in the United States and Europe, affecting thousands of patients yearly [7,10].
The COVID-19 pandemic has further exacerbated the global burden of candidiasis. Causal factors such as prolonged admittance in critical care units, prolonged administration of antibiotics and corticosteroids, and micronutrient deficiencies such as that of iron and zinc have led to an increased predisposition to COVID-19-associated candidemia (CAC) [17]. In the context of co-infection, it is of paramount importance to understand the pathogenesis and mechanism of virulence to understand disease progression, especially in the case of co-infections. Reports exist of a direct link between the development of candidiasis and the use of antibiotics and corticosteroids by COVID-19 patients [18]. It is of utmost importance to comprehend the pathogenesis and mechanisms of virulence for a better understanding of disease progression, particularly in cases involving co-infections. To this end, further research is imperative to unravel the intricate mechanisms governing each stage of Candida spp.pathogenesis. Although the complete molecular pathophysiology remains incompletely understood, certain factors, such as a weakened immune system, deficiencies in iron and zinc, and the potential for nosocomial and iatrogenic transmissions, increase the susceptibility of COVID-19 patients to candidiasis [17]. Understanding the etiology and pathogenesis of candidiasis is essential for gaining deeper insights into its connection with and transmission in COVID-19 patients.
A study conducted in Cairo, Egypt, demonstrated the effective treatment of a COVID-19 patient with oropharyngeal candidiasis (OPC) through the concurrent administration of miconazole (four times a day) and fluconazole (three times a day), with no distinct side effects [19]. Furthermore, investigating substances capable of impeding biofilm formation may be a novel area of research. Additionally, targeting the host's micronutrient acquisition for treatment shows potential as a promising avenue of study. Forty-one percent of COVID-19 patients in a hospital-based study were found to be co-infected with C. albicans, but the number of azole-resistant pathogens such as C. auris also increased. The early diagnosis and treatment of candidiasis were key factors in improving the survival of patients with IC. The survival rate was increased by empirical therapy (66% vs. 44%), the early (within 24 h) prescription of antifungal therapy (56% vs. 38%), and the use of echinocandin (64% vs. 39%) [20]. The magnitude of challenges posed by candidiasis and comparable fungal infections requires a fundamental shift in antifungal treatment research, emphasizing the exploration of secure and preventive approaches like anti-Candida vaccines. A primary challenge in the development of vaccines against Candida species is the high incidence in immunocompromised population. The immune system's impairment in these individuals raises concerns both in terms of vaccine efficacy and safety. Unlike viral and bacterial vaccines, which often target surface proteins, fungal pathogens like Candida have complex cell walls and lack easily recognizable antigens. Live vaccines generally elicit better immune response, making them more favorable vaccine candidates. However, their use must be cautious due to the potential risk of causing infections in already vulnerable individuals. Conversely, inactivated whole organism and subunit vaccines are considered safer, but they may be less effective in immunocompromised individuals because of their weakened immune systems. Additionally, the diversity of Candida species and their ability to cause a wide range of severe infections further complicate vaccine development [7]. Extensive research is being undertaken to enhance adjuvants and vaccine formulations to stimulate more robust protective responses while targeting immune response pathways that may not be compromised [21].
Ensuring the safety and efficacy of such vaccines requires rigorous testing and clinical trials, which can be time-consuming and resource-intensive [22]. Typically, vaccine efficacy assessments begin with inbred mice due to their cost-effectiveness and well-defined immune systems. However, bridging the gap between murine and human immune responses poses a concern, and careful consideration is required when extrapolating efficacy data across species, which poses a great challenge in translating preclinical studies. Furthermore, navigating regulatory processes and scaling up production to meet global demand are substantial logistical hurdles. Converting a vaccine candidate into one approved for human use necessitates substantial financial investments for clinical trials and product manufacturing. Additionally, fungal infections that predominantly affect populations in resource-limited areas often require the involvement of governmental and non-governmental organizations to support vaccine commercialization efforts, ensuring the successful development of an anti-Candida vaccine.
Vaccines targeting invasive pathogens constitute a significant advancement in disease prevention, and the development of a successful anti-Candida vaccine emerges as a clear strategy for candidemia prevention. Candida infections, particularly those caused by drugresistant strains like Candida auris, are on the rise and have a significant impact on healthcare systems globally. With the increasing fatality rate attributed to drug-resistant fungal strains and the growing population of immunocompromised individuals, the development and successful deployment of an effective fungal vaccine hold immense potential. Vaccines offer a promising solution to this problem by potentially preventing Candida infections or reducing their severity. By stimulating the immune system to recognize and respond effectively to Candida pathogens, vaccines can provide a proactive approach to combatting these infections, potentially mitigating their global impact and reducing the reliance on antifungal medications.
Despite continuing efforts, there is currently no commercially available anti-Candida vaccine that has been approved for human use [23,24]. Recently, preclinical and clinical studies of potential vaccine candidates have been published, which report good immunogenicity and a functional response against Candida spp. [ [25][26][27]. With this in mind, the development of an effective and successful vaccine against Candida infections could be a crucial step in preventing and controlling these potentially life-threatening fungal diseases. This review outlines the collective efforts made so far in developing diverse categories of Candida vaccines, including vaccines undergoing clinical trials. It also addresses the challenges associated with creating a successful anti-candidiasis vaccine in the future.
Rationale for Developing Vaccines against Candidiasis
The treatment of invasive candidiasis in immunocompromised individuals presents a formidable clinical challenge. The current arsenal of antifungal strategies, including polyenes, echinocandins, and azoles, though primary options, suffers from various limitations. These drawbacks encompass restricted efficacy against specific fungal species, adverse effects such as cytotoxicity, hepatotoxicity, allergic reactions, and inflammation, and the emergence of increasing drug-resistant strains. The effectiveness of these antifungal drugs relies on several factors, including the host's immune system, disease location, severity of infection, and the pharmacokinetics of the drug [28]. The mechanisms of action of antifungal drugs like azoles, polyenes, and echinocandins target specific components in Candida and other fungal cells, disrupting their integrity and causing cell death. However, these mechanisms can also lead to adverse effects in humans due to structural similarities or off-target effects. Azoles inhibit the synthesis of ergosterol, a crucial component of the fungal cell membrane. This disruption weakens the fungal cell membrane's integrity, leading to cell death. However, since human cells also contain cholesterol, which is structurally similar to ergosterol, azoles can have off-target effects on human cells, potentially leading to side effects such as liver toxicity or interactions with other medications. Polyenes, like amphotericin B, form pores by binding to ergosterol, but they can also interact with human cell membranes, leading to kidney damage. Amphotericin B deoxycholate formulations have been associated with severe adverse events, such as nephrotoxicity and infusion-related adverse effects. Echinocandins, by inhibiting beta-glucan synthesis, weaken the fungal cell wall without direct toxicity to human cells, though they can still cause side effects such as gastrointestinal disturbances, elevated liver enzymes, and other adverse effects such as fever, headaches, and allergic responses. Careful monitoring and management of patients are crucial during antifungal therapy to mitigate these adverse effects [29].
Importantly, the emergence of drug-resistant strains of Candida species against the majority of antifungal medications and the significant increase in the number of immunocompromised individuals globally (due to various factors like age and increased susceptibility to infections post-COVID-19) emphasize the urgency for innovative antifungal medications, immunotherapies, and antifungal vaccines. In particular, the emergence of C. auris, a new drug-resistant species, presents a grave threat to the effectiveness of candidiasis treatment. To address these challenges, experts fervently advocate improved diagnostics, more effective antifungal agents, and immunotherapies against candidiasis [30].
The current situation effectively emphasizes the need for innovative approaches to combat candidiasis. Amidst the call for enhanced therapeutic approaches, the concept of an anti-Candida vaccine emerges as a pivotal tool to alleviate the burden of systemic and vaginal candidiasis on a global scale. Such a vaccine holds the potential to significantly reduce mortality rates and, by extension, generate substantial socioeconomic benefits [30]. An anti-Candida vaccine, as a preventive measure, holds the promise of not only enhancing public health but also yielding substantial socioeconomic benefits. Firstly, it has the potential to lead to considerable cost reductions within the healthcare sector. Candida infections, especially in immunocompromised individuals, often require prolonged hospital stays, costly antifungal medications, and sometimes, surgical interventions. Additionally, vaccines will contribute to resource and time savings by averting the need for the development or discovery of new antifungal compounds.
Additionally, by curbing the incidence of drug-resistant Candida strains, an anti-Candida vaccine can help mitigate the societal costs associated with antimicrobial resistance. Furthermore, preventing fungal infections in high-risk populations, such as transplant recipients and those with certain medical conditions, can enable individuals to lead healthier lives and contribute more actively to society, ultimately boosting overall productivity and economic growth. The utilization of modern vaccine development technologies, as demonstrated by the rapid and safe development of mRNA-based vaccines against COVID-19, exemplifies an efficient and effective approach to combat fungal infections. These innovations not only enhance our capacity to address fungal diseases but also optimize healthcare efforts and resource utilization. As these efforts converge, a future where candidiasis is under control could become a reality. This would lighten the burden on healthcare systems, which currently bear the weight of managing and treating recurrent cases of candidiasis. Additionally, patients would experience a reduction in the physical and emotional toll caused by the condition. By taking on these challenges head-on and exploring innovative avenues, we pave the way for a future where candidiasis ceases to be a menacing threat.
Description of Current Vaccine Candidates against Candidiasis
The current armamentarium of vaccines targeted against bacterial and viral vaccines is highly successful in reigning the global burden of the respective diseases. Diseases like smallpox and polio have been nearly eradicated, and the incidence of pneumococcal infections, diphtheria, and measles has been significantly reduced. WHO attributes the success against 25 infectious diseases to successful vaccination campaigns. Additionally, the world was a witness to the urgency with which COVID-19 vaccines were developed and administered, leading to a significant reduction in the morbidity and mortality caused by COVID-19 infections l [31,32]. As mentioned earlier, no antifungal vaccine has yet been licensed for human use. The current research is limited to preclinical and clinical testing only. Several vaccine candidates have demonstrated safety and immunogenicity against Candida in these studies [25,33]. Hereafter, we describe the state-of-the-art vaccine candidates with potential for tackling Candida infections (Figure 2).
Live Attenuated Vaccines
Live attenuated vaccines, a cornerstone of effective immunization strategies against infectious diseases, have demonstrated their potency through successful examples targeting various viruses like polio, measles, mumps, rubella, and influenza and other infections [34,35]. While live attenuated vaccines have historically centered on viruses, recent investigations have revealed their potential in combating fungal infections, particularly candidiasis [34,36].
The principle of live attenuated vaccines revolves around inducing robust immune responses akin to natural disease immunity through the replication of pathogens at the site of infection [10,23,34]. This approach has paved the way for the development of potential vaccines against pathogenic fungi like Histoplasma capsulatum, Blastomyces dermatitidis, and Paracoccidioides brasiliensis. Notably, these vaccines have demonstrated efficacy in stimulating defensive immune responses [10,36].
In the realm of fungal vaccines, a genetically modified C. albicans tet-NRG1 strain has been engineered in which the filamentation repressor NRG1 can be overexpressed by doxycycline. The presence or absence of doxycycline (DOX) in the growth medium controls the expression of NRG1, leading to the manipulation of both morphology and virulence. This vaccine construct has shown to protect mice from lethal systemic infection with a virulent fungal strain [37]. Similarly, various attenuated strains of C. albicans, such as CNC13 (deleted in the MAP kinase HOG1), RML2U (deleted in the cell wall protein (CWP) gene ECM33 and defective in its interaction with endothelial and epithelial cells), CM1613 (a Mitogen Activated Protein Kinase MKC1 mutant), CNC13 (a MAP kinase HOG1 mutant), and a defective mutant 92′, were able to protect mice from a lethal challenge of Candida spp. [30,38]. A genetically engineered C. albicans strain PCA-2 is a caspofungin resistant mutant of its isogenic 3153A strain. Immunization of mice with PCA-2 Candida triggered an innate immune reaction, with a significant increase in the count of peripheral blood polymorphonuclear cells (PMNs) and significant protection against candidiasis. Further, adoptive transfer of macrophages derived from PCA-2-
Live Attenuated Vaccines
Live attenuated vaccines, a cornerstone of effective immunization strategies against infectious diseases, have demonstrated their potency through successful examples targeting various viruses like polio, measles, mumps, rubella, and influenza and other infections [34,35]. While live attenuated vaccines have historically centered on viruses, recent investigations have revealed their potential in combating fungal infections, particularly candidiasis [34,36].
The principle of live attenuated vaccines revolves around inducing robust immune responses akin to natural disease immunity through the replication of pathogens at the site of infection [10,23,34]. This approach has paved the way for the development of potential vaccines against pathogenic fungi like Histoplasma capsulatum, Blastomyces dermatitidis, and Paracoccidioides brasiliensis. Notably, these vaccines have demonstrated efficacy in stimulating defensive immune responses [10,36].
In the realm of fungal vaccines, a genetically modified C. albicans tet-NRG1 strain has been engineered in which the filamentation repressor NRG1 can be overexpressed by doxycycline. The presence or absence of doxycycline (DOX) in the growth medium controls the expression of NRG1, leading to the manipulation of both morphology and virulence. This vaccine construct has shown to protect mice from lethal systemic infection with a virulent fungal strain [37]. Similarly, various attenuated strains of C. albicans, such as CNC13 (deleted in the MAP kinase HOG1), RML2U (deleted in the cell wall protein (CWP) gene ECM33 and defective in its interaction with endothelial and epithelial cells), CM1613 (a Mitogen Activated Protein Kinase MKC1 mutant), CNC13 (a MAP kinase HOG1 mutant), and a defective mutant 92 , were able to protect mice from a lethal challenge of Candida spp. [30,38]. A genetically engineered C. albicans strain PCA-2 is a caspofungin resistant mutant of its isogenic 3153A strain. Immunization of mice with PCA-2 Candida triggered an innate immune reaction, with a significant increase in the count of peripheral blood polymorphonuclear cells (PMNs) and significant protection against candidiasis. Further, adoptive transfer of macrophages derived from PCA-2-immunized animals conferred protection against lethal infection in naïve animals [39]. Another attenuated C. albicans strain gpi7 effectively protected mice from disseminated invasive candidiasis. In the gpi7 strain, the β-glucan layer of the cell wall is exposed on the surface that facilitates dectin-1 receptor-dependent nuclear translocation of RelB in macrophages, the release of IL-18, and the production of protective antibodies [40].
To combat the problem of safety in immunocompromised hosts, the utilization of viable non-pathogenic fungi Saccharomyces cerevisiae has been investigated as a means to serve as carriers for immunogens, encompassing disparate antigens. Saccharomyces cerevisae are being explored as vehicles for delivering immunogens in Candida infection models. Also, it is being pursued as a vaccine vehicle for pathogens such as the dengue virus, SARS-CoV2, H5N1, Helicobacter pylori, Toxoplasma gondii, etc. The ease of genetic manipulation in S. cerevisae and the potent adjuvant properties of the yeast β-glucan favor this strategy for live attenuated anti-Candida vaccine development. Preclinical data have endorsed the activation of DCs and CD4+ T cells, as well as cross-priming CD8+ T cells in response to S. cerevisae-based vaccines. In fact, the S. cerevisiae-based vaccine vehicles are being pursued as oral vaccines and will be discussed in detail later in the section on oral vaccines [34,41]. "Trained immunity" is an important concept wherein non-specific vaccines like BCG elicit an augmented protective response during subsequent infections, whether caused by the same pathogens or different ones. This heightened defense is primarily facilitated by the innate immune system and provides an enhanced protective response to a secondary infection, caused either by the same or different pathogens, and may lead to beneficial non-specific boosting of immune responses, which may protect against various infections such as Mycobacterium tuberculosis and COVID-19 [42,43]. Studies have shown protective innate immune response against pathogens like C. albicans, Staphylococcus aureus, and Mycobacterium tuberculosis upon BCG immunization via epigenetic modifications leading to an enhanced immunological state. BCG-induced enhanced immune response was seen in vivo and in vitro experiments via NOD2-mediated epigenetic changes and histone 3 lysine 4 trimethylation, which ultimately provided increased protection against disseminated candidiasis in SCID mice [44]. Similarly, enhanced protection was observed against an experimental infection caused by an attenuated yellow fever virus vaccine strain via BCG-induced epigenetic reprogramming of innate cells [45]. Therefore, efforts should be made to assess whether BCG vaccination/induction of trained immunity-based vaccines might lower the susceptibility to candidiasis among high-risk individuals.
However, challenges persist in the translation of these developments into clinical applications. Concerns regarding the stability of attenuated vaccines, vaccine specificity, and the limited use and safety in immunocompromised individuals need careful consideration [30,46]. The key factors withholding the clinical success of the discussed live attenuated vaccines against Candida spp., especially in immunocompromised individuals, are as follows: strong immune response that can progress unregulated in immunocompromised individuals; several instances of reversion of avirulent attenuated viral vaccines to virulent forms discouraged the use of live attenuated forms of Candida spp.; the constraints of manufacturing, transport, and storage are similar to that of other immunologics. Additionally, in instances of use of a live attenuated vaccine in case of other pathogenic fungi such as Coccidioides immitis, no significant benefit was observed in the case of vaccinated and placebo groups [47]. Thus, although the potential of live attenuated vaccines is immense in the case of bacterial infections, in the case of fungal vaccines, prioritizing patient safety is essential due to the inherent risks associated with live fungal infection [48].
Conclusively, the road to clinical implementation of antifungal live attenuated vaccines remains intricate, necessitating a balance between efficacy and safety concerns [49].
Recombinant (Subunit) Vaccine
Concerns over traditional live Candida vaccines' complexity and potential unwanted immunological responses have led to the emergence of recombinant protein vaccines. These vaccines leverage genetic engineering to create targeted immunity, often involving the transfer of genes encoding immunogenic antigens. Recombinant protein vaccines are far safer alternatives due to the lack of infectious agents and ease of administration [36]. Agglutinin-like sequence proteins of Candida spp.(e.g., Als1p and Als3p) play a crucial role in adhesion and infection [50]. Vaccines based on Als1p and Als3p proteins displayed protection against various forms of candidiasis; indeed, formulations using recombinant Als1p and Als3p demonstrated considerable immunogenic potential and protected against invasive candidiasis [30,36]. The proteins Als1 and Als3, either alone or in conjunction with various adjuvants, have been suggested as potential vaccine candidates for combating invasive candidiasis. When mice were subcutaneously immunized with the recombinant N-terminus of Als1 (rAls1p-N), it protected 50% to 57% of animals against a lethal challenge from C. albicans [51]. Positive outcomes have emerged from clinical trials involving NDV-3A, a vaccine based on rAls3p-N. The vaccine candidate NDV-3 consists of recombinant Als3p invasion protein with alum as an adjuvant and was found to be efficacious against candidiasis by inhibiting attachment to epithelial/endothelial surfaces of the Candidiasis. The candidate vaccine NDV-3 displayed significant immunogenicity, eliciting a strong B-cell response and robust T-cell responses in a mouse model, thus successfully preventing both mucosal and hematogenously disseminated candidiasis in mice [25,52]. Further, phase I trials were conducted which included a control group and forty healthy adult subjects that received one dose of NDV-3 containing either 30 or 300 µg of Als3p. Progressing through phase I clinical trials, NDV-3 was found to be secure and effective, generating specific T cells that produce IFN-γ and IL-17A cytokines and also inducing anti-Als3p total IgG and IgA1 levels [25] . Another important group of secretory proteins of C. albicans that play an important role in fungal cell adhesion, epithelial as well as endothelial invasion, and metabolism are secreted aspartyl proteases (SAP), with Sap2 being the most abundant. When rats were intravaginally or intranasally immunized with recombinant Sap2, whether administered alone or with cholera toxin as an adjuvant, it led to the clearance of Candida vaginal infection [53,54]. Amidst the quest for groundbreaking pharmaceuticals and immunomodulatory treatments against C. dubliniensis infections, novel approaches such as immunotherapy and in silico investigation for vaccine designs have emerged. SAP proteins within the domain of C. dubliniensis have garnered attention as potential vaccine candidates. Computational tools were harnessed to predict epitopes for vaccine development, particularly focusing on SAP proteins [55].
Next, De Bernardis et al. developed a novel vaccine candidate containing a modified version of aspartyl proteinase-2 from Candida albicans, enclosed in influenza virosomes (PEV-7). The vaccine elicited a strong antibody in mice and rats following an intramuscular administration. Antibodies were also detected in vaginal fluid after both intravaginal and combined intramuscular and intravaginal administrations in mice and rats. In a rat model of candida vaginitis, PEV7 demonstrated substantial and enduring protection, most likely mediated by antibodies, when administered via the intravaginal route. Furthermore, a repeated-dose toxicological study in rats confirmed the safety of PEV7 [33].
C. albicans malate dehydrogenase (Mdh1p) also showed effective results in animals and is being considered for a C. albicans vaccine [56]. Additionally, heat shock protein (Hsp90p) and hyphal-regulated cell wall protein1 (Hyr1p) have also shown potential as vaccine candidates [50]. Mycograb, a human genetically recombinant antibody against heat shock protein 90 (rP-HSP90C) was developed. Mycograb, in combination with Amphotericin B, was found to provide complete protection against C. albicans, C. krusei, and C. glabrata but was not approved on grounds of safety and quality [57]. Hyphal-regulated cell wall protein1 (Hyr1) is a GPI-anchored mannan protein present on the fungal cell. Subcutaneous immunization using a recombinant N-terminus Hyr1 (rHyr1-N) with either complete Freund's adjuvant (CFA) or aluminum hydroxide demonstrated significant protection against C. albicans, C. glabrata, C. krusei, C. parapsilosis, and C. tropicalis in immunocompetent mice [58].
The use of potent adjuvants in subunit vaccines is a key determinant for vaccine efficacy and conferred protection. Incorporation of the right adjuvant can facilitate multiple advantages such as the targeted delivery of antigens to antigen presenting cells and the formation of antigen depots, aiding cellular chemotaxis, and the stimulation of dendritic cells, B and T cells, etc. They can further activate innate immunity mediators such as Toll-like receptor (TLR) ligands [33,36]. In addition to the traditional adjuvants discussed in the aforementioned section, Wüthrich and co-workers demonstrated the potency of a combination adjuvant, comprising inulin (plant-derived polysaccharide, trade name-Advax) and TLR agonists, in improving the efficacy of a recombinant subunit vaccine Blastomyces endoglucanase 2 in a model of respiratory Blastomyces dermatitidis infection.
The combination adjuvant greatly enhanced the antifungal immunity [59]. In light of these findings, combination adjuvants can also be evaluated for anti-Candida vaccine candidates for improving efficacy.
The field of candidiasis vaccine development is progressing with a range of promising recombinant protein candidates. These candidates, targeting distinct antigens, demonstrate immense potential in combating candidiasis and offer avenues for further exploration and development.
Conjugate Vaccines
Conjugate vaccines, a promising avenue in immunology, involve the strategic fusion of weaker antigens (usually cell wall polysaccharides) with strong immunogenic proteins as a carrier, thereby eliciting robust immune responses. This approach capitalizes on the capacity of polysaccharides to independently stimulate B cells, fostering T-independent immune reactions, while also presenting antigens to T cells through polysaccharide-bound peptides [10]. This synthesis leads to enduring immunity, particularly when protein carriers are affixed to polysaccharides, facilitating binding to MHC molecules and inducing potent T cell responses.(One significant application of conjugate vaccines is their focus on targeting shared polysaccharide epitopes, such as β-glucans found in fungal cell walls.)This innovative strategy holds immense potential for pan-fungal vaccines, crucial for individuals with compromised immune systems and heightened susceptibility to invasive fungal infections. Early attempts included a C. neoformans vaccine, merging capsular polysaccharide GXM with tetanus toxoid (TT), yielding significant antibody responses in animals [36]. Further efforts involved conjugating β-glucans with diphtheria toxin (CRM) to counter invasive candidiasis and aspergillosis [50].
The exploration extends to the use of fungal cell wall glycans as alternative proteinbased vaccine targets [60]. Mannans and their peptide conjugates emerged as promising Candidates, exploiting the recognition of these glycans by various receptors [30]. The conjugation of mannans with antigenic peptides enhances antigen presentation, yielding effective synthetic glycopeptide vaccines against C. albicans. Similarly, β-glucans, crucial components beneath mannans, invoke innate immune responses vital for host defense. Coupling β-glucans with diverse carriers and adjuvants highlights their immunostimulatory potential against candidiasis and aspergillosis [23,61].
Xin and co-workers recently demonstrated a synthetic glycopeptide conjugate vaccine candidate against C. albicans. Previous reports by the same group have shown a glycopeptide-tetanus toxoid vaccine candidate that is potentially effective for clinical evaluation [62,63]; however, the glycan component is difficult to synthesize, expensive, and requires significant technical inputs for large-scale manufacturing. This hindered the conventional glycoconjugate vaccines against Candida. Additionally, carbohydrate-based vaccines are setback by limited affinity of anti-carbohydrate antibodies as compared to anti-peptide or anti-protein antibodies [64,65].
Mimotopes are peptide mimics of glycan epitopes that are easier and cheaper to synthesize and can be explored easily in polyvalent formulations. The glycan part of the glycopeptide vaccine evaluated by Xin and co-workers, β-(Man)3, is ubiquitously expressed across several Candida species. Thus, novel peptide mimotopes with structural homology to glycan epitope β-(Man)3 were used in the conjugate vaccine candidate as surrogate epitopes. Specifically, a total of five mimotopes were immunogenic in mice, and out of this, three mimotopes could confer protection in mice against disseminated candidiasis. Furthermore, immunization with three mimotope-peptide conjugate vaccines was also able to induce specific antibody responses, and importantly, protection against disseminated candidiasis in mice. The mechanism of protection via this novel vaccine candidate was indicated to be similar to the protective action of MAbs B6.1-mediated murine neutrophil activity and the B6.1/murine IgG3 isotype variant mediated via recruitment of the host complement system [66]. Xin and coworkers also investigated various mechanistic targets and adjuvants for vaccine formulations; initially, the b-(Man)3-Fba conjugate, combined with dendritic cells (DCs) and complete Freund's adjuvant (CFA), provided promising mouse protection. However, the use of CFA is not suitable for human use. Alternative adjuvants resulted in weaker immune responses. To boost the immune response and increase vaccine efficacy, the antigen was linked to tetanus toxoid (TT) to form the conjugate (Man)3-Fba-TT. This formulation induced a strong antibody response and a shift from IgM to IgG, indicating development of immunological memory. When adjuvanted with alum or Monophosphoryl Lipid A (MPL), b-(Man)3-Fba-TT showed comparable protection to the original DC/CFA approach [63].
In sum, conjugate vaccines stand as a compelling avenue in the fight against fungal infections, leveraging the fusion of antigens with polysaccharides to elicit potent and enduring immune responses. The unique strategy of targeting shared epitopes offers hope for pan-fungal vaccines, benefiting vulnerable populations.
Killed Whole Cell Vaccines
The landscape of vaccine development against candidiasis showcases a range of strategies, with the whole-cell killed vaccine approach standing out for its stability and safety benefits [10]. Whole-cell killed vaccines are stable and non-pathogenic, as compared to live attenuated vaccines, because they cannot revert. The cost of manufacture, ease in handling, and comparative safety make it a preferred choice. This choice is underscored by its simplicity and cost-effectiveness, although challenges in achieving complete protection for mucosal candidiasis have been observed. Diverse avenues have been explored, including the utilization of heat-killed C. albicans and a genetically modified toxin as an adjuvant, highlighting potential cross-protection against various fungal infections. Immunization with heat-inactivated C. albicans combined with a heat-labile, genetically modified toxin derived from Escherichia coli, specifically R192G, as an adjuvant, resulted in a significant level of protection in animal models by intranasal vaccination but did not provide protection via the mucosal route [67].
An innovative strategy explored in killed whole-cell vaccine candidate development is the use of S. cerevisiae yeast in the form of heat-killed yeast. Subcutaneous immunization with heat-killed S. cerevisiae yeast emerges as a promising vehicle for this type of vaccine as it has been shown to confer cross-protection against infections caused by C. albicans, A. fumigatus, and Coccidioides posadasii. Specifically, vaccination with heat-killed yeast cells in a 3, 4, or 6-day schedule conferred protection against a Candida challenge via an antibody response directed against glycans common to the cell wall of Saccharomyces and Candida, in addition to activating cellular responses via Th1 and Th17. Alum was included in the study as an adjuvant, however, it failed to show any significant impact on the strength of immune response [68-70]. Yeasts are in fact a standard model for protein expression as they are easy to culture and manipulate and scale up the culture/protein expression. The cellular physiology and molecular biology of S. cerevisae is well deciphered. Another advantage of using yeast cells is the ability of yeast surface components to stimulate the human immune response and hence act as an adjuvant. The immunogenicity of S. cerevisae is based on the polysaccharides beta-1, 3-D-glucan, and mannan, due to which the dendritic cells are activated and phagocytosed, which, in turn, also generate the danger signals. The Toll-like receptors and mannan receptors are able to recognize the cell wall carbohydrates and aid the adjuvanticity of yeast cells. These factors have aided the use of S. cerevisae for developing vaccine candidates via various innovative approaches such as yeast display, whole recombinant yeast cells, heat-killed yeast cells, and preparations of virus-like particles. The yeast-based vaccines are also favorable candidates for oral or edible vaccines [68, 69,71].
Expanding on the use of heat-killed cells as a vaccine candidate, intranasal application of heat-killed C. albicans, plus a heat-labile genetically engineered toxin from Escherichia coli as an adjuvant R192G, provided a substantial degree of protection in animal models [67]. Further, a combined vaccine formulation of MV140 and V132 has been tested to prevent both bacterial as well as fungal genitourinary tract infections [72]. MV140 is a polyvalent bacterial preparation based on whole heat-inactivated components used to prevent recurrent urinary tract infections and V132 is heat-inactivated Candida albicans vaccine against recurrent vulvovaginal candidiasis. The vaccine combination effectively stimulates human dendritic cells (DCs) to induce IFN-γ and IL-17A-producing T cells, as well as FOXP3+ regulatory Treg cells. Furthermore, MV140/V132 triggers epigenetic reprogramming in human DCs, facilitating the induction of trained immunity. This innovative direction in vaccine research offers a new perspective on fungal infection prevention, as demonstrated by these studies. The complexity of achieving comprehensive protection, especially in mucosal candidiasis, and the interplay between efficacy and toxicity underscore the ongoing challenges in vaccine development. The unique potential of heat-killed yeast-based vaccines to address diverse fungal infections points towards a promising avenue for future research.
Oral Vaccines
Oral vaccines have significant potential in responding to urgent health challenges due to their convenience, cost-effectiveness, and potential for mass administration. Continued efforts into novel antigenic candidates and adjuvants are crucial in developing effective vaccines to prevent candidiasis, especially in at-risk populations. Significant strides have been achieved through the ingenious utilization of microbial cell surfaces for the display of immunogenic proteins. Notably, the antigen Eno1p from C. albicans has been effectively presented on both E. coli and S. cerevisiae cells, exhibiting promising potential for oral vaccine creation. This transformative approach eliminates the arduous purification process typically associated with traditional vaccine development. Immunization of Enop1 protein in mice via oral or intranasal delivery or through subcutaneous injection resulted in elevated levels of anti-Eno1p antibodies and protected mice against C. albicans infection. In one study, Eno1p-expressing S. cerevisiae cells protected 60% of the mice against candidiasis when orally administered. A similar effect was observed with L. casei cells, where the display of Eno1p yielded 20% protective efficacy in mice [73,74]. While diverse immune adjuvants are available to boost immune response in conventional vaccines, the options for mucosal adjuvants in oral immunization remain limited. This limitation arises because most adjuvants designed for injection cannot withstand the harsh conditions of the gastrointestinal mucosa. Thus, there is a need for novel oral vaccine adjuvants that are both safe and resilient in demanding environmental conditions.
The innovative fusion of immunogenic proteins and microbial cells presents a promising avenue for the development of oral vaccines against candidiasis. The protective effects witnessed in murine models, coupled with the simplicity and speed of the molecular display methodology, showcase its potential in responding to urgent health challenges. The ongoing pursuit of novel antigenic candidates through proteomic analysis holds the prospect of further elevating the effectiveness of these vaccines [75].
Bacterial Ghost Vaccines
Hollow bacterial dead cells with pores that are used to deliver the immunogen or drug to the disease site are called bacterial ghost delivery systems. For the preparation of inactivated vaccines, using formaldehyde or heat treatment compromises the surface structures of the pathogen; however, bacterial ghosts retain the structural features of the antigens (pathogen-associated molecular patterns) displayed on their surface because of the specialized methods based on genetic engineering or chemical treatment. Bacterial ghosts, in fact, have adjuvanticity and can highly boost the host innate immune responses in response to vaccination [76]. Maii and co-workers are pioneers in using the concept of bacterial ghosts in Candida vaccine development. They prepared silver and gold nanoparticles using Candida albicans ghosts via a modified sponge-like reduced protocol. Sprague-Dawley albino rats aged 4-6 weeks were vaccinated intraperitoneally. The rats were challenged with the hyphal form of Candida albicans in artificial ulcers, subcutaneously and intraperitoneally. The authors reported an enhancement in humoral and cellular immunity in the treated rats, along with accelerated ulceration wound healing and controlled inflammation in the ulcer. C. albicans ghosts induced the proliferation of all subtypes of white blood cells (WBCs), indicating the activation of cellular immunity, and agglutination tests indicated a positively boosted humoral response. A noticeable outcome was increased systemic IFN-gamma levels, especially even in the Candida ghost-administered rats, indicating the benefit of activating phagocytes to mediate the elimination of fungal cells in the absence of any hyperallergenic responses (IgE levels) [77]. Thus, the use of ghost cells is a potentially favorable but unexplored strategy in development of anti-Candida vaccines.
Preclinical and Clinical Data on Vaccine Efficacy and Safety
Preclinical and clinical data on vaccine efficacy and safety against Candida infections represent crucial stepping stones in the journey toward effective fungal vaccines. As studies continue to unravel the challenges associated with vaccine development against Candida and understanding immune response, the ultimate goal is to provide a valuable tool for preventing and mitigating the impact of Candida-related diseases, particularly in vulnerable populations. The effectiveness of anti-Candida vaccines targeting virulence factors and various forms of Candida, such as hyphae and cell wall antigens, has been showcased by mouse models. Diverse formulations are available for these vaccines, including strains that are live and attenuated, recombinant proteins, and glycoconjugates (Table 1). Nevertheless, the challenge posed by the variation among Candida species, both genetically and morphologically, is evident and has a major consequence on the success of the vaccine. The majority of previous studies have primarily regarded C. albicans as a pathogenic yeast responsible for both primary and secondary infections. However, given its commensal nature, it may potentially maintain a mutualistic relationship with the host, an aspect that has remained largely unexplored. A recent study suggests a mutualistic relationship between C. albicans and mice, suggesting that C. albicans plays a role in shaping the gut microbiota, influencing metabolism, and bolstering host immunity to the host's advantage [78]. Designing a vaccine that can cater to a wide array of diseases becomes complicated due to multiple sites of infection and the immune deficiencies prevalent in high-risk groups. Two promising recombinant vaccines, namely, PEV7 and NDV-3, have reached phase I clinical trials in humans. Another important vaccine to reach clinical trials phase 1 is PEV7. It encompasses the recombinant Sap2 protein of C. albicans in virosomal formulation. The vaccine was found to be safe and demonstrated specific and effective production of memory B cells. NDV-3A is the first vaccine to show success in preclinical tests for protecting against diseases caused by both fungal and bacterial pathogens. NDV-3A exhibited heightened antigen-specific titers and cytokine production during a phase 1 trial. After the first dose of immunization, there was an increased expression of anti-Als3p IgG antibodies, in comparison to the control group [25]. Further, an exploratory phase 1b/2a-trial was conducted for NDV-3A against recurrent vulvovaginal candidiasis (RVVC). RVVC is a major concern globally as it affects 138 million women worldwide annually and approx.492 million are affected once in their lifetime. The vaccine showed a decrease in episodes of vulvovaginal candidiasis for a period of twelve months in women aged under 40 [82]. Nonetheless, in some RVVC cases, therapeutic vaccines might exacerbate the disease due to an overly pronounced inflammatory response. Crucially, the development of a vaccine capable of combatting RVVC holds the potential to pave the way for a vaccine against severe Candida infections, even those induced by drug-resistant strains. Thus, it is imperative to continue the progression of the NDV-3A vaccine and provide valuable insights into defining clinically relevant endpoints for the immunotherapeutic management of candidiasis.
Discussion of the Various Immune Responses Elicited by Different Vaccine Types
The immune responses stimulated by fungal vaccines are crucial for advancing vaccine strategies against fungal infections. T cell-mediated responses, supported by various cellular components and soluble components like cytokines, play a central role in combating fungi. Vaccines, especially when used with adjuvants, not only initiate but also enhance immune responses, with some increasing antibody responses, while others primarily intensify T cell responses. Most vaccines boost both types of immune responses. Ensuring the efficacy and safety of antifungal vaccines is essential, especially in individuals with compromised immunity. An ideal antifungal vaccine should be immunogenic and protective without exacerbating immunopathology or worsening underlying diseases.
The understanding of the pathogen infection cycle, structural features on the pathogen, and site of infection are key determinants to balancing the cellular and humoral responses generated by the vaccine candidate under exploration. Inducing a cellular response is preferred for pathogens causing chronic or mucosal infections. Vaccines can utilize attenuated or killed microorganisms, pathogen-specific proteins, or polysaccharide-protein conjugates. Antigen presentation by antigen-presenting cells (APCs) to T cell receptors (TCRs) and the cytokine cocktail generated by pattern recognition receptors (PRRs) is crucial for effective immune response stimulation. Adjuvants like aluminum and calcium salts enhance the humoral response by activating PRRs, while novel adjuvants based on PAMP-PRR interactions aim to enhance specific cellular immunity [83].
Adjuvants play a pivotal role in disrupting immune tolerance during Candida infections by enhancing the immune system's recognition and response to this pathogen. In the context of the human gut, the presence of adjuvants becomes particularly relevant, as Candida species are commonly found in the gastrointestinal tract as part of the normal microbiota. By co-administering antigens from Candida with adjuvants, the immune system is not only primed to recognize and respond to the pathogen but is also encouraged to break immune tolerance, ensuring a robust and protective immune response. Numerous studies provide substantial evidence supporting the incorporation of pathogen-associated molecular patterns (PAMPs) derived from fungi as potent adjuvants to customize the immune responses elicited by vaccines. For example, various TLRs, Dectin-1, Dectin-2, and Mincle have been recognized for their significant roles, not only in activating innate defense mechanisms but also in regulating the differentiation of T cells toward protective Th1 and Th17 lineages [84]. Further, carbohydrates like glucans, dextrans, lentinans, glucomannans, galactomannans, chitin/chitosan, levans, and xylan derived from fungal components have the potential to be used as antigens as well as adjuvants, as they can stimulate innate cells [85]. This feature presents a promising avenue for investigating molecules with shared structures across different pathogenic fungi as a foundation for designing universally applicable antifungal vaccines. Formulations of fungal vaccines that encompass combinations of fungal PAMPs have exhibited immense potential in mouse models, acting as mediators of precisely tailored antifungal immunity. Among these formulations, glucan particles, enriched with β-glucans and chitin, have demonstrated versatility as carriers for various antigens in antifungal vaccines [86]. A significant hurdle lies in deciphering the components that define effective immunity against diverse pathogenic fungi and in devising strategies to trigger the adaptive immune response effectively, while minimizing any adverse reactions. Clearly, a critical determinant for the success of future vaccine development efforts lies in the judicious choice of adjuvants, which can steer the immune response in a manner that aligns with the specific protective requirements for each fungal pathogen. This approach holds great promise for preventing and controlling Candida infections, especially in individuals with compromised immune systems or those at risk due to medical conditions or treatments that disrupt the delicate balance of gut microorganisms.
Action of vaccine candidates employing cellular immunity mostly rely on Th1 responses; Th1 response is vital for resolving diseases caused by opportunistic pathogens like A. fumigatus. Effective vaccination should prioritize directing the immune response to-wards a Th1 response, inducing key cytokines like IL-12 and IFN-γ. Some protein antigens stimulate the production of IL-12 or IFN-γ after vaccination, suggesting their importance in vaccine efficacy [34].
The Role of Innate and Adaptive Immunity in Protection against Candidiasis
The initial stride in formulating a potent vaccine candidate against fungal infections entails enhancing our understanding of various arms and elements involved in antifungal immune response. Once a deep understanding of the mechanisms governing immune responses to fungal infections is achieved, it will be easy to develop and target new avenues for the success of vaccine strategies. In protection against fungal infections, the roles of innate and adaptive immunity are emphasized, with their relative contributions varying based on the anatomical location of the disease. Innate immunity is crucial on mucosal surfaces, while adaptive immunity becomes essential for protection against mucosal candidiasis in humans. Tissue-specific immunity is evident in systemic candidiasis, where IL-17 plays a role, and Th1 and NK cells are critical. In oral and dermal candidiasis, IL-17A-mediated antifungal immunity is paramount, involving both hematopoietic and non-hematopoietic cells, such as phagocytes, dendritic cells, and mucosal epithelial cells [87]. Tissue macrophages and dendritic cells are pivotal in restricting fungal growth and dissemination, and neutrophils are rapidly recruited to infection sites to combat fungal spores and hyphae [88]. The pivotal phase in triggering an immune response against fungal infections involves the identification of distinct fungal elements known as pathogen-associated molecular patterns (PAMPs) through pattern recognition receptors (PRRs). Various cells of the innate immune system detect fungal components through pattern recognition receptors (PRRs) like TLRs and dectin-1, leading to the activation of immune responses to control fungal infections [89].
Cell-mediated immunity, rather than humoral immunity, is considered the primary protective response against candidiasis [90]. Polymorphonuclear leukocytes and macrophages are essential for protection against candidemia through innate immunity, while CMI through T cells and cytokines predominates in mucosal tissues. The significance of humoral immunity remains inconclusive. CMI is crucial in host defense against mucosal candidiasis in immunocompromised patients and those on corticosteroid therapy. CD4+ cells play a significant role in protection against gastrointestinal Candida infections [88].
The significance of Th1 cell responses and the production of IFNγ in facilitating the fungicidal activities of both neutrophils and macrophages is well established [91]. The activation and differentiation of IFN-γ-producing Th1 cells is important in vaccine-mediated protection against multiple fungal pathogens. Furthermore, Th17-derived cytokines like IL-17 and IL-22 also play a crucial role in the defense against Candida species and in recruiting and activating neutrophils, as well as triggering the activation of epithelial cells and the secretion of antimicrobial peptides [92]. In contrast, cytokines associated with Th2-type immunity have exhibited conflicting roles in the context of a protective host response to Candida species [93,94].
Humoral immune mechanisms have been implicated in host defense against Candida infections; their contribution to antifungal defense is likely to be more modest compared to cellular mechanisms. Activated complement proteins play a significant role in initiating an appropriate cytokine response but are not capable of directly eliminating Candida hyphae in the course of an infection. Studies involving mice deficient in complement factor C3 or C5 display heightened mortality rates due to a compromised ability to resist Candida infections or dysregulated inflammatory response [95,96]. A study demonstrated that antibody genes cloned from B cell cultures obtained from patients with C. albicans infections exhibited the ability to stimulate opsonophagocytic macrophage activity in laboratory settings. Furthermore, these antibodies were found to confer protection against disseminated candidiasis when tested in live animal models [97]. While the studies are still uncovering the role of the humoral arm of immunity in Candida infections, the elicitation of protective antibodies through vaccination may be considered a viable strategy to enhance resistance to such infections [98,99].
The indispensable commensal nature of Candida albicans (C.albicans) in the human microbiota is not merely a consequence of passive co-existence but rather the outcome of the host's robust innate and adaptive immune responses, which serve to curtail the proliferation of this potentially hazardous microorganism on epithelial surfaces, a critical defense mechanism against C. albicans is the Th17 functional subset of T helper cells. The progressive depletion of these cells in individuals with advancing HIV infection leads to the failure to control fungal infection on oral epithelial tissues, enabling C. albicans to unleash its pathogenic capabilities. A significant facet of this pathogenic potential lies in C. albicans' ability to elude host immunity and intensify inflammation and immune activation. Furthermore, HIV infection creates an environment that promotes the overexpression of virulence factors by C. albicans, particularly in regard to the secretion of aspartyl proteinases (Saps). These enzymes possess the capacity to degrade vital components of the host's defense system, such as complement proteins and defensive epithelial proteins [100,101]. The effectiveness of current antifungal therapy over the long term is dependent upon effective collaboration between the host's immune system and the treatment. Consequently, novel therapeutic strategies aimed at targeting virulence factors and specific immune interventions should be considered. Among these innovative approaches, vaccination holds promise as a potential solution to mitigate the impact of C. albicans infections. Targeting antigens that trigger both T cell and antibody responses can be effective in preventing infection. Further, the use of adjuvants that help activate both T cells and B cells, leading to a more robust and protective response, should be considered. Combining multiple antigens from different Candida species can create a broader-spectrum vaccine that leads to a robust immune response.
However, immune mechanisms encompassing a range of responses, including those driven by humoral and cell-mediated immunity, or even better, a combination of both these primary components of the acquired immune system, may seem to be a novel and effective approach. The challenge here remains whether the immune response induced by such vaccines can protect against pathogenic Candida infections without affecting the role of C. albicans as a part of normal microbiota.
Challenges in Developing Effective Vaccines against Candidiasis
The development of effective vaccines against candidiasis is intricate due to challenges rooted in the long-standing co-evolution of Candida albicans with humans. The fungus' presence in the human gastrointestinal tract since birth implies mechanisms of immune evasion via morphological, genetic, and phenotypic adaptability. The development of vaccines against Candida species presents a multifaceted challenge, primarily due to the remarkable genetic and morphological variability among these fungi.
The fungi have the remarkable ability to display morphological and phenotypic plasticity [102]. C. albicans is a polymorphic fungus capable of transitioning reversibly between yeast, pseudohyphal, and hyphal forms. This ability is closely tied to its evolutionary adaptation within the human host. The unicellular yeast form C. albicans is typically considered a harmless colonizer. However, the transformation to the hyphal form is associated with pathogenesis, as hyphal structures adhere to and invade epithelial cells. Consequently, proteins specific to hyphal or hyphal-associated forms, such as Hyr1, Hwp2, Plb5, and Sod5, have been proposed as potential vaccine targets [103]. Although an advantage of targeting the invasive hyphal antigens/epitopes is highlighted by several research groups, it should be based on the selective targeting of the pathogenic hyphal form, while minimizing the impact on the normal commensal yeast forms and the normal microbiota [104,105]. Nonetheless, thorough evaluation of the utility of hyphal antigens and associated proteins can only lead to a specific vaccine candidate overriding the hindrances due to genetic variations between different Candida species. Additionally, C. albicans undergoes a heritable white-to-opaque phenotypic switch, potentially aiding in immune evasion. Opaque cells are less susceptible to phagocytosis by macrophages and can evade neutrophil killing. Beyond morphological and phenotypic variations, C. albicans displays significant genomic plasticity. This includes extensive chromosome rearrangements, aneuploidy, and loss of heterozygosity in response to various stresses, subtelomeric hypervariation, and aneuploidy [106,107]. Such genomic changes enable rapid adaptation to adverse environments by altering the copy number of specific genes on particular chromosomes.
Given the extensive antigenic and genetic diversity within C. albicans, a more effective approach to anti-Candida vaccine development may involve targeting multiple unrelated virulence-associated antigens simultaneously. Multivalent vaccines, capable o carrying multiple antigens from various strains or serotypes of the same pathogen, could offer a more comprehensive and adaptable strategy against the diverse array of antigens and virulence factors spread over different organs and at different times [108]. This strategic approach involves the amalgamation of well-known immunogenic antigens, and the identification of dominant antigens through computational analysis may play an important role. This approach holds promise in addressing the challenges posed by Candida species' genetic and morphological variations.
Further, the situation is complicated by the development of immune tolerance towards Candida, as they are commensal organisms. The diversity in C. albicans morphological forms-yeast, pseudohyphal, and hyphal-each contributing to pathogenesis, further complicates vaccine design. Immune tolerance, existing vaccine univalence, and the balance between colonization and infection pose notable challenges in vaccine development. Our complex gut microbiome requires immunological tolerogenic responses to maintain gut homeostasis and prevent chronic inflammation. It is likely that Candida species, the most common fungal species in the gastrointestinal tract throughout life, have evolved tolerance mechanisms similar to other gut bacteria like Bacteroides fragilis and certain Clostridia species to regulate the relationship between humans and fungi. One mechanism, likely resulting from the co-evolution of bacterial microbiota, commensal fungi, and the host immune system, relies on the metabolic tryptophan-AhR pathway and 2,3indoleamine dioxygenase (IDO) [109]. Candida albicans induces IDO expression in dendritic cells (DCs) and promotes tolerogenic Treg responses, possibly facilitating its transition from pathogenicity to commensalism.
Maintaining immune tolerance toward human gut commensals such as Candida represents a challenging task in the way to the development of a potent vaccine against C. albicans. The presence of immune tolerance towards C. albicans, and potentially other Candida species, presents two significant obstacles in the development of vaccines against Candida. Firstly, it hinders the development of strong and lasting immunological memory. Secondly, many of the clinical signs of Candida-related infections result more from the host immune system's response than directly by the pathogen [110,111]. Striking the right balance between immunity and tolerance is critical to maintaining gut homeostasis, as breaking host tolerance could lead to unintended consequences, including worsening fungal infections or exacerbating underlying inflammatory or autoimmune conditions. Further, understanding the complex interactions between Candida and other commensal microorganisms and their impact on vaccine efficacy may be a significant challenge. The major challenges faced in development of anti-Candida vaccines are summarized in Figure 3. Utilizing live attenuated strains in fungal vaccines is challenged by safety concerns, particularly among immunocompromised individuals. Achieving protection in such contexts and addressing potential autoimmunity against commensal fungi is essential. The majority of avirulent strains have faced obstacles preventing their progression to clinical trials. These include factors concerning the possibility of reversion to virulence, infections in people with compromised immune systems, unpredictable immune reactions, risks of horizontal transmission, a lack of information on the safety of vaccines in immunocompromised populations, difficulties with strain stability, genetic modification or mutation stability of these attenuated strains, failure to reproduce results in human volunteers at a clinical trial, and the requirement for stringent safety monitoring and stability during transport. Another significant concern related to conventional vaccines is the uninterrupted preservation of the cold chain, which must be maintained consistently. Clinical trials for live attenuated Candida vaccines may exclude immunocompromised individuals, leading to limited data on their safety and efficacy in this population. Ensuring the genetic stability of these attenuated strains over time is also a challenge, and rigorous safety monitoring is essential. As a result, alternative vaccine approaches, such as subunit or inactivated vaccines, are often preferred for immunocompromised individuals in Candida vaccine development. Despite these challenges, efforts are underway for antifungal vaccines with alternative routes of administration, lower cost, and adaptability for resource-poor settings, driven by the increasing immunocompromised population and the pressing need to counter fungal infections. A symbiotic relationship between C. albicans and mice influences gut microbiota, metabolism, and immunity. Antifungal drugs induce dysbiosis, emphasizing mutual recognition and protective mechanisms. Immune memory can fail against evolved pathogens, particularly in immunocompromised settings. Efficacy testing with humanized mice models offers a solution. Vaccine strategies must encompass C. albicans and non-albicans species. Balancing vaccine effectiveness between healthy and weaker individuals is complex, potentially requiring passive immunotherapy. Adjuvants enhance immune responses, but balancing vaccine safety and effectiveness remains a challenge.
Immunological impairments pose challenges to vaccine effectiveness and safety. Live vaccines exhibit high immunogenicity but carry infection risks. Inactivated whole organism and subunit vaccines enhance safety but may not be effective in immunocompromised individuals. Strategies include enhancing adjuvants and formulations and targeting individuals with robust immune systems. Transitioning from animal studies to human application adds further complexity. Despite several recent concerns about possible risks, alum is still the gold standard for the use of new adjuvants for human use in vaccines. Since the majority of systemic fungal infections gain entry into the host through mucosal surfaces such as the upper respiratory, gastrointestinal, vaginal, or urinary tracts, there is considerable interest in exploring the use of adjuvants for mucosal immunization (via oral, intranasal, and other routes). This approach focuses on stimulating mucosal immune responses, particularly secretory IgA, which is pivotal for delivering antigens effectively to the mucosal-associated lymphoid tissue. This method not only offers a safer and more cost-effective alternative but also simplifies large-scale vaccination efforts. Nevertheless, a significant challenge in creating an efficacious mucosal vaccine lies in the necessity to breach the mucosal epithelial barrier, ensuring efficient antigen presentation to the mucosal immune system, and surmounting the natural tolerance mechanisms at mucosal surfaces. In essence, the judicious selection of adjuvants and delivery systems is paramount to achieving the optimal protective mucosal immune response.
Progress in vaccine development has been notable, yet several challenges persist. Key issues include potential differences in vaccine efficacy between animal models and humans, limited vaccine persistence and efficacy, possible vaccine-related toxicity, lack of standardized manufacturing processes, resource-intensive clinical trials, and the complexity of ensuring vaccine stability during production, transportation, and storage. Enhanced comprehension of C. albicans-host interaction mechanisms is pivotal for identifying new targets. Preclinical trials should encompass various animal species, diverse host statuses, and experimental parameters like susceptibility and dosing. Real-time feedback in clinical trials is essential for determining dosing regimens based on safety, pharmacokinetics, and therapeutic effects. The critical need to assess vaccine-drug synergy and overcome immune response obstacles in immunocompromised individuals underscores the ongoing quest for a systemic C. albicans infection.
Potential Directions for Future Research
The advancement of an effective anti-Candida vaccine demands innovative strategies that encompass diverse facets of research [112]. The ideal vaccine necessitates heightened immunogenicity, broad-spectrum protection, addressing superficial and bloodstream infections, and effectiveness in immunocompromised individuals. As conventional approaches targeting a single antigen prove limited, the current focus leans towards multivalent formulations composed of multiple antigens from various strains. This strategy not only mirrors successes in other vaccine domains but also addresses the challenges posed by C. albicans' virulence factors. In lieu of univalent vaccines, the notion of targeting multiple virulence-associated antigens concurrently has emerged to prevent the emergence of "escape mutants" and a greater specificity against the pathogen. This would help preserve the commensal, and potentially beneficial, aspects of this gut inhabitant. One strategy may be to design a multivalent vaccine using antigens showing effective results in univalent vaccines such as Als3 and Sap2. Such multivalent vaccines would address the challenge of genetic and phenotypic diversity shown by Candida. The antigenic targets can encompass various Candida strains and different forms of the fungus, such as yeast and hyphal forms. Combining major antigens associated with critical C. albicans virulence or biological functions can have a synergistic effect on immune responses. This approach can broaden the range of protective antibodies generated by the immune system and decrease the likelihood of fungal immune evasion. By targeting multiple aspects of the fungus' virulence and biology simultaneously, a more comprehensive and robust defense can be mounted against C. albicans infections [80].
Trained immunity, induced by certain vaccines, is mediated by innate cells like monocytes, macrophages, or NK cells. Traditionally, the focus has been on adaptive immune responses, but recent research has unveiled the potential of the innate immune system's memory. Recognizing the potential of harnessing trained immunity, induced by specific vaccines, could be of paramount importance, particularly given Candida's genetic diversity. A significant portion of research in this domain has focused on inducing trained immunity using bacterial products like lipopolysaccharide (LPS) or Bacillus Calmette-Guérin vaccine (BCG), which can activate distinct trained immunity pathways guarding against subsequent infections. This concept holds potential in reducing candidiasis vulnerability. Some of the earliest instances of trained immunity were demonstrated in murine studies through lowdose Candida spp.infections in T and B cell-depleted animals, which exhibited protection against subsequent lethal fungal infections. This immune response was mediated by the fungal cell wall component β-glucan and relied on functional circulating monocytes [113,114]. The significance of trained immunity could potentially open up an important avenue for achieving cross-protection against various other infections. Systemic infection of mice with avirulent C. albicans provided protection against virulent strains [39]. The protective effect was mediated by macrophage-like cells and was found to be non-specific, as it equally extended to cross-protection against C. tropicalis and Staphylococcus aureus [42]. Promising antigens like β-glucan have been explored in this context, as similar effects were observed, mediated by increased secretion of cytokines like TNF-α and IL-6 from the β-glucan trained monocytes [30]. A preliminary study was conducted to explore the feasibility of employing β-glucan as a vaccine in humans. This involved the oral administration of β-glucan, followed by the assessment of innate immune reactions in PBMCs of healthy volunteers that were later stimulated with C. albicans in vitro [115]. Unfortunately, the findings yielded less favorable outcomes in human subjects. Further studies using different administration routes are still worth exploring to assess the role of β-glucan as a vaccine candidate to induce trained immunity against subsequent Candida infections. By inducing trained immunity through exposure to avirulent Candida species or antigens, the body can develop a heightened state of readiness to combat subsequent Candida infections. This approach offers several advantages, including the capacity to provide broad-spectrum protection against different Candida strains and the ability to enhance the immune response in immunocompromised individuals [116].
Antigens derived from Candida molecules show promise for animal model and clinical trial experimentation. Preventive vaccines for superficial candidiasis are of particular interest, targeting recurrent vaginal candidiasis and denture stomatitis. Recombinant protein-based vaccines, rAls3p-N (NDV-3) and rSap2t (PEV7), hold potential against recurrent vaginal candidiasis and even antibiotic-resistant S. aureus infections. Focus also extends to systemic vaccines against invasive candidiasis, potentially synergizing with antifungal drugs. MNs or BGs, combined with appropriate adjuvants, is a promising area for the development of pan-fungal vaccines [117]. Phylogenetically conserved antigens like MNs and BGs are found in the cell walls of many fungal species. These components exhibit common structural elements and functions across various fungal pathogens. For example, MNs are critical for cell wall integrity in numerous fungi, making them attractive targets for pan-fungal vaccines. Further, these antigens can induce immune responses with cross-reactivity against multiple fungal species. Pan-fungal vaccines are cost-effective and efficient, offering a broader spectrum of protection against fungal infections like candidiasis, aspergillosis, and pneumocystosis. A "pan-fungal" peptide vaccine, NXT-2, was developed by utilizing a previously identified recombinant pan-fungal protein (NXT-2) in multiple models of invasive fungal disease. Vaccination with NXT-2 significantly enhanced immunogenicity and showed protection in severely immunosuppressed animal models of aspergillosis, candidiasis, and pneumocystosis, compared to controls, thus suggesting that immunization with a pan-fungal vaccine could provide broad, cross-protective antifungal immunity for at-risk individuals, potentially addressing a critical gap in fungal infection prevention [81]. However, challenges such as selecting the right conserved antigens, designing effective formulations, and achieving desired immunogenicity are the key for a successful vaccine candidate.
In crafting the ideal vaccine, the emphasis lies on high immunogenicity, encompassing protection against various fungal pathogens and conditions. The pursuit of a pan-fungal vaccine capable of addressing both superficial and bloodstream infections for individuals with compromised immunity is of paramount importance. The utilization of DNA polymerase subunit knockouts to generate whole-cell vaccines shows promise.
A pan-fungal vaccine strategy is essential, supported by DNA polymerase subunit knockouts in C. albicans, promising systemic candidiasis protection, even in non-albicans species. These advances mirror those directed against viruses and bacteria, fostering hope for multivalent vaccine development, and broadening protecting immunity against multiple infections.
Resolving the intricacies of host-fungus interaction, particularly concerning cell wall glycans, offers potential for novel treatments [118]. Amidst the challenges of drug-resistant fungal pathogens, efforts in diagnostic and therapeutic innovation continue. Ongoing clinical trials targeting high-risk groups, including the immunocompromised, strive to combat emerging fungal threats [119]. Understanding antifungal drug resistance mechanisms through chemical genomic approaches aids treatment strategies. The exploration of diverse chemical libraries could lead to targeted antifungal solutions [120]. It is vital to keep in mind that the vaccine being developed must not disrupt Candida species in their advantageous commensal state because doing so could impede the host's ability to grow. These therapies ought to be similarly effective in those with compromised immune systems. Instead of depending on inbred animals, vaccine candidates can be tested during the preclinical stage using humanized and immunodeficient mouse models like SCID or nude mice, and further validated through bigger animal models. A whole-cell vaccine that is specially tailored to target the pathogenic form offers distinct advantages in instances where antigenic peptide-based vaccines could find it difficult to differentiate between commensal and pathogenic Candida forms. Given the ease of attenuated and whole-cell vaccines, highly stable and antigenic live attenuated strains of C. albicans should be investigated, as they could potentially offer broad-spectrum protection.
Further, oral vaccines offer numerous advantages, such as ease of administration, cost-effectiveness, and potential for mass immunization. Such vaccines are increasingly attracting interest because of their convenient mode of delivery, reduced invasiveness, typically enhanced safety profile, and cost-effectiveness compared to injectable vaccines. The development and utilization of oral vaccines are key in addressing urgent health challenges and are exemplified by recent advances in the COVID-19 pandemic. The identification of unique antigenic proteins within Candida species and the role of cell wall glycans and other virulence factors in Candida may lead to the development of a novel oral vaccine candidate.
Further, addressing the challenges in developing effective candidiasis vaccines requires thorough exploration for adjuvant optimization. The strategic choice of potent adjuvants and delivery systems is instrumental in steering immune responses towards desired outcomes, an indispensable facet of successful vaccine development. It is important here that we explore novel, biocompatible, and potent adjuvants such as phytocompounds (ginsenosides like Rg1, Re, and Rd) or strategies such as combination adjuvants against C. albicans for a more effective and cost-efficient Candida vaccine [59,121].
An emerging new area which can accelerate the development of an anti-Candida vaccine is that of in silico epitope prediction using immunoinformatic approaches-the development and use of algorithms and tools for predicting B and T cell epitopes. The pathogen genome can be analyzed to identify potential antigenic proteins and the epitope sequences. Protein 3D structures and antigenic features can also be screened using a structural and sequence homology search, followed by protein-protein interaction analysis and docking. These approaches can increase the efficiency of a vaccine candidate search, especially in terms of cost and time taken for identifying potential epitopes [122]. Akhtar and coworkers recently demonstrated the use of immunoinformatic approaches in predicting a potent vaccine candidate against Candida tropicalis. They identified eleven conserved antigenic but non-allergic and non-toxic epitopes on secreted aspartic protease 2 protein. Additionally, they paired the epitopes RS09 (LPS peptide mimic/TLR agonist), flagellin sequences, and PADRE sequence to create a vaccine candidate that can potentially elicit humoral and cell-mediated immune responses. The construct was also indicated to be non-homologous to any human protein and hence potentially safe [79].
Collectively, these multidimensional pathways chart the course for effective anti-Candida vaccine development. Innovative strategies such as multivalent formulations, trained immunity, and exploring for immunogenic but safe Candida antigens and/or ad-juvants underscore the possibilities. Pioneering experimental initiatives that focus on particular infections and specific populations, along with a detailed understanding of drug resistance mechanisms, drives the journey forward. While there is certainly a requirement for safe and efficient vaccine, the obstacles are significant in terms of both conceptual understanding and technical aspects in developing successful vaccines against conditions like candidiasis and other fungal infections. Nonetheless, the comprehensive approach, along with the pursuit of an effective anti-Candida vaccine, holds the potential to combat these fungal infections and enhance global health outcomes.
Conclusions
Candidiasis-the diverse spectrum of fungal infections caused by various Candida species-is a substantial global health challenge. Increasing reports of multidrug resistance in Candida species and the expanding immunocompromised population underscore the urgency for inventive approaches like a novel prophylactic vaccine. However, an anti-Candida vaccine is yet unrealized due to the challenges posed by the variability and adaptability of Candida species, pre-existing immunological tolerance, and impaired adaptive immunity in certain patients. The advantages of an anti-Candida vaccine are likely to benefit a large socio-economic demographic and represent long-term measures to reduce the disease incidence. The authors endorse a multifaceted strategy to facilitate vaccine development against Candida spp.-screening various pathogenic species' and strains' of Candida genome and proteome for novel, safe and potent epitopes, identifying stable and antigenic strains, the optimization of vaccine formulations with a special focus on novel adjuvants, and rigorous clinical trials to ensure a safe and efficacious vaccine. Stress is also to be laid on meticulous animal model selection, stability assessments, and safety protocols as a crucial step before entering clinical phases. Additionally, striking a balance between cellular and humoral responses based on pathogen and infection site is crucial for safe and effective antifungal vaccines, especially for immunocompromised individuals. Furthermore, harnessing the emerging insights into trained innate immunity, multivalent vaccines and pan-fungal vaccines offers promising prospects. The enlisted strategies can not only facilitate tackling candidiasis but also potentially provide cross-protection against other opportunistic infections. Collaborations are warranted between researchers, healthcare providers, regulatory bodies, and industries to pave the way to the successful development of a novel anti-Candida vaccine.
Figure 1 .
Figure 1. Species distribution for invasive candidiasis per 100,000 cases from
7 of 29 Figure 2 .
Figure 2. The state-of-the-art anti-Candida therapeutics comprise antifungal drugs and potential antifungal vaccines. The advantages conferred by anti-Candida vaccines outweigh the antifungal drugs.
Figure 2 .
Figure 2. The state-of-the-art anti-Candida therapeutics comprise antifungal drugs and potential antifungal vaccines. The advantages conferred by anti-Candida vaccines outweigh the antifungal drugs.
Figure 3 .
Figure 3. Major challenges in anti-Candida vaccine development.
Table 1 .
List of various vaccine candidates in development against candidiasis.
==
Domain: Biology Medicine
|
[{"added": "2023-11-01T15:26:37.464Z", "created": "2023-10-29T00:00:00.000", "id": "264804587", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": null, "oa_url": null, "pdf_hash": "a92b6fbe40f6965c070e39bb6a912002e402b457", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:7928", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "9fc9defdb943566d476c49a8ecc3eba2f73a5891", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Overexpression of Nrf2 attenuates Carmustine-induced cytotoxicity in U87MG human glioma cells
Malignant glioma is one of the most devastating tumors in adults with poor patient prognosis. Notably, glioma often exhibits resistance to conventional chemotherapeutic approaches, complicating patient treatments. However, the molecular mediators involved in tumor chemoresistance remain poorly defined, creating a barrier to the successful management of glioma. In the present study, we hypothesized that the antioxidant transcription factor, Nrf2 (nuclear factor erythroid-derived 2 like 2), attenuates glioma cytotoxicity to Carmustine (BCNU), a widely used chemotherapeutic agent known to modulate cellular oxidative balance. To test the hypothesis, we employed human malignant glioma cell line, U87MG and overexpression of Nrf2 in glioma cells was achieved using both pharmacological and genetic approaches. Notably, induction of Nrf2 was associated with increased expression of heme oxygenase-1 (HO-1), a stress inducible enzyme involved in anti-oxidant defense. In addition, over expression of Nrf2 in U87MG cells significantly attenuated the cytotoxicity of Carmustine as evidenced by both cellular viability assay and flow cytometry analysis. Consistent with this, antioxidants such as glutathione and N-acetyl cysteine significantly reduced Carmustine mediated glioma cytotoxicity. Taken together, these data strongly implicate an unexplored role of Nrf2 in glioma resistance to Carmustine and raise the possible use of Nrf2 inhibitors as adjunct to Carmustine for the treatment of malignant glioma.
Background
Malignant glioma is one of the most devastating tumors in adults. The worldwide annual incidence of malignant glioma is approximately 6 cases per 100,000 people [1] and each year, more than 14,000 new cases are being diagnosed in the United States. In contrast to other solid tumors, glioma presents various therapeutic challenges that include its intracranial location, aggressive biological behavior and infiltrative growth. Though multimodal treatment regiments are being used for the treatment of malignant glioma, it is often associated with poor patient prognosis and the mean life expectancy of patients is still less than 14 months [2].
Carmustine or bis-chloroethylnitrosourea (BCNU) wafer is the only FDA approved intracerebral chemotherapeutic agent for the treatment of newly diagnosed and recurrent malignant glioma [3]. After maximal surgical resection of tumors, biodegradable wafers of Carmustine (Gliadel®) are implanted inside the tumor cavity, providing an innovative way of delivering chemotherapy directly to the brain tumors with minimal systemic toxicity and greater efficacy than systemic Carmustine administration [4]. However, the recent studies demonstrated that the efficacy of Carmustine is substantially limited by chemoresistance [5]. Carmustine exerts tumor cytotoxicity via multiple mechanisms and it often interferes with DNA replication and transcription [6,7]. In addition, Carmustine is known to carbamylate lysine residues on proteins [8] causing protein carbamylation, a post translational protein modification that could irreversibly inactivate enzymes including glutathione reductase [9][10][11]. Therefore, by inhibiting glutathione reductase an enzyme that plays critical roles in cellular oxidative balance, Carmustine treatment may modulate the cellular oxidative status.
Nrf2 is a key redox-sensitive transcription factor that regulates the expression of endogenous antioxidants, phase II detoxification enzymes, and other cellular defensive proteins in response to cellular stress. The transcriptional activity of Nrf2 is negatively regulated by the cytoplasmic protein, Kelch-like ECH-associated protein 1 (Keap1) [12,13]. Under homeostatic conditions, Keap1 constitutively targets Nrf2 for ubiquitin conjugation and subsequent proteasome degradation in the cytoplasm by acting as a substrate adaptor for the Cul3-based E3 ubiquitin ligase complex [14]. Upon exposure of cells to oxidative stress or Nrf2 inducers such as tertbutylhydroquinone (TBHQ), multiple cysteine residues on Keap1 are alkylated, compromising the ability of Keap1 to efficiently ubiquitinate Nrf2 and resulting in elevated Nrf2 protein levels and transcriptional activity. Though recent studies demonstrated a role of Nrf2 in glioma invasion [15], angiogenesis [16,17], the selfrenewal of glioma stem cells [18], and temozolomidemediated cytotoxicity [19,20] its precise role in tumor progression remains largely controversial. Moreover, the functional role of antioxidant transcription factor Nrf2 in malignant glioma resistance to Carmustine remains largely uncharacterized. Altogether, given the role of Nrf2 in antioxidant defense mechanisms coupled with the potential modulation of cellular oxidative status by Carmustine treatment, we hypothesized that Nrf2 may functionally regulate tumor cell sensitivity to the cytotoxic effects of Carmustine, a widely used intracerebral chemotherapeutic agent.
Materials
All cell culture reagents, sera, and media were purchased from Hyclone Laboratories (Logan, UT). Carmustine (BCNU) and tert-butylhydroquinone (TBHQ) were purchased from Sigma-Aldrich Co (St. Louis, USA). TBHQ was dissolved in dimethyl sulfoxide (DMSO) and DMSO was used as a vehicle in all studies. MTT was purchased from Calbiochem (USA).
Cellular viability assay
MTT reduction assay was performed as an estimate of cellular viability, as described earlier [21]. Briefly, cells (3 × 10 4 cells/well) were plated overnight in 24-well plates and treated with vehicle or TBHQ or Carmustine as detailed in respective figure legends. Following treatments, MTT (5 mg/mL; 50 μl/well) was added to each well and incubated for 4 h at 37°C. The wells were then emptied and the blue formazan salts were dissolved in acidic isopropanol (400 μl/well) and absorbance was measured using a plate reader (Biotek) at 540 nm using a reference wavelength of 690 nm. Cellular viability was normalized to vehicle treated control wells, which represented 100% viability.
BrdU (bromodeoxyuridine) incorporation assay
Cell proliferation was measured by estimating BrdU incorporation using a Proliferation Assay kit (Calbiochem, Merk, Darmstadt, Germany), as per the manufacturer's instructions. Briefly, U87MG cells were cultured overnight in 96-well plates at a density of 10 4 cells/100 μl/well in complete growth media and treated with TBHQ/Vehicle for 24 h. BrdU label solution (Calbiochem) was added 4 h prior to the completion of TBHQ treatment. The anti-BrdU antibody was added and incubated for 1 h at room temperature and this was followed by 30 min incubation with the respective secondary antibody. The absorbance was read at 450 nm on a Synergy HT Biotech Elisa reader.
Flow cytometry
Cell death was quantified by flow cytometry, as described previously by our group [22]. Briefly, the cells were plated overnight at density of 100,000 cells/well and treated with either vehicle or TBHQ (30 micromoles) for 6 h. The wells were then emptied and vehicle or Carmustine (50 μg/ml) was added and incubated for 18 h. Afterwards, the adherent and non-adherent cells were collected and washed and the cell suspensions were stained for 15 min at room temperature with annexin V-PE (BD Pharmigen, San Diego, CA), an early apoptotic marker, and with 7aminoactinomycin D (7-AAD), a fluorescent marker that labels dead cells. The percentage of apoptotic or necrotic cell death was quantified using a FACScan flow cytometry.
Immunocytochemistry
U87MG cells after treatment with either vehicle/TBHQ (30 μM) for 6 h were fixed with ice cold methanol for 5 minutes. Cellular fixation was followed by washing twice with PBS and a 10 min treatment with 0.1% Triton-X 100 in PBS. Cells were then incubated with 12% donkey serum for 1 h at room temperature to block any nonspecific binding of antibodies. Primary antibody [Nrf2 (1:100; Santa Cruz Biotechnology, Santa Cruz, CA)] incubation was carried out for 18 h at 4°C, and this was followed by secondary antibody (Alexa Fluor) incubation for 2 h at room temperature. Finally, cells were cover slipped with a mounting medium containing nuclear stain DAPI and immunofluorescent analysis was performed using a LSM510 Meta confocal laser microscope (Carl Zeiss, Thornwood, NY, USA).
Western blotting
Western blotting was performed as described by our laboratory [21,22]. Briefly, cells after respective treatment were washed with phosphate buffered saline (PBS) and whole cell lysates were collected in radioimmunoprecipitation (RIPA) buffer containing protease inhibitor cocktail, and phenyl methane sulfonyl fluoride (PMSF). Cell lysates were sonicated, centrifuged for 5 min at 14,000 rpm at 4°C, and protein concentrations were quantified by BCA protein assay kit (Pierce, Rockford, IL). Thirty micrograms of protein was resolved on a 4-20% sodium dodecyl sulfate-polyacrylamide gel and transferred onto a polyvinylidene difluoride (PVDF) membrane. Blots were incubated overnight at 4°C in respective primary antibody [Nrf2 (1: 250) Santa Cruz Biotechnology, Santa Cruz, CA), heme oxygenase-1 (1: 1000; Abcam, Cambridge, MA), or β-actin (1:3000; Sigma, St Louis, MO)] followed by a 2-h incubation with a corresponding Alexa Fluor secondary antibody. Blots were visualized using the Li-Cor Odyssey near-infrared imaging system and quantified using Quantity One software (Bio-Rad, Foster City, CA).
Over expression of Nrf2
Precision LentiORF lentiviral particles (Thermo scientific Open Biosystems) were used to overexpress Nrf2 in U87MG cells as per manufacturer's recommended protocol. Briefly, U87MG cells were transduced with lentiviral particles at MOI (Multiplicity of infection) of 1.8. Media was replaced 72 h later with growth media and after 48 h, cells were challenged with blasticidin S (5 μg/mL; the minimum concentration required to kill non-transduced U87MG cells). Blasticidin S selection continued for one week, with media replenishment thrice weekly. The Blasticidin S resistant cells were collected and Western blotting was performed to ensure stable Nrf2 overexpression. Control cells were stably transduced with lentiviral particles containing Red Fluorescent Protein (RFP).
ELISA-based measurement of Nrf2 activity
The TransAM Nrf2 Kit (Active Motif; California, USA) was used to assay the DNA-binding activity of Nrf2 in the nuclear extracts of both the RFP and Nrf2overexpressed cells. In brief, 5 μg of nuclear extract prepared using nuclear extraction kit (Active Motif, USA) was incubated in a 96-well plate that was coated with oligonucleotide containing a consensus binding site for Nrf2. After 1 h of incubation, the wells were incubated with 100 μl of a 1:1000 dilution of Nrf2 antibody. This was followed by incubation with 100 μl of a 1:1000 dilution of horseradish peroxidase-conjugated secondary antibody at room temperature. The wells were developed using 100 μl of developing solution for 10 min before the addition of 100 μl of stop solution. Optical density was read at 450 nm with a reference wavelength of 650 nm using a Synergy HT Biotech Elisa reader.
Statistical analysis
For cellular viability studies, n = 4 wells/group were used within each experiment for analysis. For western blotting, all experiments were performed at least in triplicate using independent cell cultures. All experiments were repeated at least three times for the validation of results. Data was analyzed using a one-way analysis of variance (ANOVA), followed by Student-Newman-Keul's or Dunnett's post-hoc test. A P value, p < 0.05 was considered to be statistically significant.
TBHQ up regulated transcription factor Nrf2 in U87MG glioma cells
To establish the role of transcription factor Nrf2 in glioma cells, we employed an Nrf2 inducer, TBHQ. We found that TBHQ treatment (30-120 μM) significantly augmented the protein expression of Nrf2 in U87MG cells ( Figure 1A and B). A 6 h treatment with 30 μM of TBHQ resulted in 133% ± 30 (p < 0.05 vs. vehicle) increase in Nrf2 levels in U87MG cells in comparison to vehicle treated cells ( Figure 1B). Immunocytochemical analysis reaffirmed the induction of Nrf2 upon TBHQ treatment. Notably, the TBHQ treatment also resulted in enhanced nuclear translocation of Nrf2 as evidenced by increased colocalization of Nrf2 with the nuclear stain, DAPI in comparison to control ( Figure 1C). The expression of Heme oxygenase 1 (HO-1), one of the potential downstream targets of Nrf2, was next analyzed. The results showed that U87MG cells constitutively express the HO-1 protein ( Figure 1D). Moreover, U87MG cells treated with TBHQ exhibited a significant increase in HO-1 expression ( Figure 1D and E), suggesting TBHQ mediated upregulation of Nrf2 transcriptional activity. Given the role of redox mechanisms in tumor cell proliferation coupled with the role of Nrf2 in cellular antioxidant defense mechanisms [23], we first questioned whether Nrf2 induction in glioma cells modulates cellular proliferation. To this end, the effect of TBHQ on the proliferation of U87MG cells was evaluated by BrdU incorporation assay. TBHQ treatment augmented BrdU incorporation in glioma cells by 17.5% (p < 0.01 vs. vehicle) as compared to vehicle treatment (Figure 2A). Moreover, the MTT proliferation assay further validated the increase in glioma cell proliferation by TBHQ ( Figure 2B) and demonstrated 11.5 and 14.7% increase in proliferation upon 30 and 60 μM of TBHQ treatment, respectively.
TBHQ attenuated Carmustine induced cytotoxicity in glioma cells
To delineate the role of Nrf2 in chemoresistance to Carmustine, we pre-treated U87MG cells with TBHQ for 6 h and the cytotoxic response to Carmustine was studied. TBHQ pre-treated U87MG cells exhibited significantly lower cytotoxicity to Carmustine in comparison to vehicle treated cells ( Figure 3A). Along these lines, Carmustine (50 μg/ ml) induced 85.26 ± 0.8761% cytotoxicity in U87MG cells and upon TBHQ pretreatment the Carmustine mediated cytotoxicity was reduced to 44.64 ± 2.325% (p < 0.001 vs. Carmustine treatment alone) ( Figure 3A). Moreover, the 6 h pre-treatment with TBHQ did not significantly increase the proliferation of U87MG cells in comparison to vehicle treated controls (n = 8; data not shown) suggesting the role of Nrf2 mediated antioxidant signaling independent of proliferation in attenuating Carmustine mediated cytotoxicity. Furthermore, similar results were obtained in another human malignant glioma cell line, U118 (Additional file 1: Figure S1), reaffirming the role of Nrf2 in Carmustine mediated cytotoxic effects in glioma. The cellular viability studies were further verified using flow cytometry analysis ( Figure 3B), which demonstrated a significant reduction in Carmustine mediated cytotoxicity upon TBHQ treatment ( Figure 3C). . TBHQ treatment of glioma cells also resulted in the induction of HO-1 (MW :~31 kDa), one of the Nrf2 regulated molecular targets, as evidenced by (D) western blotting followed by (E) densitometry analysis. Densitometry is expressed as the mean ± SEM from three independent trials and data were analyzed using One-way ANOVA followed by Dunnett's post-hoc test (* p < 0.05, ** p < 0.01, *** p < 0.001 vs. vehicle-treated cultures).
NRF2 over expression in glioma cells induced resistance to Carmustine mediated cytotoxicity
To further establish the role of Nrf2 in Carmustine resistance, we performed genetic overexpression of Nrf2 in U87MG cells using lentiviral particles. The induction of Nrf2 by lenti viral particles was confirmed by western blotting analysis ( Figure 4A). Cells transduced with lentiviral particles containing Red Fluorescent Protein (RFP) served as the experimental control ( Figure 4D; left panel). In addition, a TransAM ELISA was performed to validate the DNA binding activity of Nrf2 upon genetic overexpression. As shown in Figure 4B, a 103.7% increase in DNA binding activity of Nrf2 was found in the nuclear extracts derived from Nrf2 overexpressed cells in comparison to RFP overexpressed cells. In addition, genetic over expression of Nrf2 was associated with the induction HO-1 further confirming Nrf2 mediated regulation of HO-1 in U87MG glioma cells ( Figure 4C). More importantly, Nrf2 overexpressed cells exhibited significantly lower cytotoxicity to Carmustine in comparison to RFP over expressed cells (Figure 4D and E). To further explore the role of antioxidant mechanisms in tumor cell resistance to Carmustine, we studied the role of antioxidants such as glutathione and N-acetyl cysteine in Carmustine mediated cytotoxic effects. Interestingly, we found that both glutathione and N-acetyl cysteine significantly attenuated Carmustine mediated cytotoxicity in U87MG cells. Carmustine treatment alone induced 39.4% cytotoxicity in U87MG cells, whereas co treatment of Carmustine with glutathione or N-acetyl cysteine the induction of cytotoxicity was reduced to 6.01 and 4.87% respectively ( Figure 5A and B).
Discussion
Though surgical resection is one of the prime treatment options for malignant glioma, the complete surgical removal of the tumor is often a challenge owing to the infiltrative nature of glioma. Thus, the treatment strategy frequently demands chemotherapeutic approaches for improved patient outcomes and a better understanding of the underlying mechanisms of chemoresistance is therefore critical. The intracerebral delivery of Carmustine using Carmustine wafers is found to be very well tolerated in patients and it allows drug release in a constant manner with minimal systemic toxicity [24]. However, Carmustine failed to substantially prolong median survival of GBM patients. The reason for this unsatisfactory clinical outcome remains unclear but may involve intrinsic/acquired chemoresistance of the tumor cells. To this end, variations in multidrug resistance genes [25][26][27] and DNA repair activity [28] have been demonstrated to play roles in glioma resistance to Carmustine. Although, several studies have shown that a deficiency of DNA repair enzyme, O6methylguanine methyl-DNA transferase (MGMT) can increase the sensitivity of glioma to Carmustine [29][30][31], many tumors with low levels of MGMT are nevertheless chemoresistant [32,33], suggesting the involvement of other unknown mechanisms of chemoresistance.
Though Nrf2 has been implicated in chemoresistance to 5-fluorouracil, carboplatin, cisplatin and temozolomide [20,[34][35][36][37], the role of Nrf2 in glioma resistance to Carmustine remained largely uncharacterized. Herein, we report for the first time that both pharmacological and genetic upregulation of Nrf2 in U87MG grade IV malignant glioma cells significantly attenuate Carmustine mediated cytotoxicity. This finding has significant clinical implications, given that tumor cells often exhibit elevated metabolic rate and may have augmented Nrf2 level as a result of tumor cell adaptation to high metabolic demand. Along these lines, a number of malignant tumors such as lung, ovarian, colon, breast and pancreatic cancer, exhibit an increased transcriptional activity of Nrf2 [38][39][40][41][42][43]. Our findings also demonstrate that as a consequence of enhanced Nrf2 expression, glioma cells could acquire augmented cellular proliferation. In addition, a recent study also identified a role of Nrf2 in promoting tumor angiogenesis through the HIF-1 α/ VEGF pathways [16]. Altogether, strategies to pharmacologically attenuate the Nrf2 levels and/or activity may reduce glioma growth and resistance to Carmustine.
Keap1, a BTB-Kelch protein, is regarded as the principal and negative regulator of Nrf2 and several protein kinase pathways, including mitogen-activated protein kinase and protein kinase C, have been implicated in transducing signals that control Nrf2 dependent gene expression [44,45]. Promoter methylation of KEAP1 gene is found in malignant glioma [46] and a strong inverse correlation is discovered between methylation levels and KEAP1 mRNA transcript in tumor tissue [46] suggesting a reduced expression of KEAP1 in glioma. In line with this, both U87MG and T98G glioma cells (data not shown) express basal protein expression of Nrf2. Nrf2 is believed to exert its transcriptional function by forming heterodimers with small Maf (v-maf musculoaponeurotic fibrosarcoma oncogene family) proteins, and binding to ARE-containing gene promoters [14]. Our studies demonstrate that Nrf2 regulates the expression of HO-1 in glioma cells. HO-1 is a rate-limiting enzyme that catalyzes heme degradation in which oxidative cleavage of the porphyrin ring results in the generation of biliverdin (antioxidant), carbon monoxide (anti apoptotic), and free iron [47,48]. HO-1 belongs to heat shock protein family and its expression is triggered by various . The U87MG cells stably over expressing either RFP or Nrf2 were subjected to ether vehicle or Carmustine treatment for 18 h and the cellular viability was assessed by (E) MTT reduction assay and the Figure 4D demonstrates the cellular morphology of cells upon Carmustine treatment using bright field microscopy (Scale bar =200 μm). Data are representative of at least three independent trials (n = 3/trial) and are expressed as mean ± SEM. ** p < 0.01, *** p < 0.001 vs. vehicle-treated cultures. cellular stress stimuli such as reactive oxygen species, hypoxia and heavy metals [49,50]. Owing to the antioxidant and cytoprotective nature of the enzymatic products of HO-1, elevated HO-1 expression due to deregulated Nrf2 signaling could protect tumor cells from oxidative stress-related injury and function as a key component of tumor cell adaptation to oxidative stress induced by chemotherapeutic agents [51]. Therefore, further characterization of Nrf2-HO-1 signaling is highly warranted and may lead to the development of novel therapeutic strategies for malignant glioma.
Apart from heme oxygenase 1 (HO-1), the bestcharacterized Nrf2 downstream genes include glutathione biosynthesizing enzymes such as glutathione S-transferases A1 and glutamate-cysteine ligase. For instance, Nrf2 knockout mice exhibit reduced expression of both detoxification enzymes and antioxidants [52,53]. Oxidative stress is widely implicated in the etiology of cancer and results from an imbalance in the production of Reactive Oxygen Species (ROS) and cell's own antioxidant defenses. ROS are found elevated during cancer and have been shown to activate signaling pathways involved in cellular proliferation and migration [23]. Carmustine-mediated malignant glioma cell death was significantly attenuated by cotreatment with antioxidants such as N-acetyl-L-cysteine and glutathione, suggesting the prominent role of oxidative stress mechanisms in conferring Carmustinemediated cytotoxicity. Many chemotherapeutic agents produce cytotoxic effects via generation of ROS and/or electrophilic actions, which lead to oxidative stress [54][55][56]. To this end, Carmustine is known to inhibit glutathione reductase [57,58], an integral component of the antioxidant defense mechanisms. Therefore, antitumor agents may also activate Nrf2 antioxidant signaling in tumor cells in a ROS-dependent manner, leading to the development of acquired chemoresistance. Though antioxidant vitamins such as retinoids, vitamin C, vitamin E and carotenoids have been extensively investigated in cancer prevention, the role of these vitamins in attenuating the efficacy of chemotherapeutic agents needs to be elucidated. In addition, Nrf2 is also known to regulate the expression of cysteine/glutamate exchange transporter, xCT that maintains extracellular glutamate levels [59]. Therefore, future studies are warranted to demonstrate the role of Nrf2 in regulating xCT expression and/or activity in glioma, as augmented levels of glutamate may facilitate tumor growth by eliciting neuronal damage. Altogether, Nrf2 may represent a very effective and potent therapeutic target for glioma and pharmacological inhibitors of Nrf2 may serve as a useful adjunct to Carmustine for the treatment of malignant glioma.
Conclusion
In conclusion, we have demonstrated a novel role of Nrf2 in malignant glioma cell resistance to Carmustine. Altogether, the data suggest that antioxidant transcription factor Nrf2 might be a potent and viable molecular target for the treatment of malignant glioma.
Additional file
Additional file 1: Figure S1. Human U118 malignant glioma cells (kindly donated by Dr. Raghavan Raju, Allied Health Sciences, Georgia Regents University) were cultured (3 × 10 4 cells/well in 24 well plate) in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal bovine serum, 5% bovine growth serum, and antibiotics in a 37°C humidified incubator at 5% CO2 and were treated with either vehicle or TBHQ (30 μM) for 6 h. After respective treatment, the media were removed; cells were replenished with media containing either vehicle or Carmustine at indicated concentrations and incubated for 18 h. The cell viability was measured using MTT reduction assay. Data from MTT Assay are representative of three independent experiments and are expressed as mean ± SEM. *** p < 0.001 vs. vehicle treated cells.
==
Domain: Biology Medicine
|
[{"added": "2023-01-16T14:08:32.516Z", "created": "2015-03-13T00:00:00.000", "id": "255829892", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1186/s12885-015-1134-z", "pdf_hash": "cce090ecc88223569ecc20abcc5e242a843e1721", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:9925", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "cce090ecc88223569ecc20abcc5e242a843e1721", "year": 2015}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Sex Differences in X-ray-Induced Endothelial Damage: Effect of Taurine and N-Acetylcysteine
Ionizing radiation (IR) can induce some associated pathological conditions due to numerous cell damages. The influence of sex is scarcely known, and even less known is whether the effect of antioxidants is sex-dependent. Given the increased use of IR, we investigated whether male human umbilical vein endothelial cells (MHUVECs) and female human umbilical vein endothelial cells (FHUVECs) respond differently to IR exposure and whether the antioxidants 10 mM taurine (TAU) and 5 mM N-acetylcysteine (NAC) can prevent IR-induced damage in a sex-dependent way. In untreated cells, sex differences were observed only during autophagy, which was higher in FHUVECs. In non-irradiated cells, preincubation with TAU and NAC did not modify viability, lactate dehydrogenase (LDH) release, migration, or autophagy, whereas only NAC increased malondialdehyde (MDA) levels in FHUVECs. X-ray irradiation increased LDH release and reduced viability and migration in a sex-independent manner. TAU and NAC did not affect viability while reduced LDH release in irradiated cells: they have the same protective effect in FHUVECs, while, TAU was more protective than NAC in male cells.. Moreover, TAU and NAC significantly promoted the closure of wounds in both sexes in irradiated cells, but NAC was more effective at doing this in FHUVECs. In irradiated cells, TAU did not change autophagy, while NAC attenuated the differences between the sexes. Finally, NAC significantly decreased MDA in MHUVECs and increased MDA in FHUVECs. In conclusion, FHUVECs appear to be more susceptible to IR damage, and the effects of the two antioxidants present some sex differences, suggesting the need to study the influence of sex in radiation mitigators.
Introduction
It is well established that sex is a major determinant in physiology and pathology [1] which influences many cellular processes, including the cellular redox balance [2][3][4][5][6]. However, it is not yet clearly known whether antioxidant activities are influenced by sex, although some phenolic antioxidants [7,8] and glutathione [9,10] display sex-gender-specific activities.
In humans, there has been an increase in ionizing radiation (IR) exposure because of defence sectors, the nuclear power industry, and health care's use of IR [11]. Importantly, a personalized risk assessment of IR exposure for health professionals and other work sectors, including spaceflight, is still missing [12]. Demographic factors, such as sex, seem to influence IR sensitivity; unfortunately, only a few relevant studies are available [13].
Donors
Umbilical cords from healthy human male and female neonates which were vaginally delivered at term (37-42 weeks) at the Obstetrics and Gynecology Clinic at the University of Sassari were selected from healthy, non-obese, and non-smoking mothers who were drug-free, except for folic acid and iron supplementation. HUVECs were obtained only from the umbilical cords of normal-weight neonates, according to Bertino et al. [54] (2430-4050 and 2550-4190 g for males and females, respectively, which represented the 10th and 90th centiles in Ines charts). Informed consent was obtained from the mothers of all subjects donating umbilical cords following the Declaration of Helsinki.
As previously described [55], cultured cells were characterized as endothelial cells using the exhibition of cobblestone morphology when they were contact-inhibited and an evaluation of the expression of the von Willebrand factor, which is a glycoprotein that is constitutively stored in intra-endothelial Weibel-Palade granules.
FHUVECs and MHUVECs were used at passage 3 to ensure their endothelial characteristics, and all experiments were conducted in duplicate or triplicate. Twenty-four hours before experiments, 50,000 cells at P3 for each experimental condition were suspended in M199 medium without phenol red (Life Technologies, Monza, Italy) and supplemented with 5% FBS and 5% new born calf serum (NBCS) (Life Technologies, Monza, Italy), 1% antibiotic/antimycotic (Sigma-Aldrich, Milano, Italy), and 2 mM of L-glutamine (Sigma-Aldrich, Milano, Italy) to minimize the potential effect of sex hormones contained in the bovine serum.
Experimental Procedures
The experimental groups were: a) non-irradiated HUVECs (basal cells, 10 mM TAUand 5 mM NAC -pre-treated cells) and b) irradiated HUVECs (untreated cells irradiated with 1.6 and 3.2 Gy X-rays; 10 mM TAU-pre-treated and 1.6 and 3.2 Gy irradiated cells; and 5 mM NAC-pre-treated and 1.6 and 3.2 Gy irradiated cells).
Cells were pre-treated with TAU 10 mM or NAC 5 mM (Sigma-Aldrich, Milano, Italy) 24 h before irradiation. Concentrations of the antioxidants were chosen based on the data available in the literature on HUVECs [56][57][58][59][60]. Untreated and pre-treated cells were irradiated. The irradiation was performed through an X-ray tube working at 80 kV and 5 mA (Gilardoni S.p. A, Italy). A Plexiglas layer 1 cm thick filtered the low energy part of radiation. The dose rate of about 0.2 Gy/min was continuously monitored by a DAP camera (Dose Area Product, VacuDAP by VacuTEC, Germany) placed together with the cell holder (microvials or multiwells). The following X-ray doses were used for the experiments: 1.6, 3.2, 6, and 12 Gy. At the highest doses (6 and 12 Gy), the decrease in the viability was higher than 50%, and the increase in LDH release was about 80% in male and female HUVECs. Therefore, these doses were not used for subsequent analysis.
After irradiation, the cells from each vial were seeded in a 96 well for each experimental condition (about 15,000 cells/well in triplicate). Crystal violet assay and LDH release were used to assess cell viability and cytotoxicity 24 h after the seeding. Basal cells were subjected to the same experimental conditions, except for irradiation and pre-treatments.
Cell Viability
Cell viability was determined using the crystal violet assay according to [61]. The absorbance was recorded at 540 nm, and the percentage of viability was calculated compared with basal cells, for which a value of vitality equal to 100% was assumed.
LDH Assay
LDL release was measured in a culture medium from irradiated and non-irradiated cells pre-treated or not treated with TAU and NAC using the LDH Cytotoxicity Detection kit (Roche Diagnostics, Monza, Italy) and following the manufacturer's instructions. LDL release was expressed as the percentage of the LDH measured in the medium divided by the LDH release measured after cell treatment with 2% Triton X-100 (positive control, 100% LDH release).
Wound Healing Assay
Cells were grown to confluence in gelatine-coated 12-well plates in a complete medium. Confluent cells were manually scratched in each well using a p10 pipette tip, and the cells were cultured for 48 h. Photographs were taken just after scratching and after 6, 9, 24, and 48 h of incubation at a × 4 magnification. The percentage of wound closure was calculated using ImageProPlus software (Media Cybernetics, Inc, Rockville, MD, USA) by measuring the wound area at each time point compared with the initial area measured at the time of the scratch. Each sample was assayed in duplicate.
MDA Determination
MDA was determined as previously described [62] using 10 µg of cell lysates. The quantification was performed spectrophotometrically at 535 nm by measuring the absorbance produced by 100 µL of the sample. Calibration curves were built with standards of MDA at 5, 10, 25, and 50 µM. Each sample was assayed in duplicate.
Statistical Analysis
Data were reported as the mean ± standard deviation (SD). Statistical analysis was performed using Two Way Analysis of Variance followed by the Pairwise Multiple Comparison Procedures to analyze the effect of sex, X-rays, and treatments using Sigma-Stat 3.1 software (Systat Software, Erkrath, Germany). The distribution of samples was assessed via the Kolmogorov-Smirnov and Shapiro tests.
Linear regression analysis was performed by plotting time against the percentage of wound closure and comparing slope variations through a global test of coincidence using Sigma-Stat 3.1 software (Systat Software, Erkrath, Germany). A p ≤ 0.05 was considered statistically significant.
Characteristics of Donors
The mothers of female and male neonates did not differ significantly in age and body weight, and neonates of both sexes did not diverge significantly in body weight (Table 1). Table 1. Physical data of the enrolled cohorts.
Body Weight of Mothers (kg)
Body Weight of Neonates (kg) Values are reported as the mean ± SD.
Effect of X-rays on HUVECs Viability and Lactate Dehydrogenase (LDH) Release
Viability and LDH release in basal cells, a measure of cytotoxicity, did not present sexual dimorphism. The irradiation of cells with X-rays at doses of 1.6 and 3.2 Gy reduced cell viability in a statistically significant manner, but this occurred regardless of cell sex ( Figure 1A,C), and irradiation increased LDH release in a dose-dependent manner, regardless of cell sex ( Figure 1B,D). ly), and to understand whether TAU and NAC can prevent IR-induced damage in a sex-dependent way given their high safety profiles and low costs [52,53].
Characteristics of Donors
The mothers of female and male neonates did not differ significantly in age and body weight, and neonates of both sexes did not diverge significantly in body weight (Table 1). Values are reported as the mean ± SD.
Effect of X-rays on HUVECs Viability and Lactate Dehydrogenase (LDH) Release
Viability and LDH release in basal cells, a measure of cytotoxicity, did not present sexual dimorphism. The irradiation of cells with X-rays at doses of 1.6 and 3.2 Gy reduced cell viability in a statistically significant manner, but this occurred regardless of cell sex ( Figure 1A,C), and irradiation increased LDH release in a dose-dependent manner, regardless of cell sex ( Figure 1B,D).
Effect of X-rays on HUVECs Migration
Migration, expressed as the percentage of wound closure that did not diverge between the sexes, and closure were completed in 48 h in basal cells. X-rays, instead, significantly (Table 2). In detail, at 1.6 Gy, both male and female cells showed a slower and partial recovery compared with basal cells and had no significant sex difference (Table 2), which was also confirmed by linear regression analysis that showed similar slopes when 1.6 Gy irradiated MHUVEC were compared with FHUVECs (y = 1.351x + 0.784 and y = 1.454x + 2.131 for MHUVECs and FHUVECs, respectively). A 3.2 Gy X-ray produced a longer delay in wound closure than 1.6 Gy, and its effects were associated with sex, which was also confirmed by linear regression analysis which showed that the slopes significantly diverged between 3.2 Gy-irradiated MHUVECs and 3.2 Gy-irradiated FHUVECs (Figure 2), suggesting that wound repair was more rapid in FHUVECs than in MHUVECs.
Effect of X-rays on HUVECs Migration
Migration, expressed as the percentage of wound closure that did not diverge between the sexes, and closure were completed in 48 h in basal cells. X-rays, instead, significantly reduced wound closure in both MHUVECs and FHUVECs (Table 2). In detail, at 1.6 Gy, both male and female cells showed a slower and partial recovery compared with basal cells and had no significant sex difference (Table 2), which was also confirmed by linear regression analysis that showed similar slopes when 1.6 Gy irradiated MHU-VEC were compared with FHUVECs (y = 1.351x + 0.784 and y = 1.454x + 2.131 for MHUVECs and FHUVECs, respectively). A 3.2 Gy X-ray produced a longer delay in wound closure than 1.6 Gy, and its effects were associated with sex, which was also confirmed by linear regression analysis which showed that the slopes significantly diverged between 3.2 Gy-irradiated MHUVECs and 3.2 Gy-irradiated FHUVECs (Figure 2), suggesting that wound repair was more rapid in FHUVECs than in MHUVECs.
Effect of X-rays on Autophagy
Autophagy is a catabolic process that delivers cellular constituents, including damaged or superfluous organelles and long-lived proteins, to lysosomes for degradation and recycling [21]. In basal conditions, it was measured via a LC3II/I ratio and was significantly higher in FHUVECs. After 3.2 Gy irradiation, the LC3II/I ratio significantly increased (about 110%) only in FHUVECs ( Figure 3A), while it was similar to those of basal cells in irradiated MHUVECs.
Autophagy is a catabolic process that delivers cellular constituents, including damaged or superfluous organelles and long-lived proteins, to lysosomes for degradation and recycling [21]. In basal conditions, it was measured via a LC3II/I ratio and was significantly higher in FHUVECs. After 3.2 Gy irradiation, the LC3II/I ratio significantly increased (about 110%) only in FHUVECs ( Figure 3A), while it was similar to those of basal cells in irradiated MHUVECs.
Effect of X-rays on Lipid Peroxidation
Malondialdehyde (MDA) levels did not significantly diverge in basal male and female HUVECs, but they significantly increased in the irradiated cells of both sexes. In particular, the increase was significantly more pronounced in FHUVECs than in MHU-VECs ( Figure 3B).
Effect of Treatments on HUVECs Viability and LDH Release
In basal FHUVECs and MHUVECs, 10 mM TAU and 5 mM NAC did not affect viability and LDH release ( Figure 1A,B), whereas they reduced LDH release in irradiated cells ( Figure 1C,D). In detail, 10 mM TAU and 5 mM NAC had both a significant protective effect in LDH reduction in FHUVECs, and TAU was more protective in male cells than NAC ( Figure 1C,D). In fact, after irradiation, TAU reduced LDH release in MHU-VECs, while NAC had no effect.
Effect of Pre-Treatments on HUVECs Migration
TAU and NAC did not modify the migration of non-irradiated HUVECs for both sexes ( Figure 4A,B). However, TAU significantly promoted the closure of the wound in 1.6 and 3.6 Gy irradiated cells, especially in the late phase (48 h), but this occurred independently of cell sex ( Figure 4A,B).
Moreover, linear regression analysis showed that the slopes significantly diverged between 3.2 Gy irradiated and non-irradiated TAU-pretreated MHUVECs ( Figure 4C). No other statistically significant differences emerged from the linear regression analysis.
Effect of X-rays on Lipid Peroxidation
Malondialdehyde (MDA) levels did not significantly diverge in basal male and female HUVECs, but they significantly increased in the irradiated cells of both sexes. In particular, the increase was significantly more pronounced in FHUVECs than in MHUVECs ( Figure 3B).
Effect of Pre-Treatments on HUVECs Viability and LDH Release
In basal FHUVECs and MHUVECs, 10 mM TAU and 5 mM NAC did not affect viability and LDH release ( Figure 1A,B), whereas they reduced LDH release in irradiated cells ( Figure 1C,D). In detail, 10 mM TAU and 5 mM NAC had both a significant protective effect in LDH reduction in FHUVECs, and TAU was more protective in male cells than NAC ( Figure 1C,D). In fact, after irradiation, TAU reduced LDH release in MHUVECs, while NAC had no effect.
Effect of Pre-Treatments on HUVECs Migration
TAU and NAC did not modify the migration of non-irradiated HUVECs for both sexes ( Figure 4A,B). However, TAU significantly promoted the closure of the wound in 1.6 and 3.6 Gy irradiated cells, especially in the late phase (48 h), but this occurred independently of cell sex ( Figure 4A,B).
Moreover, linear regression analysis showed that the slopes significantly diverged between 3.2 Gy irradiated and non-irradiated TAU-pretreated MHUVECs ( Figure 4C). No other statistically significant differences emerged from the linear regression analysis.
NAC pre-incubation did not affect male and female HUVECs migration when compared with basal cells (Figure 5A,B). The regression analysis evidenced a significant difference in slopes between NAC-irradiated MHUVECs versus non-irradiated NAC-pretreated MHUVECs ( Figure 5C), which indicated a positive effect of NAC on cellular migration. Antioxidants 2023, 12, x FOR PEER REVIEW 6 of 15 NAC pre-incubation did not affect male and female HUVECs migration when compared with basal cells (Figure 5A,B). The regression analysis evidenced a significant difference in slopes between NAC-irradiated MHUVECs versus non-irradiated NAC-pretreated MHUVECs ( Figure 5C), which indicated a positive effect of NAC on cellular migration. Moreover, linear regression analysis showed that slopes significantly diverged between MHUVECs and FHUVECs pretreated with NAC and exposed to 1.6 Gy ( Figure 5D), which indicated that NAC is more effective in FHUVECs. No other statistically significant differences emerged from the linear regression analysis.
Moreover, linear regression analysis showed that slopes significantly diverged between MHUVECs and FHUVECs pretreated with NAC and exposed to 1.6 Gy ( Figure 5D), which indicated that NAC is more effective in FHUVECs. No other statistically significant differences emerged from the linear regression analysis.
Effect of Pre-Treatments on Autophagy
In non-irradiated male and female cells, 10 mM TAU and 5 NAC pre-incubation did not significantly affect autophagy (expressed as the LC3II/I ratio), but in irradiated FHUVECs, they attenuated the autophagy in a non-statistically significant manner as they were practically inactive in irradiated MHUVECs ( Figure 3A).
Effect of Pre-Treatments on Lipid Peroxidation
Ten mM TAU pre-incubation did not significantly affect lipid peroxidation in both nonirradiated and irradiated MHUVECs and FHUVECs compared with basal cells (Figure 3B), although a non-significant reduction in MDA was observed in FHUVECs.
However, NAC pre-incubation highlighted sex differences in the lipid peroxidation as MDA levels were statistically significantly higher in FHUVECs than in MHUVECs ( Figure 3B).
Discussion
The biological and molecular mechanisms underlying IR damage are still not fully understood [19], and it is even less known whether IR damage is influenced by sex. In this study, we show that viability, LDH release, cell migration, and lipid peroxidation do not vary between sexes in basal conditions, while autophagy is higher in female cells than in male ones. Globally, this indicates that sex differences are parameter specific, which has already been shown in other experimental models [55,63,64]. Indeed, the data regarding autophagy are not in line with previous results [55], but the discrepancy probably depends on differences in serum concentrations in the culture medium.
Sexual polymorphism is also related to cell migration and MDA levels in irradiated cells, whereas changes in cell viability and cytotoxicity are sex-independent. These findings indicate that IR amplifies sex differences in a parameter specific manner. Radiation-induced autophagy may have a different role in cell fate depending on the dose and duration of radiation leading to survival or death [65]. As a close link between oxidative stress and autophagy was described [63,66], the observed increase in lipid peroxidation could explain, at least in part, the increase in autophagy observed in irradiated cells. However, it does not explain the results obtained for the cells preincubated with NAC where both irradiated and non-irradiated lipid peroxidation are higher in female cells, but the autophagy is scarcely affected. Other sex differences after irradiation have been described in human male lymphocytes, which are less sensitive than female cells when exposed to 30 Gy X-rays [67], but other studies do not observe any significant sex differences in human hematopoietic stem cells irradiated with X-rays (0.5 and 2 Gy) [68]. This suggests that sex differences are related to the cell type, the radiation dose, and the studied parameter.
The development of non-toxic agents to combat radiation-induced endothelial dysfunction is of paramount importance because alterations in endothelial function affect the control of vascular tone, angiogenesis, hemostasis, inflammation, vascular integrity, and vessel repairing and the provision of an antioxidant, anti-inflammatory, and antithrombotic interface [69]. Some of these processes appear to be sex-dependent [70]. In non-irradiated cells, 10 mM TAU does not modify any of the studied parameters compared with basal cells, except for the LC3II/I ratio. In fact, TAU attenuates the sex difference in the autophagic response. In irradiated cells, TAU reduces cytotoxicity in male and female cells, and is ineffective regarding viability. In addition, it promotes cell migration after radiation at 24 and 48 h and decreases the autophagic process expressed as the LC3II/I ratio. It does not affect MDA levels. In particular, TAU increases the migration capacity, especially in the late phases in both sexes. TAU prevents apoptosis induced by hyperglycemia [57,71], lipopolysaccharide, and tumor necrosis factor-alpha stimulation and reduces oxidative stress [72] in HUVECs and other human endothelial cells not stratified for sex. In vitro, it declines high potassium-induced contraction in rabbit ear arteries [73]. Furthermore, TAU administered in vivo attenuates low-density lipoprotein-induced endothelial dysfunction [74]. Globally, the above data also suggest that the activity of TAU is target-specific, and its effect may be due to a combination of different mechanisms as proposed by Christophersen [36], although the author does not focus on the sex effect. The small beneficial effects observed regarding TAU could be of great relevance in cancer irradiated patients who appear to be TAU-depleted after cytotoxic chemotherapy and/or radiotherapy [37].
Pre-treatment with the glutathione precursor NAC [29,40] reduces cytotoxicity in male and female cells and promotes wound closure at 24 h and 48 h after radiation, particularly in FHUVECs. Moreover, NAC cancels sex differences in autophagy in irradiated and non-irradiated cells. Finally, in both irradiated and non-irradiated cells, NAC exposure brings out a sex difference in MDA levels, which are higher in females. Millimolar NAC leads to a higher rate of wound closure than the controls 36 h after wounding in human skin fibroblast cell lines not stratified for donor sex [75]. The sex differences observed with NAC are not surprising because glutathione metabolism shows significant sex differences [10]; for example, intracellular glutathione synthesis requires glutamate-cysteine ligase, which is less expressed in the female liver than in the male liver, at least in rats [62].
Globally, our data show some small protective and sex-specific effects of TAU and NAC. In particular, both promote a decrease in X-ray-mediated cytotoxicity. TAU is more effective in promoting wound closure in MHUVECs, while NAC is more effective in FHUVECs. Moreover, TAU does not affect autophagy, while NAC attenuated the differences between the sexes observed in the autophagic response.
Finally, a sex-specific effect of NAC on MDA levels can be noted as it increases levels for females. On the contrary, TAU does not modify this parameter. Overall, these data suggest that the two antioxidants may mediate sex-specific protective effects through different mechanisms, although the effect of NAC seems to be more influenced by sex, and this aspect could be in line with sex differences described in glutathione metabolism and glutathione cycle [10].
In conclusion, TAU and NAC have similar safety and tolerability in non-irradiated MHUVECs, while NAC is less safe than TAU in non-irradiated FHUVECs because it increases lipid peroxidation. Cell irradiation increases autophagy only in FHUVECs where it produces a more marked elevation in MDA and a more rapid wound closure than in MHU-VECs. In irradiated cells, NAC preincubation has a positive effect on cellular migration and LDH release, which is more effective in FHUVECs. However, TAU significantly promoted the closure of the wound and a decrease in LDH release independently of cell sex in the same experimental conditions. Thus, taurine appears to be more protective than NAC in male cells.
A further understanding of radiation-induced endothelial dysfunction could lead to progress in the development of countermeasures, such as antioxidant or mitigator therapies, for cardiovascular diseases in subjects exposed to radiation.
Finally, our results confirm and stress the importance of reporting cell sex in experiments and including the sex-gender variable in preclinical and clinical research [2] to understand sex-specific mechanisms and create personalized diagnostic and therapeutic approaches. Moreover, these results allow us to lay the groundwork for a sex-specific use of antioxidants. Data Availability Statement: Data will be made available on request.
Acknowledgments: This article is dedicated to the memory of our beloved and esteemed full professor Annalisa Romani who recently passed away.
==
Domain: Biology Medicine
|
[{"added": "2023-01-01T16:16:21.030Z", "created": "2022-12-29T00:00:00.000", "id": "255333454", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2076-3921/12/1/77/pdf?version=1672367282", "pdf_hash": "88478fad6128f78ce630a30d9dd2f6fc74da23aa", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:3417", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "ec3b551ed390cbeea3086b881f0f68f5b090befa", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Identification of a novel intermittent hypoxia-related prognostic lncRNA signature and the ceRNA of lncRNA GSEC/miR-873-3p/EGLN3 regulatory axis in lung adenocarcinoma
Background Lung adenocarcinoma (LUAD) is still the most prevalent type of respiratory cancer. Intermittent hypoxia can increase the mortality and morbidity associated with lung cancer. Long non-coding RNAs (lncRNAs) are crucial in lung adenocarcinoma. However, the effects of intermittent hypoxia-related long non-coding RNAs (IHRLs) on lung adenocarcinoma are still unknown. Method In the current research, eight IHRLs were selected to create a prognostic model. The risk score of the prognostic model was evaluated using multivariate and univariate analyses, and its accuracy and reliability were validated using a nomogram and ROC. Additionally, we investigated the relationships between IHRLs and the immune microenvironment. Result Our analysis identified GSEC, AC099850.3, and AL391001.1 as risk lncRNAs, while AC010615.2, AC010654.1, AL513550.1, LINC00996, and LINC01150 were categorized as protective lncRNAs. We observed variances in the expression of seven immune cells and 15 immune-correlated pathways between the two risk groups. Furthermore, our results confirmed the ceRNA network associated with the intermittent hypoxia-related lncRNA GSEC/miR-873-3p/EGLN3 regulatory pathway. GSEC showed pronounced expression in lung adenocarcinoma tissues and specific cell lines, and its inhibition resulted in reduced proliferation and migration in A549 and PC9 cells. Intriguingly, GSEC manifested oncogenic properties by sponging miR-873-3p and demonstrated a tendency to modulate EGLN3 expression favorably. Conclusion GSEC acts as an oncogenic lncRNA by interacting with miR-873-3p, modulating EGLN3 expression. This observation underscores the potential of GSEC as a diagnostic and therapeutic target for LUAD.
INTRODUCTION
Among the most prevalent malignancies, lung adenocarcinoma (LUAD) has a higher incidence in the population (Sung et al., 2021). Due to the difficulty of early identification and therapy, LUAD remains one of the tumors with the poorest prognosis (Thai et al., 2021). Despite the discovery of various biomarkers and diagnostic equipment, there are still numerous gaps in the accurate diagnosis and therapy of LUAD Yu et al., 2021). Intermittent hypoxia (IH) is a condition marked by intermittent oxygen deprivation, which may result in oxidative stress and inflammation (Shobatake et al., 2022). The primary trigger of intermittent hypoxia is obstructive sleep apnea (OSAS), which causes a combination of apnea and hypopnea due to a decrease in the size of the pharyngeal cavity. At the same time, the patient is asleep (Labarca et al., 2020). There has been less investigation into the association between intermittent hypoxia and lung cancer in previous studies. The severity of OSAS is associated with a relationship between OSAS and lung cancer, a molecular pathway involved in hypoxia-induced lung cancer progression (Li et al., 2017). Several studies show an association between intermittent hypoxia and cancer, which promotes carcinogenesis and progression through oxidative stress and inflammatory translation (Navarrete-Opazo & Mitchell, 2014;Mateika et al., 2015;Almendros & Gozal, 2018). Intermittent hypoxia in lung adenocarcinoma has been associated with a more aggressive tumor phenotype, increased metastatic potential, and resistance to therapy. It promotes the survival and proliferation of cancer cells, stimulates angiogenesis, and enhances the ability of tumor cells to invade and metastasize to distant sites. Furthermore, intermittent hypoxia has been shown to contribute to therapy resistance and poor patient outcomes.
Long non-coding RNAs (lncRNAs) are a class of RNA molecules that do not code for proteins but play critical regulatory roles in gene expression. They have been increasingly recognized as key players in various biological processes, including cancer development and progression (Peng, Koirala & Mo, 2017). In lung adenocarcinoma, aberrant expression and dysregulation of lncRNAs have been observed. These lncRNAs can act as oncogenes or tumor suppressors, influencing tumor growth, invasion, metastasis, and drug resistance. They can interact with other molecules, such as DNA, RNA, and proteins, to modulate gene expression and cellular processes involved in cancer development. Recent research has shown that lncRNAs regulated the early development of LUAD in multiple ways, including through different signalling pathways (Song et al., 2021). Downregulated DGCR5 expression was powerfully associated with smaller tumor size, indicating LUAD patients have a higher survival rate (Dong et al., 2018). Furthermore, the risk model of 12 ferroptosisrelated lncRNAs has significant prognostic value for LUAD and may be ferroptosis-related therapeutic targets in the clinic (Lu et al., 2021). The Lactate Metabolism-Related lncRNA signature has also been related to immune cell infiltration and the immune checkpoint blocker CTLA-4 (Mai et al., 2022). Patients diagnosed with LUAD can have their prognosis accurately predicted using lncRNAs associated with cuproptosis (Wang et al., 2022). The above results indicate that lncRNAs may also play a role in the immune microenvironment of LUAD.
In recent years, there has been growing interest in understanding the interplay between lncRNAs and intermittent hypoxia in lung adenocarcinoma. Emerging evidence suggests that specific lncRNAs are responsive to hypoxia and are differentially expressed under low oxygen conditions. These hypoxia-responsive lncRNAs can modulate gene expression patterns associated with tumor growth, metastasis, and therapy resistance. They can act as mediators or regulators of hypoxia-induced signalling pathways, influencing the cellular response to intermittent hypoxia. Therefore, we investigate the association between intermittent hypoxia-related long non-coding RNAs (IHRLs) and LUAD.
In this work, we present a series of experiments to investigate the expression levels of IHRLs and their association with the immune microenvironment in LUAD. The potential regulatory mechanisms, including target genes and miRNAs, were further investigated. The results of this study will be helpful in the investigation of potential prognostic biomarkers associated with LUAD.
Data collection and processing
The Cancer Genome Atlas (TCGA) data ( [URL]/) were used to obtain the transcriptional data and clinical characteristics of LUAD patients; patients with incomplete survival data were excluded. The clinical data was constructed utilizing Perl programming language ( [URL]/5.32.1). The downloaded data is a single FPKM file for each sample. We use Perl language to merge all samples into a matrix for subsequent analysis. All the data applied in this research were freely accessible in a public database. There was no need for approval from an ethics commission.
Identification of intermittent hypoxia-related genes (IHRGs)
Thirty-nine IHRGs were obtained from the GeneCards ( [URL]/) database according to the screening criterion: Relevance score >10. Thirty-two mRNAs were finally selected as differentially expressed genes (DEGs) by R software (version 4.3;R Core Team, 2023). The 32 DEGs were then sorted into a PPI network by STRING ( [URL]/), and the network was visualized using Cytoscape 3.4.0 software.
Functional enrichment analysis of intermittent hypoxia-related DEGs
The enrichment analyses were performed by Kyoto Encyclopedia of Genes and Genomes (KEGG) ( [URL]/) and gene ontology (GO) ( [URL]. org/). The biological characteristics of intermittent hypoxia-related DEGs were identified using these two databases.
Construction of intermittent hypoxia-related lncRNAs prognosis model
Bioinformatics was used to identify lncRNAs DEGs associated with intermittent hypoxia. Patients (n = 504) were randomly allocated to training and validation groups in a 1:1 ratio. IHRLs were screened using univariate Cox regression analysis, the LASSO Cox algorithm, and multivariate analysis. Then, a prognostic model was established using the regression coefficient (β), the risk score =Expression 1 * β 1 +Expression 2 * β 2 +. . . +Expression n * β n ).
Then, we assessed the risk scores and clinical characteristics of LUAD using univariate and multivariate Cox regressions and ROC curves assessment.
LncRNA risk score model correlation with tumor microenvironment infiltration
Based on gene expression profiles, CIBERSORT was used to calculate immune cell composition. The association between the risk model and LUAD immune infiltration was investigated utilizing R software. Furthermore, we apply the cor test function to calculate the relationship between risk score and immune cells.
Construction of competing endogenous RNA axis
The Mircode database ( [URL]/) investigated the relationship between miRNAs and lncRNAs. The TargetScan database ( [URL]/) was used to identify the downstream target mRNAs of miRNAs. The GEPIA database ( [URL]/) was used to analyze the expression of the downstream target mRNAs.
Cell culture and cell transfection
LUAD cell lines A549 and PC9 were provided by Procell Life Science & Technology Co,. Ltd (Wuhan, China). The LUAD cells were cultured in DMEM medium (Gibco, Billings, MT, USA) with 10% FBS and maintained at 37 degrees Celsius in 5% CO 2 in an incubator. Si-GSEC, miR-873-3p inhibitor, and miR-873-3p mimics were synthesized by GenePharma (Shanghai, China). The LUAD Cells were plated in 96-well plates and transfected with plasmid by Lipo3000 (Invitrogen, Carlsbad, CA, USA). These cells were incubated at 37 degrees Celsius for 48 h and harvested for the subsequent experiments.
Cell proliferation assay
For cell viability experiments, HBE, A549, PC9, and NCI-H1299 cells were seeded in 96-well plates and incubated for 24 h at 37 degrees Celsius and 5% CO2 in the incubator. After treating with the CCK-8 solution for 2 h, we measured the absorbance at 450 nm using a PerkinElmer EnSight (Waltham, MA, USA)
Wound healing assay
The migration of cells was quantified using a wound-healing assay. After incubating the cells in 6-well dishes at 37 degrees Celsius, a line was drawn with a 200ul pipette tip. Images were captured at 0 and 24 h post-injury, and ten distinct sites were randomly selected for annotation and quantification.
Colony formation assays
LUAD cells (A549 and PC9) were seeded into 6-well dishes, transfected the following day, and incubated for approximately two weeks. After two weeks, the cells were extracted, washed three times with PBS, and dyed with 4% paraformaldehyde and crystal violet solutions, respectively. Finally, the data is stored by photographing it with the camera for subsequent research.
Dual-luciferase assay
The sequences of GSEC (GSEC wt and GSEC mut) and EGLN3 (EGLN3 wt and EGLN3 mut) containing the homologous binding sites of miR-873-3p were amplified and uniformly inserted into the vector pGL3 (Promega, Madison, WI, USA). The miR-873-3p mimics were then co-transfected with GSEC wt, GSEC mut, EGLN3 wt, or EGLN3 mut using Lipo3000. The DualLuciferase Reporter Assay Kit was used to measure luciferase activity.
RNA immunoprecipitation assay
The RNA immunoprecipitation (RIP) experiment was conducted with the EZ-Magna RIP Reagent (Millipore, USA). The LUAD cells were lysed with RIP buffer and incubated with Anti-Ago2 or negative control (Anti-IgG). RT-qPCR identified the extracted RNAs after treatment with immunoprecipitated RNA and Proteinase K (Roche).
Western blot
RIPA assay lysis buffer (RIPA: PMSF =100:1) was used to extract the protein of LUAD cells. After measuring all densities, the prepared proteins (30 µg/lane) were separated by SDS-PAGE and transferred into pre-cut PVDF membranes. After blocking in skimmed experimental milk (5%) for two h at 37 degrees Celsius, the targeted membranes were incubated with preconfigured primary anti-EGLN3 (ab30782; 1:1000; Abcam; Cambridge, UK) and β-actin (ab207327; 1:2000; Abcam; Cambridge, UK) overnight at 4 • C. The following day, the incubation with the associated secondary antibody was applied for 2 h before getting washed with PBS solution. Immunoreactive bands were displayed with the ECL Chemiluminescence Kit (G2020; Service) and visualization of the image capture system (ChemiDoc MP; Bio-Rad, Hercules, CA, USA).
Construction of predictive risk model in LUAD patients
The risk score was calculated according to our formula as follows: Risk Score = (0.2276 *
Prognosis value of model lncRNAs in LUAD
To determine how the risk score of the prognostic model affects survival, we performed univariate and multivariate Cox proportional hazards analyses. Our risk score of this prognostic model predicted LUAD OS independently in these investigations (Figs. 6A-6B). Risk scores were more specific and sensitive than other clinical features (AUC =0.782) (Fig. 6C). In summary, these results highlight the immunomodulatory effects of the risk model.
Expression of lncRNA GSEC, miR-873-3p, EGLN3 in LUAD
RT-qPCR analysis was used to detect the expression of lncRNA GSEC, miR-873-3p, and EGLN3 in LUAD. Expression of lncRNA GSEC and EGLN3 in LUAD cell lines was higher than in normal cell lines (Figs. 9A, 9C). Expression of miR-873-3p in LUAD cell lines was lower than in normal cell lines (Fig. 9B). Then, we found that the expression of EGLN3 in LUAD cell lines was higher than in normal cell lines at the protein level (Fig. 9D). Figure 9E indicated that the transfection was successful. Meanwhile, lncRNA GSEC
DISCUSSION
Our research systematically identified IHRLs based on intermittent hypoxia-correlated mRNAs in LUAD. Current research shows intermittent hypoxia is of great significance in the occurrence and development of tumors (Vilaseca et al., 2017;Chen et al., 2018;Li et al., 2021). Scientists are also interested in this field of research. In various cancers, lncRNAs have been shown to play a crucial role in intermittent hypoxia, which has been confirmed by multiple research (Chen et al., 2022;Hao et al., 2022). Therefore, we focused on the interaction of IHRLs with LUAD tumors. In this research, we screened 32 DEGs for variations in mRNA expression levels using the LUAD data of the TCGA database, which was then analyzed using bioinformatics methods. According to GO enrichment analysis, the DEGs were mainly associated with hypoxia, chemical stress, neuron death, carboxylic acid binding, oxidoreductase activity, and cytokine receptor binding. The investigations show intermittent hypoxia causes oxidative stress due to mitochondrial response inside mouse neuron death (Douglas et al., 2010). Pro-inflammatory cytokines play a role at several levels of hypoxic chemical reflex and cardiovascular control pathways, which may contribute to CIH-induced cardiorespiratory alterations (Del Rio et al., 2012). The above studies imply that intermittent hypoxia can result in multisystem illness. We utilized KEGG pathway analysis to identify that the DEGs were involved in the HIF-1 signalling pathway, AGE-RAGE signalling pathway, IL-17 signalling pathway, and ferroptosis. HIF-1A is a crucial pathway in intermittent hypoxia. Recent studies suggest intermittent hypoxia may interfere with mtROS in lung cancer cells to play a carcinogenic role through the HIF-1 α/ATAD2 pathway (Hao et al., 2022). Ferroptosis is a recently discovered type of cell death, and the ferroptosis inhibitor Fer-1 has been shown to reduce intermittent hypoxia-induced lung injury in rats (Shah, Shchepinov & Pratt, 2018;Chen et al., 2022). Numerous studies have demonstrated that molecular pathways and processes are intimately associated with the formation and progression of cancers (Alinejad et al., 2017; Waghela et al., 2021;Lei, Zhuang & Gan, 2022;Lin et al., 2022). In addition, eight key IHRLs were identified after univariate Cox regression and Lasso-Cox regression analysis, and a prognostic model based on them was established. The combination of multivariate COX regression and univariate COX assists in identifying the LncRNA signature in lung cancer (Mo et al., 2022). Finally, the intermittent hypoxia-related prognostic lncRNA signature could effectively predict the survival risk of LUAD patients. Numerous research has demonstrated that lncRNAs can function as competing endogenous RNAs and sponge microRNA (miRNA) sites (ceRNAs) (Wang et al., 2019;Conte et al., 2021;Yang et al., 2021b). Furthermore, investigations have revealed that lncRNAs associated with ceRNAs play a critical role in the occurrence and progression of malignancies Ye, Li & Zhao, 2021;Zhao, Feng & Tang, 2021). lncRNA SNHG16 enhances lung cancer cell proliferation, migration, and invasion by modulating the miR-520/VEGF axis (Chen et al., 2020). LncRNA SPINT1-AS1 stimulates the growth of breast cancer cells through sponge let-7a/b/i-5p (Zhou et al., 2021). Through stabilizing GLUT1, the LncRNA GAL enhances colorectal cancer liver metastasis (Li et al., 2022). These results indicated that ceRNAs had been extensively studied and reported in various cancers (Yang et al., 2021a). According to prior investigations, we hypothesized a new axis of lncRNA GSEC/miR-873-3p/EGLN3 using bioinformatics approaches. Meanwhile, the axis was confirmed by experimentation. GSEC performs a vital function in tumor formation and progression. LncRNA GSEC has an essential role in the occurrence and progression of numerous forms of cancer. LncRNA GSEC promotes osteosarcoma's proliferation, migration, and invasion by targeting the miR-588/ EIF5A2 axis (Liu et al., 2020). LncRNA GSEC promotes breast cancer progression by sponging the miR-202-5p/AXL Axis (Zhang et al., 2021). LncRNA GSEC encourages the progression of hepatocellular carcinoma by targeting the miR-101-3p/SNX16/PAPOLG axis (Hu et al., 2022). We observed that lncRNAs GSEC and EGLN3 were over-expressed in LUAD, and miR-873-3p was downregulated, similar to previous results on lncRNA GSEC. Furthermore, Cheng et al. (2019) revealed that miR-873-3p expression was suppressed and acted as a tumor suppressor in LUAD. The above results confirmed our conclusions about the low word of miR-873-3p in LUAD. In our investigation, the knockdown of miR-873-3p attenuated the effect of lncRNA GSEC interference on A549 and PC9 cells. Our data revealed that miR-873-3p worked through the ceRNA regulatory mechanism. Utilizing the TargetScan database, the intermittent hypoxia-related mRNA (EGLN3) was identified as the downstream target of miR-873-3p. EGLN3 is associated with various tumor conditions, and USP9X relieves cholangiocarcinoma by upregulating EGLN3 (Chen et al., 2021;Jin et al., 2022). Our research using RT-qPCR and western blotting methods revealed that EGLN3 was significantly more upregulated in A549 and PC9 cell lines than in the normal group. MiR-873-3p was a mediator in the regulation process between lncRNA GSEC and EGLN3. Consequently, we concluded that lncRNA GSEC acts as an oncogene in lung adenocarcinoma by targeting miR-873-3p to modulate EGLN3, which might be The relationship between EGLN3 and miR-873-3p was performed by Dual-Luciferase reporter assay. * * p < 0.01; vs. ctrl mimics. (D-F) Expression of EGLN3 following miR-873-3p overexpression. * * p < 0.01; * * * p < 0.001; vs. ctrl mimics. (G-I) Expression of EGLN3 following si-GSEC or si-GSEC+miR-873-3p inhibitor. * * * p < 0.001; vs. ctrl inhibitor; # p < 0.05; ## p < 0.01 vs. si-GSEC+ctrl inhibitor.
Full-size DOI: 10.7717/peerj.16242/ fig-12 the mechanism of LUAD progression. This study strongly supports increasing evidence advocating for measuring tumor hypoxia before treatment to select patients most likely to benefit from hypoxia modification in lung cancer. Our research suggests that this polygenic test may help provide treatment options for patients with hypoxic lung cancer. Unfortunately, there are still some limitations to our study. Our raw data are from publicly available databases. Using publicly available databases can be beneficial and limiting in research studies. While these databases provide a vast amount of data that can be accessed and analyzed, some potential limitations and biases could affect the findings. One limitation is the quality and accuracy of the data within these databases. Another potential limitation is the representativeness of the data in publicly available databases.
Furthermore, the data in publicly available databases may be subject to publication bias. It is also important to note that publicly available databases may not always be up-to-date or comprehensive. The application of mutual validation demonstrates the viability of this risk model, and more data may be needed for further verification. Meanwhile, we preliminarily validated the ceRNA of intermittent hypoxia-related lncRNA GSEC/miR-873-3p/EGLN3 regulatory axis in vitro. More experiments need to be performed in vivo.
CONCLUSIONS
Identifying a prognostic model related to intermittent hypoxia and specific lncRNAs provides a foundation for further investigation. The predictive risk model developed in this study can potentially enhance patient care and treatment decision-making. Furthermore, using this predictive risk model can have implications for healthcare policy-making. The findings of this study have a significant impact on future research, clinical practice, and policy-making in the field of LUAD. Further research can build upon these findings to deepen our understanding of LUAD progression and identify potential therapeutic targets. The predictive risk model can potentially improve clinical decision-making and patient outcomes while also informing healthcare policies to optimize resource allocation and patient care. Overall, this study has the potential to significantly impact the field of LUAD and contribute to advancements in patient care and treatment strategies.
==
Domain: Biology Medicine
|
[{"added": "2023-10-14T15:57:30.533Z", "created": "2023-10-10T00:00:00.000", "id": "264061609", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.7717/peerj.16242", "pdf_hash": "f1fabe7523d25d84b52235a62e19c4987d666d12", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:8850", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "76c7520fbb3692969c5dd9e301e15d537931b17d", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Wild ungulates as sentinels of flaviviruses and tick-borne zoonotic pathogen circulation: an Italian perspective
Background Vector-borne zoonotic diseases are a concerning issue in Europe. Lyme disease and tick-borne encephalitis virus (TBEV) have been reported in several countries with a large impact on public health; other emerging pathogens, such as Rickettsiales, and mosquito-borne flaviviruses have been increasingly reported. All these pathogens are linked to wild ungulates playing roles as tick feeders, spreaders, and sentinels for pathogen circulation. This study evaluated the prevalence of TBEV, Borrelia burgdorferi sensu lato, Rickettsia spp., Ehrlichia spp., and Coxiella spp. by biomolecular screening of blood samples and ticks collected from wild ungulates. Ungulates were also screened by ELISA and virus neutralization tests for flaviviral antibody detection. Results A total of 274 blood samples were collected from several wild ungulate species, as well as 406 Ixodes ricinus, which were feeding on them. Blood samples tested positive for B. burgdorferi s.l. (1.1%; 0-2.3%) and Rickettsia spp. (1.1%; 0-2.3%) and showed an overall flaviviral seroprevalence of 30.6% (22.1–39.2%): 26.1% (17.9–34.3%) for TBEV, 3.6% (0.1–7.1%) for Usutu virus and 0.9% (0-2.7%) for West Nile virus. Ticks were pooled when possible and yielded 331 tick samples that tested positive for B. burgdorferi s.l. (8.8%; 5.8–11.8%), Rickettsia spp. (26.6%; 21.8–31.2%) and Neoehrlichia mikurensis (1.2%; 0-2.4%). TBEV and Coxiella spp. were not detected in either blood or tick samples. Conclusions This research highlighted a high prevalence of several tick-borne zoonotic pathogens and high seroprevalence for flaviviruses in both hilly and alpine areas. For the first time, an alpine chamois tested positive for anti-TBEV antibodies. Ungulate species are of particular interest due to their sentinel role in flavivirus circulation and their indirect role in tick-borne diseases and maintenance as Ixodes feeders and spreaders. Supplementary Information The online version contains supplementary material available at 10.1186/s12917-023-03717-x.
Background
Vector-borne diseases are described as emerging infections due to their massive spread in recent decades. Tickborne diseases (TBDs) and mosquito-borne diseases represent the most threatening vector-borne infections worldwide [1]. The rise of vector populations, together with the pathogens they may carry, is a deeply studied topic, but there are many aspects yet to be revealed [2][3][4][5].
The diffusion of TBDs is driven by several factors, mainly consequent to human activities. Modern agriculture has led to changes in land use leading to an increase in mountain abandonment, usually followed by natural reforestation, which is a frequently reported phenomenon in the Eastern Alps [6,7]. In turn, land abandonment stimulates an increase in wild animals, which can boost tick presence [2].
Wild ungulates are abundant in northern Italy [8,9] and are known as preferential hosts for tick feeding and reproduction, especially for Ixodes ricinus (I.ricinus) ticks [9,10]. In Europe, as well as in northern Italy, I. ricinus is one of the most studied tick species due to its high competence in the transmission of zoonotic pathogens, including viruses, bacteria, and parasites, to humans [1,[11][12][13].
The epidemiological cycles of these infections involve wild ungulates that mainly act as tick feeders and spreaders in the environment, as well as reservoir hosts of A. phagocytophilum [21]. In contrast, I. ricinus acts as both a vector and reservoir for Rickettsia spp. [22], and with regard to the other infectious agents, different vertebrate species serve as reservoir hosts [21,23,24].
Coxiella burnetii (C.burnetii) has been identified in I. ricinus, although its vector capacity is still debated [25,26]. It may cause disease in humans and livestock [27], and similar to domestic ruminants, red deer and wild boar are susceptible to C. burnetii and have been suggested to play a role in its maintenance [28].
Flaviviruses are currently the most widespread vectorborne zoonotic viruses in Europe and are transmitted by hard ticks (TBEV) or mosquitos (Usutu virus-USUV and West Nile virus-WNV). Both USUV and WNV are considered endemic in Italy first being detected in black birds (1996) and horses (1998), respectively [29,30]. The epidemiological cycle of these flaviviral species includes Culex mosquitos and wild birds. Mammals, including humans, can be infected but are considered dead-end hosts [31][32][33].
Regarding TBEV, Italy is considered a low-risk country, and most cases of human infection are found in the northeastern regions [20]. The main vector is I. ricinus [34], but the detection of TBEV in ticks is challenging, time-consuming and expensive since the prevalence is restricted to small foci of viral circulation [24]. In areas where TBEV incidence is high in humans, prevalence in ticks rarely exceeds 1% [35,36]. Recently, wild ungulates have been considered TBEV sentinels, noting that they seroconvert but usually do not develop disease, even though a fatal case has been recently described in a roe deer [34,[37][38][39]. In European countries, TBEV antibodies were found in several wild ungulate species [34,[40][41][42], with seroprevalence ranging between 0.74% in Finland and 63.5% in Poland. Despite the comparable clinical and epidemiological scenarios, fewer studies have investigated the seroprevalence in ungulates for WNV and USUV; seroconversion in asymptomatic animals was reported, suggesting the role of ungulates as sentinels in viral diffusion [36,43,44].
The involvement of wild ungulates in the epidemiology of all the abovementioned vector-borne pathogens is noticeable, and although several of these etiological agents are considered endemic to the alpine area of northern Italy, data derived from ungulate populations are currently scarce. This research aimed to (i) assess the prevalence of a zoonotic virus (TBEV) and zoonotic bacteria (B.burgdorferi s.l., Rickettsia spp., Ehrlichia spp., C. burnetii) in wild ungulate blood and associated tick samples using biomolecular tools; (ii) identify the selected species and genospecies through sequencing; (iii) investigate the sentinel role of ungulates in flaviviral (TBEV, WNV, USUV) circulation by antibody detection; and (iv) evaluate any the statistical significance of the obtained prevalence and seroprevalence related to several variables of interest.
The zoonotic bacterium A. phagocytophilum was purposely excluded from this study since it was previously investigated in the same ungulate and vector population [21].
Three out of 274 (1.1%; 0-2.3%) blood samples collected from two red deer (from Lombardy) and a wild boar (from Veneto) were positive for B. burgdorferi s.l.. Due to the low quantity of bacterial DNA (Cq > 35), none of the samples were confirmed by PCR and sequencing. Three out of 274 (1.1%; 0-2.3%) blood samples collected from a roe deer (from Lombardy), a red deer and a wild boar (from Friuli Venezia Giulia) tested positive for Rickettsia spp. Sequencing analysis confirmed the detection of an unidentified Rickettsia spp.from roe deer blood, with only 96% identity with known rickettsiae. Furthermore, R. helvetica was identified from red deer, while positivity of the wild boar blood was not confirmed by PCR. All blood samples tested negative for Ehrlichia spp., Coxiella spp., and TBEV.
Tick samples
A total of 406 ticks were collected and extracted in pools of two ticks (n = 75) or individually (n = 256). Thus, the total number of screened tick samples was 331 (Table 2). All ticks were identified as I. ricinus and most of the tick samples consisted of adult females (n = 231) or males (n = 90), while few immature stages were found, i.e., nymphs (n = 6) and larvae (n = 4). Most of the ticks were collected from roe deer, red deer, and mouflons. All tick samples tested positive for the DNA/RNA internal control, showing no PCR inhibition.
Twenty The prevalence of B. burgdorferi s.l. and Rickettsia spp. in tick samples did not show any significant difference when considering the ungulate species from which the ticks were collected. All tick samples tested negative for Coxiella spp.and TBEV. More details and data on the host species from which positive ticks were collected are shown in Table 2.
Ten out of 331 samples (3%; 1.2-4.9%)were positive for more than one pathogen. The presence of concurrent bacterial species was confirmed by Sanger sequencing in 6 of 10 specimens: coinfection between B. afzelii and R. helvetica was found in 2 pooled samples, one pool of male ticks and one of females, and in 3 individually extracted ticks, one female and two males. One of the males also tested positive for a third pathogen, N. mikurensis. A different male tick tested positive for both B. afzelii and R. monacensis.
In total, 34 out of 111 blood samples tested positive by ELISA showing an overall anti-flavivirus seroprevalence of 30.6% (22.1-39.2%); in detail, 24 samples were positive, and 10 were borderline. All sampled species were positive, except for mouflon. The detailed seroprevalence of the tested species is shown in Table 3.
The virus neutralization test (VNT) confirmed that most of the ELISA-positive sera were positive for TBEV (29/34), resulting in an overall prevalence of 26.1% (17.9-34.3%). Four ungulate species had TBEV antibodies: chamois, wild boar, roe deer and red deer (Table 3). Positive animals were mainly from the alpine and prealpine areas of both investigated regions. Of note, some were shot in a hilly area (altitude is approximately 170-180 m above see level) close to the flatlands. In addition, a roe deer tested positive for both TBEV and WNV.
Of the five TBEV-negative samples, one was negative for all investigated flaviviruses, while the remaining four were USUV positive. These were wild boars culled in a hilly area of the Friuli Venezia Giulia region; three were culled between June and October of the same year in different municipalities but in a range of 10-15 km 2 . Flavivirus antibody positivity did not show any statistical association with "species", "sex", "age", and "season" variables. Detailed results of ELISA and VNT tests are available in Additional file 3.
Discussion
The present study evaluated the presence and frequency of several vector-borne pathogens using molecular biology and serology. Except for C. burnetii, all investigated pathogens were identified through direct or indirect testing, highlighting the wide diffusion of both tick-borne and mosquito-borne infections in northern Italy.
Overall, the medical and veterinary importance of I. ricinus was confirmed by its high vector occurrence and remarkably frequent infection with zoonotic pathogens. According to the present results, wild ruminants (except for chamois) appear to be highly infested compared to wild boars, but they seem to play a minor role as reservoirs for tick-borne infectious agents. On the other hand, since most of the feeding I. ricinus were adults, the role of ungulates as tick amplifiers was confirmed. In addition, ungulates appear to be promising sentinels of flavivirus infections.
B. burgdorferi s.l. was detected in both blood (1.1%) and tick (8.8%) samples, although the low amount of Borrelia DNA (high Cq) in blood samples hampered further characterization. The low levels of bacteria are in accordance with previous investigations that outlined how wild ungulates can act as Borrelia dilution hosts, and thus, blood positivity is only sporadically identified without any apparent epidemiological relevance [45]. However, the role of artiodactyls is controversial: on the one hand, they represent an important feeding source for ticks; on the other hand, they have shown a limited contribution to the transmission of the Lyme spirochete, decreasing the spread of infection [45,46]. Moreover, I. ricinus tick samples showed a noticeable prevalence of Borrelia spp.and several zoonotic species were identified. The detection of four genospecies (B.garinii, B. afzelii, B. burgdorferi s.s., and B. valaisiana) indirectly highlights the presence of suitable reservoirs in the study area, such as wild birds and small mammals, and confirms the endemicity of these pathogens in the study area. A previous study [23] conducted in northeastern Italy on questing ticks reported the same zoonotic species, although at a lower prevalence, and similar identifications were found in northwestern Italy [47,48]. The prevalence reported herein is also higher than that described in Spain (2.3%) and Poland (3.3%) [49,50], likely due to the diversity of the ecological niches in different areas.
Evidence of Ehrlichia, particularly N. mikurensis, was limited to the vectors, with a prevalence of 1.2%, in agreement with a previous study conducted in a nearby area on questing ticks [23]. Despite the recent discovery of N. mikurensis, it seems that rodents could be reservoirs, while wild ungulates may act as tick spreaders rather than amplifiers [51].
In contrast, Rickettsia spp. was detected in both blood samples and feeding ticks. The prevalence in blood was low (1.1%), and the detection of R. helvetica (in red deer) and an unidentified Rickettsia (in roe deer) is likely to be an occasional finding because wild ruminants are not considered reservoirs of infection [52]. In contrast, the prevalence in tick samples was high (26.6%). Most Rickettsia spp.were R. monacensis and R. helvetica, which are typically found in inland and continental areas [15]. A lower (Slovakia -6.8%) and similar (Poland -26.8%) prevalence in I. ricinus collected from wild ungulates was reported by other authors, reflecting the heterogeneity between habitats [12,50]. Although these rickettsial species are considered pathogenic only in immunocompromised patients, a recent case described the onset of disease in an immunocompetent patient in Portugal [53].
Two ticks were positive for Ca. R. mendelii, which was first identified in 2016 in I. ricinus in Eastern Europe [54] and has been reported in I. ricinus questing ticks in Poland and the Czech Republic and in feeding ticks on migratory birds in Italy. The zoonotic potential of Ca. R. mendelii is still unknown [15,55,56].
C. burnetii was not found in blood and tick samples, whereas a prevalence of approximately 5% had been reported both in questing and wildlife-collected I. ricinus [25,26]; other in vivo experimental studies demonstrated the shedding of C. burnetii in I. ricinus faeces, suggesting its potential role as a reservoir in the wild [27]. In the studied area, I. ricinus and wild ungulates do not appear to have a pivotal role in C. burnetii maintenance in the wild. To date, their role in the sylvatic cycle has not yet been clarified: however, several cow and goat farms were found to be positive in the same regions [57,58].
Similarly, all blood and tick samples were negative for TBEV based on direct identification, in line with the role of wild ungulates as tick hosts instead of reservoirs [34]. Moreover, direct investigation in ticks is a sensitive tool for TBEV identification only when testing high amounts of specimens since the prevalence is usually low. Similar results have been reported by other authors in the same area [23], where the prevalence in questing ticks was 0.21% after testing more than 2300 samples [59].
Therefore, other more cost-effective methods should be considered to assess the risk of new foci. The present study shows the importance of serological surveys, which yielded a high flavivirus seroprevalence (i.e., 30.6%). Most of the samples (26.1%) were positive for antibodies against TBEV (Table 3) and were collected from chamois, red deer, roe deer and wild boar, revealing a high seroprevalence of TBEV in the investigated area. As reported by other authors, wild ungulates can be considered suitable sentinels for TBEV circulation [41,60]. TBEV antibodies were found -at a lower percentage -in many artiodactyls all over Europe: in roe deer from 2.1 to 22.9% [40,43,60,61]; in wild boar from 5.6 to 20% [40,43,62]; in moose at 0.74%; in white-tailed deer at 0.74% [42]; and in red deer at 1.4% [41]. Only Krzysiak et al. ( 2021) found a higher seroprevalence in European bison (62.7%) [34].
Comparing the different seroprevalence values, a great influence of the investigated area seems in place. Knowing that TBEV foci are localized and not homogeneous, a different pattern between and within nations can be expected. In fact, when sampling a larger area, a lower prevalence is expected, and when sampling smaller areas, high variability is likely [41,60]. The present study focused on two of the most affected areas in Italy for TBEV; thus, a higher seroprevalence was expected compared to other Italian regions. However, the investigated region is considered a low-risk area in the European context, and the remarkable seroprevalence described is in contrast with the scarce human reports, suggesting probable under reporting of cases [20]. Imhoff et al. (2015) highlighted some critical aspects of the use of TBEV serological techniques as screening methods, such as haemolysis of sera (related to the shot) and the scarce precision of the geographical data, due to the wide foraging area of wild ungulates [36]. Despite haemolysis in some samples, antibody detection was confirmed by two different diagnostic tests: ELISA and VNT, the gold standard test. As proposed by other authors, precise testing of TBEV antibodies in wild ungulates could be useful to establish risk maps in areas where data based only on human incidence could be biased, considering the high human TBEV vaccination coverage [60,61]. Indeed, finding positive animals in areas where no human cases were previously reported would be important for the identification of other potential risk areas [24,60]. Of note, TBEV seroconversion in alpine chamois has never been described in Europe before. This alpine species could be relevant for the early detection of new TBEV foci as an indirect sentinel of positive ticks in areas where other ungulate species are less frequently observed.
The results of the VNTs also highlighted the presence of USUV and WNV antibodies, detected in four wild boars (USUV -3.6%) and one roe deer (WNV -0.9%).
The preference of Culex mosquitos for wild boars rather than ruminants, when present in the same area, was thus confirmed as already proposed [43]. USUV was known to be present in the studied region: in fact, a human case has been recently described [63], but it had never been identified in wild boars before [32]. Few studies researched USUV antibodies in wild boars and found similar (3.4% in Serbia) or higher (8% in France) seroprevalences [43,64].
Regarding WNV antibodies in roe deer, both higher (23.5% in Serbia, 4.8% in Czech Republic) and lower seroprevalence (0% in Spain) were described [44,64,65]. The positive roe deer identified in the present study were found in Belluno Province, where until 2020, no human cases were reported [33]. West Nile disease cases were mainly located along the Po Valley and the description of WNV at northern latitudes may reflect the effects of climate change.
Conclusion
When dealing with vector-borne diseases, all epidemiological data may be useful to better understand their prevalence and diffusion. Emerging pathogens are often identified in vectors and/or animals first, and only later the same pathogens are diagnosed in humans [29,33,35,66,67]. Emerging infectious agents such as N. mikurensis and Ca. R. mendelii were identified, but they have never been diagnosed in humans in the studied area; thus, their presence should be acknowledged going forward. Tick sampling from wild ungulates has proven effective for tick-borne pathogen surveillance, as well as serological surveys on wild ungulates, due to the remarkable susceptibility of these wild species to flaviviral infections. Further epidemiological studies should take these aspects into account, since wild ungulate monitoring could serve as an early warning system for the detection of viral diffusion in areas considered lacking but at risk of vector, both mosquito and tick, expansion.
Area description
The study area encompassed the Friuli Venezia Giulia (FVG), Veneto (V), and Lombardy (L) regions, located in northern Italy (Fig. 1). Prealpine and alpine areas located in the provinces of Udine (UD), Belluno (BL) and Varese (VA) were investigated. These areas are approximately located at latitudes between 45°-46° N and are characterized by a wide altimetric excursion, ranging from 180 m a.s.l. in hilly areas up to more than 3000 m a.s.l. of the highest mountain peaks. The climate of the area under study is characterized by the climatic features of the alpine region with relevant temperature excursions between seasons in relation to different altitudes. Higher altitudes are characterized by continental weather: winter months are cold and snowy, while the temperature is mild during warm seasons. In the prealpine area, the climate is overall milder, and during the hunting seasons, the temperatures range from 0 °C to approximately 30 °C, with a maximum humidity reaching 70-98% in several periods (data extrapolated from ARPA Veneto, ARPA FVG and kindly provided by ARPA Lombardia) [68][69][70].
In addition, the alpine region is facing continuous and dramatic changes due to global warming, which highlights not only a change in temperature but also several other modifications regarding snow cover, humidity, precipitation, vegetation and natural hazards [71,72].
Sample collection
Sampling procedures were carried out between May 2017 and September 2020 during licenced hunting seasons. Blood samples were collected by hunters from wild ungulates in 9 mL Vacumed® tubes with K 3 EDTA (FL Medical srl, Italy) in hunting check stations or directly in the field. Ungulate carcasses were carefully inspected for ectoparasites by hunters and/or a veterinarian. When present, ticks were collected up to a maximum of 10 specimens per carcass, and each of them was placed in a 1.5 mL sterile microtube. An anamnestic form was filled out for each animal recording species, age, sex, date and place of culling, weight, health condition, and presence/ absence of ticks. Both blood and tick samples were stored at 4 °C and sent to the Laboratory of Microbiology and Infectious Diseases of the Animal Medicine, Production and Health Department (Legnaro, Italy). Once in the laboratory, blood samples were divided into 200 µL aliquots while the ectoparasites were first morphologically identified using the identification keys of Manilla and Cringoli [74,75] and then placed, individually or in pools of a maximum of two specimens, in a new 1.5 mL sterile microtube. Pools included ticks from the same host with the same features (species, sex, stage), and a low engorgement stage. Both blood samples and ectoparasites were stored at -80 °C until processing.
Molecular analyses
Nucleic acids were extracted from 200 µL of whole blood and tick samples using the All Prep DNA/RNA Mini Kit (QIAGEN GmbH, Germany) following the manufacturer's instructions. Each tick sample was ground thoroughly with a sterile disposable plastic pestle in a 1.5 mL microtube, resuspended in 350 µL of Buffer RLT Plus, and homogenized by repetitive pipetting.
Before the lysis step during nucleic acid extraction, both DNA and RNA internal controls, supplied by the Quantinova Pathogen + IC kit (QIAGEN GmbH, Germany), were added to all specimens to assess both extraction efficiency and the presence of PCR inhibitors. Extracted nucleic acids were stored at -80 °C (RNA) or -20 °C (DNA) and then screened using previously published PCR and real-time PCR methods to detect the following viral and bacterial pathogens: TBEV [76], B. burgdorferi s.l. [77,78], Rickettsia spp. [79,80], Coxiella spp. [81], Ehrlichia spp. [82] and N. mikurensis [83,84]. Positive and negative controls were included in each run. Details about the methods and procedures are provided in Additional file 1.
Internal control detection and pathogen screening were performed using a Quantinova Pathogen + IC kit (QIAGEN GmbH, Germany) on a LightCycler96 Instrument (Roche, Switzerland) with the Internal Control Assay kit and genus-specific real-time PCR assays (Additional file 1). N. mikurensis HRM real-time PCR assays were performed using 5x HOT FIREPol EvaGreen qPCR Mix Plus (Solis Biodyne) on a MyGo Pro instrument (IT-IS, United Kingdom). Ehrlichia spp.endpoint PCR screening was performed using 1× Phire Hot Start II PCR Master Mix (Thermo Fischer Scientific Baltics, Lithuania) on a Biometra TGradient thermal cycler (Analytic Jena GmbH, Germany).
All samples yielding a positive signal in both the internal control assay and pathogen screening were further investigated by specific end-point PCR assays (Additional file 1), followed by Sanger sequencing of amplicons. PCRs were performed with the same reagents and instrument used for Ehrlichia spp.screening.
PCR products were visualized by electrophoresis on 2% agarose gels stained with SybrSafe DNA Stain (Invitrogen by Thermo Fischer Scientific, USA) and subsequently purified using ExoSap-IT Express PCR Product Cleanup (Thermo Fischer Scientific Baltics, Lithuania) according to the manufacturer's instructions. Bidirectional Sanger sequencing of all specific PCR products was carried out at the StarSEQ® GmbH facilities (Mainz, Germany) using the same PCR primers. Nucleotide sequences were assembled and edited using ChromasPro v.2.1.8(Technelysium Pty Ltd, Australia) and were then deposited in GenBank (Accession numbers in Additional file 2) and analysed using the Nucleotide BLAST [85] search engine (National Center for Biotechnology Information, Bethesda, MD).
ELISA and virus neutralization for flaviviruses
Serum samples were collected in 9 mL Vacumed® tubes without anticoagulants (FL Medical srl, Italy). To assess the seroprevalence of TBEV, serum samples were screened with the two-step ELISA test Immunozym FSME IgG all species (PROGEN, Biotechnik GmbH, Heidelberg, Germany). According to the manufacturer's recommendations, sera were diluted 1:50. The results of the test were expressed in Vienna International Units and samples were considered positive with > 126 Vienna units/mL, borderline if between 63 and 126 Vienna units/ mL, and negative when < 63 Vienna units/mL. Positive and borderline samples were further tested with the gold standard confirmatory test, i.e., the virus neutralization test. VNTs were performed at the National Reference Laboratory for Arboviruses of the Ostrava Public Health Institute (Ostrava, Czech Republic). Anti-TBEV VNTs were performed using sterile 96-well plates. The TBEV strain (Hypr) was cultivated in intracerebrally infected suckling mice, and PS cells (porcine stable kidney cell line) were used as the susceptible cell line; VNTs were performed and the results were expressed as previously described [86] with minor changes consisting of the use of 25 µL of PS cell suspension (600,000 cells per mL) instead of the CV-1 cell line (African green monkey kidney fibroblasts). For anti-WNV and anti-USUV tests, an identical procedure was used, choosing the CV-1 cell line as the susceptible cell line for both viruses. WNV lineage 2 and USUV lineage Eur3 were used as virus suspensions. In each VNT, the endpoint titre was assessed as the higher serum dilution that inhibited the viral cytopathic effect. Samples showing a titre of anti-Flavivirus antibodies equal to 1:8 or higher were considered positive.
Statistical analysis
Data regarding sample collection and laboratory analyses were organized in a database on a Microsoft Excel Worksheet and descriptive statistics (counts, percentage and CI95%) were used to summarize results according to pathogen detection or serological results with respect to "species", "sex", and "age" of the animals, and "season" and "region" of the collection variables; the statistical analyses to detect significant differences in infection rates were conducted by means of chi-square test or Fisher's exact test, when appropriate. The level of statistical significance was set for alpha = 0.05.
Fig. 1
Fig. 1 Descriptive map highlighting the study area of sample collection. On the left, the three investigated regions (FVG, V, L) are red. On the right, a detailed image highlights the three provinces where sampling procedures were carried out, Udine (UD), Belluno (BL) and Varese (VA). Map created with mapchart.net[73] and modified with Microsoft PowerPoint (Microsoft office 365)
Table 1
Details on ungulate species tested in the study divided by the region of origin Seven samples tested negative at the internal control screening and were excluded from the analysis for TBP detection; thus, the analysed blood samples (n.274) are less than the initial number of sampled animals (n.281) A total of 88 out of 331 (26.6%; 21.8-31.4%) of the tick samples were positive for Rickettsia spp. Two zoonotic species were identified, namely, Rickettsia monacensis (R. monacensis) (35/88) and Rickettsia helvetica (R. helvetica) (36/88). Candidatus Rickettsia mendelii (Ca. R. mendelii) was found in two samples from ticks feeding on roe deer. Four out of 331 tick samples (1.2%; 0-2.4%) collected from roe deer tested positive for Neoehrlichia mikurensis (N.mikurensis). No other Ehrlichia species were identified.*
Table 3
ELISA and VNT test results of screened wild ungulate sera
==
Domain: Biology Medicine
|
[{"added": "2023-09-14T13:36:50.983Z", "created": "2023-09-14T00:00:00.000", "id": "261707893", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://bmcvetres.biomedcentral.com/counter/pdf/10.1186/s12917-023-03717-x", "pdf_hash": "840e5578edbed3e656bf061fd8dd814cba3159d4", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:6693", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "19ff82cf16faa9437068dc6d333862df670c17e0", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Biologics for Reducing Cardiovascular Risk in Psoriasis Patients
Psoriasis is a chronic inflammatory skin disease with a high prevalence of cardiovascular disease (CVD), obesity, dyslipidemia, hypertension, diabetes mellitus, and metabolic syndrome. Among them, CVD is the most common cause of morbidity and mortality in psoriasis patients. Since CVD is associated with considerable morbidity and mortality, primary care clinicians are increasingly committed to reducing the risk of CVD in patients with psoriasis. Biologics targeting TNF-α, IL-12/23, and IL-17 are systemic therapies that can dramatically improve the condition of psoriasis. Recent studies have reported that these inflammatory cytokine signals may promote atherosclerosis, suggesting that biologics might be effective for improving psoriasis as well as reducing the risk of CVD. Here, we reviewed cardiovascular risk in psoriasis patients, the association between psoriatic inflammation and atherosclerosis, and the efficacy of biologics for reducing the risk of cardiovascular diseases.
Introduction
Psoriasis is a chronic inflammatory skin disease characterized by erythema with scaling. It affects 2-3% of the world's population and about 0.5% of Asians [1][2][3][4]. Innate and acquired immunity is involved in the pathogenesis of psoriasis. Moreover, psoriasis is an independent risk factor for cardiovascular events, and the cardiovascular risk is particularly high in patients with severe psoriasis [5]. The increased prevalence of cardiovascular risk factors such as obesity, dyslipidemia, hypertension, diabetes mellitus, and metabolic syndrome in patients with psoriasis is also associated with an increased risk of developing cardiovascular diseases such as myocardial infarction, angina pectoris, and stroke [6][7][8][9]. Chronic inflammation is considered a strong link between psoriasis and associated cardiovascular events [10]. Various cytokines and inflammatory cells play a central role in developing psoriatic lesions, resulting in endothelial dysfunction [11,12]. Recently, the concept of the "psoriasis march" has been proposed, in which systemic inflammation caused by psoriasis and obesity leads to insulin resistance and vascular endothelial dysfunction, which in turn promotes atherosclerosis and the development of cardiovascular disease (CVD) [13]. It is essential to note the high incidence of CVD in patients with psoriasis, since CVD is directly related to morbidity and mortality. This means that cardiovascular risk should be assessed in patients with psoriasis, and lifestyle modifications should be made to manage blood pressure, blood glucose, and lipids. In addition, strict therapeutic control is also important to control the systemic inflammation of psoriasis. The advent of biologics has dramatically changed the treatment of psoriasis. TNF-α inhibitors, IL-23 inhibitors, and IL-17 inhibitors are highly effective against psoriatic skin lesions [14][15][16][17][18][19][20][21]. Recently, clinical studies and basic research have suggested that these biologics, which target inflammatory cytokines, effectively reduce cardiovascular risk [11,[22][23][24][25][26][27][28][29]. This article discusses cardiovascular risk in psoriasis, the association between psoriatic inflammation and atherosclerosis, and the cardiovascular-risk-reducing effects of biologics.
However, it seems almost evident that managing both diseases and controlling symptoms would decrease the incidence of CVD.
Psoriasis relates to dyslipidemia, which is also a factor of risk for CVD. In a systemic review, 20 out of 25 studies, including 265,512 psoriasis patients, reported that psoriasis was significantly associated with dyslipidemia. The OR for dyslipidemia ranged from 1.04 to 5.55 in 238,385 psoriasis patients [15]. The serum levels of triglyceride, cholesterol, and LDL were significantly higher in psoriasis patients, but not HDL levels [16]. Furthermore, TNF-α and other proinflammatory cytokines promote dyslipidemia by increasing the levels of LDL-C and oxLDL-C, decreasing the quality of lipoprotein, and reducing the level of HDL-C [17,18]. Using nuclear magnetic resonance spectroscopy, the lipid profile in psoriasis patients is similar to that observed in diabetes patients [19]. These lipid abnormalities in psoriasis patients drive systemic inflammation and promote insulin resistance, finally leading to the development of CVD.
Furthermore, many studies have found psoriasis to be an independent risk factor for atherosclerosis, myocardial infarction, stroke, and diabetes [5,. In particular, CVDs such as atherosclerosis, myocardial infarction (MI), and stroke are among the most critical complications because they are fatal. To investigate the causes of death in psoriasis, a large cohort study was conducted in the United Kingdom from 1987 to 2002. Comparing 3603 severe psoriasis patients with 14,330 healthy controls, the study found that life expectancy was approximately six years shorter in patients with severe psoriasis. Cardiovascular events were the most common cause of death [52]. Table 1 lists epidemiological studies published between 2006 and 2021 that examined the association between psoriasis and CVD [5,9,49,.
CVD included MI, coronary artery disease, angina, atherosclerosis, peripheral vascular disease, stroke, ischemic heart disease, cerebrovascular disease, CVD mortality, and coronary heart disease. Most studies found that the presence of psoriasis increased the risk of CVD, although some studies found no association between psoriasis and CVD risk. The risk of MI and stroke in mild psoriasis and severe psoriasis requiring systemic therapy was generally increased in severe psoriasis. As described above, many epidemiologic studies have been conducted. A systematic review including these epidemiological studies revealed that psoriasis increases the risk of CVDs such as MI and stroke [77][78][79], and statin administration is recommended to reduce CVD risk in patients at risk for CVD [80,81]. However, in practice, only a small percentage of physicians prescribe statin administration to those who need statin treatment [81]. Therefore, Barbieri et al. investigated measures to improve CVD prevention through specialist-led care from the perspective of healthcare professionals and patients. Results of the study showed that dermatologists and psoriasis patients had a positive view of participating in a specialist-led model of care to improve CVD prevention [82]. This suggests that dermatologists need to be more proactive in communicating with psoriasis patients and engaging in statins and other treatments to reduce CVD risk in psoriasis patients. The American Academy of Dermatology and National Psoriasis Foundation recommend assessing the risk of CVD in psoriasis patients. The screening of hypertension and diabetes and the assessment of CV risk every 3-6 years are encouraged for psoriasis patients [83]. In an effort to reduce CVD risk in patients with psoriasis, Garchick and Berger et al. proposed an algorithm to be used in patients with psoriasis who have a BSA >10% or who require biologic agents or phototherapy ( Figure 1) [84]. If the patient is at high risk, an appropriate CVD risk assessment should be performed, and the initiation of statin therapy should be considered. Even if the patient is not at high risk, performing a standard cardiovascular risk assessment is essential. Since dermatologists are the primary point of contact for psoriasis care, it is important for us to keep in mind that patients with severe psoriasis should be screened for a thorough CVD risk assessment. patients with psoriasis who have a BSA >10% or who require biologic agents or photo-therapy ( Figure 1) [84]. If the patient is at high risk, an appropriate CVD risk assessment should be performed, and the initiation of statin therapy should be considered. Even if the patient is not at high risk, performing a standard cardiovascular risk assessment is essential. Since dermatologists are the primary point of contact for psoriasis care, it is important for us to keep in mind that patients with severe psoriasis should be screened for a thorough CVD risk assessment.
Atherosclerosis in Psoriasis Patients
It is known that coronary artery plaques are more common in patients with psoriasis than in healthy individuals [50]. In the pathogenesis of psoriasis, inflammatory cytokines such as TNF-α, IL-23, and IL-17 form the primary pathogenesis and cause systemic inflammation. These inflammatory cytokines may cause vascular damage not only in the skin, but also in other parts of the body, and increase the risk of developing CVD via atherosclerosis. In fact, psoriasis and atherosclerosis have much in common [3,4,85,86]. It has been reported that treatment targeting IL-1β reduces atherosclerosis [24]. Therefore, it has been suggested that treatment targeting TNF-α, IL-23, and IL-17, which play a major role in shaping psoriasis pathology, may also contribute to the reduction in atherosclerosis.
Risk factors associated with atherosclerosis, such as hypertension and diabetes mellitus, increase vascular endothelial damage. High levels of LDL cholesterol in the blood gradually lead to excessive accumulation of LDL cholesterol in the intima of blood vessels. The accumulated LDL cholesterol is oxidized by reactive oxygen species and converted to oxidized LDL. Macrophages take up the oxidized LDL via scavenger receptors and become foam cells. Over time, the foam cells promote atherosclerosis by promoting atheroma plaque formation [87]. Macrophages in the plaque release inflammatory cytokines
Atherosclerosis in Psoriasis Patients
It is known that coronary artery plaques are more common in patients with psoriasis than in healthy individuals [50]. In the pathogenesis of psoriasis, inflammatory cytokines such as TNF-α, IL-23, and IL-17 form the primary pathogenesis and cause systemic inflammation. These inflammatory cytokines may cause vascular damage not only in the skin, but also in other parts of the body, and increase the risk of developing CVD via atherosclerosis. In fact, psoriasis and atherosclerosis have much in common [3,4,85,86]. It has been reported that treatment targeting IL-1β reduces atherosclerosis [24]. Therefore, it has been suggested that treatment targeting TNF-α, IL-23, and IL-17, which play a major role in shaping psoriasis pathology, may also contribute to the reduction in atherosclerosis.
Risk factors associated with atherosclerosis, such as hypertension and diabetes mellitus, increase vascular endothelial damage. High levels of LDL cholesterol in the blood gradually lead to excessive accumulation of LDL cholesterol in the intima of blood vessels. The accumulated LDL cholesterol is oxidized by reactive oxygen species and converted to oxidized LDL. Macrophages take up the oxidized LDL via scavenger receptors and become foam cells. Over time, the foam cells promote atherosclerosis by promoting atheroma plaque formation [87]. Macrophages in the plaque release inflammatory cytokines such as IL-1 and TNF-α. These inflammatory cytokines are thought to promote further atherosclerosis [88,89] (Figure 2).
Various studies have shown that acquired immunity also plays a vital role in atherosclerosis. It has been found that CD4+ T cells are present in atheroma plaques [90]. CD4+ T cells also play a significant role in the pathogenesis of psoriasis, especially Th1 cells and Th17 cells. Th1 cells differentiate and produce IFN-γ under the action of IL-12. IFN-γ is also involved in the production of proinflammatory cytokines, the upregulation of gene expression of adhesion molecules, the production of psoriasis-related cytokines, and the activation of macrophages and vascular endothelial cells, leading to the development of atherosclerotic lesions [91,92]. such as IL-1 and TNF-α. These inflammatory cytokines are thought to promote further atherosclerosis [88,89] (Figure 2). Various studies have shown that acquired immunity also plays a vital role in atherosclerosis. It has been found that CD4+ T cells are present in atheroma plaques [90]. CD4+ T cells also play a significant role in the pathogenesis of psoriasis, especially Th1 cells and Th17 cells. Th1 cells differentiate and produce IFN-γ under the action of IL-12. IFN-γ is also involved in the production of proinflammatory cytokines, the upregulation of gene expression of adhesion molecules, the production of psoriasis-related cytokines, and the activation of macrophages and vascular endothelial cells, leading to the development of atherosclerotic lesions [91,92].
IL-17 is a cytokine produced by CD4+ T cells. Th17 cells are activated mainly through cytokine signaling from dendritic cells [3]. The IL-17 receptor is expressed on epidermal cells in the skin, and IL-17 activates epidermal cells to cooperate with inflammatory cells to form the psoriasis pathology. The IL-17 receptor is known to be expressed on vascular endothelial cells, and its action on vascular endothelial cells is believed to promote the production of inflammatory cytokines such as granulocyte colony stimulating factor (G-CSF) and IL-6, promoting atherosclerosis [93]. On the other hand, other reports indicate that IL-17 does not promote atherosclerosis. In addition, IL-17 stabilizes plaque and may have an anti-atherogenic effect. IL-17 is a cytokine produced by CD4+ T cells. Th17 cells are activated mainly through cytokine signaling from dendritic cells [3]. The IL-17 receptor is expressed on epidermal cells in the skin, and IL-17 activates epidermal cells to cooperate with inflammatory cells to form the psoriasis pathology. The IL-17 receptor is known to be expressed on vascular endothelial cells, and its action on vascular endothelial cells is believed to promote the production of inflammatory cytokines such as granulocyte colony stimulating factor (G-CSF) and IL-6, promoting atherosclerosis [93]. On the other hand, other reports indicate that IL-17 does not promote atherosclerosis. In addition, IL-17 stabilizes plaque and may have an anti-atherogenic effect.
The Effect of Psoriasis Treatments on Cardiovascular Risk
The main strategy to reduce the risk of CVD has been the prevention of atherosclerosis through strict lipid control. In fact, the risk of CVD can be reduced by statin administration [80,94]. More recently, it has been found that statin administration reduces vascular endothelial inflammatory markers [95]. Thus, it is clear that lipid control is important in reducing CVD risk. In addition, the recent success of the CANTOS (Canakinumab Antiinflammatory Thrombosis Outcomes Study) trial has established evidence that the suppression of inflammatory cytokines reduces CVD risk. In this study, the anti-inflammatory effects of IL-1β inhibitor canakinumab administration were compared between three different doses and placebo in 10,061 patients with a history of myocardial infarction and high-sensitivity C-reactive protein levels. The results validated that anti-inflammatory therapy with canakinumab reduced the recurrence of CVD events [24]. Moreover, a recent study showed that biologics could reduce coronary inflammation assessed as perivascular fat attenuation index [96]. Furthermore, biologics can decrease intima-media thickness (IMT), an indicator of subclinical atherosclerotic plaque development, by reducing inflam-mation [80,[86][87][88][89][90][91][92][93][94][95][96][97][98][99][100]. In fact, cardiovascular events were reduced in psoriasis patients with biological treatment (HR 0.58 (0.30-1.10)) in a Danish nationwide cohort study [101]. Therefore, anti-inflammatory therapy targeting TNF-α, IL-23, and IL-17, the main pathological factors in psoriasis, can potentially contribute to the reduction in CVD risk and psoriatic skin rash.
Weight gain and higher BMI have been reported with TNF-α inhibitors; weight loss should be requested at the same time when using TNF-α inhibitors [102,103]. TNF-α inhibitors have been reported to be effective in treating insulin resistance in psoriasis patients [104], but no such reports have been reported for other biologic agents. Metaanalysis suggests that TNF-α inhibitor treatment reduces CVD risk [105,106]. The risk of myocardial infarction was also reduced with TNF-α inhibitors [107]. While these clinical studies suggest that TNF-α inhibitors may contribute to CVD risk reduction, some studies have shown the opposite. In summary, there is currently disagreement as to whether TNF-α inhibitors can reduce the risk of CVD, and further accumulation of evidence is needed.
IL-23 is a cytokine composed of two subunits, IL-12p40 and IL-23p19. IL-23 induces Th17 cell differentiation. The IL-23/IL-17 axis is central to psoriasis pathogenesis, and blocking IL-23 is highly effective for psoriasis lesions. Some studies showed the proatherogenic role in IL-23. IL-23 is highly expressed in atherosclerotic lesions [108], serum levels of IL-23 are elevated in patients with carotid atherosclerosis compared to healthy controls, and IL-23 and IL-23R mRNA levels are significantly elevated in carotid plaques. It has also been found that patients with elevated serum IL-23 levels have higher mortality [25]. Granulocyte macrophage colony-stimulating factor (GM-CSF) promotes plaque progression, which is mediated by IL-23, and increases apoptosis susceptibility in macrophages by promoting proteasomal degradation of the IL-23-mediated apoptosis susceptibility in macrophages by promoting proteasomal degradation of the cell survival protein B-cell lymphoma 2 (Bcl-2) and by increasing oxidative stress [26]. Information on IL-23 inhibitors' CVD risk reduction remains at the research level, with some studies suggesting that IL-23 is protective against atherosclerosis by acting to maintain the intestinal barrier and homeostasis of the intestinal microbiota [109]. On the other hand, some studies using LDL receptor-deficient mice have shown that IL-23 does not affect atherosclerosis [110].
IL-17A, IL-17B, IL-17C, IL-17D, IL-17E, and IL-17F are known as IL-17 family members [111]. Secukinumab, ixekizumab, bimekizumab, and brodalumab as IL-17 inhibitors have been used to treat psoriasis [14,16,18,19]. IL-17A and IL-17F are mainly associated with psoriasis, forming homodimers of each or heterodimers of both subunits [111]. Secukinumab and ixekizumab target IL-17A, and bimekizumab targets both IL-17A and IL-17F. On the other hand, brodalumab targets the IL-17 receptor, IL-17RA. These IL-17 inhibitors have also shown efficacy against psoriatic skin lesions in direct comparative studies with etanercept and ustekinumab [18,112,113]. The effect of IL-17A on vascular dysfunction has been examined in a mouse model; administration of angiotensin II increases IL-17 protein from T cells and aortic media. Mice showed improvement in blood pressure elevation and vascular dysfunction associated with administering angiotensin II [29]. Another laboratory has shown that blocking IL-17A reduces peripheral oxidative stress levels, proinflammatory cytokines, and vascular inflammation [23]. Furthermore, there is a report that IL-17A expression is high in atherosclerotic plaques from patients with ischemic symptoms [108]. These results show that IL-17A increases plaque instability and causes atherosclerosis. On the other hand, the results from other researchers demonstrated that IL-17A has a protective role in atherosclerosis. IL-17A levels are not involved in atherosclerotic plaque formation because there is no correlation between serum IL-17A levels and carotid intima-media thickness [27]. Thus, reports on IL-17A and atherosclerotic plaque formation have yielded different views. In 2019, the benefit of secukinumab, an IL-17A antagonist, for cardiovascular markers was investigated in the CARIMA (Evaluation of Cardiovascular Risk Markers in Psoriasis Patients Treated with Secukinumab) study. In this study, patients with moderate to severe plaque psoriasis without clinical CV disease were treated with secukinumab, and endothelial function was measured by flow-mediated dilation (FMD) as the primary endpoint. Baseline FMD was predominantly lower in psoriatic patients compared to healthy volunteers; secukinumab did not make a difference at 12 weeks, but there was an increase in FMD in psoriatic patients receiving secukinumab at 52 weeks [11]. Furthermore, another recent study, a prospective observational study for coronary artery plaque characteristics in psoriasis patients with biologics, showed IL-17 inhibitors reduced non-calcified plaque burden in psoriasis patients, suggesting the crucial role of IL-17 in atherosclerotic pathways [22]. Although future extensive clinical studies are needed, this is an important study that suggests a reduction in CVD risk by suppressing IL-17A.
Conclusions
Psoriasis is not only a cutaneous but also a systemic inflammatory disease, a group of diseases that carries a very high cardiovascular risk. Therefore, it is necessary to break away from treating psoriasis as a skin-only target and treat it as a systemic disease to avoid highly lethal cardiovascular disorders. We need to focus on obesity, dyslipidemia, diabetes, hypertension, and metabolic syndrome, which are frequently associated with psoriasis, and provide integrated treatment in close collaboration with other medical departments. Inflammatory cytokines, mainly TNF-α, IL-17, and IL-23, which occur in the pathogenesis of psoriasis, are thought to increase the risk of CVD, leading not only to psoriasis but also to vascular dysfunction. Therefore, it is suggested that biologics targeting these molecules may have a positive impact not only on the pathogenesis of psoriasis but also on CVD risk.
Conflicts of Interest:
The authors have declared no conflict of interest.
==
Domain: Biology Medicine
|
[{"added": "2023-02-04T16:16:17.167Z", "created": "2023-02-01T00:00:00.000", "id": "256561083", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.3390/jcm12031162", "pdf_hash": "4c4c08007a2c01bdccbf06672a3dbb00b6b94f01", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:8057", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "eb3cbb35f9849d4685ce4602363d07faf055c34b", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
An Assessment of the Anti-Inflammatory, Antimicrobial, and Antioxidant Activities of Ficus sur Stem-Bark
Ficus sur (Moraceae), is a plant that has found use in traditional African medicine in the treatment of sickle cell disease, epilepsy, pain and inflammations. The present study was aimed at investigating hexane and methanol stem-bark extracts of Ficus sur for their phytoconstituents, anti-inflammatory, antimicrobial and antioxidant activities. Phytochemical screenings were performed using standard protocols. In-vitro anti-inflammatory activities were assessed using egg albumin denaturation method. In-vitro antimicrobial (agar and broth dilution method) and antioxidant [total antioxidant capacity (TAC), DPPH and H 2 O 2 scavenging] assays were carried out on the extracts. Thin layer chromatography was employed in the separation of the components of both extracts. The phytochemical investigation revealed the presence of secondary metabolites such as anthraquinones, terpenoids, flavonoids, steroids, saponins, phenols and tannins. The extracts showed anti-inflammatory activity comparable to that of diclofenac sodium. The extracts showed antimicrobial activity against test organisms with MICs ranging from 2.5- 40 mg/ mL. The IC 50 values for methanol and hexane extracts in the DPPH and H 2 O 2 assays were 89.95 ± 0.30 and 350.70 ± 0.72 μg/mL and 708.51 ± 0.28 and 682.76 ± 0.20 μg/mL, respectively. The TAC (gAAE/100 g) for methanol and hexane extracts were 23.560 ± 0.014 and 17.863 ± 0.037 g, respectively. The results suggest that the stem bark of Ficus sur could be exploited as potential therapeutic candidate for the treatment of bacterial infections, inflammations and diseases associated with oxidative-stress.
Introduction
Medicinal plants are known to have antioxidant, antimicrobial, anthelminthic, anti-inflammatory and wound healing activities amongst others [1][2][3]. They are classified as the richest bio-resource, because they are the source of modern medicines, nutraceuticals, functional foods, food supplements, folk medicines, pharmaceutical intermediates and chemical entities for synthetic drugs [4]. Without specific knowledge of their cellular actions or mechanisms, various parts of plants have been used in traditional medicine to treat a variety of diseases and even as poisons (phytotoxins) [5].
Ficus is a genus of woody plants that comprises about 850 plant species and belongs to the family Moraceae. They are collectively called Figure trees or Figures and are distributed throughout the tropics and temperate zones, including most African countries such as Ghana, Burkina Faso, Nigeria and Cameroon. Ficus sur, a medicinal species of the family Moraceae has many therapeutic applications. A powdered preparation of the bark has been used to treat skin rashes and mouth sores in most parts of Africa.
A leaf preparation by maceration is used to cure chest problems. A decoction of the leaf has been used as a disinfectant wash and as a cure to ophthalmia [6]. Studies have shown antispasmodic and antiplasmodial activities from aqueous extracts of bark and leaves [7].
Other parts of F. sur have proven to be potent against a wide variety of ailments including gonorrhoea, sore throat, toothache, eye problems and many more [8]. F. sur is used in folk medicine for the treatment of sickle cell disease in Burkina Faso [9]. Traditional medicine practitioners in Nigeria use F. sur for effective management of epilepsy [10]. F. sur is used in the treatment of leprosy, infertility, gonorrhea, rickets, circumcision, oedema, respiratory disorders and many more [11]. Both root and bark decoctions of F. sur are recorded to have caused death, due to toxic substances [12]. The application of this plant species in the treatment of microbial diseases (example gonorrhoea), and in the treatment of sickle cell disease and management of epilepsy portrays it (Ficus sur) as a candidate for research.
Organic and Medicinal Chemistry International Journal
Most research studies conducted on the pharmacological potential of Ficus sur were mainly focused on crude extracts of the leaves, roots, barks [12,13] and fruits [14]. Nevertheless, it is also important to identify the bioactive compounds responsible for each one of the ascribed bioactivities, especially for the stem-bark. At the time of carrying out this research, next to nothing had been reported on the anti-inflammatory activities of the stem-bark. The aim of this study was to examine the efficacy of Ficus sur methanol and hexane extracts as an anti-inflammatory, antimicrobial and antioxidant using in vitro assays.
Sample collection and identification
The stem-barks of Ficus sur were collected in the month of October, 2018 at Kwahu-Asakraka, (Latitude: 6°37'44" N and Longitude: 0°41'29" W) in the Eastern Region of Ghana with the help of a local herbalist. They were taxonomically identified and authenticated by Mr. Clifford Asare at the Department of Herbal Medicine, Faculty of Pharmacy and Pharmaceutical Sciences, KNUST, Kumasi, Ghana. A voucher specimen number (KNUST/HMI/2019/ S042) was deposited in the Herbarium of Faculty of Pharmacy and Pharmaceutical Sciences for reference purposes.
Chemicals and reagents
All chemicals were purchased from Sigma Aldrich Co. Ltd, Irvine, U. K., except the standard drugs. The organic solvents were of analytical grade and procured from BDH Laboratory Supplies (England).
Extraction of plant material
The stem-barks of Ficus sur were thoroughly washed, first under running water to remove any form of debris and subsequently rinsed in distilled water to exclude dissolve heavy metals in tap water [1,2]. The stem-barks were chopped into smaller pieces, air dried under shade for two weeks, pulverized into coarse powder, and stored in a desiccator until analysis.
Preparation of methanol and hexane extracts
Maceration was used for the extraction of the phytoconstituents of the pulverized sample. A mass of 100 g of the pulverized sample of F. sur was soaked separately in 500 mL of methanol and hexane and macerated with gentle stirring for 72 hours at ambient temperature. The methanol and hexane extracts were condensed and evaporated to dryness using the rotary evaporator at 50 o C (BUCHI Rota vapor R-114). The extracts were dried and the percentage yield of extracts with respect to powdered plant material determined. The extracts were then stored at 4 o C in a refrigerator.
Phytochemical screening of extracts
The pulverized sample and the crude extracts obtained were screened to assess the presence of phytoconstituents using the methods described by Trease and Evans (2009) [15].
In-vitro anti-inflammatory assay using egg albumin denaturation
Anti-inflammatory assay was carried out according to a modification of the standard methods by Kumari [16]. Stock solutions of 1000 μg/mL of both extracts were prepared by using sterile distilled water as a solvent. From the stock solutions, various concentrations of 800, 600, 200 and 100 μg/mL were prepared using sterile distilled water as a solvent.
The reaction mixtures of total volume 5 mL were prepared by dissolving 0.2 mL of egg albumin (fresh egg of a hen), 2.8 mL of phosphate buffer saline (PBS, pH of 6.4) and 2 mL of the various concentrations of extract solutions. A volume of 2 mL of 200 μg/mL of diclofenac sodium was used as the standard reference drug and 2 mL of double distilled water solution served as negative control. The mixtures were incubated at 37 °C in Bio-Oxygen Demand (BOD) incubator for 15 minutes.
The mixtures were then heated in a water bath at 70 o C for 5 minutes to induce denaturation. The absorbance of the solutions was measured in triplicate at 660 nm using UV-vis spectrophotometer. The procedure was independently repeated to obtain three independent sets of data for the analysis in triplicate. The percentage inhibition of protein denaturation was calculated as follows: Where, A 0 = absorbance of negative control; A = absorbance of test solution
Antimicrobial activity
Agar well diffusion and Broth micro-dilution (minimum inhibitory concentration) assays were employed to assess the antimicrobial activities of the extracts.
Sources of microorganisms
Four bacteria and one fungus were used as test organisms. There were two Gram positive bacteria (Staphylococcus aureus and Enterococcus faecalis) and two Gram negative bacteria (Escherichia coli, Pseudomonas aeruginosa). The fungus was Candida albicans. The microbial strains were provided by the Pharmaceutical Microbiology Section of the Department of Pharmaceutics, Faculty of Pharmacy and Pharmaceutical Science, KNUST, Kumasi. The microbial strains were sub-cultured on nutrient agar slants and incubated at 37 °C for 24 hours.
Inoculum preparation
Bacterial isolates were streaked onto nutrient agar (Oxoid, United Kingdom) plates and incubated for 18-24 hours at 37 o C. Using the direct colony suspension method, suspensions of the organisms were made in nutrient broth and incubated overnight at 37 o C. For the tests, colony suspensions in sterile saline was adjusted to 0.5 McFarland standard and further diluted in sterile double strength nutrient broth (∼2 × 10 5 CFU/mL ) [17].
Agar well diffusion
The antimicrobial activities of the different extracts were determined using a modification of the agar well diffusion standard method previously described [1,18]. Ciprofloxacin (0.05 mg/mL) and clotrimazole (0.05 mg/mL) were used as the standard reference antimicrobial drug. The extracts and antibiotics were tested in triplicates and mean zones of inhibition were calculated for each extract and the standard antibiotic.
Broth micro-dilution
In the determination of the minimum inhibitory concentration (MIC), the method used was a modification of micro-well dilution standard method previously described [1,18]. Ciprofloxacin and clotrimazole were used as positive control. The experiment was carried out in triplicate.
DPPH radical scavenging assay
The DPPH free radical scavenging activity of the two extracts were examined using the standard methods previously described [1,19]. Ascorbic acid was used as reference standard. The experiment was independently repeated to obtain three independent sets of data for the analysis. The absorbance was measured at 517 nm. DPPH radical scavenging (%) was calculated using the formula: Where, A 0 = absorbance of control; A = absorbance of test solution
Hydrogen peroxide scavenging assay
Determination of hydrogen peroxide scavenging potential of the extracts were carried out employing the standard methods previously described [1,20]. Gallic acid was used as reference standard. Absorbance was taken at 510 nm using a UV-vis spectrophotometer. The experiment was independently repeated to obtain three independent sets of data for the analysis. The percentage scavenging activity was calculated using the formula below Where A test is absorbance of the test samples and A control is the absorbance of the negative control. The results were further reported in IC 50 .
Total antioxidant capacity (TAC) assay
A methodology previously described was used to study the total antioxidant capacity of the extracts of F. sur [1,21]. Ascorbic acid was used as the reference standard antioxidant and distilled water was used as the blank. The absorbance of the solutions was measured in triplicates using a UV-visible spectrophotometer at 695 nm. The experiment was independently repeated to obtain three independent sets of data for the analysis. From the linear equation of the ascorbic acid concentration-absorbance plot, the corresponding independent variables as ascorbic acid equivalents (AAE) were determined, and the results expressed as gAAE/100g ascorbic acid.
Thin layer chromatography (TLC)
The number of components present in the extracts were determined by the analytical TLC method. The pre-coated silica gel plates (0.25 mm) with a fluorescent indicator (F254) were spotted with the extracts about 1 cm from the bottom edge of plates, with the aid of capillary tubes and allowed to dry [1,22]. Various Solvent systems of petroleum ether/ethyl acetate and hexane/ ethyl acetate in the ratio of 9:1 and 8:2 respectively were used. The ratio of 8:2 (hexane/ethyl acetate) gave the best separation of components for all the extracts. The plates were dried and visualized by a 254 nm UV lamp. The separated spots were then marked and their sample and solvent fronts were measured.
The retardation factor (R f ) of the eluted spots was calculated as follows: Distance travelled by spot Distance travelled by solvent front f R =
Data analysis
Microsoft Excel 2016 and GraphPad Prism 6.0 for Windows (GraphPad Sofware, San Diego, CA, USA) were used for all data analyses and graphs.
Extraction of plant material
The yields of the extract in relation to the powdered plant material were calculated as percentages. The yields were 2.71 and 1.62% for methanol and hexane extracts respectively.
Phytochemical screening
The therapeutic activities of plants are as a result of the presence of complex chemical constituents in different parts [23]. The phytochemical screening revealed the presence of seven secondary metabolites out of the nine tested for in the pulverized sample and the methanol extract, with alkaloids and carotenoids being absent. Alkaloids, carotenoids and phenols were absent in the hexane extract ( Table 1).
The methanol and hexane extracts had six phytochemicals in common, that is anthraquinones flavonoids, saponins, steroids, tannins and terpenoids. The absence of alkaloids in the stem-bark of F. sur corroborates the work of Adebayo [24] who investigated haematinic properties of methanolic stem-bark and fruit extracts of Ficus sur in rats pre-exposed to phenylhydrazine induced haemolytic anaemia [24]. Secondary metabolites of plants which include phenolics and flavonoids, have been shown to exhibit several biological activities such as antioxidant, antiaging, antidiabetic, antimutagenic, anticarcinogenic, anti-inflammatory and antimicrobial [25]. Saponins have a wide range of pharmacological properties, including antifungal, antiparasitic, molluscicidal and anti-inflammatory [26]. The presence of these phytochemicals in the extracts of F. sur stem-bark indicate that they will play a key role in the prevention of various bacterial infections, inflammations and diseases associated with oxidative-stress.
In vitro anti-inflammatory assay (egg albumen denaturation method)
Denaturation of proteins is a well-documented cause of inflammation and rheumatoid arthritis. Several anti-inflammatory drugs have shown concentration-dose-dependent ability to inhibit thermally induced protein denaturation. The denaturation of albumin protein leads to formation of antigens which initiate type III hypersensitive reaction leading to inflammation [27]. The ability of plant extract to inhibit thermal denaturation of protein (egg albumin) is a reflection of its anti-inflammatory activity [28].
At the concentration of 200 μg/mL, percentage inhibition of the standard, methanol and hexane extracts were 73.870, 47.176 and 20.500% respectively as shown in Table 2. The anti-inflammatory activity shown by the extracts could be attributed to the presence of saponins, terpenoids and steroids in the methanol and hexane extracts of F. sur, which have been reported to exhibit anti-inflammatory activity [29]. The presence of polyphenols including tannins and flavonoids in F. sur have been reported to reduce inflammation and suppress several stages of angiogenesis, including endothelial cell migration, invasion, matrix metalloproteinase activity, and tube formation [30]. 73.870 ± 0.002* Diclo: Diclofenac Sodium. Results were expressed as mean (n = 3) ± standard deviation; The data were anaylsed using one-way ANOVA compared to diclofenac (reference drug). *P < 0.001.
Agar well diffusion
The antimicrobial activities of the extracts were determined at two concentration levels of 50 and 100 mg/mL for the agar well diffusion assay as shown in Table 3. The agar well diffusion is carried out to test for the sensitivity of the organisms to the antimicrobial agent (plant extract). The diameter of the zone of inhibition determines the effectiveness of the extract against the microorganism. The larger the diameter, the greater the sensitivity of the microorganism to the extract. The sizes of the zone of inhibition are compared to standards to determine if the microorganism is sensitive or resistant to the plant extract. From the results obtained, the methanol and hexane extracts recorded zones of inhibition at the lower concentration of 50 mg/ mL. At this concentration, the methanol extract recorded inhibition against E. faecalis, S. aureus and P. aeruginosa but showed no inhibition against E. coli and C. albicans. However, the hexane extract showed inhibition against only S. aureus at this concentration. At a concentration of 100 mg/mL, all the tested organisms were susceptible to both the methanol and hexane extracts.
Generally, susceptibility increased with the increased concentration of extract as the zones of inhibition increased for all organisms. P. aeruginosa was the most susceptible to the methanol extract at 100 mg/mL with C. albicans being the least susceptible. At the same concentration, S. aureus was the most susceptible to the hexane extract with E. faecalis being the least susceptible. All the four tested bacteria were susceptible to the ciprofloxacin (standard drug) with the gram-positive bacteria S. aureus showing the highest susceptibility. Both extracts and clotrimazole (standard drug) showed activity against the fungus C. albicans.
Broth microdilution
The extracts showed broad spectrum antimicrobial activity against the tested organisms. The methanol extract showed a better antimicrobial activity (at MIC of 2.5 mg/mL to 10.00 mg/ mL) against the test organisms than the hexane extract (at MIC of 20.00 to 40.00 mg/mL). The results are shown in Table 4. The results from the antimicrobial assay performed showed that the two extracts of F. sur stem-bark exhibited varying inhibitory effects against the five selected microorganisms (two Gram-positive, two Gram-negative and one fungus). The best results were observed with the use of the methanol extract against all the selected microorganisms. The minimum inhibitory concentrations (MICs) were between the range of 2.5 mg/mL to 10.0 mg/mL. The highest activity observed with the use of methanol extract was against P. aeruginosa with MIC of 2.5 mg/mL. The antimicrobial activity shown by the extracts could be attributed to the presence of terpenoids, saponins and polyphenols such as flavonoids and tannins in the methanol and hexane extracts of F. sur which have been reported to exhibit antimicrobial activity [31,32].
In vitro antioxidant capacity
The total antioxidant potential of a plant extract depends largely on both the constituent of the extract and the test system. Different factors can also influence the activity of the extract, therefore when carrying out a study related to the antioxidant and antiradical properties of plant products, more than one method is usually used to evaluate the antioxidant capacity/activity [33]. Considering the various mechanisms of antioxidant actions, the antioxidant properties of the extracts were evaluated by (DPPH) free radicals scavenging, Hydrogen Peroxide scavenging and the Total Antioxidant Capacity assays.
DPPH radical scavenging capacity
The DPPH radical scavenging activity of the extracts was used to determine and study the ability of the extracts of F. sur to mop up free radicals that may be found in animals and humans. Methanol and hexane extracts of F. sur and ascorbic acid (reference standard) scavenged DPPH radical in a dose dependent manner (Figure 1). The reference antioxidant (ascorbic acid), hexane and methanol extracts of F. sur showed antioxidant activity in the DPPH free radical scavenging assay with IC 50 of ascorbic acid, hexane and methanol ranged from 3.17 ± 0.32 to 350.70 ± 0.72 μg/mL, as shown in Table 5. The results implied that the potency of the tested samples of extracts as antioxidants decreased in the order: ascorbic acid > methanol > hexane (Figure 1). Methanol extract showed better antioxidant activity compared to the hexane probably due to the presence of the polyphenols that act as anti-aging agent by neutralizing the effect of free radicals [34]. Polyphenols including tannins and flavonoids have many favourable effects on human health, such as the inhibition of the low density proteins oxidiza-tion [35]. Though hexane and methanol extracts which comprise of a mixture of compounds were not as potent as the ascorbic acid, F. sur stem-bark extracts may be useful in the manufacture of drugs to help prevent or cure health problems that could arise from the systemic actions of oxidative agents, thus its usage in folk medicine for the treatment of sickle cell diseases in Burkina Faso [9]. Non-radical oxidizing agents scavenging potential of the hexane and methanol extracts of F. sur were evaluated by the use of hydrogen peroxide (H 2 O 2 ) scavenging method. The results are shown in Table 6. Methanol and hexane extracts of F. sur and gallic acid (reference standard) exhibited H 2 O 2 scavenging capacity in a dose dependent manner (Figure 2).
Hydrogen peroxide scavenging assay
The IC 50 of a sample is the concentration of the sample required to scavenge 50% of the peroxide in a system. It is used to evaluate the antioxidant capacity of a sample. The lower the IC 50 , the better the antioxidant potential of the sample under examination [1,36]. Results showed that, hexane and methanol extracts demonstrated a significant antioxidant activity in concentration-dose dependent manner. The IC 50 values of gallic acid (standard drug), hexane and methanol extracts ranged from 204.40 ± 0.01 to 708.51 ± 0.28 μg/mL as shown in Table 6. From the results, both methanol and hexane extracts which comprise of a mixture of compounds showed slightly lower activity than gallic acid (reference standard), even though they are all good antioxidants. Bioactive isolates from these extracts responsible for antioxidant activity could be attributed to the terpenoids and polyphenols, such as tannins and flavanoids in F. sur and could be exploited for the treatment of oxidative-stress diseases [34].
Total antioxidant capacity (TAC)
Ascorbic acid also known as Vitamin C is an electron donor antioxidant and this property is responsible for all its known functions. Vitamin C is a potent reducing agent and scavenger of free radicals in biological systems. It is a cofactor for enzymes involved in regulating photosynthesis, hormone biosynthesis, and regenerating other antioxidants [37].
Concentrations of ascorbic acid ranging between 6.125 to 100 μg/mL showed antioxidant activity and mean absorbances between 0.059 ± 0.003 to 0.932 ± 0.002 at wavelength of 695 nm (Figure 3). The TAC was found to be proportional to the concentration of extract. TAC of the extracts were examined by Phosphomolybdenum method and the results were expressed as gram ascorbic acid equivalent per 100 grams (gAAE/100g) [1]. The gAAE/100g, represents the fraction of the plant extract that can act as ascorbic acid in 100 g of the extract. The hexane and methanol extracts had 17.863 ± 0.037 and 23.560 ± 0.014 gAAE/100g, respectively, (Table 7). Generally, the TAC increased with increasing concentration, thus the higher the TAC, the better the activity of the sample. Polyphenols including flavonoids are important natural antioxidants, which are basically associated with curing of various diseases and disorders including cancer, diabetes, gout, urolithiasis, obesity, and other diseases associated with ageing [38,39]. Both extracts demonstrated appreciable antioxidant activities due to the pres-ence of the various phytochemicals such as flavonoids, phenols, tannins, terpenoids among others in F. sur.
Thin layer chromatography (TLC)
The number of components present in the extracts were determined by the analytical TLC method. The chromatographic spots which were representative of compounds in the various extracts were observed and their Rf values determined. Table 8 gives the results of the TLC analysis. The hexane extract showed four spots and methanol five spots with R f values between 0.113 to 0.950 and 0.050 to 0.900, respectively. The number of spots indicating the separated components in the two extracts were less for both extracts when compared to the phytoconstituents identified to be present in each stem-bark extract. This means that some of the components existed as isomers, did not elute due to the polarity of the mobile phase or co-eluted in mixtures and it may be necessary to employ two dimensional TLC, HPLC or column chromatography to achieve complete separation of the components [1].
Conclusion
The hexane and methanol extracts of F. sur showed the presence of varying secondary metabolites including saponins, tannins, terpenoids, steroids, flavonoids, phenols and anthraquinones. The study demonstrated that the hexane and methanol extracts of F. sur possess a variety of anti-inflammatory, antibacterial, antifungal and antioxidant activities. This implies the extracts could be effective against inflammations, infectious and diseases associated with oxidative-stress, and could become a potential therapeutic agent for their treatment. Further studies are ongoing in our laboratory towards isolation, characterization, identification and determination of biological activities present in the stem-barks of F. sur.
Disclosure
Part of this work was presented as a poster at the "8 th Ghana Science Association, Research Seminar and Poster Presentations" held at the Kwame Nkrumah University of Science and Technology, Kumasi, Ghana, in May 2019.
==
Domain: Biology Medicine
|
[{"added": "2021-09-13T22:14:21.128Z", "created": "2020-05-27T00:00:00.000", "id": "243268688", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "http://juniperpublishers.com/omcij/pdf/OMCIJ.MS.ID.555766.pdf", "pdf_hash": "e26d4d798c1bc211d8f2e83421a9826133bf0b62", "pdf_src": "ScienceParsePlus", "provenance": "peS2o-0028.json.gz:668", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "5daf5ecff817deb59e34f7a187fc425549a7c6c6", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Cisplatin-based chemoradiation decreases telomerase-specific CD4 TH1 response but increases immune suppressive cells in peripheral blood
The synergistic effect of chemoradiation (CRT) has been previously demonstrated in several cancer types. Here, we investigated the systemic immune effects of CRT in patients with lung or head and neck cancer. Peripheral blood mononuclear cells were collected at baseline and 1 month after treatment from blood samples of 29 patients treated with cisplatin-based chemoradiotherapy for lung or head and neck cancer. Circulating anti-tumor Th1 response was assessed by the ELISpot assay using a mixture of human leucocyte antigen (HLA) class II restricted peptides derived from telomerase (TERT). Phenotyping of circulating immunosuppressive cells (Treg and MDSC) was performed by flow cytometry. A significant increase of circulating Treg was observed in 60% of patients after CRT The mean rate of Treg was 3.1% versus 4.9% at baseline and after CRT respectively, p = 0.0015). However, there was a no significant increase of MDSC rate after CRT. In contrast, a decrease of tumor-specific Th1 response was documented in 7 out of 10 evaluated patients. We found high frequency of pre-existing tumor-specific Th1 response among patients with objective response after CRT compared to non-responders. Cisplatin-based CRT promotes expansion of Treg and decrease of circulating anti-tumor Th1 response in peripheral blood. The balance towards a sustained specific anti-tumor T-cell response appears to be associated with response to CRT.
Background
Concurrent chemoradiation (CRT) represents a standard curative treatment for several locally advanced cancers [1]. The addition of chemotherapy (CT) to radiation therapy (RT) improves locoregional control via a synergistic effect through the induction of irreversible DNA damages [1]. During the past decade, major findings have described the immunological effects of cytotoxic anticancer therapy. Indeed, CT and RT used as monotherapy exert their anti-tumor effect not only directly by creating DNA lesions that lead eventually to cell killing, but also indirectly by stimulating an anti-tumor immune response via the innate and adaptive immunity [2][3][4]. Tumor cells exposed to RT and/or CT release tumorassociated antigens (TAA) which are captured by dendritic cells (DCs) for processing and presentation on MHC class I and II molecules to T cells [5]. This leads to the priming and activation of effector T-cell responses against the TAA. The activated effector T cells traffic to tumor site where they specifically recognize and kill their target cancer cells. Also, RT may cause regression of tumors distant from the irradiated site, a phenomenon known as abscopal effect. Despite being rarely observed in daily practice, preclinical and clinical evidence have suggested that this effect may be immune-mediated, translating the systemic anti-tumor effect of local RT. [6][7][8] On the other hand, RT was shown to drive the accumulation of immunosuppressive cells such as regulatory T cells (Tregs), myeloid-derived suppressor cells (MDSCs) and type 2 macrophages in the tumor micro microenvironment [9][10][11][12]. Furthermore RT can induce PD-L1 expression on both tumor cells and immune cells as well as upregulation of immune checkpoint receptors (TIGIT, TIM3 …) on tumor infiltrative lymphocytes, hence limiting the anti-tumor immunity [13][14][15][16][17].
Although the effect of each of RT and CT on immune response has been widely studied, little is known about the impact of their combination on the immune system [18,19]. Understanding the immune modulatory properties of concurrent CRT has gained a great interest in the field of the combination with cancer immunotherapy [20][21][22].
CD4 T cells play a central role in orchestrating the adaptive immune response [23]. They can kill tumor cells that express MHC-II molecules either directly via MHC-II/peptide recognition [24] or indirectly by inducing MHC-II expression via IFN-γ [25][26][27]. Since MHC-II peptides have a lower MHC binding affinity than MHC-I peptides [28][29][30], CD4 T cells could have a wider range of regulation of the antitumor response. Telomerase reverse transcriptase (TERT) is a self-tumor antigen that plays a major role in tumor development and progression [31,32], and is overexpressed in more than 90% of human tumors. Naturally occurring HLA-II-restricted CD4 T cell responses against TERT peptides were detected in patients with various types of cancer and were associated with a good prognosis [33][34][35]. Thus, the assessment of anti-TERT CD4 Th1 cell immunity in circulating lymphocytes has been used as a tool for monitoring antitumor CD4 Th1 response in cancer patients [33][34][35].
In this study, we assessed the effect of concurrent CRT on peripheral tumor-specific CD4 Th1 response and immunosuppressive cells in patients with lung or head and neck cancer.
Increase of circulating immunosuppressive cells after CRT
The impact of CRT on Treg and MDSC was evaluated in 20 patients by flow cytometry from PBMCs collected before CRT and 1 month after. We assessed the percentages of circulating myeloid-derived suppressor cells (MDSC) and regulatory T cells (Treg) in viable PBMCs using flow cytometry (Supplementary Figure S1). MDSC were defined as HLA-DR low Lin − CD33 + CD14 + CD11b + and Treg were defined as CD3 + CD4 + CD25 + CD127 low-FoxP3 + . A significant increase in Treg was observed in 12 out of 20 patients (60%) after CRT ( Fig. 1A and B). The mean Treg rate was 2.7% before CRT and 4.9% after CRT (p = 0.0015) (Fig. 1C). Furthermore, the rate of CTLA4 + Treg significantly increased after CRT (p = 0.003) (Fig. 1D-F). Next, we evaluated the effect of CRT on MDSC and found that overall MDSC rate increased significantly in 8 out of 20 patients (40%) after CRT ( Fig. 2A and B). Although the mean MDSC rate increased after CRT, the difference was not statistically significant (Fig. 2C). Altogether, these results suggest that CRT promotes high Treg expansion in the peripheral blood.
Decrease of peripheral anti-telomerase CD4 Th1 response after CRT We further evaluated the impact of CRT on tumorspecific T cell response. To this end, we measured by IFN-γ ELISpot the CD4 T cell response directed against telomerase (TERT), a shared-tumor antigen ( Fig. 3A) [36]. We previously showed that circulating anti-TERT CD4 T cell response is a surrogate marker of the host's antitumor Th1 immunity and that the presence or induction of circulating anti-TERT CD4 T cell response was associated with a good prognosis in several cancers such as renal carcinoma, anal carcinoma, and NSCLC [33-35, 37, 38]. T cell responses against viruses such as CMV, EBV and FLU measured concurrently were used as control. Anti-TERT Th1 response was found in 12 out of 26 patients (46%) at baseline. Data was missing in three patients. Response against TERT in a representative patient with loss of anti-TERT Th1 response after CRT is shown in Fig. 3B. In 19 patients with available samples before and after CRT, we found that 10/19 had anti-TERT Th1 response at baseline, among which seven (7/10) had a significant decrease of their response, whereas only two patients (2/10) had a significant increase of their anti-TERT Th1 response after CRT (Fig. 3C). In contrast, no obvious change of the frequency or intensity of antiviral recall responses was observed after CRT (Fig. 3D). These results suggest that CRT may induce a decrease of tumor-specific T cells in peripheral blood. Thus, we assessed whether there is a relationship between the decrease of anti-TERT T response and the increase of immunosuppressive cells. Overall, 17 patients had available data for both anti-TERT response and immunosuppressive cells ( Table 2). In patients with a decrease in anti-TERT response (6/17) or no response (8/ 17), we found 10/14 patients (71%) who had an increase in Treg and/or MDSC rates in peripheral blood. Two patients with an increase in anti-TERT response had an increase in Treg rates as well.
Influence of CRT-related immune response on clinical response
After a median follow-up of 12 months (range, 3-30 months), the median overall survival and progressionfree survival were 28 and 17 months, respectively, similar to previously reported outcomes after CRT in these cancers [39][40][41]. The clinical response was assessed in 24 patients. The objective response (OR) rate was 10/24 (42%). Progressive disease (PD) was seen in 14 patients (58%). Next we studied the association of naturally occurring anti-TERT immune response and clinical outcome ( Table 2). We found a high frequency of anti-TERT Th1 response among the majority of CRT-responders, Fig. 1 Circulating Treg cells before and after CRT. PBMCs from 20 patients treated with CRT for lung or head and neck cancer were taken before CRT and 1 month after. CD4 + CD25 + CD127 low FoxP3 + Treg were analyzed by flow cytometry. A and D Representative plots for Treg (A) and CTLA-4 + Treg (D) in one patient. B and E Treg (B) and CTLA4 + Treg (E) rates variation after CRT, lines in red representing significant increase (> 20%) from baseline (n = 20). C and F Treg (C) and CTLA-4 + Treg (F) rates before and after CRT (n = 20). Results are shown as mean (standard deviation). **, p < 0.005 (Wilcoxon test) compared to non-responders. Indeed, the anti-TERT response was found either at baseline or after CRT in 5/7 patients (71%) with OR, while 4/9 patients (44%) with PD exhibited peripheral anti-TERT CD4 T cell response either at baseline or after CRT. Therefore, patients who were able to mount specific anti-tumor T-cell responses were probably more likely to respond to treatment. Furthermore, we evaluated overall survival according to TERT responses and immunosuppressive cells (Supplementary Figure S2). The median value of TERT response, MDSC and Treg rates was used as a cut-off for TERT low/high, MDSC low/high, and Treg low/high, respectively. There was no significant difference in patients with TERT low or high response (Supplementary Figure S2A), nor between MDSC low or high (Supplementary Figure S2B), and Treg low and high levels (Supplementary Figure S2C) measured before and after CRT.
Next, we evaluated the relationship between clinical response and immunosuppressive cells. There was a significant increase in Treg and/or MDSC in 7/8 patients (88%) with OR, and 7/9 (78%) patients with PD (Table 2). Thus, there was no difference between responders and non-responders with regard to immunosuppressive cells.
Our results suggest that the clinical response was mostly influenced by the peripheral anti-TERT CD4 T cell response and not by immunosuppressive cells.
Discussion
In this study, we wanted to determine the impact of CRT on anti-tumor specific responses in cancer patients. To this end, we assessed T-cell responses directed specifically against TERT, known for its frequent expression in various cancer types and its high immunogenicity [42]. In our cohort of patients presenting predominantly with a non-small cell lung cancer (NSCLC), we found that T-cell responses against TERT were naturally present in 46% of the cases. This was in line with previous results showing TERT-specific CD4 T cell responses in 45% of patients with non-metastatic NSCLC at baseline [33]. The prognostic value of specific immune responses in the peripheral blood of cancer patients have been reported in several malignancies. For instance, Masterson et al. demonstrated that the presence of E7-specific immune responses in the peripheral blood of HPV + head and neck squamous cell carcinoma patients was associated with better overall survival [43]. Interestingly, our results suggested that patients who were able to mount specific anti-tumor T-cell responses were more likely to respond to treatment. However, we demonstrated a significant decrease of anti-TERT responses after CRT in most of the patients. The loss of anti-tumor specific immune responses could not be related to a global T-cell anergy, as illustrated by the unchanged antiviral recall responses' frequency. We hypothesized that the decrease of tumor-specific-T cell responses after CRT was mainly related to RT by promoting suppressive cells' expansion.
Indeed, evidence support the ambivalent role of RT in activating the host antitumor immunity while promoting immunosuppression [3,9,21]. The induction of suppressive Treg and MDSCs after CRT has been previously reported. Hence, Schuler et al. reported the amplification of highly suppressive, cisplatin-resistant Treg after CRT and these cells persist long-term after treatment and could be responsible for suppression of antitumor immune responses and recurrence in HNSCC [44]. Recently, Hanoteau et al. showed that removal of MDSC in vivo improves CRT effectiveness [45]. Furthermore, [46] studied the impact of RT and CRT in patients with cervical cancer and showed that RT alone or with [47]. Here, we observed a significant expansion of circulating Treg and MDSC after CRT. Although, the suppressive functions of these cells were not formally explored, we speculated that these cells could be involved in the decrease of antitumor Th1 response observed after CRT.
Cisplatin or carboplatin-based CT was commonly used in combination with RT, both in head and neck and lung cancer. These drugs have been shown to stimulate host antitumor immunity either by increasing tumor cells sensibility to immune effector cells attack or through elimination of immune suppressive cells [48,49]. In line with this, we previously reported that cisplatin-based CT reinvigorates TERT-specific Th1 response by promoting MDSC depletion [35,37,50]. Thus, our data also suggest that the inhibitory effect of RT rather than platinumbased CT was responsible of the attenuation of tumorspecific T cell responses.
Our study has several limitations. First, our analyses were limited to peripheral immunity rather than the tumor microenvironment. Second, the relatively small number of patients included makes it difficult to perform Fig. 3 Spontaneous anti-tumor and antiviral responses before and after CRT. A PBMCs from 26 patients treated with CRT for a lung cancer or a head and neck cancer were collected before CRT and 1 month after. After short stimulation (1 week) with a mixture of HLA class II peptides derived from TERT or viral peptides, the presence of TERT-specific T cells was detected by IFNγ ELISPOT assay. The results represented specific IFNγ spots after subtraction of background. Responses were positive when IFNγ spots were more than 10 and more than 2-fold the background. B Response against TERT in a representative patient with loss of anti-TERT Th1 reponse. Bottom: histograms represented specific IFNγ spots number in medium (grey) and TERT (black). Top: illustration of medium and TERT ELISPOT wells. C individual variation of the intensity of anti-TERT Th1 response in patients with available data at baseline and 1 month after CRT (n = 19). Lines in red represent significant decrease (> 20%) from baseline. D intensity of specific anti-viral response in 10 patients with available data at baseline. Number of patients with anti-TERT response is shown between brackets robust statistical analysis. Third, analyses were performed before and after CRT without providing an interim analysis during treatment which could have allowed the understanding of CRT's early impact on anti-tumor immune responses. Currently, we are recruiting patients treated with CRT for locally advanced inoperable lung or head and neck cancer to study the mechanisms underlying the immunomodulation induced by CRT in a prospective large cohort (iRTCT cohort, NCT 03117946).
Conclusion
This study emphasized the role of CRT in the modulation of systemic immune responses. We found that after CRT there was a decrease in anti-TERT response in most of the patients that could be explained by the concomitant increase in immunosuppressive cells, which was predictive of the clinical response. These preliminary results have implications in clinical practice particularly in combination strategy with immune checkpoint inhibitors.
Patients and blood samples
Lung cancer patients and head and neck cancer patients treated with CRT at the department of radiation oncology of the University Hospital of Besancon (France) were Table 2 Clinical response, anti-TERT response and immunosuppressive cells in all patients (n = 29) Abbreviations: HN Head & neck cancer, SCC Squamous cell carcinoma, NSCLC Non-small cell lung cancer, SCLC Small cell lung cancer, RT Radiotherapy, CT Chemotherapy, OR Objective response, PD Progressive disease, TERT Antitumor response, C Cisplatin, CE Cisplatin + etoposide, CN Cisplatin + navelbine, CP Carboplatin + paclitaxel. Δ: evolution of anti-TERT response, Treg and MDSC rates after CRT, defined as stability ( ), increase ( ) or decrease ( ). (+) sign represents presence of anti-TERT response, (−) sign represents absence of anti-TERT response enrolled. All patients were included after the signature of informed consent, in accordance with the French regulation and after approval by the local ethics committee. Blood samples were collected prior to treatment and 1 month after. Peripheral Blood Mononuclear Cells (PBMCs) were Ficoll-isolated (Amersham, Biosciences, France) and frozen in aliquots in liquid nitrogen. Approximately 30 ml of blood were collected before and after treatment (1 month later). This allowed the isolation of 15-20 × 10 6 PBMCs at baseline and around 10 7 PBMCs 1 month after CRT. After thawing, cell viability was estimated around 90%.
Clinical response to treatment was evaluated 3 months after the end of CRT with CT scan based on RECIST criteria. Objective response was defined as a complete response, a partial response or a stable disease. Otherwise, progressive disease was stated. Patients with progressive disease after CRT have been treated according to the standard of care. In this limited cohort, no patient received adjuvant immunotherapy (eg, Durvalumab) following CRT at the time of the study.
In vitro stimulation for the detection of tumor-specific CD4+ Th1 responses in blood
Telomerase-specific CD4+ T-cell responses were assessed in PBMCs using a standard IFNγ ELISPOT assay, following in vitro stimulation. Briefly, PBMCs (3.10 6 cells per well) were cultured for 6 days in a 24well plate in RPMI containing 5% human serum and 1% penicillin-streptomycin, along with the mixture of TERT-derived peptides (5 μg/mL). To assess anti-viral T cell responses, cells were stimulated with a mixture of peptides derived from CMV, EBV and Flu (1 μg/mL). Recombinant interleukine (IL) 7 (5 ng/mL, R&D Systems, France) was added on day 1, and recombinant IL-2 (50 U/mL, Proleukin, Chiron, France) was added on day 3. Plates were incubated at 37°C.
IFNγ ELISPOT assay
The presence of peptide-specific T cells was measured by IFNγ ELISpot assay at day 7 according to the manufacturer's instructions (Diaclone, France), as previously reported [36]. Briefly, lymphocytes from in vitro stimulation were incubated for 17 h at 37°C in duplicates or triplicates (10 5 per well) in a precoated 96-well ELISpot plate with anti-human IFNγ monoclonal antibody, with 5 μg/mL of the peptide mixtures derived from TERT and CEF in the X-vivo 15 medium (Lonza). Cells cultured with medium alone or phorbol myristate acetate (PMA, 100 ng/mL; Sigma-Aldrich) and ionomycin (10 μmol/L; Sigma-Aldrich) were used as negative and positive controls, respectively. The IFNγ spots were revealed following the manufacturer's instructions (Diaclone, 856051020P). IFNγ secreting cells i.e., spot-forming cells were counted using the C. T. L. Immunospot system (Cellular Technology Ltd). After subtracting the negative control values (background), we calculated the number of IFNγ spots per 10 5 cells. A response was considered positive if the number of IFNγ spots per 10 5 cells was > 10 and more than two times the background.
Analysis of circulating immunosuppressive cells by flow cytometry
To discriminate live from dead cells, PBMCs were first washed in 1× PBS (Gibco) and stained with Fixable viability dye (FvD)-eFluor 506 according to the manufacturer's instructions. For MDSC analysis, 10 6 cells were surface-stained in the dark for 30 min at 4°C with a mixture of the following antibodies: PerCP-Cy5.5 antihuman HLA-DR, BV421 anti-human CD14, APC antihuman CD33, and PE-Cy7 anti-human CD11b. Lineage cocktail (Lin-) was composed of anti-human CD19 APC Alexa Fluor 750, CD56 APC Alexa Fluor 750, and CD3 APC Alexa Fluor 750. The following isotype controls were used for anti-CD11b: PE-Cy7 mouse IgG1, and for anti-CD33: APC mouse IgG1.
All antibodies used are referenced in Supplementary Table S1. The stained samples were acquired on a FACS CantoII cytometer and analyzed with Diva software (Franklin Lakes, NJ, USA). Around 100,000 events in viable cells were measured for each sample. According to our previous works, an increase or decrease of 20% of Treg or MDSC rate after treatment was considered as significant [34,35,51].
Statistical analysis
Data are presented using mean values +/− standard deviation (SD). Statistical comparison between groups was based on Wilcoxon test using Prism 6 GraphPad Software. Survival curves were calculated with the Kaplan-Meier method. A p value ≤0.05 was used as the cutoff for significance.
==
Domain: Biology Medicine
|
[{"added": "2023-01-15T14:40:47.550Z", "created": "2021-06-18T00:00:00.000", "id": "255813267", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1186/s12865-021-00429-5", "pdf_hash": "ae479779b984174194f7da3326a2e26c51dd4676", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:7474", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "ae479779b984174194f7da3326a2e26c51dd4676", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
YRNA expression predicts survival in bladder cancer patients
Non-coding RNAs play an important role in human carcinogenesis. YRNAs (Ro-associated Y), a novel class of non-coding RNAs, have been identified as biomarker in various malignancies, but remain to be studied in urinary bladder cancer (BCA) patients. The expression of all four YRNAs (RNY1, RNY3, RNY4, RNY5) was determined in archival BCA (urothelial carcinoma, n = 88) and normal urothelial bladder (n = 30) tissues using quantitative real-time PCR. Associations with clinicopathological parameters and prognostic role for overall and cancer-specific survival were analysed. All YRNAs were significantly downregulated in BCA tissue. A low expression of RNY1, RNY3 and RNY4 was associated with muscle-invasive BCA, lymph node metastases and advanced grade. Furthermore, expression of RNY1 and RNY3 was predictive for BCA patients’ overall (also RNY4) and cancer-specific survival as estimated using Kaplan-Meier and univariate (but not multivariate) Cox regression analyses. RNY1, RNY3 and RNY4 show good discriminative ability between tumor and normal tissue, as well as between muscle-invasive and non-muscle-invasive urothelial carcinoma. The expression of YRNAs is altered in BCA and associated with poor prognosis. Possible diagnostic role of YRNAs should be investigated in further studies.
Background
Urinary bladder cancer (BCA) is among the most common malignancies worldwide; approximately 430.000 new cases and 165.000 deaths were estimated for 2012 [1]. An important step in BCA progression is the invasion of the detrusor muscle and metastatic spread. BCA symptoms are sometimes non-specific leading to delayed diagnoses at an invasive stage, which is accompanied with an unfavorable outcome. To improve the therapeutic management a better understanding of the molecular biology of BCA is necessary.
The vast majority of the human genome (98%) consists of non-coding genes [2]. Non-coding RNA (ncRNAs) do not encode proteins, but have a putative regulative func-tion of gene expression. The ncRNAs are classified according to their size in nucleotides (nt) into small-ncRNAs (sncRNA <200 nt) and long-ncRNAs (lncRNA >200 nt) [3]. Much effort has been spent to identify and functionally characterize dysregulated microRNAs [4,5] and lncRNAs [6] in BCA in the past years, but few is known about other subtypes of the ncRNAs. YRNAs (Ro-associated Y) were recognized as a component of soluble ribonucleoproteins (Ro RNPS) in the blood of patients with rheumatic autoimmune diseases [7]. Nowadays, four highly conserved human YRNAs (RNY1, RNY3, RNY4, and RNY5) are known. YRNAs have a size of 80-110 nt and a stem-loop structure due to their complementary 5′ and 3′ ends [8]. They are functionally relevant for DNA replication [9] and Ro60 inhibition [10]. YRNAs are overexpressed in various cancer cells [11], and RNY1 and RNY3 inhibition was shown to decrease cell proliferation [11,12]. YRNA-derived fragments are involved in caspase-3-dependent cell death and NF-κB-dependent inflammation and may have an inflammatory role [13]. It was also shown that RNY5 fragments in extracellular vesicles trigger cell death, and thereby may help cancer cells to optimize the microenvironment for proliferation and invasion [14]. YRNAs have not been investigated in a large cohort of BCA so far; we therefore studied the expression profile of YRNAs in BCA and normal urothelial tissue.
Patients
Formalin-fixed, paraffin embedded (FFPE) bladder tissues were randomly selected from the archive of the Institute of pathology at the University Hospital Bonn from patients (n = 112) who underwent transurethral resection of the bladder (TURB) or radical cystectomy for BCA from 1990 until 2009. Follow-up information was available for all patients; median follow-up time was 51 months (range 1-210). The detailed clinicopathological parameters are reported in Table 1.
Ethics, consent and permissions
All patients gave written informed consent for the collection of biomaterials within the framework of the Biobank at the University Hospital Bonn. The study was approved by the ethic committee (280/12) at the University Hospital Bonn.
Tissue samples acquisition
A first 5 μm thick section from the FFPE block was stained with haematoxylin and eosin and used for histological control and mapping of the block content. BCA (n = 88) and normal urothelial tissue (n = 30) samples were then macrodissected using a scalpel from five consecutive 20 μm sections of the block. The absence of significant inflammation as well as absence of any signs of dysplasia/atypia was ensured morphologically in normal tissues. Some samples with normal urothelial tissue stemmed from patients with BCA, when spatial divergence of these samples could be guaranteed and carcinoma in situ did not coexist.
RNA isolation and quantitative real-time PCR
The RNA was isolated using the Recover All Total Nucleic Acid Isolation Kit (Ambion, Foster City, CA, USA) according to the suppliers recommendations. Afterwards, the DNA-free DNA Removal Kit (Ambion) was used to digest DNA contaminants. RNA purity and concentration were determined using the NanoDrop 2000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA). The isolated total RNA was stored at −80°C until further use.
Finally, the relevance of YRNAs for patients' prognosis was determined using Kaplan Meier estimates. Expression of YRNA was significantly correlated with BCA patients' overall (RNY1, RNY3, RNY4) and cancerspecific (RNY1, RNY3) survival (all log rank p < 0.05; Fig. 2). We also performed univariate and multivariate Cox regression analyses: RNY1 and RNY3 were significantly predictive for cancer-specific and overall survival (all p < 0.05), but lost their predictive value in a multivariate model (see Tables 3 and 4 for details).
Discussion
YRNAs have been identified as novel non-coding class of RNA molecules which may be used as biomarker for cancer [17][18][19]. So far, little information exists about the expression of YRNA in BCA patients. In 2008, Christov et al. demonstrated an increase of YRNA expression [11]. However, his study cohort was very small (n = 4) and thereby limiting any meaningful statistical conclusion. Thus, we investigated the expression of all four YRNAs in an enlarged cohort of BCA patients to allow a robust statistical analysis. Interestingly, YRNA expression levels were significantly downregulated in our dataset, mean expression levels in BCA tissue were 2-to 4-fold lower than in normal tissue. It should be noted that Christov et al. [11] normalized the YRNA expression to the mRNA HPRT1, whereas our study used RNU6-2 and SNORD43; RNU6-2 and SNORD43 were earlier shown to be useful reference genes for the analysis of BCA samples [16]. Notably, as expected from the experiments of Christov et al. [11], RNY1, RNY3 and RNY4 expression was highly correlated, whereas the degree of correlation of RNY5 to the other YRNAs was less pronounced. The analysis of 88 tissue samples with urothelial carcinoma allowed as to correlate the expression of YRNAs with clinicopathological parameters. RNY1, RNY3 and RNY4 expression was associated with advanced stage (muscle invasive BCA, lymph node metastasis) and grade (G3 tumors when opposed to G1 and G2). Importantly, the expression levels of RNY1 and RNY3 were significantly predictive for cancer-specific and overall survival of BCA patients with a clear trend for RNY4. However, the strong correlation of YRNA expression and muscle-invasiveness of the tumor impaired achieving an independency in the multivariate Cox regression analyses within a cohort of 88 BCA patients.
Although YRNA are non-coding RNAs, they are also of functional relevance and do not represent transcriptory garbage. YRNAs are essential factors for chromosomal DNA replication [9], whereby they execute their function during the initiation of DNA replication [20]. siRNA mediated knock-down of RNY1 and RNY3 reduced the number proportion of S phase cells in the HeLa cells; degradation of RNY3 reduced also the number S-phase cells in EJ30 bladder cancer cells. Furthermore, the mitotic index and the cell density was reduced after treatment with RNY3 siRNAs [11]. Within this context it is interesting that we have observed a decrease of YRNA levels in BCA patients. Seemingly, Similar trends with decreased abundance of several YRNAs in tumor patients (in serum) were observed in other tumor types (head and neck squamous cell carcinomas [18], breast cancer [19]), which support our findings, even given the fact that conclusions from cell line studies are suggesting the upregulation could be associated with tumor growth and proliferation This could be related, from one side, to different YRNA effects in different primary tumors and, from the other side, to the artificial construct and well known limitations of cell cultures. Also, the effect of YRNA overexpression was not studied in the above mentioned cell culture study [11]. The biological functions of the YRNA are still understudied and could be multidirectional. Some studies show that YRNAs demonstrating decreased levels during mitosis and high levels during S and G2 phases of the cell cycle, partially through association with chromatin [21]. This may be a possible explanation for decreased expression in highly proliferating tumor tissues. Many microRNAs are known for "managing" the cell fate and cell proliferation through interactions with p53 and other members of p53-family [22], which is highly deregulated in tumor Fig. 1 The expression of YRNAs (ΔCq Expression) was determined in a cohort of 30 normal urothelial (CTRL) and 88 bladder cancer (BCA) tissue samples. All YRNAs were significantly downregulated in BCA (all p < 0.001, see Table 2). a Receiver operator characteristic (ROC) analysis for YRNAexpression to discriminate between normal (CTRL) and tumor (BCA) tissue. b ROC-analysis for YRNA-expression to discriminate between muscleinvasive (MIBC) and non-muscle invasive bladder cancer (NMIBC). c-f RNY1-, RNY3-, RNY4-and RNY5-expression in normal, MIBC and NMIBC tissue samples (p-level < 0.001 for RNY1, RNY3 and RNY4; p-level = 0.739 for RNY5). Short horizontal red line with number = Expression median tissue of patients with urinary bladder cancer compared to normal tissue [23]. The interactions with this pathway were not studied for YRNA to date and would probably also provide the explanations for aberrant YRNA expression. It may also be speculated that YRNAs are secreted by BCA cells to act as mediator of immunoescape: extracellular YRNAs fragments activate TLR7 to promote apoptosis in macrophages and monocytes [13].
YRNAs are expected to be a suitable non-invasive biomarker because approximately 25 to 33 nt large YRNA fragments have been identified using small RNA sequencing in human serum and plasma [17]. It was further shown that changes of specific YRNA fragments in serum are associated with ER-negative breast cancer [19]. Similarly, specific YRNA fragments were also circulating at altered levels in head and neck cancer patients [18]. However, specific identification of these small (25 to 33 nt size) fragments implies application of small RNA sequencing procedures and is therefore at least today not suited for daily routine. In our study we were able for the first time to show that RNY1, Fig. 2 Kaplan-Meier curves and log-rank test for YRNAs expression dichotomized based on the best cut-off, in each case separately for cancerspecific and overall survival, respectively: a/b RNY1, c/d RNY3, e/f RNY4, and g/h RNY5. Kaplan Meier estimates indicate that expression of YRNAs is statistically significant prognostic for cancer-specific survival (RNY1, RNY3) and overall survival (RNY1, RNY3, RNY4) in BCA patients (all log-rank p > 0.05). Abbreviations: OSoverall survival, CSScancer-specific survival RNY3 and RNY4 could very good discriminate between the normal and tumor tissue with a maximal AUC of 0.863 (RNY3), and to a lesser extent between muscle-invasive and not-muscle-invasive tumors (maximal AUC 0.780 for RNY3). These findings could support the diagnostic value of YRNA, which certainly warrants further investigations.
Some limitations of our study should be acknowledged: The RNA integrity was not determined after RNA isolation. Formalin-fixed, paraffin embedded tissues are usually degraded to approximately >200-400 bp sized RNA fragments [24,25], and thus amplifying PCR products of approximately 100 bp size is feasible. We randomly picked samples obtained over a period of approximately 20 years and long-term storage may alter the RNA integrity [26]. However, relative YRNA expression levels were not correlated with the year of surgery (data not shown). The normal urothelial tissue samples were in several patients derived from patients with BCA, however necessary precautions were undertaken to prevent contamination of normal samples with tumor tissue (see Materials and methods). Although even in this case we cannot exclude molecular alterations occurred in the microscopically normal urothelium. The tissue was macrodissected with a scalpel, thus RNA some degree of inevitable contamination with other cells like inflammatory, stromal or endothelial cells could have affected the YRNA expression studies.
Conclusions
The expression of all four YRNAs is downregulated in tumor tissue in patients with urinary bladder urothelial carcinoma. Expression changes are associated with advanced disease, higher grade and metastatic disease and may have prognostic relevance for cancer-specific and overall survival.
==
Domain: Biology Medicine
|
[{"added": "2023-01-16T14:18:28.811Z", "created": "2017-11-10T00:00:00.000", "id": "255828386", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://bmccancer.biomedcentral.com/track/pdf/10.1186/s12885-017-3746-y", "pdf_hash": "2317ceab6a32bb48f4e9d343ff116243c4f3d4d5", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:8343", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "2317ceab6a32bb48f4e9d343ff116243c4f3d4d5", "year": 2017}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Dicarbonyl stress in clinical obesity
The glyoxalase system in the cytoplasm of cells provides the primary defence against glycation by methylglyoxal catalysing its metabolism to D-lactate. Methylglyoxal is the precursor of the major quantitative advanced glycation endproducts in physiological systems - arginine-derived hydroimidazolones and deoxyguanosine-derived imidazopurinones. Glyoxalase 1 of the glyoxalase system was linked to anthropometric measurements of obesity in human subjects and to body weight in strains of mice. Recent conference reports described increased weight gain on high fat diet-fed mouse with lifelong deficiency of glyoxalase 1 deficiency, compared to wild-type controls, and decreased weight gain in glyoxalase 1-overexpressing transgenic mice, suggesting a functional role of glyoxalase 1 and dicarbonyl stress in obesity. Increased methylglyoxal, dicarbonyl stress, in white adipose tissue and liver may be a mediator of obesity and insulin resistance and thereby a risk factor for development of type 2 diabetes and non-alcoholic fatty liver disease. Increased methylglyoxal formation from glyceroneogenesis on adipose tissue and liver and decreased glyoxalase 1 activity in obesity likely drives dicarbonyl stress in white adipose tissue increasing the dicarbonyl proteome and related dysfunction. The clinical significance will likely emerge from on-going clinical evaluation of inducers of glyoxalase 1 expression in overweight and obese subjects. Increased transcapillary escape rate of albumin and increased total body interstitial fluid volume in obesity likely makes levels of glycation of plasma protein unreliable indicators of glycation status in obesity as there is a shift of albumin dwell time from plasma to interstitial fluid, which decreases overall glycation for a given glycemic exposure.
Abstract
The glyoxalase system in the cytoplasm of cells provides the primary defence against glycation by methylglyoxal catalysing its metabolism to D-lactate. Methylglyoxal is the precursor of the major quantitative advanced glycation endproducts in physiological systems -arginine-derived hydroimidazolones and deoxyguanosine-derived imidazopurinones. Glyoxalase 1 of the glyoxalase system was linked to anthropometric measurements of obesity in human subjects and to body weight in strains of mice. Recent conference reports described increased weight gain on high fat dietfed mouse with lifelong deficiency of glyoxalase 1 deficiency, compared to wild-type controls, and decreased weight gain in glyoxalase 1-overexpressing transgenic mice, suggesting a functional role of glyoxalase 1 and dicarbonyl stress in obesity. Increased methylglyoxal, dicarbonyl stress, in white adipose tissue and liver may be a mediator of obesity and insulin resistance and thereby a risk factor for development of type 2 diabetes and non-alcoholic fatty liver disease. Increased methylglyoxal formation from glyceroneogenesis on adipose tissue and liver and decreased glyoxalase 1 activity in obesity likely drives dicarbonyl stress in white adipose tissue increasing the dicarbonyl proteome and related dysfunction. The clinical significance will likely emerge from on-going clinical evaluation of inducers of glyoxalase 1 expression in overweight and obese subjects. Increased transcapillary escape rate of albumin and increased total body interstitial fluid volume in obesity likely makes levels of glycation of plasma protein unreliable indicators of glycation status in obesity as there is a shift of albumin dwell time from plasma to interstitial fluid, which decreases overall glycation for a given glycemic exposure.
Keywords Methylglyoxal . Glycation The obesity epidemic and related complications In the last 30 years the causes of global premature death and loss of productive life, as assessed in disability adjusted life years reported by the World Health Organization, has changed from communicable diseases in children towards noncommunicable diseases in adults. Impaired metabolic health development of insulin resistance leading to type 2 diabetes (T2DM) is now a leading cause of disability-adjusted life years. A major risk factor for this is being overweight and obese (body mass index >25 kg/m 2 ) in which insulin resistance is a common feature and contributor to development of T2DM [1]. The number of people who are overweight and obese worldwide is now >2.1 billion, a prevalence of ca. 37 % in the adult population and 13 % in adolescents and children in developing countries and 23 % in developed countries [2]. In 2010, overweight and obesity were estimated to cause 3.4 million deaths worldwide. Most deaths attributable to overweight and obesity are cardiovascular deaths; 36 % of the increased risk of coronary heart disease (CHD) mortality and 59 % of the increased risk of stroke mortality risk associated with obesity was linked to increased blood pressure and cholesterol, 14 % and 25 % of CHD and stroke excess mortality were directly linked to glucose and the remaining excess risk is unexplained [3]. A further complication of overweight and obesity is nonalcoholic fatty liver disease (NAFLD) leading to hepatic steatosis (NASH), cirrhosis, liver failure and cancer [4]. Obesity-induced insulin resistance is a major driver of development of T2DM and NAFLD. NAFLD affects 20-40 % of the population in Westernised countries and will likely increase direct and indirect medical costs by 25 % in the next 5 years. There is an urgent requirement for improved understanding of the risk factors of insulin resistance, obesity and NAFLD and to guide interventions to decrease incidence and health impact.
Obesity and the glyoxalase system Recent reviews have described the impact of obesity on the glyoxalase system [5][6][7][8]. Herein consideration of the subject is advanced with description of recent experimental studies and findings with discussion of new underlying concepts, interpretation and inferences.
Genetic factors, dietary factors and early-life nutrition influence risk of insulin resistance and obesity. A gene functionally linked to obesity and diabetes is the glyoxalase 1 (Glo-1) gene, GLO-1 [9]. Glo-1 is part of the cytosolic glyoxalase system present in the cytoplasm of all mammalian cells - Fig. 1a. The glyoxalase system catalyses the metabolism of methylglyoxal (MG) to D-lactate via the intermediate S-Dlactoylglutathione and thereby suppresses the spontaneous modification of proteins and DNA by MG forming advanced glycation endproducts (AGEs). MG is the major precursor of AGEs in vivo, modifying mainly arginine residues in proteins to form hydroimidazolone MG-H1 - Fig. 1b, and deoxyguanosine residues in DNA to form imidazopurinones MGdG [10,11] - Fig. 1c. Accumulation of MG adducts would otherwise cause protein dysfunction and mutagenesis. The function of the glyoxalase system is the enzymatic defence against MG glycation where Glo-1 catalyses the key step of removing the potentially damaging MG. Periods of increased MG concentration are called Bdicarbonyl stress^produced by increased MG formation and/or decrease Glo-1 activity.
In human subjects, GLO-1 was linked to anthropometric measurements of obesity -upper-arm circumference and supra-iliac skinfold thickness [12]. In mice, meta-analysis of 34 mouse cross-breeding experiments linked GLO-1 to body weight [13]. Mice with a preference for a high energy-rich diet without marked health impairment have a relatively high expression of Glo-1 [5]. In the mouse overeating model of obesity, leptin mutant (ob/ob) mice, Glo-1 protein was decreased 80 % in the liver [14]. Recent conference reports described increased weight gain on high fat diet (HFD)-fed mouse with through-life expression of GLO-1 siRNA and mild Glo-1 deficiency, compared to wild-type controls [15], and decreased weight gain in Glo-1 overexpressing transgenic mice [16], suggesting a functional role of Glo-1 and dicarbonyl stress in obesity. We found increased MG concentration in hepatocyte-like hepatoma G2 cells in vitro incubated with saturated fatty acid and mono-unsaturated fatty acid, palmitic acid and oleic acid, respectively, suggesting that fatty acid metabolism may drive increased MG formation [17] see below. HFD-fed wild-type mice had increased MG-H1 content of heart and liver, as judged by immunoassay [18]. Dicarbonyl stress may be a mediator of obesity and insulin resistance and thereby a risk factor for development of T2DM and NAFLD. Moreover, in a mouse model of hepatocellular carcinoma, Glo-1 was a tumour suppressor protein [19]. Hence, decrease of Glo-1 activity and hepatic dicarbonyl stress in NAFLD with progression to NASH may also increase risk of hepatocellular carcinoma.
Obesity and dicarbonyl stress
Several studies have attempted to model dicarbonyl stress in obesity by administration of exogenous MG. Difficulties performing such studies are: (i) lack of commercial availability of suitable high purity MG, (ii) interference-free assay of MG, and (iii) and judgement of an appropriate dose to administer. Commercial 40 % MG contains 9-17 mol% formaldehyde and many other compounds that potentially interfere in studies of dicarbonyl stress [20]. Methods for preparation of high purity MG and interference-free assay of MG have been described [21,22]. The flux of endogenous formation of MG has been estimated at ca. 3-6 mg/kg (ca. 0.05 % glycolytic rate, which we find relatively constant in many cell types) [23]. Experimental studies have often used 10-20 fold higher than thiswhich is likely similar to and exceeds the upper limit of severe dicarbonyl stress of poorlycontrolled clinical diabetes and end stage renal disease [24,25]. MG formation of cells with GLUT1 glucose transport increased only 2-3 fold in the high glucose concentration characteristic of T2DM and MG concentration in blood of patients with T2DM showed a similar 2-3 fold increase [24,26].
Infusion of MG (60 mg/kg/day) into healthy rats induced impaired glucose tolerance, decreased glucose transporter GLUT-4, phosphoinositide-3-kinase activity, and insulinstimulated glucose uptake in adipose tissue [27]. Administration of exogenous MG (50-75 mg/kg, daily, i.p.) induced insulin resistance in mice [28], inhibited insulinstimulated phosphorylation of protein kinase B and extracellularly-regulated kinase, contributing to insulin resistance in muscle cells [29]. It also inhibited insulin-induced insulin receptor substrate tyrosine phosphorylation and phosphatidylinositol 3-kinase/protein kinase B pathway activation in pancreatic beta-cells [30], increased free fatty acids, hypoadiponectinemia, hypoxia and macrophage recruitment of adipose tissue [31]. These levels of MG exposure also arrested growth of rats, impaired renal function, induced hypercholesterolaemia and impaired vasodilation. There were also degenerative changes in cutaneous capillaries with loss of endothelial cells, basement membrane thickening, luminal occlusion and inflammatory responseincreased receptor for AGE (RAGE), interleukin-1ß, tumour necrosis factor-α and connective tissue growth factor in medial layers of arteries, and transforming growth factor-ß in glomerular tufts, tubular epithelial cells and interstitial endothelial cells [32]. These MG administration models to date, therefore, explore features of MG intoxication. Some of the features produced may be similar to those developing in obesityalthough they are likely markedly less severe.
Moderate dicarbonyl stress in clinical obesity
To investigate dicarbonyl stress in clinical obesity we recruited obese and non-obese healthy human subjects and placed them on an isocaloric diet for 2 weeks. Blood samples were collected after overnight fasting and plasma prepared. Plasma MG was determined by stable isotopic dilution analysis liquid chromatography-tandem mass spectrometry (LC-MS/MS) [22] and plasma D-lactate concentration by endpoint enzymatic assay [33]. Plasma MG was increased 35 % in obese subjects and D-lactate was increased ca. 2-fold - Table 1. Plasma MG levels were intermediate between those found in nonobese healthy subjects and patients with type 1 or type 2 diabetes [22,34,35], suggesting clinical obesity is a state of moderate dicarbonyl stress. Plasma D-lactate levels, a surrogate and qualitative indicator of MG flux, also supported this [24,35,36] and suggest the flux of formation of MG is increased in obesity. (D-Lactate is metabolised in human subjects [37]). The MG-H1 residue content of plasma protein was determined in obese and non-obese subjects by exhaustive enzymatic hydrolysis of plasma protein and MG-H1 detection and quantitation by and stable isotopic dilution analysis LC-MS/MS [38] and no significant difference was found. Estimates of MG and in white adipose tissue (WAT) in experimental models of overfeeding, HFD fed mice, were ca. 5000 nmol per g tissue versus ca. 2000 nmol per g tissue in controlsequivalent to ca. 5 mM and 2 mM MG, respectively [39]. Our estimates of MG in WAT of the same experimental model by the reference protocol [22] are markedly lower (mean ± SD): 2.91 ± 0.98 (n = 7) versus 1.47 ± 0.65 (n = 8, P < 0.01); equivalent to ca. 3 and 1.5 μM, respectively. We also found markedly lower levels of glyoxal in WAT but similar levels of 3-deoxyglucosone (Masania, J., Rabbani, N., Rossmeisl, M., Kopecky, J. and Thornalley, P. J.; unpublished observations). Mathematical modelling of MG formation and protein glycation in mouse tissue predicted MG concentrations in the 1-2 μM range in normal metabolism and are likely increased 2-3 fold in obesity [22]. Hence markedly higher estimates are not sustainable metabolically and may have been caused by interference -MG and glyoxal formation in pre-analytic processing. Trichloroacetic acid deproteinization and azide blocking of acid-stable peroxidase is required to avoid overestimation of MG and glyoxal in mouse tissues [40]. This requires further investigation.
Plasma protein glycation adducts as reporters of glycation exposure in obesity Glycated albumin In clinical obesity as insulin resistance develops, there is progressive decline in glycemic control with impaired fasting and postprandial hyperglycemiaas detected by continuous glucose monitoring. There is a moderate increase in glycated hemoglobin (A1C) as prediabetes develops [41] but surprisingly there is often a decrease in glycated albumin with increased BMI although A1C is increased. This is found both in obese adults and children [42,43] and lower than expected glycated albumin in obesity also extends into the patients with T2DM [43]. The steady-state level of albumin glycation in plasma depends on the increased glucose concentration in plasma and duration over which it occurs, and also the residence time of albumin in the plasma compartmentas judged by the albumin transcapillary escape rate (TER). Until degraded with a half-life of ca. 20 days [44], albumin cycles from plasma into interstitial fluid, lymph and returns to plasmawith some leakage through renal glomeruli and return to venous circulation by the renal albumin retrieval pathway [45]. The rate of glycation by glucose normally is 4-fold higher in the plasma compared to interstitial fluid so that decrease in dwell time in plasma by increased TER may decrease glycation without change in plasma glucose concentration [45,46] - Fig. 2. The rate of glycation of albumin by glucose, r Glycation , is directly proportional to the concentration of glucose and the concentration of albumin; r Glycation = k Glycation [Glucose][Albumin], where k Glycation is the rate constant for glycation of albumin by glucose. The concentrations of albumin and glucose are 2.7-fold and 1.4 fold higher in plasma than interstitial fluid, which multiplied together indicate that the rate of glycation by glucose normally is 3.8 fold or ca. 4fold higher in the plasma compared to interstitial fluid. [45,46]. Relative glycation kinetics deduced from: r Glycation = k [Glucose][albumin], k is the glycation rate constant and assuming r Glycation in the plasma compartment =100 % Albumin TER is increased by hypertension (from 5.6 % per hour to 7.6 % per h) and plasma volume decreases by up to 10 % [47]. It is also increased in overweight/obese subjects with metabolic syndrome [48]. Obesity also increases the total body interstitial fluid volume [49]. This suggests that the explanation for decreased plasma glycated albumin in obesity with increased glucose exposure, as indicated by decreased glycated albumin/A1C ratio, may be due to a shift of albumin residence time from plasma to interstitial fluid in favour of the latter. This suggests glycated albumin is an unreliable marker of glycemic control in obesity.
The rate of degradation of albumin in obesity and diabetes may also influence the level of glycated albumin if changed. In obese subjects and patients with diabetes, with normal renal function, the synthesis of albumin and plasma concentration of albumin are unchanged from those of lean healthy controls, suggesting that the rate of degradation of albumin is also normal [50,51]. In experimental diabetes early studies of Baynes and co-workers and others showed that the rate of degradation of albumin was not increased but slightly lower in streptozotocin-induced diabetic rats than in healthy controls [52,53]. So there is no evidence that an increased rate of albumin degradation that contributes to decreased glycated albumin or AGE-modified albumin in obesity and diabetes. This may change, however, with impaired renal function with increased urinary loss of albumin and compensatory increased albumin synthesis [54].
AGEs in plasma/serum protein in obesity
A shift of albumin residence time from plasma to interstitial fluid may also influence AGEs. Nε-Carboxymethyl-lysine (CML) residue content of serum protein was inversely linked to BMI and body fat mass [55,56]. In studies where plasma early glycated (glycated albumin), CML and florescent AGE content were determined, all were decreased in obese subjects compared to controls [57]. Decrease of CML residue content of plasma protein in obesity was confirmed in an independent study and association with central obesity and inflammation [58]. Decreased CML residue content of plasma protein in obesity has been explained as due to enlargement of adipose tissue mass in obesity [59] and a contributory feature to this is likely decreased glycation of albumin due to the shift of albumin from plasma to interstitial fluid. An association with inflammation is also expected as albumin TER increases with increased capillary permeability in vascular inflammation [60]. Decreased residence time of albumin in the vascular lumen in obese subjects may also explain herein how plasma MG is increased without increase in MG-H1 residue content of plasma protein in our study - Table 1. This is supported by a recent study with a Glo-1 inducer where plasma MG concentration and whole body endogenous formation of MG-H1 adduct flux was decreased without decrease in plasma protein MG-H1 residue content. This was attributed to improved vascular function, decreased albumin TER and increased plasma dwell time of albumin with Glo-1 inducer intervention [61]. Given this, it is also likely that plasma protein AGE content in obesity is not a reliable indicator of AGE tissue exposureas found previously [59].
Source of dicarbonyl stress in obesity and its likely effects
Our finding of increased plasma MG and D-lactate concentrations in obese human subjects compared to non-obese subjects on an isocaloric diet suggests the source of dicarbonyl stress in obesity is not of dietary origin. Indeed, recent studies of metabolic transit of MG indicate that dietary MG is metabolised and/or reacts with protein in the intestinal lumen and has limited bioavailability [62]. Moreover the insulin resistance in obesity suggests the usually dominant source of MG formation from increased flux through anaerobic glycolysis may not be increased. Triosephosphates, however, are not limited to intermediates of anaerobic glycolysis but are also intermediates of glyceroneogenesis and gluconeogenesis. Increased glyceroneogenesis is associated with adipocyte expansion in obesity supporting increased fatty acid esterification for triglyceride deposition [63]. Most triglyceride synthesis involves glyceroneogenesis via triosephosphate intermediates [64]. Glyceroneogenesis is not limited, however, because it uses pyruvate as the carbon source, which has unrestricted entry and metabolism into major cell types experiencing insulin resistancesuch as adipocytes, hepatocytes and skeletal muscle cells. As flux of glyceroneogenesis increases to support triglyceride synthesis, there is proportionate increase in MG formation by related increase in flux of triosephosphate formation. Most triosephosphate intermediates go on to form glycero-l-3-phosphate but with 0.05 % of triosephosphate degrading to MG, when the flux through glyceroneogenesis increases then concomitantly the flux of MG formation also increases. Increased glyceroneogenesis, therefore, is likely a major driver of increased MG formation in obesity - Fig. 3.
There is no inconsistency with previous findings of increased MG from red blood cells in high glucose concentration as therein MG is formed from increased concentrations of triosephosphates driven by increased uptake and metabolism of glucose in anaerobic glycolysis. Red blood cells do not suffer impaired glucose metabolism in insulin resistance. MG is mainly formed non-enzymatically from the same precursors in all cells and tissuestriosephosphatesbut the pathways that sustain triosephosphate concentrations differ and may be multiple in cells and tissues.
In experimental models of obesity there is evidence of decreased Glo-1 activity in visceral adipose tissue [16]. Human Glo-1 protein is a dimer of 42 kDa. It undergoes posttranslational modifications: C139 may form a mixed disulfide with GSH, inhibiting Glo-1 activity in vitro. Glo-1 may be Snitrosylated on C139 and is a substrate for calcium, calmodulindependent protein kinase II, with phosphorylation at T107. There is acetylation at K148 [65] and likely de-acetylation by cytosolic sirtuin-2 [66]. Glo-1 down regulation in obesity may be driven through hypoxia signalling by hypoxia-inducible factor-1α (HIF1α), which down-regulates Glo-1 expression [67]. As adipose tissue expands, interstitial oxygen tension decreases. HIF1α protein is highly enriched in expanding adipocytes as it drives increased adipose tissue vascularization. Adipocyte-specific deletion of HIF1α decreased HFD-induced adipose tissue inflammation and insulin resistance [68]. Increased MG formation by glyceroneogenesis and decreased Glo-1 expression through HIF1α signalling therefore provides the conditions for dicarbonyl stress in obesity.
The consequences of dicarbonyl stress in white adipose tissue (WAT) are unknown but contribute to insulin resistance. Obesity is associated with decreased activity of the insulin sensitising effects of fibroblast growth factor-21 (FGF21) due to down regulation of the FGF21 receptor cofactor β-Klotho [69]. Decreased FGF21 activity also impairs uncoupling protein-1 expression in brown adipose tissue, energy utilisation for thermogenesis [70] and facilitating fat deposition and weight gain. MG-driven protein glycation decreased expression of β-Klotho [71] and thereby likely contributes to insulin resistance. Decreased β-Klotho is also permissible for induction of pro-inflammatory mediators, interleukin-8, monocyte chemotactic protein-1, intracellular adhesion molecule-1 and receptor for AGEs, RAGE [71]. Overexpression of Glo-1 prevented insulin resistance and inflammation in HFD-fed mice, suggesting a functional role of dicarbonyl stress in obesity. This will be tested clinically by evaluation of Glo-1 inducer therapeutics.
In dicarbonyl stress there is increased protein glycation by MG. MG modification is directed to arginine residues often at functional sites of proteins and leads to functional change or inactivation. Protein targets of MG modification are called the dicarbonyl proteome [72]. The extent of protein glycation by MG is usually 1-2 % but low level increases can have profound physiological effect [73]. Examples are: MG modification of mitochondrial proteins, which increases formation of reactive oxygen species and oxidative damage [74]; modification of extracellular matrix proteins produces endothelial cell detachment with exposure of the sub-endothelium, platelet activation and thrombosis [72]; and modification of apolipoprotein B100 of low density lipoprotein (LDL) producing a atherogenic transformation to small, dense LDL [75]. Moreover, MG modification of proteins stimulates their proteolysis, decreasing protein half-life and thereby concentrations of the unmodified protein unless there is compensatory increased transcription -as found for apolipoprotein A1 of high density lipoprotein [76]. In relation to this, increasing endogenous MG by Glo-1 silencing in aortic endothelial cells changed expression of >400 genes [77]. Dicarbonyl stress, therefore, has proteome, dicarbonyl proteome and transcriptome signatures.
Increased MG formation from glyceroneogenesis on adipose tissue and liver and decreased Glo-1 activity in obesity likely drives dicarbonyl stress in WAT increasing the dicarbonyl proteome and related dysfunction. The functional significance of this is indicated by protection from insulin resistance, inflammation and weight gain in Glo-1 transgenic mice in experimental model of over-eating induced obesity. The clinical significance will likely emerge from on-going clinical evaluation of inducers of Glo-1 expression in overweight and obese subjectsfor example, Clinicaltrials.gov; NCT02095873.
Summary
MG metabolism and the glyoxalase system are disturbed in obesity leading to dicarbonyl stress. Functional genomics studies with Glo-1 in the overfeeding model of HFD-fed mice Fig. 3 Increased formation of methylglyoxal in the triglyceride/ free fatty acid cycle. Percentage flux of glyceroneogenesis in triglyceride formation in liver and adipose tissue is from [64] suggest dicarbonyl stress is a risk factor for health impairment and complications of obesity. Likely drivers of dicarbonyl stress in obesity are: increased formation of MG from increased glycerogenesis in triglyceride synthesis and decreased Glo-1 expression and activity through hypoxia and inflammatory signalling. A recent clinical intervention study with a Glo-1 inducer produced a profound improvement of insulin resistance, improved glycemic control and arterial function and decreased vascular inflammation, suggesting that Glo-1 inducer therapeutics may have a future key role in alleviating complications of obesity. Increased albumin TER in obesity associated with increased fat mass, hypertension and inflammation suggest glycated albumin is not a reliable measure of glycemic control and formation of AGEs in obesity.
==
Domain: Biology Chemistry Medicine
|
[{"added": "2022-12-11T15:17:44.428Z", "created": "2016-06-24T00:00:00.000", "id": "254506552", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://link.springer.com/content/pdf/10.1007/s10719-016-9692-0.pdf", "pdf_hash": "6ce15138f70208386f0ff69fbf38f59d7fb0f3e2", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:5958", "s2fieldsofstudy": ["Biology", "Medicine", "Chemistry"], "sha1": "6ce15138f70208386f0ff69fbf38f59d7fb0f3e2", "year": 2016}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Effects of osteoprotegerin from transfection of pcDNA3.1(+)/chOPG on bioactivity of chicken osteoclasts
Background Osteoprotegerin (OPG) has been reported to prevent bone resorption by inhibiting the formation, function, and survival of osteoclasts in a variety of animal models. However, the effects of OPG on bone metabolism in avian species have not been described. The objective of this study was to investigate the effects of chicken OPG (chOPG) expressed in chicken embryo fibroblasts (CEFs) on chicken osteoclast function in vitro. Methods The chOPG sequence containing the open reading frame (ORF) was amplified from chicken embryo frontal bone and inserted into the pcDNA3.1 (+) vector. PcDNA3.1 (+)/chOPG was transiently transfected into CEFs by lipofectamine 2000. Transcription of OPG mRNA and expression of chOPG recombinant protein were detected by reverse transcription polymerase chain reaction (RT-PCR) and indirect immunofluorescence. The level of chOPG recombinant protein was detected by enzyme-linked immunosorbent assay. The suspension of osteoclasts was separated from chicken embryos and divided into three groups (control group, pcDNA3.1 (+) group and pcDNA3.1 (+)/chOPG group). The percentage of osteoclast apoptosis was detected by flow cytometry. The tartrate-resistant acid phosphatase (TRAP) secreted by osteoclasts was measured by the diazol method. The resorbing activity of osteoclasts was evaluated by the area of lacunae on bone flaps and the concentration of calcium in the supernatant liquid of osteoclasts. Results 48 h after transfection, the exogenous OPG gene transcription was detected by RT-PCR. After 72 h, the CEFs transfected from pcDNA3.1 (+)/chOPG displayed green fluorescence and the concentration of chOPG protein was 15.78 ± 0.22 ng/mL. After chicken osteoclasts were cultured for 5 d in a medium containing supernatant from transfected CEFs, the percentage of osteoclast apoptosis was increased significantly, the concentration of TRAP, the area of lacunae on bone flaps and calcium concentration were decreased significantly in the pcDNA3.1(+)/OPG group compared to the control group and the pcDNA3.1 (+) group. Conclusion Constructed pcDNA3.1 (+)/chOPG transfected into CEFs expressed bioactive OPG protein that was able to inhibit osteoclast function.
Background
Osteoporosis in laying hens is a condition that involves a progressive loss of bone resulting in bone fragility and increased risk of fracture. Surveys of laying flocks in Europe have indicated that about 30% of the birds experience one or more bone fractures due to osteoporosis during their lifetime. The high fracture rates show that osteoporosis not only leads to production losses, but also to severe welfare problems in hens [1].
In laying hens, the main types of bone providing structural integrity are cortical and trabecular bone. In addition to these, medullary bone, an extremely labile source for calcium that develops in specific bones of female birds at the onset of sexual maturity, provides a labile source of calcium for shell formation. Bones undergo a constant process of remodelling, which at the cellular level involves a coordinated regulation of osteoblasts and osteoclasts. As hens mature sexually, bone formation of osteoblasts switches from structural bone to medullary bone [2]. In the absence of structural bone formation, continued osteoclastic resorption of structural bone will result in a depletion of structural bone, ultimately leading to osteoporosis.
The differentiation and function of osteoclasts are regulated by soluble cytokines from osteoblasts, such as osteoprotegerin (OPG) and the receptor activator of nuclear factor ligand (RANKL; also called OPG ligand) [3]. OPG is a soluble decoy receptor that inhibits osteoclast formation, function, and survival by preventing the binding of RANKL to the receptor activator of nuclear factor B (RANK), a membrane-bound protein that is found on chondrocytes, dendritic cells, osteoclast precursors, and mature osteoclasts [4]. Many cytokines and effectors are known to influence the osteoclastic bone resorption via the OPG/RANK/RANKL trio of proteins [5,6]. Changes of expression levels of OPG/RANK/ RANKL would be expected to cause bone disorders such as postmenopausal osteoporosis, glucocorticoidinduced osteoporosis, and sporadic Paget's disease in man [7].
Although the importance of OPG in the osteoclastogenesis has been established in mammalian models, it is not yet clear how OPG regulates the function of osteoclasts in avian species. To elucidate the function of OPG in laying hens in vivo, we amplified the open reading frame (ORF) of the chicken OPG (chOPG) sequence, constructed the pcDNA3.1 (+)/chOPG plasmid and transiently transfected it into chicken embryo fibroblasts (CEFs). We tested whether pcDNA3.1 (+)/chOPG expressed OPG protein at a level able to inhibit the biological activity of osteoclasts in vitro.
Cloning of the ORF of chOPG
Total RNA was extracted from chicken embryonic frontal bone (Animal Husbandry Industry Co., Nanjing, China) with TRIzol ® Reagent (Invitrogen, Inc. Carlsbad, CA, USA) according to the manufacturer's instruction. RNA purity was determined by 260 nm and 280 nm absorbance ratios and integrity was checked by 1% agarose/formaldehyde gel electrophoresis. A Biometra DNA Thermal Cycler was used for reverse transcription polymerase chain reaction (RT-PCR). RT-PCR was performed in the presence of DTT, oligo(dT)18, dNTP, RNase inhibitor, first-strand buffer and Moloney murine leukaemia virus reverse transcriptase (TakaRa Bio Inc. Japan). The final mixture was reacted at 42°C for 60 min and at 70°C for 15 min to denature the enzyme. On the basis of the published nucleotide sequence of chOPG (DQ098013), one pair of PCR primers (Invitrogen) were designed. Primers P1 and P2 were used to amplify the ORF of chOPG sequence. Nhe| and Xho| (TakaRa Bio Inc.) restriction sites were inserted into primers P1 and P2, respectively: P1: 5'-CATGCTAGCATGAACAAGTTCCTGTGC-3' (sense strand, positions 10-27 of cDNA sequence); P2: 5'-CCGGCTCGAGTTAGACACATCTTACTTT-3' (antisense strand, positions 1,201-1,218 of cDNA sequence).
PCRmix (TakaRa Bio Inc.), primers P1 and P2, and cDNA were mixed and amplified for 30 cycles under the following conditions: denaturation for 30 s at 94°C, annealing for 45 s at 47°C, and extension for 50 s at 72°C. The products were subsequently sequenced (Invitrogen) after 1% agarose electrophoresis, recovery and purification.
Cell culture and DNA transfection
CEFs were prepared from two 10-days old chicken embryos (Animal Husbandry Industry Co) and were grown according to standard procedures, cultured in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen-Gibco) supplemented with 5% fetal bovine serum (Invitrogen-Gibco). The number of cells was adjusted to 2 × 10 5 cells/ml and incubated in 24-well tissue culture plates (Bo Quan Sci&Tech. Co. Ltd. Nanjing, China) containing cover glass at 37°C in a humid atmosphere of 5% CO 2 for 24 h. Prior to each test, CEFs were washed three times with phosphate buffered solution (PBS), transfected with 1 μg/well pcDNA3.1 (+)/OPG plasmid and pcDNA3.1 (+) vector using 3 μl/well lipofectamine 2000 (Invitrogen), respectively, followed by incubation at 37°C in 5% CO 2 for 48 h and 72 h. The culture medium was renewed every 2nd day.
RT-PCR analysis of chOPG mRNA
The cells (both floating and adherent cells) were harvested 48 h post transfection. The total RNA was extracted with TRIzol ® Reagent according to the manufacturer's instruction. RNA samples were then treated with DNase I (1 U/μg) (TakaRa Bio Inc.) before the RT step to avoid the interference with contaminating genomic DNA. P5 (5'-ATGAACAAGTTCCTGTGC-3') and P6 (5'-TTAGACACATCTTACTTT-3') were subjected to PCR using upstream and downstream primers.
Immunocytochemical analysis of chOPG product in CEFs
CEFs (2 × 10 5 ) cultured for 72 h on glass coverslips (6 mm × 6 mm) were replated into 24-well plates. Glass coverslips were washed with 0.01 M PBS and fixed in 4% formaldehyde for 45 min. Detergent extraction with 3% Triton X-100 was performed for 10 min. Coverslips were saturated with PBS containing 5% bovine serum albumin (Wuhan Boster Biotechnology Company, China) for 1 h at room temperature with gentle rocking, processed with rabbit anti-chOPG polyclonal antibody (Nanjing Agricultural University) for 1 h at 37°C and followed by FITC-goat-anti-rabbit IgG (Wuhan Boster Biotechnology Company) for 1 h at 37°C and then stained by DAPI staining solution (Wuhan Boster Biotechnology Company). Coverslips were washed by PBS for 30 min prior to each treatment. Finally, coverslips were mounted on slides and fluorescence signals were analyzed by a Fluoview microscopy (Olympus, Japan).
ELISA analysis of chOPG product in supernatant
The concentration of the chOPG product in the supernatant was determined using an enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems, Minneapolis, MN, USA) according to the manufacturer's instructions. The concentration was determined for three wells of each sample by measuring the optical density (OD) at 450 nm wavelength by an ELISA reader (Immuno Mini NJ-2300, InterMed, Japan).
Effect of chOPG on osteoclast bioactivity
Tibias and humeri were isolated from 15 18-days old chicken embryos. Osteoclast cultures were prepared as previously described [8]. Briefly, a cell suspension was seeded at a concentration of 2 × 10 5 cells per well in 24well dishes containing either glass coverslips or bovine bone slices (4 mm × 4 mm × 50 μm) (Nanjing Agricultural University). Non-adherent cells were washed off after 2 h. The adherent cells were grown for another 2 d and then cultured in DMEM containing OPG supernatant. The medium was changed every 48 h. Glass coverslips, bovine bone slices and supernatant were harvested after 5 d. The percentage of osteoclast apoptosis was detected by flow cytometry. TRAP secreted to the supernatant by osteoclasts was measured at a OD of 530 nm by the diazol method using TRAP test kit (Bo Quan Sci&Tech. Co. Ltd. Nanjing, China). The resorption lacunae on the bone slice was visualized by toluidine blue staining after removal of osteoclasts using 50 mM NH 4 OH and brief sonication [9]. The concentration of calcium in the supernatant was determined by atomic absorption spectrometry (wavelength 422.7 nm, electric current 3.0 mA, spectrum width 0.4 nm) after 5 times dilution.
Statistical analysis
All values were expressed as means ± the standard deviation (SD). Differences between mean values of normally distributed data were assessed by the one-way ANOVA test and Student's t-test. Statistical difference was accepted at P < 0.05.
Results
Cloning of the ORF of chOPG and construction pcDNA3.1 (+)/chOPG The size of the specific gene fragment amplified was, as expected, about 1.2 kbp ( Figure 1A). The positive clones were identified by PCR amplification and the double restriction digestion with Nhe| and Xho| ( Figure 1B and Figure 1C). Analysis of the PCR products by agarose gel electrophoresis showed that both constructs contained a DNA insert of the correct size and in the correct orientation. The result of sequencing showed that it had 100% homology with that reported in GenBank (DQ098013) indicating that the OPG gene has an extensive hereditary conservation and that no mutations were present in this region of the vector.
Immunofluorescence studies showed that chOPG protein was distributed in the cytoplasma and CEFs showing green fluorescence were observed in the pcDNA3.1 (+)/chOPG group, but were not present in the other groups (Figure 2A).
In the culture supernatant of the pcDNA3.1 (+)/chOPG group transfected from CEFs, the concentration of chOPG was 15.78 ± 0.22 ng/ml, whereas chOPG was not be demonstrated in media from the control group or the pcDNA3.1 (+) group.
Effect of product from transfected CEFs on chicken osteoclast bioactivity in vitro
The morphology of osteoclasts after culturing for 5 d is shown in Figure 2B. Osteoclasts grew well in the control and pcDNA3.1 (+) transfected CEF groups, whereas major nuclei disappeared, many vacuoles and lipid droplet appeared in the cytoplasm and many non-adherent and dead osteoclasts were observed in the culture solution of the pcDNA3.1 (+)/chOPG transfected CEF group. The percentage of osteoclast apoptosis in the control, pcDNA3.1 (+) and pcDNA3.1 (+)/chOPG groups was 10.32%±1.50%, 12.61%±0.95%, 20.59% ±2.83%, respectively ( Figure 2C). TRAP enzyme activity in the pcDNA3.1 (+)/chOPG group was significantly decreased compared to the control group (P < 0.01) ( Figure 3A). An individual resorption event was seen as a dark border of toluidine blue stain surrounding an excavation. The data were recorded for each resorption event separately ( Figure 2D). The quantity and area of lacunae reflected bone resorption by osteoclasts (Table 1). DMEM culture solution did not contain Ca 2+ until after culturing thus suggesting osteoclast activity ( Figure 3B).
Discussion
Bone is an exceedingly complex tissue with multisystemic regulation. Skeletal metabolism depends on the dynamic balance of bone formation by osteoblasts and bone resorption by osteoclasts. The discovery of the OPG/RANKL/RANK system in the mid 1990s has led to major advances in our understanding of how bone modeling and remodeling are regulated [10]. Current research has focused on OPG in humans and mice, while reports on avian OPG are lacking. In our laboratory, chOPG mRNA was extracted from chicken embryo frontal bone. The OPG coding region was successfully amplified and sequence analysis indicated that OPG is highly conserved evolutionary. The sequence reported here had a 68.76%, 68.60% and 68.29% homonology to human, rat and mouse OPG, respectively. The sequence similarity suggests a similar function across species.
Bone is particularly intriguing in laying hens because of the huge demands for calcium for eggshell formation and the occurrence of medullary bone. On the surface of the medullary bone, osteoclasts undergo cyclical functional modifications during the egg-laying cycle [11].
In this study, chOPG induced osteoclast apoptosis after in vitro incubation for 5 d. This result was similar to that reported by Lacey et al. [12], who demonstrated that OPG inhibited bone resorption and induced osteoclast apoptosis though inhibition of F-actin ring formation of mature osteoclasts or altered interaction between stroma cell and osteoclasts.
The results suggest that the secretion of TRAP by osteoclasts was significantly decreased; further demonstrating that recombinant chOPG could inhibit the activity of osteoclasts in vitro. Chamber et al. [13] and Boyde et al. [14] provided evidence for a direct association between the quantity, area and depth of absorption and the capability of osteoclasts to resorption bone. The present study showed that chOPG inhibited osteoclast bone resorption and consequently the concentration of Ca 2+ in the supernatant was significantly reduced. However, the mechanisms by which OPG exerts its biological activity as well as the nature of its molecular interactions with osteoclasts are not well defined. Hakeda et al. [15] reported the first evidence of a direct biological activity of OPG on isolated osteoclasts via a 140 kDa OPG-binding protein. The exact nature of osteoclastic OPG receptors was not further characterized. Direct biological activities of OPG on osteoclasts were recently showed by Wittrant et al. [16] demonstrating OPG enhanced proMMP-9 activity along with several other parameters (TRAP, TIMP, cathepsin K) in purified osteoclasts. Theoleyre et al. [17] showed that OPG stimulates proMMP-9 activity of osteoclasts by the ras/ MAPK pathway involving p38 and ERK1/2 phosphorylations. Moreover, OPG-induced MAPK pathway depends on RANKL. In general, OPG is not only a soluble decoy receptor for RANKL as described in the literature but may be also considered as a direct effector of osteoclast functions.
Conclusions
ChOPG is capable of inhibiting bone resorption as well as promoting osteoclast apoptosis. The study also indicates that pcDNA3.1 (+)/chOPG may be a target for regulating bone metabolism in chicken bone metabolic diseases such as osteoporosis.
==
Domain: Biology Chemistry Medicine
|
[{"added": "2014-10-01T00:00:00.000Z", "created": "2011-03-24T00:00:00.000", "id": "12691893", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Chemistry"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://actavetscand.biomedcentral.com/track/pdf/10.1186/1751-0147-53-21", "pdf_hash": "f5bee835696fff1a0fe47af98e1d914209040d4e", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:4534", "s2fieldsofstudy": ["Biology"], "sha1": "f5bee835696fff1a0fe47af98e1d914209040d4e", "year": 2011}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Dual functions of Aire CARD multimerization in the transcriptional regulation of T cell tolerance
Aggregate-like biomolecular assemblies are emerging as new conformational states with functionality. Aire, a transcription factor essential for central T cell tolerance, forms large aggregate-like assemblies visualized as nuclear foci. Here we demonstrate that Aire utilizes its caspase activation recruitment domain (CARD) to form filamentous homo-multimers in vitro, and this assembly mediates foci formation and transcriptional activity. However, CARD-mediated multimerization also makes Aire susceptible to interaction with promyelocytic leukemia protein (PML) bodies, sites of many nuclear processes including protein quality control of nuclear aggregates. Several loss-of-function Aire mutants, including those causing autoimmune polyendocrine syndrome type-1, form foci with increased PML body association. Directing Aire to PML bodies impairs the transcriptional activity of Aire, while dispersing PML bodies with a viral antagonist restores this activity. Our study thus reveals a new regulatory role of PML bodies in Aire function, and highlights the interplay between nuclear aggregate-like assemblies and PML-mediated protein quality control.
The predicted secondary structure for the mAire CARD (based on the homology model in Fig. 1e) is shown on the top. The indicated residue numbers are based on mAire sequence. Green triangles indicate the locations of mAire mutations used in (b-d) and Fig. 1f-h. Blue diamonds indicate the APS-1 patient mutations analyzed in (e-h) and Fig. 1i, j. Conserved CARD-CARD interfaces identified for hNOD1 and hMAVS CARDs are shown at the bottom. Note that Ia, IIa and IIIa surfaces interact with Ib, IIb and IIIb, respectively. Dotted line for IIIa indicates shifted position in the predicted a-helix in Aire CARD relative to hNOD1 and hMAVS CARD. b. Purification of mAire CARD. Left, schematic of the protein purification protocol; mAire CARD was expressed as a NusA fusion in E. coli and was purified by Ni-NTA affinity purification. 3C protease was used to release CARD from the fusion construct. Right, SDS-PAGE gel of 3C protease-cleaved WT and mutant mAire CARD used in EM imaging of filaments in Fig. 1b, f. Note that the NusA tag alone does not form filaments. c. Transcriptional activity of WT mAire CARD and the mutants in 4D6 cells, as measured by the relative mRNA levels of Aire-dependent genes (represented by CD4, CELF2, IGFL1 and S100A9). The relative mRNA level of an Aire-independent gene, HPRT1, is shown as a negative control. Data are presented as mean ± s.d., n = 3. P-values (two-tailed t-test) were calculated in comparison to WT mAire. * p < 0.05; ** p < 0.01; p > 0.05 is not significant (ns). Exact p-values are provided in the Source Data File. Right, western blot (WB) showing the expression levels of FLAG-tagged mAire. d. Representative fluorescence microscopy images of WT and mAire mutants in 4D6 cells. e. Purification of hAire CARD. Left, schematic of the protein purification protocol. Unlike mAire CARD, hAire CARD fused to NusA-tag yielded very low levels of purified protein.
Successful purification of hAire CARD from E. coli required a minimal N-terminal His6 tag and a refolding step using 6 M guanidine hydrochloride (GdHCl). 3C protease was used to remove the His6 tag. Right, SDS-PAGE gel of 3C protease-cleaved WT and mutant hAire CARD used in (f). f. Representative EM images of WT and hAire CARD with APS-1 patient mutations indicated in (a). g. Transcriptional activity of WT and mutant hAire in 293T cells. Experiments were performed as in (c) and are presented as mean ± s.d., n = 3. P-values (two-tailed t-test) were calculated in comparison to WT hAire. * p < 0.05; ** p < 0.01; p > 0.05 is not significant (ns). Exact pvalues are provided in the Source Data File. h. Representative fluorescence microscopy images of FLAG-tagged hAire in 293T using anti-FLAG.
Supplementary Figure 2: Chemically induced multimerization partially restores the transcriptional activity of AireDCARD.
a. Transcriptional activity of ∆CARD fused with tandem repeats of FKBP (FKBP1-4) in the presence and absence of chemical dimerizer (AP1903) in 293T cells. Experiments were performed as in Fig. 1g and are presented as mean ± s.d., n = 3. Note that KRT14 and S100A9 are Aire-dependent genes. P-values (two-tailed t-test) were calculated in comparison to empty vector (EV) + AP1903. * p < 0.05; ** p < 0.01. a. Examples of Aire foci that have no overlap, partial overlap, and complete overlap with endogenous PML bodies. We observed that some WT and Sp110-CARD swap foci were not completely overlapping with PML, but rather positioned adjacent to PML bodies. For systematic and quantitative analysis of the degree of co-localization of Aire foci and PML bodies, see (b). b. Fraction of Aire (WT Aire or Sp110-CARD swap) foci co-localized with PML bodies while defining co-localization as having greater than a certain threshold % overlapping area (xaxis). See Methods for details of automated image analysis workflow. Sp110-CARD swap foci are associated with PML bodies significantly more than WT mAire regardless of the choice of threshold. A total of 2022 and 383 Aire foci were examined for WT mAire and Sp110-CARD swap samples, respectively. For quantitation of Aire foci co-localized with PML, we arbitrarily chose the threshold definition of 50% overlap area. c. Representative fluorescence microscopy images of two independent monoclonal 4D6 cells stably expressing WT hAire-FLAG under a doxycycline-inducible promoter. 4D6 cells were induced at 1 µg/ml doxycycline for 24 hrs before immunostaining with anti-FLAG and anti-PML. Top right, quantification of Aire foci co-localized with PML bodies from the two clones of 4D6 cells stably expressing WT hAire-FLAG. Analysis was done as in (b) and shows minimal co-localization of WT hAire with PML bodies, independent of the choice of the threshold definition. Statistical significance comparison between WT hAire-FLAG foci overlapping with PML (50% threshold) from the two 4D6 clones was calculated using a twotailed Student's t-test (see Methods for details) and determined to be not significant (p = 0.9118). A total of 371 and 450 Aire foci were examined for WT clones A and B, respectively. Bottom right, WB analysis of WT hAire-FLAG expression in the presence of 0.2 µg/ml (+) or 1 µg/ml (++) of doxycycline.
Supplementary Figure 4: Effect of SIM-mAire, IE1 and MG132 on Aire foci localization.
a. Transcriptional activity of WT mAire (black circle) with and without co-expression of SIM-mAire (green circle) in 293T cells. Each circle represents 0.6 µg/ml DNA transfected. Experiments were performed as in Fig. 4d and presented as mean ± s.d., n = 3. Experiments were performed as in Fig. 1g and are presented as mean ± s.d., n = 3. P-values (two-tailed ttest) were calculated in comparison to WT mAire. ** p = 0.0015 and * p = 0.016. Note that KRT14 and S100A9 are Aire-dependent target genes.
b. Representative fluorescence microscopy images of SIMΔCARD-FLAG with or without coexpression of WT mAire-HA in 4D6 cells. c. Representative fluorescence microscopy images of IE1-HA in 4D6 cells. Cells that express IE1 show diffuse nuclear staining of endogenous PML, whereas cells with no IE1 expression have intact PML bodies. d. WB showing the expression levels of the proteins used in Fig. 4e, f. e. SUMO modification analysis of FLAG-tagged WT mAire and ΔCARD. mAire (0.6 µg/ml DNA) was co-expressed with HA-SUMO1 or -SUMO2 (0.6 µg/ml DNA) in 293T cells and the immunoprecipitation was performed as in Fig. 3c. f. Representative fluorescence microscopy images of FLAG-tagged WT mAire in the presence and absence of MG132 (10 µM). 4D6 cells were treated with MG132 for 4 hrs before fixation. Right, quantitative analysis of Aire foci co-localized with PML bodies. n = number of Aire foci examined per sample. Statistical significance comparison was calculated using a two-tailed Student's t-test for two population proportions where each population consists of all individual Aire foci examined.*** p = 2.159e-12. g. Transcriptional activity of WT mAire in the presence and absence of MG132 (10 µM). Cells were treated with MG132 for 16 hrs. The fold-change as a result of Aire expression was plotted in the right panel. Experiments were performed as in Fig. 1g and are presented as mean ± s.d., n = 3. Supplementary Fig. 3c. b. Transcriptional activity of hAire WT, C302Y (top panel), and C311Y (bottom panel) stably expressed in 4D6 cells in the presence of 0.2 µg/ml (+) and 1 µg/ml (++) of doxycycline for 48 hrs. Cells were harvested and RT-qPCR was performed as in Fig. 1g. Transcriptional activity was normalized to samples without doxycycline. Data are presented as mean ± s.d., n = 3. Note that IGFL1 and S100A9 are Aire-dependent genes and the Aire-independent gene HPRT1 is shown as a negative control. c. Representative fluorescence microscopy images of FLAG-tagged hAire∆CARD C302Y variant (C302Y∆CARD) in 4D6 cells. Cells were immunostained with anti-FLAG and anti-PML. C302Y∆CARD does not form nuclear foci.
==
Domain: Biology Chemistry Medicine
|
[{"added": "2020-03-05T10:19:36.620Z", "created": "2020-02-27T00:00:00.000", "id": "212854024", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Chemistry"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41467-020-15448-w.pdf", "pdf_hash": "0e1ea11225d1d35da0b6489e6461019207c23fa4", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:5967", "s2fieldsofstudy": ["Biology"], "sha1": "f78dc294fba30d920affe9dc525cdb54b6198ff7", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Positioning of nucleosomes containing γ-H2AX precedes active DNA demethylation and transcription initiation
In addition to nucleosomes, chromatin contains non-histone chromatin-associated proteins, of which the high-mobility group proteins are the most abundant. Chromatin-mediated regulation of transcription involves DNA methylation and histone modifications. However, the order of events and the precise function of high-mobility group proteins during transcription initiation remain unclear. Here we show that high-mobility group AT-hook 2 protein (HMGA2) induces DNA nicks at the transcription start site, which are required by the histone chaperone FACT complex to incorporate nucleosomes containing the histone variant H2A. X. Further, phosphorylation of H2A. X at S139 (γ-H2AX) is required for repair-mediated DNA demethylation and transcription activation. The relevance of these findings is demonstrated within the context of TGFB1 signaling and idiopathic pulmonary fibrosis, suggesting therapies against this lethal disease. Our data support the concept that chromatin opening during transcriptional initiation involves intermediates with DNA breaks that subsequently require DNA repair mechanisms to ensure genome integrity.
I n the eukaryotic cell nucleus, chromatin is the physiological template of all DNA-dependent processes including transcription. The structural and functional units of chromatin are the nucleosomes, each one consisting of~147 bp of genomic DNA wrapped around a core histone octamer, which in turn is built of two H2A-H2B dimers and one (H3-H4) 2 tetramer 1,2 . In addition to canonical histones (H1, H2A, H2B, H3 and H4), there are so called histone variants for all histones except for H4. Histone variants differ from the canonical histones in their amino acid sequence and have specific and fundamental functions that cannot be performed by canonical histones. The canonical histone H2A has a large number of variants, each with defined biochemical and functional properties 3,4 . Here we focus on the histone variant H2AFX (commonly known as H2AX, further referred to as H2A. X), which represents about 2-25% of the cellular H2A pool in mammals 5 . Phosphorylated H2A. X at serine 139 (H2A. XS139ph; commonly known as γ-H2AX, further referred to as pH2A. X) is used as a marker for DNA doublestrand breaks 6 . However, accumulating evidence suggests additional functions of pH2A. X [7][8][9][10] . The histone chaperone FACT (facilitates chromatin transcription) is a heterodimeric complex, consisting of SUPT16 and SSRP1 (Spt16 and Pob3 in yeast) that is responsible for the deposition of H2A/H2B-dimers onto DNA 11,12 . The FACT complex mainly interacts with H2B mediating the deposition of H2A/H2B-dimers containing different H2A variants 13 . Thus, the deposition of H2A. X into chromatin seems to be mediated by the FACT complex 14 .
In addition to nucleosomes, chromatin contains non-histone chromatin-associated proteins, of which the high-mobility group (HMG) proteins are the most abundant. Although HMG proteins do not possess intrinsic transcriptional activity, they are called architectural transcription factors because they modulate the transcription of their target genes by altering the chromatin structure at the promoter and/or enhancers 15 . Here we will focus on HMG AT-hook 2 protein (HMGA2), a member of the HMGA family that mediates transforming growth factor beta 1 (TGFB1, commonly known as TGFβ1) signaling 16 . We have previously shown that HMGA2-induced transcription requires phosphorylation of H2A. X at S139, which in turn is mediated by the protein kinase ataxia telangiectasia mutated (ATM) 10 . Furthermore, we demonstrated the biological relevance of this mechanism of transcriptional initiation within the context of TGFB1 signaling and epithelial-mesenchymal transition (EMT). Interestingly, TGFB signaling has been reported to induce active DNA demethylation with the involvement of thymidine DNA glycosylase (TDG) 17 . Active DNA demethylation also requires GADD45A (growth arrest and DNA damage protein 45 alpha) and TET1 (ten-eleven translocation methylcytosine dioxygenase 1), which sequentially oxidize 5-methylcytosine (5mC) to 5carboxylcytosine (5caC) 18,19 and are cleared through DNA repair mechanisms. In line with these ideas, classical DNA-repair complexes have been linked to DNA demethylation and transcriptional activation 20,21 . On the other hand, DNA double-strand breaks induce ectopic transcription that is essential for repair, supporting a tight mechanistic correlation between transcription, DNA damage, and repair 22 .
TGFB1 signaling and EMT are both playing a crucial role in idiopathic pulmonary fibrosis (IPF). IPF is the most common interstitial lung disease showing a prevalence of 20 new cases per 100,000 persons per year 23,24 . A central event in IPF is the abnormal proliferation and migration of fibroblasts in the alveolar compartment in response to lung injury. IPF patients die within 2 years after diagnosis mostly due to respiratory failure. Current treatments against IPF aim to ameliorate patient symptoms and to delay disease progression 25 . Unfortunately, therapies targeting the causes of or reverting IPF have not yet been developed. Here, we demonstrate that inhibition of the HMGA2-FACT-ATM-pH2A. X axis reduces fibrotic hallmarks in vitro using primary human lung fibroblast (hLF) and ex vivo using human precision-cut lung slices (hPCLS), both from control and IPF patients. Our study supports the development of therapeutic approaches against IPF using FACT inhibition.
Results
HMGA2 is required for pH2A. X deposition at transcription start sites. We have previously reported that HMGA2-mediated transcription requires phosphorylation of the histone variant H2A. X at S139, which in turn is catalyzed by the protein kinase ATM 10 . To further dissect this mechanism of transcription initiation, we decided first to determine the effect of Hmga2-knockout (KO) on genome wide levels of pH2A. X. We performed next generation sequencing (NGS) after chromatin immunoprecipitation (ChIPseq; Fig. 1 and Supplementary Fig. 1a) using pH2A. X-specific antibodies and chromatin isolated from mouse embryonic fibroblasts (MEF) from wild-type (WT = Hmga2 + /+) and Hmga2deficient (Hmga2-KO = Hmga2−/−) 26 embryos. The analysis of these ChIP-seq results using the UCSC Known Genes dataset 27 revealed that pH2A. X is specifically enriched at transcription start sites (TSS) of genes in an Hmga2-dependent manner (Fig. 1a), since Hmga2-KO significantly reduced pH2A. X levels. A zoom into the −750 to +750 base pair (bp) region relative to the TSS (Fig. 1b) revealed that pH2A. X levels significantly peaked (x = 0.396; n = 9522; P < 2.2E-16) at the TSS (−250 to +250 bp) in Hmga2 + /+ MEF. Further, the genes were ranked based on pH2A. X levels at the TSS (Source Data file 04) and the results were visualized as heat maps (Fig. 1c). From the top 15% of the genes with high pH2A. X levels at TSS (further referred as top 15% candidates; n = 9522), we selected Gata6 (GATA binding protein 6), Mtor (mechanistic target of rapamycin kinase) and Igf1 (insulin like growth factor 1) for further single gene analysis. Explanatory for these gene selection, we have previously reported Gata6 as direct target gene of HMGA2 10,28 , KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment based analysis of the top 15% candidates showed significant enrichment of genes related to the mammalian target of rapamycin (mTOR) signaling pathway ( Supplementary Fig. 1b), HMGA2 has been related to the insulin signaling pathway 29,30 . In addition, Rptor (regulatory associated protein of MTOR complex 1) is outside of the top 15% candidates (Fig. 1c) and was selected as negative control. Visualization of the selected genes using the UCSC genome browser confirmed the reduction of pH2A. X at specific regions close to TSS in Hmga2−/− MEF when compared to Hmga2 + /+ MEF (Fig. 1d, top) except in the negative control Rptor. Similar results were obtained after ChIP-seq using H2A. X and H3 antibodies (Fig. 1d, bottom), the last one frequently used for monitoring nucleosome position. Promoter analysis of Gata6, Mtor, Igf1 and Rptor by ChIP using pH2A. X-, H2A. X-, H3-and HMGA2-specific antibodies ( Supplementary Fig. 1c, d) confirmed the ChIP-seq data. These findings suggest that the first nucleosome relative to the TSS of the top 15% candidates contains pH2A. X and Hmga2 is required for correct positioning of this first nucleosome.
Position of the first nucleosome containing pH2A. X correlates with RNA polymerase II and the basal transcription activity of genes. Phosphorylation of specific amino acids in the C-terminal domain of the large subunit of the RNA polymerase II (Pol II) determines its interaction with specific factors, thereby regulating the transcription cycle consisting of initiation, elongation and termination 31 . To monitor transcription initiation, ChIP-seq was performed using antibodies specific for transcription initiating S5 phosphorylated Pol II (further referred to as pPol II) and chromatin isolated from Hmga2+/+ and Hmga2−/− MEF ( Fig. 2 and Supplementary Fig. 1a). Analysis of the ChIP-seq results using the UCSC Known Genes dataset revealed that pPol II was enriched at TSS in Hmga2-dependent manner (Fig. 2a). In addition, we observed that pPol II enrichment coincides with pH2A. X peaks at TSS (Fig. 2b) also in an Hmga2-dependent manner. Visualization of Gata6, Mtor and Igf1 using the UCSC genome browser (Fig. 2c) and ChIP analysis of their promoters (Fig. 2d, left) confirmed the reduction of pPol II at specific regions close to TSS in Hmga2−/− MEF when compared to Hmga2 + /+ MEF. Furthermore, the reduced pPol II levels after Hmga2-KO correlated with the reduced expression of the analyzed genes as shown by quantitative reversetranscription PCR (qRT-PCR, Fig. 2d, right). These effects were not observed in the negative control Rptor. Moreover, Hmga2 overexpression in Hmga2−/− MEF reverted the observed effects (Fig. 2d), thereby demonstrating the specificity of the changes caused by Hmga2-KO.
Further analysis of the ChIP-seq data by k-means clustering 32 revealed three clusters in the top 15% candidates (Fig. 3a). Cluster 1 (n = 3266) showed pPol II, pH2A. X, HMGA2 and H3 enrichment directly at the TSS (top), while clusters 2 (n = 3208) and 3 (n = 3075) showed enrichment of these proteins 125 bp and 250 bp 3′ of the TSS, respectively (middle and bottom). Remarkably, RNA sequencing (RNA-seq) based expression analysis in Hmga2 + /+ and Hmga2−/− MEF (Fig. 3b) revealed that the genes in the three clusters have different basal transcription activities, whereby cluster 1 has the lowest (x̄= 0.6201), cluster 2 the middle (x̄= 0.6979) and cluster 3 the highest (x̄= 1.56) basal transcription activity in Hmga2 + /+ MEF. Hmga2-KO significantly reduced the basal transcription activity in all three clusters, to 0.325 (P = 4.72E-5) in cluster 1, 0.375 (P = 1E-4) in cluster 2 and 0.867 (P = 2.47E-3) in cluster 3. Our results demonstrate a correlation between the basal transcription activity and the position of pPol II, pH2A. X, HMGA2 and H3 relative to TSS. Interestingly, the observed Hmga2- Fig. 1 HMGA2 is required for pH2A. X deposition at TSS. a Aggregate plot for pH2A. X enrichment within the gene body ±2 kb of UCSC Known Genes in Hmga2+/+ and Hmga2−/− MEF. ChIP-seq reads were normalized using reads per kilobase per million (RPKM) measure and are represented as log2 enrichment over their corresponding inputs. TSS, transcription start site; TTS, transcription termination site. Dotted square, ±750 bp region around the TSS. b Top, schematic representation of the genomic region highlighted in a. Bottom, box plot of pH2A. X enrichment in the genomic regions showed as squares at the top in Hmga2+/+ and Hmga2−/− MEF. RPKM of the pH2A. X ChIP-seq were binned within each of these genomic regions and represented as log2. Box plots indicate median (middle line), 25th, 75th percentile (box) and 5th and 95th percentile (whiskers); n = 9522 genes enriched with pH2A. X; asterisks P-values after two-tailed Wilcoxon-Mann-Whitney test, ***P ≤ 0.001. c Heat map for pH2A. X enrichment at the TSS + 0.25 kb of UCSC Known Genes in Hmga2 + /+ and Hmga2−/− MEF. Genes were ranked by pH2A. X enrichment in Hmga2 + /+ MEF. Doted square, the top 15% ranked genes, as well as Gata6, Mtor, Igf1 and Rptor were selected for further analysis. d Visualization of selected HMGA2 target genes using UCSC Genome Browser showing HMGA2 (black), pH2A. X (turquoise), H2A. X (yellow) and H3 (blue) enrichment in Hmga2+/+ and −/− MEF. ChIP-seq reads were normalized using RPKM measure and are represented as log2 enrichment over their corresponding inputs. Images show the indicated gene loci with their genomic coordinates. Arrows, direction of the genes; black boxes, exons; dotted squares, regions selected for single gene analysis. See also Supplementary Fig. 1 Fig. 2a-c).
HMGA2 is required for enrichment of the FACT complex at TSS. To gain insights into the HMGA2, ATM, and pH2A. X transcriptional network 10 , native chromatin preparations from Hmga2 + /+ and Hmga2−/− MEF were digested with micrococcal nuclease (MNase) and subsequently fractionated by sucrose gradient ultracentrifugation (SGU) (Fig. 4a-e and Supplementary Fig. 3a-c). From the obtained fractions, the proteins were extracted and analyzed either by western blot (WB; Fig. 4a, d, top), by densitometry analysis of WB ( Supplementary Fig. 3b), or by high-resolution mass spectrometry-based proteomic approach (Fig. 4b-c and Source Data file 04), while the DNA was also isolated and analyzed either by gel electrophoresis (Fig. 4d, bottom), or by DNA sequencing (MNase-seq, Fig. 4e and Supplementary Fig. 4a, b). In Hmga2 + /+ MEF (Fig. 4a, left and Supplementary Fig. 3b), WB of the obtained fractions showed that HMGA2 sedimented in fractions 4 to 9, whereas pPol II and histones mainly sedimented in fractions 1 to 4, where protein complexes of higher molecular weight (MW) are expected. Interestingly, the histone variant H2A. X and its posttranslationally modified form pH2A. X showed a similar sedimentation pattern as the core histones. However, pH2A. X sedimentation in fraction 4 was more pronounced. In Hmga2−/− MEF (Fig. 4a, right and Supplementary Fig. 3b) the levels of pH2A. X were reduced and distributed in fractions 3 to 5. The reducing effect of Hmga2-KO on pH2A. X levels was confirmed by immunostaining in MEF ( Supplementary Fig. 4c-d) Position of transcription initiating S5 phosphorylated RNA polymerase II at the TSS is Hmga2-dependent. a, b Aggregate plots for phosphorylated serine 5 RNA polymerase II (pPol II) enrichment within the gene body ±2 kb of UCSC Known Genes (a) and in a ± 4 kb region respective to pH2A. X peaks (b) in Hmga2+/+ and Hmga2−/− MEF. ChIP-seq reads were normalized using reads per kilobase per million (RPKM) measure and are represented as log2 enrichment over their corresponding inputs. TSS, transcription start site; TTS, transcription termination site. c Visualization of selected HMGA2 target genes using UCSC Genome Browser showing pPol II enrichment in Hmga2 + /+ and −/− MEF. ChIP-seq reads were normalized using RPKM measure and are represented as log2 enrichment over their corresponding inputs. Images represent the indicated gene loci with their genomic coordinates. Arrows, direction of the genes; black boxes, exons; dotted squares, regions selected for single gene analysis. d Analysis of selected HMGA2 target genes. Left, ChIP of Gata6, Mtor, Igf1 and Rptor after pPol II immunoprecipitation in Hmga2 + /+ and Hmga2−/− MEF. Right, qRT-PCR-based, Tuba1a-normalized expression analysis under the same conditions. Bar plots presenting data as means; error bars, s.e.m (n = 3 biologically independent experiments); asterisks, P-values after two-tailed t-test, ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05. See also Supplementary Figs. 1-2 . Source data are provided as a Source Data files 01 and 04.
analysis was focused on fractions 3 and 4, because fraction 4 contained all proteins monitored by WB in Hmga2 + /+ MEF, whereas in fraction 3 levels of HMGA2 and pH2A. X were significantly reduced. We analyzed these two fractions by highresolution mass spectrometry-based proteomic approach and identified proteins that were more than 1.65-fold significantly enriched in Hmga2 + /+ MEF when compared to Hmga2−/− MEF ( Fig. 4b; Source Data file 04; n = 1215 and P < 0.05 in fraction 3; n = 1729 and P < 0.05 in fraction 4). A closer look on the proteins enriched in fractions 3 and 4 of Hmga2 + /+ MEF revealed the presence of both components of the FACT complex, SUPT16 and SSRP1 (Fig. 4b), as well as proteins related to transcription regulation and nucleotide excision repair (NER; Supplementary Fig. 3c). Interestingly, Hmga2-KO significantly reduced the levels of SUPT16 (from 0.909 to 0.298; n = 3; P = 3.5E-3) and SSRP1 (from 0.831 to 0.290; n = 3; P = 3.9E-3) in fraction 4 without HMGA2 is required for enrichment of the FACT complex at TSS. a-e Native chromatin from Hmga2 + /+ and −/− MEF was digested with micrococcal nuclease (MNase) and fractionated by sucrose gradient ultracentrifugation (SGU). a The obtained fractions were analyzed by WB using the indicated antibodies. Representative images from three independent experiments. MW, molecular weight, kDa, kilo Dalton. Inp, input represents 0.5% of the material used for SGU. Square, fractions selected for further analysis. b Mass spectrometry analysis of proteins in fractions 3 and 4. Volcano plot representing the significance (−log10 P-values after one-tailed t-test) vs. intensity fold change between Hmga2 + /+ and −/− MEF (log2 of means intensity ratios from three independent experiments). Square, proteins with log2 fold change >1.65. Diamond, SUPT16; triangle, SSRP1. c Bar plots showing normalized reporter intensity of SUPT16 (left) and SSRP1 (right) in fractions 3 and 4 of the SGU in a. Data are shown as means ± s.e.m. (n = 3 biologically independent experiments); asterisks, P-values after two-tailed t-Test, **P ≤ 0.01; ns, non-significant. d Top, WB analysis as in a using antibodies specific for components of the FACT complex. Bottom, DNA was isolated from the fractions obtained by SGU in A and analyzed by agarose gel electrophoresis. Representative images from three independent experiments. Square, fractions selected for MNase-seq. e MNase-seq of fractions 3 and 4 of the SGU in a. Aggregate plots representing the enrichment over input (as log2 RPKM) of genomic sequences relative to the TSS ± 2 kb. f Box plots of ChIP-seq-based SUPT16 (left) and SSRP1 (right) enrichment analysis within the TSS + 0.5 kb of the top 15% candidates in Hmga2 + /+ and −/− MEF. Values are represented as log2 of mapped reads that were normalized to the total counts and the input was subtracted. Box plots indicate median (middle line), 25th, 75th percentile (box) and 5th and 95th percentile (whiskers); n = 9522 genes enriched with pH2A. X; asterisks, P-values after two-tailed Mann-Whitney test, ***P ≤ 0.001. See also Supplementary Figs. 2 and 3. Source data are provided as a Source Data files 01 and 02.
the UCSC Known Genes as reference dataset, we found that in fraction 4 the sequencing reads were enriched with TSS in an Hmga2-dependent manner (Fig. 4e), since this enrichment was abolished after Hmga2-KO. In fraction 3, we did not detect TSS enrichment of the sequencing reads. Our results indicate that the native chromatin in fraction 4 contains mono-and di-nucleosomes ( Supplementary Fig. 4b), which are enriched with TSS, HMGA2, pH2A. X and pPol II in WT MEF. To link the results obtained by MNase-seq and mass spectrometry after fractionation by SGU, we performed ChIP-seq using SUPT16-and SSRP1-specific antibodies and chromatin from Hmga2 + /+ and Hmga2−/− MEF ( Fig. 4f and Supplementary Fig. 4e). Confirming the results in fraction 4, an accumulation of both components of the FACT complex was detected at TSS in Hmga2 + /+ MEF, whereas Hmga2-KO reduced the levels of SUPT16 and SSRP1 at TSS. In summary, these results demonstrate that HMGA2 is required for recruitment of the FACT complex to TSS.
HMGA2-FACT interaction is required for pH2A. X deposition. The results in Fig. 4a-f suggest an interaction between HMGA2 and the FACT complex. In addition, we found in our previously published mass spectrometry based HMGA2 interactome 10 that HMGA2 precipitated SUPT16 and SSRP1 ( Supplementary Fig. 4f). Indeed, the interaction of HMGA2 with both components of the FACT complex was confirmed by coimmunoprecipitation (Co-IP) assay using nuclear protein extracts from MEF after overexpression of HMGA2 tagged Cterminally with MYC and HIS (HMGA2-MYC-HIS; Fig. 5a). To further characterize the HMGA2-FACT interaction, nuclear extracts from Hmga2 + /+ and Hmga2−/− MEF were fractionated into chromatin-bound and nucleoplasm fractions (Supplementary Fig. 5a). WB of these fractions showed that HMGA2 was exclusively bound to chromatin, whereas SUPT16 and SSRP1 were present in both sub-nuclear fractions. However, higher levels of both FACT components were detected in the chromatinbound fraction of Hmga2 + /+ MEF as compared to the nucleoplasm. Interestingly, Hmga2-KO reverted the distribution of SUPT16 and SSRP1 between these two sub-nuclear fractions, thereby supporting that Hmga2 is required for tethering the FACT complex to chromatin. These results were confirmed by ChIP of Gata6, Mtor, Igf1 and Rptor (Fig. 5b) using SUPT16-and SSRP1specific antibodies and chromatin isolated from Hmga2 + /+ and Hmga2−/− MEF that were stably transfected with a tetracyclineinducible expression construct (tetOn) either empty (−; negative control) or containing the cDNA of WT HMGA2-MYC-HIS. Doxycycline treatment of these stably transfected MEF induced the expression of WT HMGA2-MYC-HIS ( Supplementary Fig. 5b). Hmga2-KO reduced the levels of SUPT16 and SSRP1 in the promoters of the selected HMGA2 target genes (Fig. 5b), while doxycycline-inducible expression of WT HMGA2-MYC-HIS in Hmga2−/− MEF reconstituted the levels of both FACT components, thereby demonstrating the specificity of the effects caused by Hmga2-KO. These effects were not observed in the negative control Rptor.
To demonstrate the causal involvement of the FACT complex in context of HMGA2-mediated chromatin rearrangements, we analyzed the levels of pH2A. X and H2A. X at the Gata6, Mtor, Igf1 and Rptor promoters by ChIP using chromatin from Hmga2 + /+ MEF and stably transfected Hmga2−/− MEF that were treated with DMSO (control) or a FACT inhibitor (FACTin; CBLC000 trifluoroacetate) 33 and doxycycline as indicated (Fig. 5c). FACTin treatment induces negative supercoiling and the formation of lefthanded Z-DNA, which is recognized by the FACT subunit SSRP1 34 . Increasing doses of FACTin resulted in chromatin trapping of SUPT16 and SSRP1 ( Supplementary Fig. 5c, d) 35 . In addition, FACT inhibition significantly reduced pH2A. X and H2A. X levels specifically at the promoters of the selected HMGA2 target genes, confirming that the FACT complex is required for proper H2A. X deposition (Fig. 5c, left). Further, Hmga2-KO also reduced pH2A. X and H2A. X levels at the same promoters In summary, these results demonstrate that the FACT complex is required for HMGA2 function and consequently also for proper pH2A. X levels at the promoters of the HMGA2 target genes.
After confirming the loss of lyase activity in RΔA HMGA2-MYC-HIS ( Supplementary Fig. 6c, d), single-strand DNA breaks (DNA nicks) at the Gata6, Mtor, Igf1 and Rptor promoters were monitored using genomic DNA from Hmga2 + /+ MEF and stably transfected Hmga2−/− MEF that were non-treated (−) or treated with doxycycline ( Fig. 6a). We detected DNA nicks at the analyzed promoters in Hmga2 + /+ MEF, whose levels were reduced upon Hmga2-KO specifically at the promoter of the selected HMGA2 target genes. Interestingly, inducible expression of WT HMGA2 in Hmga2−/− MEF reconstituted the levels of DNA nicks, whereas RΔA HMGA2 did not rescue the effect induced by Hmga2-KO, thereby confirming that HMGA2 lyase activity is required for the DNA nicks detected at the promoters of HMGA2 target genes. Further, we decided to demonstrate the requirement of the lyase activity for the function of HMGA2. Thus, the levels of pH2A. X and H2A. X at the Gata6, Mtor, Igf1 and Rptor promoters were analyzed by ChIP using chromatin from Hmga2 + /+ and stably transfected Hmga2−/− MEF (Fig. 6b). Confirming the results in Figs. 1 and 5c, Hmga2-KO reduced pH2A. X and H2A. X levels specifically at the promoter of the selected HMGA2 target genes. In addition, doxycyclineinducible expression of WT HMGA2 in Hmga2−/− cells reconstituted the levels of pH2A. X and H2A. X, demonstrating the specificity of the effect observed after Hmga2-KO. However, doxycycline-inducible expression of RΔA HMGA2 in Hmga2−/− MEF did not rescue the effect induced by Hmga2-KO. Although the mutations inducing the loss of the lyase activity did not affect the HMGA2-FACT complex interaction ( Supplementary Fig. 6e), these results show that the lyase activity is required for proper pH2A. X and H2A. X levels at the promoter of the selected HMGA2 target genes.
Interestingly, crossing the NONCODE database with our top 15% candidates revealed that 79% of the candidates have annotated noncoding RNAs (ncRNAs) in close proximity (n = 7535), including Gata6, Mtor, Igf1 and Rptor (Fig. 6c, left top). Mapping the identified ncRNAs to the murine genome allowed us to identify 2,106 unique ncRNAs (7.4%) that mapped to loci close to promoters controlling the expression of adjacent mRNAs ( Supplementary Fig, 7b). From these promoter related ncRNAs 36,37 more than half (1401; 67%) were in the antisense strand (as) in divergent (div; 621 ncRNAs) or convergent (con; 780 ncRNAs) orientation 36,37 relative to the corresponding promoter and mRNA ( Supplementary Fig. 7c, d). Interestingly, Hmga2-KO significantly reduced the median expression levels of these antisense divergent ncRNAs from 0.085 to 0.067 (P = 0.039; Supplementary Fig. 7e), without significantly affecting the levels of antisense convergent and sense ncRNAs. In silico analysis allowed us to detect putative binding sites of the identified ncRNAs at the TSS of the corresponding mRNAs (Fig. 6c, right) with favorable minimum free energy (MFE < − 55 kcal/mol) and relatively high consensus (cons > 41%;), supporting the formation of DNA-RNA hybrids containing a nucleotide sequence that favors DNA nicks 38 . In the same genomic regions, we also identified strand asymmetry in the distribution of cytosines and guanines, so called GC skews (Fig. 6c, left middle; Supplementary Fig. 7f, g), that are predisposed to form R-loops, which are threestranded nucleic acid structures consisting of a DNA-RNA hybrid and the associated non-template single-stranded DNA 39 . Supporting this hypothesis, published genome-wide sequencing experiments after DNA-RNA immunoprecipitation (DRIP-seq) in NIH/3T3 mouse fibroblasts 40 confirmed the formation of DNA-RNA hybrids in the top 15% candidate genes (n = 9522; Supplementary Fig. 7f, g), including Gata6, Mtor, Igf1 and Rptor (Fig. 6c, left bottom). This correlated with high amounts of GC skews at their TSS with at least 38.5% of the TSS and downstream region having a GC skew higher than 0.05 (n = 3669). All these observations prompted us to investigate the role of HMGA2 during R-loop formation at the TSS. Thus, we analyzed by DRIP assays the levels of R-loops at the Gata6, Mtor, Igf1 and Rptor promoters using the antibody S9.6 41 and nucleic acids isolated from Hmga2 + /+ and stably transfected Hmga2−/− MEF (Fig. 6d). Hmga2-KO increased R-loops levels at the promoters analyzed, whereas doxycycline-inducible expression of WT HMGA2 in Hmga2−/− MEF reduced R-loop levels back to similar levels as in Hmga2 + /+ MEF. Interestingly, doxycyclineinducible expression of RΔA HMGA2 in Hmga2−/− MEF did not rescue the effect induced by Hmga2-KO. In parallel, treatment of the samples before IP with RNase H1 (RNH1), which degrades RNA in DNA-RNA hybrids, reduced the levels of R-loops in all tested conditions, demonstrating the specificity of the antibody S9.6 41 . In summary, these results demonstrate that HMGA2 and its lyase activity are required to solve R-loops at the analyzed promoters, including the negative control Rptor. The fact that we detected R-loops in Rptor (Fig. 6d) without significant changes in the levels of DNA nicks ( Fig. 6a) or pH2A. X (Fig. 6b) suggest a different regulatory mechanism for Rptor when compared to the selected Hmga2 target genes.
HMGA2-FACT-ATM-pH2A. X axis is required to solve R-loops and induce DNA demethylation. The inducible expression of RΔA HMGA2 in Hmga2−/− MEF did not decrease R-loops levels at TSS that were increased after Hmga2-KO (Fig. 6d), supporting that the lyase activity of HMGA2 is required to solve R-loops. To further investigate these results, the levels of doublestranded DNA (dsDNA) at the Gata6, Mtor and Igf1 promoters were analyzed by DNA immunoprecipitation (DIP) assays ( Fig. 7a, left). Inversely correlating with the effects on R-loops, Hmga2-KO reduced dsDNA levels at the promoters analyzed. Further, doxycycline-inducible expression of WT HMGA2 in Hmga2−/− MEF reconstituted dsDNA levels, whereas RΔA HMGA2 failed to rescue the effect induced by Hmga2-KO. These results further support the requirement of HMGA2 and its lyase activity for solving R-loops. Since DNA methylation alters chromatin structure and is associated with R-loop formation 18,42 , we also analyzed the levels of 5-methylcytosine (5mC) at the Gata6, Mtor and Igf1 promoters by DIP assays using 5mC- Fig. 6 HMGA2-lyase activity is required for pH2A. X deposition and solving of R-loops. a Analysis of single-strand DNA breaks at promoters of selected HMGA2 target genes using genomic DNA from Hmga2 + /+, Hmga2−/− MEF, as well as Hmga2−/− MEF that were stably transfected with a tetracycline-inducible expression construct (tetOn) either for WT Hmga2-myc-his or the lyase-deficient mutant RΔA Hmga2-myc-his. MEF were treated with doxycycline as indicated. b ChIP-based promoter analysis of selected HMGA2 target genes using the indicated antibodies and chromatin from MEF as in a. c Left, Genome-browser visualization of selected HMGA2 target genes showed nascent RNA (GRO-seq) in WT MEF 68 , GC-skew and RNA-seq on plus-strand after DNA-RNA hybrid immunoprecipitation (DRIPc-seq) in NIH/3T3 mouse fibroblasts 40 . Images represent mapped sequence tag densities relative to the indicated loci. Genomic coordinates are shown at the bottom. Arrow heads, non-coding RNAs in antisense orientation; Arrows, direction of the genes; black boxes, exons. Right, in silico analysis revealed complementary sequences between the identified antisense ncRNA (green) and genomic sequences at the TSS of the corresponding mRNAs (red) with relatively favorable minimum free energy (MFE) and high percentage of complementarity (cons), supporting the formation of DNA-RNA hybrids containing a nucleotide sequence that favors DNA nicks (squares). d Analysis of selected HMGA2 target genes by DNA-RNA immunoprecipitation (DRIP) using the antibody S9.6 41 and nucleic acids isolated from MEF treated as in a. Prior IP, nucleic acids were digested with RNase H1 (RNH1) as indicated. In all bar plots, data are shown as means ± s.e.m. (n = 3 biologically independent experiments); asterisks, P-values after two-tailed t-Test, ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05; ns, non-significant. See also Supplementary Figs 43 (Fig. 7a, right). Correlating with the effects on R-loops, Hmga2-KO increased 5mC levels, which in turn were reduced by inducible expression of WT HMGA2 in Hmga2−/− cells but not by RΔA HMGA2. These results showed that HMGA2 and its lyase activity are required for proper 5mC levels at the analyzed promoters. In addition, the results using FACTin (Fig. 5c) showed that the FACT complex is required for HMGA2 function and consequently for proper pH2A. X levels at TSS.
To demonstrate the sequential order of events of the molecular mechanism proposed here ( Fig. 7b), additional experiments were performed ( Fig. 7 and Supplementary Fig. 8). We first analyzed the Gata6, Mtor and Igf1 promoters by DRIP and DIP using nucleic acids isolated from Hmga2 + /+ MEF that were nontreated (-) or treated with FACTin as indicated ( Supplementary Fig. 8a, b). FACTin treatment in Hmga2 + /+ MEF increased Rloop and 5mC levels, whereas dsDNA levels were reduced, thereby supporting that the FACT complex is required to solve Rloops and for proper levels of 5mC at the analyzed promoters, similarly as the HMGA2 lyase activity (Figs. 6d and 7a). Previously, we have shown that ATM loss-of-function (LOF) blocks TGFB1-induced and HMGA2-mediated transcription activation 10 . To confirm the causal involvement of ATM in the mechanism of transcription regulation proposed here (Fig. 7b), the levels of pH2A. X and H2A. X at the Gata6, Mtor and Igf1 promoters were analyzed by ChIP using chromatin from Hmga2 + /+ MEF and stably transfected Hmga2−/− MEF that were treated with DMSO (control) or an ATM inhibitor (ATMi; KU-55933) and doxycycline as indicated (Fig. 7c). Interestingly, ATMi treatment counteracted the rescue effect on pH2A. X levels that was mediated by inducible expression of WT HMGA2 in Hmga2−/− MEF, without significantly affecting H2A. X levels, thereby supporting that ATM is required for the post- . X axis is required to solve R-loops and induce DNA demethylation. a DNA immunoprecipitation (DIP) based promoter analysis of selected HMGA2 target genes using antibodies specific for double-stranded DNA (dsDNA) or 5-methylcytosine (5mC) and genomic DNA from Hmag2 + /+, Hmag2−/− MEF, as well as Hmga2−/− MEF that were stably transfected with a tetracycline-inducible expression construct (tetOn) for either WT Hmga2-myc-his or the lyase-deficient mutant RΔA Hmga2-myc-his. MEF were treated with doxycycline as indicated. b Schematic representation of the sequential order of events during transcription activation mediated by the HMGA2-FACT-ATM-pH2A. X axis. c ChIP-based promoter analysis of selected HMGA2 target genes using the indicated antibodies and chromatin from MEF treated as in a. In addition, MEF were treated with ATM inhibitor (ATMi; KU-55933) as indicated. d DIP-based promoter analysis as in a, using 5mC-specific antibodies. In addition, MEF were treated with ATMi as indicated. e ChIP-based promoter analysis of selected HMGA2 target genes using the indicated antibodies and chromatin from MEF treated as in a. In addition, MEF were transfected with control (Ctrl) or Gadd45a-specific small interfering RNA (siRNA) as indicated. f DIP-based promoter analysis as in a, using 5mC-specific antibodies. In addition, MEF were transfected with Ctrl or Gadd45a-specific siRNA as indicated. g ChIP-based promoter analysis of selected HMGA2 target genes using the indicated antibodies and chromatin from MEF treated as in a. h WB analysis using antibodies specific for HIS-tag, HA-tag and H3 after DRIP using the antibody S9.6 41 and chromatin isolated from MLE-12 cells that were stably transfected either with a control (scramble, scr) or an Hmga2-specific short hairpin DNA (sh) construct and transiently transfected with WT Hmga2-myc-his or the lyase-deficient mutant RΔA Hmga2-myc-his and Gadd45-HA as indicated. Representative image from two independent experiments. Input (Inp), 5% of IP starting material; immunoglobulin G (IgG), negative control. In all bar plots, data are shown as means ± s.e.m. (n = 3 biologically independent experiments); asterisks, P-values after one-tailed t-test, ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05; ns, non-significant. See also Supplementary Fig. 8. Source data are provided as a Source Data files 01 and 02.
translational modification of H2A. X rather than for the deposition of H2A. X into the analyzed promoters. In addition, we monitored 5mC levels by DIP at the Gata6, Mtor and Igf1 promoters (Fig. 7d) and found that ATM-LOF also counteracted the rescue effect on 5mC levels mediated by inducible expression of WT HMGA2 in Hmga2−/− MEF, thereby supporting that phosphorylation of H2A. X at S139 is required for proper 5mC levels. The results obtained after ATM-LOF (Fig. 7c, d) support that ATM acts downstream of HMGA2 and the FACT complex.
Gadd45a has been reported to promote transcriptional activation by repair-mediated DNA demethylation 19 . Thus, we investigated the potential involvement of Gadd45a in the order of events proposed here (Fig. 7b). To this purpose, we analyzed the effect of Gadd45a-specific LOF using small interfering RNA (siRNA; siG45a; Supplementary Fig. 8c) on pH2A. X, H2A. X and 5mC levels at the Gata6, Mtor and Igf1 promoters in Hmga2 + /+ MEF and stably transfected Hmga2−/− MEF (Fig. 7e, f). While siG45a transfection counteracted the rescue effect on 5mC levels mediated by inducible expression of WT HMGA2 in Hmga2−/− MEF (Fig. 7f), it did not significantly affect pH2A. X and H2A. X levels (Fig. 7e), confirming that GADD45A is required for proper 5mC levels but not for pH2A. X and H2A. X levels. Further, we found that GADD45A gain-of-function (GOF) after transfection of a human GADD45A expression construct into mouse lung epithelial (MLE-12) cells reduced 5mC levels in HMGA2dependent manner ( Supplementary Fig. 8d, e). Our results (Fig. 7e, f and Supplementary Fig. 8d, e) indicate that GADD45A acts downstream of the HMGA2-FACT-ATM-pH2A. X axis (Fig. 7b). Confirming this interpretation, ChIP-seq using GADD45A-specific antibodies and chromatin isolated from Hmga2 + /+ and Hmga2−/− MEF (Supplementary Fig. 8f) revealed that GADD45A and pH2A. X are enriched at similar regions respective to TSS of the top 15% candidates. Moreover, ChIP analysis of the Gata6, Mtor and Igf1 promoters using GADD45A-or TET1-specific antibodies and chromatin from Hmga2 + /+ and stably transfected Hmga2−/− MEF that were treated with DMSO (control) or doxycycline (Fig. 7g) showed that Hmga2-KO abrogated GADD45A and TET1 binding to the analyzed promoters. Strikingly, inducible expression of WT HMGA2 reconstituted GADD45A and TET1 binding to the analyzed promoters, whereas RΔA HMGA2 did not rescue the effect induced by Hmga2-KO.
We have shown that genetic ablation of Hmga2 increased Rloop levels (Fig. 6d) and reduced GADD45A binding (Fig. 7g) at the Gata6, Mtor and Igf1 promoters. Interestingly, Arab and colleagues recently reported that GADD45A preferentially binds DNA-RNA hybrids and R-loops rather than single-stranded (ss) or double-stranded (ds) DNA or RNA 18 . To elucidate these at first glance contradictory results, we performed DRIP using the antibody S9.6 41 and chromatin from MLE-12 cells that were stably transfected with a scrambled (scr) or a Hmga2-specific short hairpin RNA construct (shHmga2) and non-treated (-) or treated with doxycycline to induce transient expression of WT or RΔA HMGA2 (Fig. 7h). WB analysis of the precipitated material revealed that both WT and RΔA HMGA2 bind to R-loops. Further, exogenous GADD45A also binds to R-loops, confirming the results by Arab and colleagues 18 . However, GADD45A binding to R-loops increased after inducible expression of WT HMGA2, but not after RΔA HMGA2, suggesting that DNA nicks in the R-loops increase the affinity of GADD45A to the R-loops. The results by WB after DRIP (Fig. 7h) correlate with the ChIP analysis of the Gata6, Mtor and Igf1 promoters (Fig. 7g) and were confirmed by DRIP and sequential ChIP (DRIP-ChIP; Supplementary Fig. 8g). Taking together, our results demonstrate that the HMGA2-FACT-ATM-pH2A. X axis acts upstream of GADD45A and facilitates its binding to R-loops at specific promoters by nicking the DNA moiety of the DNA-RNA hybrid, thereby inducing DNA repair-mediated promoter demethylation.
HMGA2-FACT-ATM-pH2A. X axis mediates TGFB1 induced transcription activation. We have previously shown that HMGA2 mediates TGFB1 induced transcription 10 . Thus, we decided to evaluate the mechanism of transcription activation proposed here (Fig. 7b) within the context of TGFB1 signaling. We performed RNA-seq in Hmga2 + /+ and Hmga2 − /− MEF that were non-treated or treated with TGFB1 and visualized the results of those genes that were induced by TGFB1 treatment as heat maps after k-means clustering (Fig. 8a and Supplementary Fig. 9a). Four clusters were identified, of which clusters 2 (n = 1471) and 4 (n = 1974) contained genes that were TGFB1 inducible in an Hmga2-independent manner. Cluster 3 (n = 381) contained TGFB1 inducible genes, whose expression increased after Hmga2-KO, while TGFB1 treatment in Hmga2−/− MEF reduced their expression. We focused on cluster 1 (n = 640) for further analysis, which contained TGFB1 inducible genes in Hmga2-dependent manner. Cross-analysis of our RNA-seq after TGFB1 treatment ( Fig. 8a and Supplementary Fig. 9a) with our ChIP-seq data ( Fig. 3a and Supplementary Fig. 8f) confirmed the existence of three gene groups based on the position of the first nucleosome 3′ of the TSS containing pH2A. X, which we called position clusters 1 to 3 to differentiate them from the TGFB1 inducible clusters. Consistent with our previous results (Fig. 3b), the genes in the position clusters 1 to 3 displayed increasing basal transcription activity, whereby position cluster 1 has the lowest (x= 0.155), position cluster 2 the middle (x̄= 0.662) and position cluster 3 the highest (x̄= 2.766) basal transcription activity in Hmga2 + /+ MEF (Supplementary Fig. 9b). In addition, the position of the first nucleosome relative to the TSS also correlated with the strength of transcriptional activation induced by TGFB1 (Fig. 8b) ChIP-seq analysis of pH2A. X levels was also performed using the same conditions as in our RNA-seq after TGFB1 treatment ( Fig. 8c and Supplementary Fig. 9c). TGFB1 treatment increased pH2A. X levels from 2.23 to 2.937 (P = 7.5E-3) at the TSS of TGFB1 inducible cluster 1 genes in Hmga2 + /+ MEF, whereas this effect was not observed in Hmga2−/− MEF, confirming the requirement of Hmga2 for the effects induced by TGFB1.
Further, to determine the causal involvement of the FACT complex during TGFB1 induced transcriptional activation, we performed a series of experiments analyzing Gata6, Mtor and Igf1 in Hmga2 + /+ and Hmga2−/− MEF that were non-treated or treated with FACTin ( Fig. 8d-f). TGFB1 treatment in Hmga2 + /+ MEF increased the expression of Gata6, Mtor and Igf1 (Fig. 8d) as well as the levels of pPol II and pH2A. X in their promoters (Fig. 8e), whereas 5mC levels were reduced (Fig. 8f). The effects induced by TGFB1 treatment were not observed in Hmga2−/− MEF confirming the requirement of Hmga2. Further, FACTin treatment counteracted the effects induced by TGFB1 in Hmga2 + /+ MEF supporting the causal involvement of the FACT complex. Interestingly, WB analysis of protein extracts from Hmga2 + /+ and Hmga2−/− MEF (Supplementary Fig. 9d) demonstrated that the effects observed after Hmga2-and FACT-LOF take place neither affecting total SMAD2/3 levels, nor changing their activation by TGFB1. Consistent with the mechanism of transcriptional regulation proposed here (Fig. 7b) and the results in Fig. 7e, f, siRNA-mediated Gadd45a-LOF counteracted the reducing effect of TGFB1 on 5mC levels in Hmga2 + /+ MEF (Fig. 8g) without affecting the increasing effect on pH2A. X levels (Fig. 8h), thereby confirming that GADD45A acts downstream of the HMGA2-FACT-ATM-pH2A. X axis.
Inhibition of the FACT complex counteracts fibrosis hallmarks in IPF. The clinical potential of the here proposed mechanism of transcription regulation (Fig. 7b) was approached by placing it into the context of the most common interstitial lung disease, IPF, in which TGFB signaling plays a key role 44 . RNA-seq in primary human lung fibroblasts (hLF) isolated from control (n = 3) and IPF (n = 3) patients revealed increased expression levels of HMGA2, SUPT16H and SSRP1 in IPF patients when compared to control donors (Supplementary Fig. 10a). Further, cross-analysis of RNA-seq in primary hLF isolated from control and IPF patients 44 with RNA-seq in Hmga2 + /+ MEF that were nontreated or treated with TGFB (Fig. 9a) allowed us to identify 923 orthologue genes that were at least 1.5 fold significantly increased (P ≤ 0.05) in IPF hLF and in TGFB treated MEF when compared to the corresponding control cells. Gene set enrichment analysis (GSEA) 45 Inputs were subtracted from the corresponding samples. Red line, average; n = 640 genes; error bars, s.e.m.; asterisks, P-values after one-tailed Mann-Whitney test, ***P ≤ 0.001; **P ≤ 0.01; ns, non-significant. d qRT-PCR-base expression analysis of HMGA2 target genes in Hmag2 + /+, Hmag2−/− MEF that were non-treated (Ctrl) or treated with TGFB1 and FACT inhibitor (FACTin; CBLC000 trifluoroacetate) as indicated. e ChIP-based promoter analysis of selected HMGA2 target genes using the indicated antibodies and chromatin from MEF treated as in d. f DIP-based promoter analysis of selected HMGA2 target genes using antibodies specific for 5-methylcytosine (5mC) and genomic DNA from MEF treated as in d. g DIP-based promoter analysis of selected HMGA2 target genes using the indicated antibodies and genomic DNA from Hmag2 + /+ or Hmag2−/− MEF that were transfected with control (−) or Gadd45a-specific small interfering RNA (siRNA) as indicated. h ChIP-based promoter analysis of selected HMGA2 target genes using pH2A. X-specific antibodies and chromatin from MEF treated as in g. In all bar plots, data are shown as means ± s.e.m. (n = 3 biologically independent experiments); asterisks, P-values after two-tailed t-test, ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05; ns, non-significant. See also Supplementary Fig. 9. Source data are provided as a Source Data files 01 and 04. Fig. 9 Inhibition of the FACT complex counteracts fibrosis hallmarks in IPF. a RNA-seq-based comparison of gene expression in IPF and after TGFB1 treatment. 2D Kernel Density plot representing the log2 fold change between gene expression in primary human lung fibroblasts (hLF) from IPF patients vs. control donors on the y-axis and log2 fold change between gene expression in Hmga2 + /+ MEF treated with TGFB1 vs. non-treated on the x-axis. Square, genes with log2 FC > 0.58 and P ≤ 0.05 in both, hLF IPF and TGFB1-treated MEF. P-values after Wald test. b Gene set enrichment analysis (GSEA) using the normalized enrichment scores (NES) of genes inside the square in a. EMT, epithelial-mesenchymal transition; resp, response. c GSEA line profile of the top two enriched pathways in b. d qRT-PCR-based expression analysis of selected HMGA2 target genes in hLF from control donors (Ctrl) or IPF patients that were non-treated (Ctrl) or treated FACT inhibitor (FACTin; CBLC000 trifluoroacetate) as indicated. e ChIP-based promoter analysis of selected HMGA2 target genes using pH2A. X-specific antibodies and chromatin from hLF treated as in d. f DIP-based promoter analysis of selected HMGA2 target genes using 5mC-specific antibodies and genomic DNA from hLF treated as in d. g qRT-PCR-based expression analysis of fibrotic markers in hLF treated as in d. FN1, fibronectin; COL1A1, collagen; ACTA2, smooth muscle actin alpha 2. h Functional assays for IPF hallmarks in Ctrl or IPF hLF treated as in d. Top, hydroxyproline assay for collagen content. Middle, proliferation assay by BrdU incorporation. Bottom, Transwell invasion assay. i Representative pictures from confocal microscopy after immunostaining using the antibody S9.6 or COL1A1-specific antibody in human precision-cut lung slices (hPCLS) from IPF patients (n = 3 biologically independent experiments). The hPCLS were treated as in d. DAPI, nucleus. Scale bars, 500 μm. In all bar plots, data are shown as means ± s.e.m. (n = 3 biologically independent experiments); asterisks, P-values after tow-tailed t-test, ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05; ns, nonsignificant. See also Supplementary Fig. 10. Source data are provided as a Source Data files 01 and 04. NATURE COMMUNICATIONS | [URL] ARTICLE NATURE COMMUNICATIONS | (2021) 12:1072 | [URL] | www.nature.com/naturecommunications (P = 6.3E-3), TGFB1 signaling pathway (P = 0.012), inflammatory response (P = 0.033), MYC target genes (P = 0.011), UV response P = 0.031), fatty acid metabolism (P = 0.029), among others (Fig. 9b). In addition, graphical representation of the enrichment profile showed high enrichment scores (ES) for EMT (ES = 0.515) and TGFB1 signaling pathway (ES = 0.706) as the top two items of the ranked list (Fig. 9c). To determine the role of the HMGA2-FACT-ATM-pH2A. X axis in IPF we analyzed GATA6, MTOR and IGF1 in Ctrl and IPF hLF (Fig. 9d-f). Correlating with the results obtained in MEF after TGFB1 treatment (Fig. 8d-f), we detected in IPF hLF increased expression of GATA6, MTOR and IGF1 (Fig. 9d), as well as increased levels of pH2A. X and H2A. X in their promoters (Fig. 9e and Supplementary 10b), whereas 5mC levels were reduced (Fig. 9f). Strikingly, FACTin treatment counteracted the effects observed in IPF hLF, supporting the involvement of the HMGA2-FACT-ATM-pH2A. X axis in this interstitial lung disease.
To test this hypothesis, we monitored various hallmarks of fibrosis in Ctrl and IPF hLF, such as expression of fibrotic markers by qRT-PCR (Fig. 9g), levels of ECM proteins by Hydroxyproline and Sircol assays (Fig. 9h, top and Supplementary 10c), cell proliferation by bromodeoxyuridine (BrdU) incorporation assay (Fig. 9h, middle) and cell migration by Transwell invasion assay followed by hematoxylin and eosin (H&E) staining (Fig. 9h, bottom). Remarkably, FACTin treatment of IPF hLF significantly reduced all hallmarks of fibrosis analyzed, thereby suggesting the use of FACTin for therapeutic approaches against IPF. Moreover, our in vitro findings in primary hLF were also confirmed ex vivo using human precision-cut lung slices (hPCLS) from 3 different IPF patients ( Fig. 9i and Supplementary Fig. 10d-g). FACTin treatment of IPF hPCLS reduced the levels of the fibrotic markers COL1A1, FN1, smooth muscle actin alpha 2 (ACTA2), the mesenchymal marker vimentin (VIM), as well as HMGA2 and pH2A. X. In contrast, the levels of DNA-RNA hybrids were increased after FACTin treatment.
Discussion
Here, we uncovered a mechanism of transcription initiation of TGFB1-responsive genes mediated by the HMGA2-FACT-ATM-pH2A. X axis. The lyase activity of HMGA2 induces DNA nicks at the TSS, which are required by the FACT complex to incorporate nucleosomes containing H2A. X at specific positions relative to the TSS. The position of the first nucleosome containing H2A. X not only correlates with the basal transcription activity of the corresponding genes, but also with the strength of their inducibility after TGFB1 treatment. Further, ATM-mediated phosphorylation of H2A. X at S139 is required for repair-mediated DNA demethylation and transcriptional activation. Our data support a sequential order of events, in which specific positioning of nucleosomes containing the classical DNA damage marker pH2A. X precedes DNA demethylation and transcription initiation, thereby supporting the hypothesis that chromatin opening involves intermediates with DNA breaks that require mechanisms of DNA repair that ensure the integrity of the genome.
Biological and clinical relevance of the HMGA2-FACT-ATM-pH2A. X axis. We demonstrated the biological relevance of our data within the context of TGFB1 signaling (Fig. 8 and Supplementary Fig. 9). TGFB1 treatment induced promoter specific increase of pH2A. X and pPol II, whereas 5mC levels were decreased, resulting in transcription activation in Hmga2-and FACT-dependent manner (Fig. 8a-f). Interestingly, Gadd45a-LOF interfered with the 5mC decrease (Fig. 8g), without affecting pH2A. X levels (Fig. 8h), supporting the sequential order of events proposed here (Fig. 7b), in which GADD45A acts downstream of the HMGA2-FACT-ATM-pH2A. X axis. Consistent with our findings, Thillainadesan and colleagues reported TGFB induced active DNA demethylation and expression of the p15 ink4b tumor suppressor gene 17 . While published reports showed the effect of TGFB on specific genes 10,17,46 , in this report we demonstrated the genome wide effect of TGFB treatment affecting the global nuclear architecture and strongly suggesting future NGS studies. Following a similar line of ideas, Negreros and colleagues recently reported genome wide changes on DNA methylation induced by TGFB1 47 . The translational potential of our work was demonstrated within the context of IPF ( Fig. 9 and Supplementary Fig. 10), in which TGFB1 signaling plays an important role. Inhibition of the HMGA2-FACT-ATM-pH2A. X axis reduced all fibrotic hallmarks in vitro (using primary hLF) and ex vivo (using hPCLS). Interestingly, the FACT complex is a potential marker of aggressive cancers with low survival rates 48 and FACTin is being tested in a clinical trial for cancer treatment (ClinicalTrials.gov Identifier: NCT01905228, NCT02931110). Our work provides the molecular basis for future studies developing therapies against IPF using FACTin.
Methods
Key resources. Please see Supplementary Table 2 containing the key resources used in this study. Cell culture. All studies were done on immortalized MEF cultivated for less than twenty passages. Hmga2 wild type (+/+) and knockout (−/−) primary mouse embryonic fibroblast (MEF) were isolated from mouse embryos at embryonic day E15.5 and subsequently immortalized using simian virus (SV) 40 10 . MEF and Human embryonic kidney cell HEK293T (ATCC, CRL-11268) were cultured at 37°C in 5% CO 2 in DMEM medium with 4.5 g/l glucose, 10% FCS 4 mM L-Glutamine, 1 mM Pyruvate, 100 U/ml penicillin and 100 U/ml streptomycin. Mouse lung epithelial cells (MLE-12, ATCC CRL-2110) were cultured in Dulbecco's Modified Eagle Medium: Ham's F-12 Nutrient Mixture (5% FCS, 100 U/ml penicillin and 100 U/ml streptomycin) at 37°C in 5% CO 2 . Primary fibroblast from Ctrl and IPF patients were cultured in complete MCDB131 medium (8% FCS, 1% L-glutamine, penicillin 100 U/ml, streptomycin 0.1 mg/ml, EGF 0.5 ng/ml, bFGF 2 ng/ml, and insulin 5 μg/ml)) at 37°C in 5% CO 2 . Because of the concern that the phenotype of the cells is altered at higher passage, cells between passages 4 and 6 were utilized in the experiments described here. All cells were washed with 1x PBS, trypsinized with 0.25% (w/v) trypsin and subcultivated at the ratio of 1:5 to 1:10.
Cell treatments, transfections and siRNA-mediated knockdown. MEF were treated with 1 μg/ml doxycycline or DMSO (used as solvent for doxycycline) for 4, 6 or 24 h to induce the expression of transgenes. Initial FACT complex and ATM kinase inhibition was performed with 5 μM CBLC000 trifluoroacetate (FACTin, Sigma Aldrich) for 2 h or 1 μM KU-55933 (ATMi, Calbiochem) for 6 h, respectively. MEF were transiently transfected either with 100 nM siCtrl (negative control; AM4611, Ambion) or siGadd45a (siG45; AM16708, Ambion) for 48 h. TGFB1 signaling was induced after a 16 h starvation (cell culture medium supplemented with 1% FCS) with 10 ng/ml human recombinant TGFβ1 (Sigma Aldrich) and chromatin changes were assayed after 3 h and gene expression alterations after 24 h incubations. For IPF resolution experiments, primary hLF were treated with 5 μM FACTin for 12 h.
Bacterial culture. For cloning experiments, chemically competent E. coli TOP10 (Thermo Fisher Scientific) were used for plasmid transformation. TOP10 strains were grown in Luria broth (LB) at 37°C with shaking at 180 rpm on LB agar at 37°C overnight.
ChIP sequencing and data analysis in Hmga2 + /+ and Hmga2−/− MEF. Libraries were prepared according to Illumina's instructions accompanying the Ovation Ultra Low Kit. Single-end sequencing was performed on an Illumina HiSeq2500 machine at the Max Planck-Genome-Centre Cologne. Raw reads were visualized by FastQC ( [URL]/) to determine the quality of the sequencing. Trimming was performed using trimmomatic 49 with the following parameters LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MIN-LEN:50 CROP:63 HEADCROP:13. High quality reads were mapped by using Bow-tie2 50 to mouse genome mm10. ChIP-seq data were represented as aggregate plot or heat maps using deeptools 51 following their instructions. Bam-files were converted to Bed-files using Bedtools' bamToBed ( [URL]/) command. Genome browser snapshots were created with Homer using makeTagDirectory ( [URL]) and makeUCSCfile ( [URL]. ucsd.edu/homer/ngs/ucsc.html). Reads were normalized to 30 million reads. Peak calling was performed by using model-based analysis for ChIP-seq (MACS) 52 with a cut-off of p < 0.01 and the following parameters: --nonmodel, -shift size 30, and an effective genome size -g of 1.87e9. Peaks were annotated by using annotePeaks.pl for mm10 from Homer ( [URL]). From the 200,051 peaks found for pH2A. X in Hmga2 + /+ MEF, 3,935 peaks were annotated in the promoter-TSS region and were used for further analysis.
Identification of position clusters and analysis of inducibility. The enrichment of pH2A. X at TSS plus 250 bp downstream in the top 15% candidates was clustered using the k-means algorithm implemented in deeptools' plotHeatmap command. From the 3 position clusters identified, genes in Hmga2 + /+ MEF with a FC more than 1.5 to Hmga2−/− were selected. For expression analysis, RPKMs equal or higher than 10 were considered as outliers (Number of outliers: C1, 5 genes; C2, 2 genes; and C3, 0 genes).
RNA isolation, reverse transcription, quantitative PCR. Total RNA was isolated with Trizol (Invitrogen) and quantified using a Nanodrop Spectrophotometer (ThermoFisher Scientific, Germany). Synthesis of cDNA was performed using 0.5-1 µg total RNA and the High Capacity cDNA Reverse Transcription kit (Applied Biosystems). Quantitative real-time PCR reactions were performed using SYBR® Green on the Step One plus Real-time PCR system (Applied Biosystems). The housekeeping genes Tuba1a and HPRT1 were used to normalize gene expression 53 . Primer pairs used for gene expression analysis are described in Supplementary Information, Supplementary Table 1.
Native chromatin fractionation and sucrose gradient ultracentrifugation. Two 15 cm dishes with MEF were washed with 1x PBS and pellets were resuspended in 1 ml of lysis buffer (10 mM HEPES pH 7.4, 10 mM KCl, 0.05% NP-40, 1 mM DTT, 25 mM NaF, 0.5 mM Na 3 VO 4 , 40 μg/ml phenylmethylsulfonyl fluoride and protease inhibitor). After incubating 20 min on ice, cells were spun down at 300 × g at 4°C for 10 min. The nuclei were washed once in lysis buffer and were then resuspended in 2 volumes of Low Salt Buffer (10 mM Tris-HCl pH 7.4, 0.2 mM MgCl 2 supplemented with protease and phosphatase inhibitors) including 1% Triton-X100. After 15 min incubation on ice, cells were spun down at 300 × g at 4°C for 10 min. The pellet was washed in 1 ml MNase digestion buffer (10 mM Tris-HCl, 25 mM NaCl, 1 mM CaCl2, 1 mM DTT, 25 mM NaF, 0.5 mM Na 3 VO 4 , 40 μg/ml phenylmethylsulfonyl fluoride and protease inhibitor) and resuspended again in 1 ml MNase digestion buffer with 1,250 Units MNase (NEB Biolabs). Chromatin-MNase mix was incubated at 37°C for 30 min. MNase reaction was stopped by adding 50 mM EDTA. Samples were sonicated for 30 sec on / 30 sec off using the Bioruptor with high amplitude. Chromatin was spun down at 18,407 ×g at 4°C for 10 min and the supernatant was used for ultracentrifugation. Sucrose gradients (5% to 40%) were prepared in 1,800 μl low salt buffer (10 mM NaCl, 10 mM Tris-HCl pH7.4, 0.2 mM EDTA, 0.2 mM DTT, 20 mM NaF, 20 mM Na 3 VO 4 , 40 μg/ml phenylmethylsulfonyl fluoride and protease inhibitor) in polyallomer centrifuge tube (Beckman). Fragmented native chromatin was loaded on top of the 9 ml 5% to 40% sucrose gradients and centrifuged for 16 h and 30 min at 168,544 ×g in a SW50.1 ultracentrifuge rotor (Beckman Coulter). Following centrifugation, 11 fractions (1000 μl each) were collected manually from the bottom of the tubes. Later, these fractions were used for western blot, mass spectrometry and NGS.
Mass spectrometry: sample preparation, methods and data analysis. Proteins were methanol/chloroform precipitated from sucrose gradient fractions and dried pellets reconstituted in 8 M urea 54 . Per fraction, 140 µg of protein (according to the 660 nm protein assay, Pierce), were subjected to in-solution digest using protein to enzyme ratios of 1:100 and 1:50 for Lys-C (Wako Chemicals GmbH) and trypsin (Serva), respectively 55 . The resulting peptide mixture was desalted and concentrated using Oligo R3 (Thermo Fisher Scientific) extraction 56 . Peptides (5 µg according to the quantitative fluorimetric peptide assay, Pierce), were subsequently labeled using 6-plex tandem mass tags (Thermo Fisher Scientific) following the manufacturer's protocol but employing a reagent to peptide ratio of four. Labeling channels were used for fractions from a replicate sucrose gradient as well as an internal standard sample consisting from an analogously treated mix of all replicate gradient input samples. After validation of labeling efficiency by liquid chromatography-tandem mass spectrometry (LC-MS2), samples were mixed by equal protein amount and 4 µg total peptides purified as well as concentrated using STAGE tips 57 . The subsequent LC-MS2 analysis of 50% of that peptide material used an in-house packed 70 μm ID, 15 cm reverse phase column emitter (ReproSil-Pur 120 C18-AQ, 1.9 μm, Dr. Maisch GmbH) with a buffer system comprising solvent A (5% acetonitrile, 0.1% formic acid) and solvent B (80% acetonitrile, 0.1% formic acid). Relevant instrumentation parameters are extracted using MARMo-SET and included in the supplementary material as Source Data File 03 58 . Peptide/ protein group identification and quantitation was performed using the MaxQuant suite of algorithms (v. 1.6.5.0) against the mouse uniprot database (canonical and isoforms; downloaded on 2019/01/23; 86695 entries) 59,60 . For downstream analysis, intensities of fractions were divided by their corresponding inputs and samples that were divided by zero were set to 0.1.
MNase-sequencing and data analysis. For DNA purification from sucrose ultracentrifugation fractions, 200 µl of fractions were resuspended with 200 µl 1× PBS and incubated with 0.5 µl RNase A (10 mg/ml, Sigma Aldrich) for 15 min at 37°C. Samples were resuspended with 400 µl Ultra-pure phenol:chloroform (Invitrogen) and incubated for 5 min at RT. After centrifugation for 5 min at 18,407g at 4°C, the clear phase containing DNA was transferred into a fresh tube. 40 µl of 3 M sodium acetate pH 4.9 and 1 ml of ethanol were added to the samples and DNA was precipitated for 30 min at −80°C. DNA-mix was spun down for 30 min at 18,407g at 4°C and the DNA pellets were washed with 70% ice cold ethanol. After centrifugation for 15 min at 18,407g at 4°C, the pellets were dried at RT and further resuspended in 50 µl of nuclease-free water and heated for 15 min at 37°C. DNA was analyzed by agarose gel electrophoresis or NGS.
For sequencing, purified DNA was quantified by Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific). 10 ng DNA was used as input for TruSeq ChIP Library Preparation Kit (Illumina) with following modifications. Instead of gel-based size selection before final PCR step, libraries were size selected by SPRI-bead based approach after final PCR with 18 cycles. In detail, samples were first cleaned up by 1x bead:DNA ratio to eliminate residuals from PCR reaction, followed by 2-sidedbead cleanup step with initially 0.6x bead:DNA ratio to exclude larger fragments. Supernatant was transferred to new tube and incubated with additional beads in 0.2x bead:DNA ratio for eliminating smaller fragments, like adapter and primer dimers. Bound DNA samples were washed with 80% ethanol, dried and resuspended in TE buffer. Library integrity was verified with LabChip Gx Touch 24 (Perkin Elmer). Sequencing was performed on the NextSeq500 instrument (Illumina) using v2 chemistry with 2x38bp paired setup. Raw reads were visualized by FastQC to determine the quality of the sequencing. Trimming was performed using trimmomatic with the following parameters LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 HEADCROP:5, MINLEN:15. High quality reads were mapped by using with bowtie2 to mouse genome mm10. For downstream analysis, fragments between 100 and 200 bp were selected and the reads were centered. Reads were normalized by an RPKM measure and represented as log2 enrichment over the corresponding inputs. To avoid division through zero, zero counts were pseudo-counted as "1". Replication-deficient lentiviruses containing doxycycline inducible Ctrl (Empty), Hmga2 WT-myc-his and Hmga2 RΔA-myc-his were produced by transient transfection of pCMVR8.74 (Addgene: #22036), pMD2. G (Addgene: #12259) and transfer plasmid into HEK293T cells in a 6-well plate. Viral supernatants were collected after 48 h, spun down at 3,488 ×g for 20 min, and then used to transduce immortalized MEF in the presence of polybrene (10 μg/ml, Sigma). GOF experiments for GADD45A require a high transfection efficiency which were not obtained by standard transfection protocols in MEF. Therefore, Hmga2 was knocked down by shRNA in MLE-12 cells which allowed transient transfection of a GADD45A expression construct. For a stable KD of Hmga2 in MLE-12 cells, lentiviruses for pLKO-scrambled or pLKO-shHmga2 were produced and transduced as described before. Forty-eight h later, MEF and MLE-12 cells were selected by stepwise increase with 1.5 to 3.0 or 4.0 μg/ml puromycin respectively and the pooled populations were used for various experiments.
For chromatin and nucleoplasm preparations, the protocol was adapted with minor modifications from 61 . Cells were washed with 1 x PBS and pellets were resuspended in 2 volumes of lysis buffer (10 mM HEPES pH 7.4, 10 mM KCl, 0.05% NP-40, 1 mM DTT, 25 mM NaF, 0.5 mM Na 3 VO 4 , 40 μg/ml phenylmethylsulfonyl fluoride and protease inhibitor). After 20 min incubation on ice, cells were spun down at 300g at 4°C for 10 min. The supernatant contains the cytoplasmic proteins. The nuclei were washed once in lysis buffer and were then resuspended in 2 volumes of Low Salt Buffer (10 mM Tris-HCl pH7.4, 0.2 mM MgCl 2 supplemented with protease and phosphatase inhibitors including 1% Triton-X100). After 15 min incubation on ice, cells were spun down at 300g at 4°C for 10 min. The supernatant contains the nucleoplasm proteins and the pellet contains the chromatin. Pellets were resuspended in 2 volumes of 0.2 N HCl and incubated on ice for 20 min. Resuspended pellets were spun down at 18,407g at 4°C for 10 min and the supernatant containing acid soluble proteins was neutralized with the same volume of 1 M Tris-HCl pH 8.0. Western blotting was performed using standard methods and antibodies specific for SUPT16 (Cell Signaling; 1:1,000 dilution), SSRP1 (Biolegend; 1:1000 dilution), HMGA2 (Abcam; 1:1000 dilution), H3 (Abcam; 1:5,000 dilution) and AKT (Cell signaling; 1:1000 dilution) were used. Immunoreactive proteins were visualized with the corresponding HRP-conjugated secondary antibodies (Jackson; 1:10,000 dilution) using the WesternBright ECL detection solutions (Biozym). Signals were detected and analyzed with Luminescent Image Analyzer (Las 4000, Fujifilm). Protein concentrations were determined using Bradford kit (Pierce).
DNA-RNA hybrid immunoprecipitation (DRIP)-qPCR. Total nucleic acids were extracted from MEF by SDS/Proteinase K treatment at 37°C followed by phenolchloroform extraction and ethanol precipitation. Free RNA was removed by RNAse A treatment. DNA was fragmented overnight using HindIII, EcoRI, BsrGI, XbaI, and SspI and pretreated, or not, with 40 U RNase H1 (NEB, 5000 U/ml). For DRIP, R-loops were immunoprecipitated using 6 μg DNA-RNA hybrids antibody (Kerafast, Cat. ENH001) per 10 μg of digested DNA in 500 µl IP buffer. Bound Rloops were recovered by addition of 50 μl pre-blocked dynabeads protein A magnetic beads (Thermo Fisher Scientific) followed by two washes and elution in an EDTA/SDS-containing buffer. DNA fragments were treated with Proteinase K and recovered with a QIAquick PCR purification kit (Qiagen). Validation of the DRIP was performed by qPCR. Primer pairs used for DRIP analysis are described in Supplementary Information, Supplementary Table 1.
DNA immunoprecipitation. DNA immunoprecipitation (DIP) analysis was performed as described earlier 62 with minor adaptations. Briefly, homogenized cells in TE buffer were lysed overnight in 20 mM Tris-HCl, pH 8.0, 4 mM EDTA, 20 mM NaCl, 1% SDS at 37°C with 20 µl Proteinase K. Genomic DNA was purified with Phenol: Chloroform, treated with RNAse A and sonicated with Diagenode Bioruptor to an average DNA length of 300-600 bp. Fragmented DNA was re-purified using Phenol: Chloroform extraction and 4 µg of re-purified DNA was was diluted in 450 µl TE buffer. DNA was denatured for 10 min at 95°C and chilled on ice for additional 10 min. Fifty-one µl of 10x IP buffer (100 mM Na-Phosphate, pH 7.0, 1.4 M NaCl, 0.5% Triton X-100) and 4 µg of antibodies specific against 5mC (Abcam), dsDNA (Abcam) and IgG as a control (Santa Cruz) were added to the DNA-TE mix. After 2 h while overhead shaking at 4°C, 40 µl of prewashed Dynabeads (0.1% BSA/PBS) were added for additional 2 h. Beads were washed 3 times in 1x IP buffer, followed by resuspension in 100 µl proteinase K digestion buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 0.5% SDS). Precipitated DNA was removed from the beads by incubating with 100 μl of 50 mM Tris-HCl pH 8.0, 10 mM EDTA, 0.5% SDS and 5 µl Proteinase K (10 mg/ml stock) for 4 h at 37°C. DNA was purified using the QIAquick PCR purification kit (Qiagen) and subjected to qPCR. For qPCR, the percentage of input was calculated after subtracting the IgG background.
Comet assay. Comet assay (Ancam) was performed as described by the manufacturer with minor modifications. MEF were treated with doxycycline for 6 h, harvested by trypzination and mixed with low-melting agarose. Mixture was immediately added onto microscopy slides. Lysis was performed in alkaline lysis solution (1.2 M NaCl, 100 mM Na 2 EDTA, 0.1% sodium lauryl sarcosinate, 0.26 M NaOH, pH >13) overnight. Slides were washed and electrophoresed in 0.03 M NaOH, 2 mM Na 2 EDTA (pH ∼12.3) at 1 V/cm for 25 min. DNA was stained with DNA Vista Dye (Abcam) and images were taken with a confocal microscope. Intensities were measured using ImageJ. The tail length and the extended tail moment were calculated as measure for DNA damage.
DNA nick assay and RNA DNA hybrid prediction. Genomic DNA of Hmga2 + /+, Hmga2−/− and doxycycline inducible MEF was extracted using the Gen-Elute DNA Miniprep kit (Sigma-Aldrich) according to the protocol provided by the manufacturer. Equal amount of total DNA was applied for Real-time PCR analysis. For detection of DNA nicks, primers for DNA nick assay were designed containing the consensus GT or CT sites specific for DNA nicking enzymes 38 (Supplementary Table 1). Nick primers were used with SYBR® Green on the Step One plus Real-time PCR system (Applied Biosystems) and normalized to the Ct values obtained within the surrounding~300 bp DNA region amplified with the flanking primers (Supplementary Table 1). The % of nick DNA was represented as the ratio between: (Nick FWD + Flank RWD) / (Flank FWD + Flank RWD). To determine the directional association of the different ncRNAs associated to the nick DNA area close to the TSS on each target mRNA, we aligned the sequences of the associated ncRNAs using Global Alignment with free end gaps (Geneious 8.1.9, Biomatters Ltd., San Diego, CA). ncRNA:gDNA hybrids were predicted using the RNA hybrid-online server with parameter (MFE < -50 kcal/mol) and consensus alignment was calculated using T-coffee.
RNA sequencing and data analysis. Total RNA of Hmga2 + /+ and Hmga2−/− MEF treated with water or TGFB1 for 24 h was isolated using Trizol (Invitrogen). RNA was treated with DNase (DNase-Free DNase Set, Qiagen) and repurified using the miRNeasy micro plus Kit (Qiagen). Total RNA and library integrity were verified on LabChip Gx Touch 24 (Perkin Elmer). One µg of total RNA was used as input for SMARTer Stranded Total RNA Sample Prep Kit-HI Mammalian (Clontech). Sequencing was performed on the NextSeq500 instrument (Illumina) using v2 chemistry with 1x75bp single end setup. Raw reads were visualized by FastQC to determine the quality of the sequencing. Trimming was performed using trimmomatic with the following parameters LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 HEADCROP:5, MINLEN:15. High quality reads were mapped using with HISAT2 v2.1.0 with reads corresponding to the transcript with default parameters. RNA-seq reads were mapped to mouse genome mm10. After mapping, Tag libraries were obtained with MakeTaglibrary from HOMER (default setting). Samples were quantified by using analyzeRepeats.pl with the parameters (mm10 -count genes -strand + and -rpkm; reads per kilobase per millions mapped). UCSC known genes with a 1.5-fold change upon TGFB1 treatment in Hmga2 + /+ MEF were classified as TGFB1 inducible and used for downstream analysis. To avoid division through zero, those reads with zero RPKM were set to 0.001.
ChIP sequencing after Ctrl versus TGFB1 treatment and data analysis. Precipitated DNA samples were purified by QIAquick PCR purification kit (Qiagen) and quantified by Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific). Two ng of DNA was used as input for TruSeq ChIP Library Preparation Kit (Illumina) with following modifications. Instead of gel-based size selection before final PCR step, libraries were size selected by SPRI-bead based approach after final PCR with 18 cycles. In detail, samples were first cleaned up by 1x bead:DNA ratio to eliminate residuals from PCR reaction, followed by 2-sided-bead cleanup step with initially 0.6x bead:DNA ratio to exclude larger fragments. Supernatant was transferred to new tube and incubated with additional beads in 0.2x bead:DNA ratio for eliminating smaller fragments, like adapter and primer dimers. Bound DNA samples were washed with 80% ethanol, dried and resupended in TE buffer. Library integrity was verified with LabChip Gx Touch 24 (Perkin Elmer). Sequencing was performed on the NextSeq500 instrument (Illumina) using v2 chemistry with 1x75bp single end setup. Raw reads were visualized by FastQC to determine the quality of the sequencing. Trimming was performed using trimmomatic with the following parameters LEADING:3 TRAIL-ING:3 SLIDINGWINDOW:4:15 HEADCROP:4, MINLEN:15. High quality reads were mapped by using with bowtie2 50 to mouse genome mm10. For downstream analysis, reads were scaled based on their read counts and normalized by subtracting reads of the corresponding inputs using deeptools.
Overrepresentation analysis of top 15% genes and proteins enriched at TSS. The top 15% candidate genes with UCSC ID were converted to refSeq using the UCSC genome Table browser. RefSeqs were submitted to DAVID Gene ID conversion tool (version 6.8) to obtain EntrezGene IDs. KEGG pathway enrichment analysis for EntrezGene IDs was performed using the enrichKEGG function from the clusterProfiler v3.0.4 package 64 with a P value cutoff of 0.05, a minimal size of genes annotated by Ontology term for testing of 10, a maximal size of genes annotated for testing of 500, a qvalue cutoff of 0.2. P Value was adjusted using the Benjamini & Hochberg method. Final plots were generated using the dotplot() function.
Overrepresented proteins identified by mass spectrometry in fractions 3 and 4 (1.65-fold enriched, P < 0.05) were identified using WEB-based GEne SeT AnaLysis Toolkit ( [URL]/) using KEGG and Reactome pathway as functional database.
Identification and classification of ncRNAs associated with the top 15% candidate genes. The distribution of the different ncRNA to the gene areas (introns, exons and 3′ UTR and TTS, transcription termination site +5 kb) and promoter (TSS, transcriptional start site, −5 kb) was analyzed using Bedtools v2.15 (intersect -wa -wb), crossing two bed files: the bed file containing the coordinates from the top 15% candidate genes ± 5 kb and the coordinates from the NONCODE database ( [URL]:// www.noncode.org/datadownload/NONCODEv5_mm10.lncAndGene.bed.gz). In total, 52,976 annotated ncRNAs were found in 7,535 of the top 15% candidate genes (79%). For classification of the ncRNAs, duplicated ncRNAs due to isoforms in the top 15% candidate genes were considered only once which decreased the number of analyzed ncRNAs to 28,279. A bed-file containing the coordinates of the unique ncRNAs was analyzed using annotePeaks.pl for mm10 from Homer. Annotations in the 5'UTR or 3'UTR were counted as promoter or TTS enrichment, respectively. For orientation analysis, 7 ncRNAs were removed because of lacking information on transcription orientation. Fasta-sequences of divergent ncRNAs were submitted to MEME-ChIP ( [URL]) and given strands were analyzed for motif enrichment using default parameters.
Crossing of murine TGFB1 with human IPF data and GSEA. IPF RNA-seq samples from GSE116086 ( [URL]? acc=GSE116086) were remapped by the help of bowtie2 to human genome version hg38. Differential gene expression was analyzed using DEseq2 (default) 65 . Human gene name was converted to mouse (mgi_symbol) by the use of getLDS from biomaRt program. IFP-RNA-seq was crossed by mgi_symbol with the Hmga2-RNA-seq after TGFB1 treatment in Hmga2 + /+ MEF. The log2 fold change (log2FC) both RNA-seq was used to perform a 2D kernel density plot by the help of the function kde2d from MASS package v7.3-51.4 with the number of grip points 50. Gene enrichment set analysis (GSEA) was obtained using fgsea (parameters minSize = 10, nperm = 1000) taken the "h.all.v7.0.symbols.gmt" as pathway database. PlotEnrichment was used to plot the two most enriched pathways from the Up-regulated genes from either IPF-RNA-seq or Hmga2-RNA-seq after TGFB1 treatment.
Migration and proliferation assays. Lung fibroblasts, to be assessed for cellular proliferation, were cultured either in 96-well or 48-well plates. Fibroblast proliferation was determined using colorimetric BrdU incorporation assay kit (Roche) according to manufacturer's instructions. Absorbance was measured at 370 nm with reference at 492 nm in a plate reader (TECAN). Depending on the experiment, proliferation of cells was plotted either as the difference of absorbance at 370 and 492 nm (A370 nm-A492 nm) or as a percentage of absorbance compared to control cells absorbance.
Collagen assays. Total collagen content was determined using the Sircol Collagen Assay kit (Biocolor, Belfast, Northern Ireland). Equal amounts of protein lysates from Ctrl and IPF human lung fibroblasts were added to 1 ml of Sircol dye reagent, followed by 30 min of mixing. After centrifugation at 10,000 × g for 10 min, the supernatant was carefully aspirated, and 1 ml of Alkali reagent was added. Samples and collagen standards were then read at 540 nm on a spectrophotometer (Bio-Rad). Collagen concentrations were calculated using a standard curve generated by using acid-soluble type 1 collagen.
Hydroxyproline measurements. Hydroxyproline levels in human lung fibroblasts were determined using the QuickZyme Hydroxyproline Assay kit (QuickZyme Biosciences). The cells and lung tissue were separately homogenized in 1 ml 6 N HCl with a Precellys tissue homogenizer (2 × 20 s, 3,800 g). The homogenate was then hydrolyzed at 90°C for 24 h. After centrifugation at 13,000 g for 10 min, 100 μl from the supernatant was taken and diluted 1:2 with 4 N HCl. 35 μl of this working dilution was transferred to a 96-well plate. Likewise, a hydroxyproline standard (12.5-300 μM) was prepared in 4 N HCl and transferred to the microtiterplate. Following addition of 75 μl of a chloramine T-containing assay buffer, samples were oxidized for 20 min at room temperature. The detection reagent containing p-dimethylaminobenzaldehyde was prepared according to the manufacturer's instruction and 75 μl added to the wells. After incubation at 60°C for 1 h, the absorbance was read at 570 nm with a microtiter plate reader (Infinite M200 Pro, Tecan) and the hydroxyproline concentration in the sample was calculated from the standard curve and related to the employed amount of lung tissue. The hydroxyproline content in lung tissue is given as μg hydroxyproline per mg lung tissue.
Experiments with human PCLS. Control PCLS used for Supplementary Fig. 10d were prepared from tumor-free lung explants from patients who underwent lung resection for cancer. IPF PCLS used for Fig. 9i and Supplementary Figs. 10d-g were prepared from explanted lungs from IPF patients. Both types of tissue were obtained from the KRH Hospital Siloah-Oststadt-Heidehaus or the Hanover Medical School (both Hanover, Germany). Tissue was processed immediately within 1 day of resection as described before 66 . Briefly, human lung lobes were cannulated with a flexible catheter and the selected lung segments were inflated with warm (37°C) lowmelting agarose (1.5%) dissolved in DMEM Nutrient Mixture F-12 Ham supplemented with l-glutamine, 15 mM HEPES without phenol red, pH 7.2-7.4 (Sigma Aldrich), 100 U/ml penicillin, and 100 µg/mL streptomycin (both from Biochrom). After polymerization of the agarose solution on ice, tissue cores of a diameter of 8 mm were prepared using a sharp rotating metal tube. Subsequently, the cores were sliced into 300-350 µm thin slices in DMEM using a Krumdieck tissue slicer (Alabama Research and Development). PCLS were washed 3× for 30 min in DMEM and used for experiments. Viability of the tissue was assessed by an LDH Cytotoxicity Detection Kit (Roche) according to the manufacturer's instruction. For IPF resolution experiments, human IPF PCLS were treated with 50 or 100 μM FACTin for 72 h and the medium with FACTin replenished every 24 h.
Statistics and reproducibility. The source data for all the plots presented in the article, including the values for statistical significance and the implemented statistical tests, are provided in Source Data file 01. Further details of statistical analysis in different experiments are included in the Figures and Figure legends. Briefly, protein enrichment on chromatin and expression analysis of samples were analyzed by next generation sequencing in duplicates of one experiment. Three independent experiments of the mass spectrometry-based proteomic approach were performed. For the rest of the experiments presented here, samples were analyzed at least in triplicates and experiments were performed three times. Statistical analysis was performed using GraphPad Prism 5 and Microsoft Excel. Data in bar plots are represented as mean ± standard error (mean ± s.e.m.). t-Tests were used to determine the levels of difference between the groups and P-values for significance. P-values after one-or two-tailed t-test, *P ≤ 0.05; **P ≤ 0.01 and ***P ≤ 0.001. In the Figs. 1b, 3b, 4f, 8b, 8c and Supplementary Figure 9b, P-values were determined using Wilcoxon-Mann-Whitney test. In the 2D Kernel Density plot presented in Fig. 9a the statistical significance was calculated using DESeq2's integrated Wald test.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.
Data availability
The data that support this study are available from the corresponding author upon reasonable request. In addition, sequencing data of ChIP, RNA and MNase have been deposited in NCBI's Gene Expression Omnibus 67 and is accessible through GEO Series with accession number GSE141272. The mass spectrometry-based interactome data have been deposited into the PRIDE archive and assigned to the project accession PXD016586. In addition, we retrieved and used a number of publicly available datasets to aid analysis of our data: pH2A. X and HMGA2 ChIP-Seq data: GSE63861 10 The source data are provided with this paper.
==
Domain: Biology Chemistry Medicine
|
[{"added": "2020-03-12T10:39:53.173Z", "created": "2020-03-08T00:00:00.000", "id": "214722432", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Chemistry"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41467-021-21227-y.pdf", "pdf_hash": "44e620854ed076137b4f1d8856475e1738dec351", "pdf_src": "SpringerNature", "provenance": "peS2o-0028.json.gz:7433", "s2fieldsofstudy": ["Biology"], "sha1": "b991c563b2e84efbb01fb6d6fd3884a2de77d244", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Histone Acetyltransferase 1 is Required for DNA Replication Fork Function and Stability
The replisome functions in a dynamic environment that is at the intersection of parental and nascent chromatin. Parental nucleosomes are disrupted in front of the replication fork. The daughter duplexes are packaged with an equal amount of parental and newly synthesized histones in the wake of the replication fork through the action of the replication-coupled chromatin assembly pathway. Histone acetyltransferase 1 (Hat1) is responsible for the cytosolic diacetylation of newly synthesized histone H4 on lysines 5 and 12 that accompanies replication-coupled chromatin assembly. Analysis of the role of Hat1 in replication-coupled chromatin assembly demonstrates that Hat1 also physically associates with chromatin near sites of DNA replication. The association of Hat1 with newly replicated DNA is transient but can be stabilized by replication fork stalling. The association of Hat1 with nascent chromatin may be functionally relevant as loss of Hat1 results in a decrease in replication fork progression and an increase in replication fork stalling. In addition, in the absence of Hat1, stalled replication forks are unstable and newly synthesized DNA becomes susceptible to Mre11-dependent degradation. These results suggest that Hat1 links replication fork function to the proper processing and assembly of newly synthesized histones.
INTRODUCTION
The central event in the division of a cell is the duplication of its chromosomes.
Chromosome duplication requires the proper functioning of two interconnected processes. The first is the replication of the genomic DNA. The second is the duplication of the chromatin structure that governs the correct packaging and architecture of the chromosomes in the nucleus. The successful coordination and completion of these processes is essential to ensure genome stability, maintain correct patterns of gene expression and properly regulate cell proliferation.
DNA replication occurs in a unique and highly dynamic chromatin environment. The replication fork must navigate through the chromatin structure in front of the fork by disrupting nucleosomes in its path. In the wake of the replication fork, the nascent daughter duplexes must be rapidly assembled into nucleosomes. Nascent chromatin on the daughter duplexes is assembled from two distinct pools of histones; parental and newly synthesized.
The parental histones are derived from the nucleosomes disrupted during the passage of the replication fork. These nucleosomes dissociate into stable H3/H4 tetramers and H2A/H2B dimers. Regulation of parental histone recycling, mediated by the histone chaperone Asf1, is critical for proper replication fork function and stability (1). Asf1 functions in conjunction with the MCM2-7 replicative helicase and RPA to remove parental histones H3 and H4 from in front of the replication fork and transfer them to the newly replicated DNA near their original genomic location (2)(3)(4)(5)(6). Loss of Asf1 or disruption of Asf1 activity through histone over-expression impedes DNA unwinding and replication fork progression (7)(8)(9). Other factors, such as FACT and the POLE3-POLE4 complex are also involved in processing parental histones at the replication fork and may be involved in the association of H2A/H2B dimers with the H3/H4 tetramers (10)(11)(12).
Several recent studies have shown that replication-coupled chromatin assembly is required for replication fork function. In these studies, the supply of histones to the replication fork was blocked, either by preventing new histone protein synthesis or disrupting histone deposition by depleting CAF-1 or Asf1 (39,(48)(49)(50)(51). There is also evidence that the post-translational modifications on newly synthesized histones can influence replication fork function. Histone H3 lysine 56 acetylation has been shown to positively regulate binding of histones to CAF-1 (38). Consistent with this, loss of H3 lysine 56 acetylation and CAF-1 have similar effects on DNA replication in S. cerevisiae (49). Histone deacetylases, HDAC1 and HDAC2, which have been proposed to deacetylate newly synthesized histones following their assembly into chromatin, have been shown to be important for replication fork function and for the stabilization of stalled replication forks in conjunction with the WRN helicase (52)(53)(54)(55)(56).
Recent evidence suggests that Hat1 also has the potential to influence replication fork function. Studies in a wide range of eukaryotes show that loss of Hat1 sensitizes cells to DNA double strand breaks and causes HU sensitivity and genome instability in mammalian cells (19,57,58). In addition, it was recently reported that Hat1 is transiently recruited to chromatin during replication-coupled chromatin assembly and affects the protein composition of nascent chromatin (59). Therefore, we investigated whether Hat1 provides a link between the processing and assembly of newly synthesized histones and replication fork function. We confirm that Hat1 transiently associates with newly replicated DNA. We show that loss of Hat1 induces a dramatic reduction in replication fork progression and increases replication fork stalling. We also demonstrate that stalling of replication forks stabilizes the association of Hat1 with newly replicated DNA and that loss of Hat1 leads to destabilization of stalled forks and MRE11-dependent degradation of newly synthesized DNA.
MATERIALS AND METHODS
Cell culture conditions. Mouse embryonic fibroblasts were prepared as previously described (19). Cells were grown in DMEM (Sigma) supplemented with 10%FBS (Sigma) and Penicillin/Streptomycin (Gibco). Images were acquired using MetaMorph version 7.8.10 and quantification was completed using ImageJ version 1.52t according to a previously described protocol (63).
DNA fiber assay. DNA was labeled with 50µM and 250µM for 20 minutes each. HU (Sigma) was used at 4mM for 5 hours; Mirin (Sigma) was used at 100µM for 5 hours.
After labeling and treatment, cells were collected by trypsinization and resuspended in PBS. 2µL of the cells suspension were spotted on a glass slide and lysed with lysis buffer (0.5% SDS, 200 mM Tris-HCl, pH 7.4, 50 mM EDTA) for 10min, slides were then tilted to 15° to stretch the DNA fibers and fixed with Methanol/Acetic Acid (3:1) overnight at 4 degrees. Next day DNA was denatured with 2.5N HCl for 30min and wash several times with PBS before blocking with 1%BSA/PBS for 30min. Rat anti-BrdU (1:50, AbD Serotec) was used to detect CldU, and mouse anti-BrdU (1:20, Becton Dickinson) to detect IdU. Antibodies were diluted in blocking buffer and incubated for 1 hour at room temperature. AlexaFluor 594-conjugated anti-rat (1:250, Molecular Probes) and AlexaFluor 488-conjugated anti-mouse (1:250, Molecular Probes) were used as secondary antibodies and incubated for 1 hour at room temperature. Slides were mounted with Vectashield with DAPI.
Immunofluorescence. Cells were seeded on coverslips and allowed to attach for 24 hours. Next day the cells were fixed with 4% PFA at room temperature for 10 minutes, washed several times with PBS and permeabilized with 0.5% Triton X-100/PBS for 15 minutes at room temperature, after several PBS washes cells were blocked with 5% BSA in PBS for 30 minutes at room temperature. Anti-phosphorylated ATR (Ser 428) (Cell Signaling #2853 1/100) was incubated overnight at 4 degrees. Next day after several washes, secondary AlexaFluor 594-conjugated anti-rabbit was diluted 1/250 and incubated 1 hour at room temperature. Antibody excess was extensively washed and slides were mounted with Vectashield with DAPI.
Comet Assay. The Comet Assay kit (Trevigen, Gaitherburg,MD) was used according to the manufacture instructions. Briefly, MEFs were resuspended in ice cold PBS (Ca2+ and Mg2+ free) to a concentration of 1 × 10 5 cells/ml. 5 µl cells were mixed with 50 µl of warm low melting Agarose and 50 µl were evenly spread onto the special comet slides.
Slides were stored at 4 °C in the dark and transferred to pre-chilled lysis solution for 60 minutes at 4 °C. Next, slides were transferred to alkali unwinding solution at room temperature for 60 minutes. Slides were transferred to electrophoresis tank which contained pre-chilled Alkaline electrophoresis solution and run at 1 Volt/cm, 300 mA for 45 minutes at 4 degrees. The slides were immersed twice in deionized water for 5 minutes intervals and washed in 70% ethanol for 5 minutes. Then cells were stained with 100 µl of SYBR Green I for 5 minutes in the dark and slides were analyzed under Zeiss Axiophot fluorescence microscope. Images were taken using Metavue software version 6.3r2 software and comet tails were analyzed using opencomet by Imagej.
RESULTS
Hat1 transiently localizes to newly replicated DNA. Current models of replicationcoupled chromatin assembly predict that Hat1 associates with, and modifies, newly synthesized histone H4 in the cytoplasm before transferring the modified histones to Asf1 for subsequent nuclear import and deposition. However, recent results using iPOND (isolation of proteins on nascent DNA) suggested that Hat1 becomes transiently associated with newly replicated DNA (59). As this has the potential to significantly expand the role of Hat1 in genome duplication, we sought to confirm this observation.
Proximity ligation assay-based chromatin assembly assays (CAAs) have recently been developed and serve as a powerful method for analyzing protein dynamics on newly replicated DNA (60)(61)(62)(63). The proximity ligation technique determines whether two molecules reside close to each other in the cell by employing two species-specific secondary antibodies that are fused to oligonucleotides. If the secondary antibodies recognize primary antibodies that are in close proximity, the oligonucleotides can both bind to a nicked circular DNA, creating a template for rolling circle replication. This amplifies sequences that can be bound by a fluorescent probe and visualized. To adapt this for use as a chromatin assembly assay, newly replicated DNA is labeled by the incorporation of the thymidine analog IdU. The proximity of proteins to newly replicated DNA is detected using antibodies against the protein of interest and antibodies recognizing IdU. To validate the CAA, we monitored the localization of PCNA, H4 lysine 5 acetylation and H4 lysine 12 acetylation to newly replicated DNA in Hat1 +/+ and Hat1 -/-MEFs (mouse embryonic fibroblasts). As seen in Figure 1A, quantitation of the CAA precisely mirrored the results previously obtained with iPOND. The localization of PCNA to newly replicated DNA was Hat1-independent and the acetylation of H4 lysines 5 and 12 required Hat1 (19,59).
Using α-Hat1 antibodies, we tested whether Hat1 is in proximity to newly replicated DNA. There is abundant CAA signal in Hat1 +/+ cells and only background in the Hat1 -/cells ( Figure 1B). We next asked whether Hat1 is transiently associated with newly synthesized DNA or whether it is stably bound to chromatin. We performed CAA assays immediately following a pulse of IdU and after 15, 30 and 60 minutes of a thymidine chase. As seen in Figure 1C, the level of Hat1 on newly replicated DNA is significantly reduced after a 15 minute chase and is completely lost after 30 minutes.
Intriguingly, if replication forks are stalled by the addition of HU, Hat1 association with newly replicated DNA is stabilized for extended periods of time (at least 5 hours). These data verify that Hat1 is transiently associated with nascent chromatin near sites of DNA replication and becomes stably associated when replication forks stall.
Hat1 is required for normal replication fork progression. The physical association
of Hat1 with the highly dynamic chromatin at sites of DNA replication greatly expands the spectrum of potential functions for this enzyme in genome duplication. In particular, this raises the possibility that Hat1 plays a direct role in replication fork function or stability. To test this, we used DNA fiber analysis in Hat1 +/+ and Hat1 -/-MEFs ( Fig. 2A).
Hat1 +/+ and Hat1 -/cells were incubated with CldU, followed by IdU incubation for equal times and replication fork progression was measured by DNA fiber analysis in which antibodies targeting the CldU (red) and IdU (green) are used to label the newly replicated DNA with different colors. The relative rates of replication fork progression were determined by measuring the lengths of the IdU tracts that are located at junctions with CldU labeled DNA, as this ensures that the replication fork was functional at the beginning of the IdU incubation. We observed a significant decrease in the length of labeled DNA fibers in the Hat1 -/cells, indicating that DNA replication progressed more slowly in the absence of Hat1. Consistent with an effect of Hat1 loss on replication fork function, analysis of PCNA dynamics at the replication fork by CAA showed that PCNA dissociation is significantly delayed in the absence of Hat1 ( Figure 2B).
Loss of Hat1 increases replication fork stalling. A decreased rate of DNA replication
can be due to decreases in the velocity of the replication fork or increases in the frequency of replication fork stalling. To test the latter possibility, we stained Hat1 +/+ and Hat1 -/cells with antibodies against phosphorylated ATR (Ser428). ATR is recruited to single stranded DNA at sites of replication fork stalling where it is activated by phosphorylation. Loss of Hat1 resulted in an increased number of cells positive for phospho-ATR foci (Fig. 3A).
To confirm the increase in replication fork stalling, we used the CAA to measure the association of Rad51 with the single strand DNA that is created at stalled replication forks (64). As seen in Figure 2B, there was a significant increase in the association of Together, these data indicate that Hat1 is necessary for proper replication fork function and the prevention of replication stress.
Hat1 is critical for the stability of stalled replication forks. As seen in Figure 1C, Hat1 is stably associated with stalled replication forks. To determine whether Hat1 is involved in maintaining the stability of stalled replication forks, we analyzed the stability of newly replicated DNA at stalled replication forks using the DNA fiber assay. Hat1 +/+ and Hat1 -/cells were treated with CldU and IdU sequentially for equal lengths of time.
HU was then added to induce replication fork stalling. After 5 hours, the lengths of the IdU and CldU tracts were measured. If the stalled replication forks remain stable, the ratio of IdU tract length to CldU tract length will be 1. If the newly replicated DNA (represented by the IdU labeled DNA) at the stalled forks is unstable, the ratio of IdU tract length to CldU tract length will be less than 1. As seen in Figure 4A, there was a significant decrease in the IdU tract length in the absence of Hat1. We conclude that Hat1 is required for the protection of newly replicated DNA at stalled replication forks.
The degradation of newly replicated DNA at stalled replication forks is the result of Mre11 nuclease activity (66)((67). To determine whether the instability of nascent DNA in the absence of Hat1 is also Mre11-dependent, Hat1 +/+ and Hat1 -/cells were sequentially treated with CldU and IdU for equal lengths of time. The cells were then treated with HU in the presence of Mirin, a specific inhibitor of Mre11 activity. As seen in Figure 4B, newly replicated DNA is equally stable in Hat1 +/+ and Hat1 -/cells when Mre11 activity is inhibited. These data indicate that Hat1 functions to protect newly replicated DNA from Mre11-mediated degradation.
We used a comet assay to determine whether Hat1-dependent replication fork instability led to a decrease in the ability of cells to recover from replication stress. Hat1 +/+ and Hat1 -/cells were treated with HU for 3 hours and then allowed to recover for 12 hours in the absence of HU. As seen in Figure 4C, Hat1 +/+ cells were better able to recover from prolonged replication stress than the knock out cells, consistent with a loss of replication fork integrity in the absence of Hat1.
DISCUSSION
Contrary to the predictions of current models of replication-coupled chromatin assembly, our results demonstrate that Hat1 localizes to chromatin at sites of DNA replication.
There are several models to explain the localization of Hat1 to nascent chromatin. First, Hat1 may not transfer H3/H4 dimers to Asf1. Rather, Hat1 may remain associated with the H3/H4 dimers throughout the entire replication-coupled chromatin assembly process and load onto newly replicated DNA through CAF-1-mediated deposition of H3/H4/Hat1 complexes. This model is consistent with numerous proteomic studies that have identified the Hat1 complex as major components of soluble H3 and H4 complexes (22,26,(68)(69)(70)(71)(72). Alternatively, following transfer of H3/H4 dimers to Asf1, Hat1 may enter the nucleus independently and bind to nascent chromatin after histone deposition. Finally, Hat1 may participate in an additional chromatin assembly pathway distinct from the Asf1/CAF-1 pathway. One potential pathway may utilize the histone chaperone NASP. A distinct nuclear yeast Hat1 complex contains histones H3 and H4 and a histone chaperone, Hif1, which is the yeast homolog of NASP (15,21,68).
Subsequent experiments have shown that NASP also interacts with the Hat1 complex in mammalian cells (22). NASP is important for buffering the pools of soluble H3/H4, particularly under conditions of replication stress, and can form a multi-chaperone complex with Asf1. Several studies have shown that NASP can function as a nucleosome assembly factor in vitro (73)(74)(75). Hence, this model predicts that Hat1 localizes to newly replicated DNA in conjunction with NASP-mediated deposition of H3/H4. It is clear that DNA replication is coupled to newly synthesized histone deposition. This link was originally suggested by studies demonstrating that DNA replication required active protein synthesis(76-78). More recently, histone supply was more directly linked to replication fork function by experiments that specifically limited histone production (48,50). The assembly of newly synthesized histones into chromatin was directly implicated in replication fork function through the identification of DNA replication defects in cells lacking components of the replication-coupled chromatin assembly pathway, such as Asf1 and CAF-1 (39,49,51). (59). Decreased levels of these proteins in the proximity of replication forks may create an altered chromatin structure that negatively affects replication fork function or they may be directly involved in replisome function. Indeed, it was recently shown that Brd2, Brd3 and Brd4 function at the replication fork to antagonize the ATAD5-mediated unloading of PCNA, which is consistent with our observation that PCNA unloading is slowed in Hat1 -/cells (79). Finally, the presence of Hat1 on nascent chromatin near replication forks suggests that Hat1 may directly modify and regulate components of the replisome.
The association of Hat1 with nascent chromatin is transient but becomes stable if replication forks stall. The physical association of Hat1 with stalled forks is likely to be functionally relevant as Hat1 is required for the stability of stalled replication forks. An attractive mechanism for the role of Hat1 in replication fork stabilization involves the recruitment of Rad51 to stalled forks. Rad51 binds to single strand DNA at stalled replication forks and plays a central role in maintaining replication fork stability. Hat1 forms an S-phase-specific complex with Rad51 and is involved in the recruitment of Rad51 to DNA double strand breaks (80). However, we do not detect any decrease in Rad51 localization to stalled replication forks in Hat1 -/cells, suggesting that the mechanisms for Rad51 recruitment to DNA double strand breaks and stalled replication forks are distinct.
It has also been suggested that Hat1 is involved in the initiation of DNA replication.
Studies in yeast showed that Hat1 physically interacts with the origin recognition complex (ORC). In addition, combining mutations in Hat1 with temperature sensitive alleles of ORC components or CDC45 resulted in synthetic growth defects. Hat1 was also recruited to origins of replication at the time of origin activation. Despite these connections, there were no defects in replication origin firing in Hat1 mutants in yeast(81).
Our results suggest an update to current models of replication-coupled chromatin assembly to incorporate the localization of Hat1 to nascent chromatin at sites of DNA replication. In addition, our results indicate that Hat1 lays a direct and integral role in both genome and epigenome duplication.
Acknowledgements
This work was support by a grant form the National Institutes of Health (R01 GM062970 to M. R. P.)). Microscopy was supported by a grant from the NIH/NINDS (P30 NS104177).
==
Domain: Biology Chemistry Medicine
|
[{"added": "2020-03-31T20:53:35.046Z", "created": "2020-03-18T00:00:00.000", "id": "214722859", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Chemistry"], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "http://www.jbc.org/article/S0021925817492313/pdf", "pdf_hash": "98f6663aef0074a5fa7c285524e95a4f06029dbd", "pdf_src": "BioRxiv", "provenance": "peS2o-0028.json.gz:5165", "s2fieldsofstudy": ["Biology", "Medicine"], "sha1": "98f6663aef0074a5fa7c285524e95a4f06029dbd", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Standing Genetic Variation and the Evolution of Drug Resistance in HIV
Drug resistance remains a major problem for the treatment of HIV. Resistance can occur due to mutations that were present before treatment starts or due to mutations that occur during treatment. The relative importance of these two sources is unknown. Resistance can also be transmitted between patients, but this process is not considered in the current study. We study three different situations in which HIV drug resistance may evolve: starting triple-drug therapy, treatment with a single dose of nevirapine and interruption of treatment. For each of these three cases good data are available from literature, which allows us to estimate the probability that resistance evolves from standing genetic variation. Depending on the treatment we find probabilities of the evolution of drug resistance due to standing genetic variation between and . For patients who start triple-drug combination therapy, we find that drug resistance evolves from standing genetic variation in approximately 6% of the patients. We use a population-dynamic and population-genetic model to understand the observations and to estimate important evolutionary parameters under the assumption that treatment failure is caused by the fixation of a single drug resistance mutation. We find that both the effective population size of the virus before treatment, and the fitness of the resistant mutant during treatment, are key-parameters which determine the probability that resistance evolves from standing genetic variation. Importantly, clinical data indicate that both of these parameters can be manipulated by the kind of treatment that is used.
Introduction
For most HIV patients, treatment with modern antiretroviral therapy leads to a rapid decline of viral load (VL) of several orders of magnitude. However, when the virus acquires resistance to one or more drugs, treatment can fail. It is still an open question whether the mutations responsible for resistance originate usually from standing genetic variation (also referred to as pre-existing mutations or minority variants), or from new mutations which occur during therapy. In fact, there is no single biological system for which the relative role of pre-existing and new mutations is well known [1]. Another important open question is whether multiple simultaneous mutations are needed for the viral population to be able to grow during therapy, or whether a single mutation allows escape. Amongst evolutionary biologists, it is commonly assumed that therapy with multiple drugs works so well because the virus needs multiple mutations to escape, which is unlikely to happen. However, patient data show that patients often fail therapy with a single resistance mutation [2,3] which suggests that a single mutation can increase the fitness of the virus to above 1, even though the virus is still susceptible to two of the drugs in the treatment. In this scenario, the main benefit of combination therapy over monotherapy would be that combination therapy reduces the population size of the virus and therefore the probability that mutations occur. In this study we will analyze patient data under the assumption that a single mutation can lead to virologic failure and thereby propose an alternative view on the evolution of drug resistance during multi-drug therapy.
We will look at the establishment of drug resistance mutations in three different situations: (1) when triple-drug therapy (ART) is started for the first time, (2) when pregnant women are treated with a single dose of nevirapine (sdNVP) to prevent infection of the baby during birth and (3) when ART is interrupted and restarted (an overview of abbreviations is given in table 1). We will argue that standing genetic variation plays a crucial role in each of these cases. We find that the probability that resistance mutations become established in each of these cases can be understood by using a simple population genetic model.
For readers who are not familiar with HIV, it is important to know that the genotype-phenotype map for drug resistance in HIV is very well known. Lists of the important resistance mutations for each drug are published (e.g., in the International AIDS Society-USA drug resistance mutations list, [4], so that doctors can compare the genotype of the virus of a patient before treatment with this list to decide which drugs to prescribe. The aim of treatment is to achieve viral suppression. If treatment fails, i.e., the viral load stays or becomes higher than a predetermined threshold, such as 50/ml, despite adherence to the regimen, a second genotypic test will be performed to see whether the virus has acquired new resistance mutations. Since the second half of the 1990s, treatment is usually with a combination of three drugs, which are chosen such that mutations which confer resistance against one of the drugs do not confer cross-resistance against the other two drugs. Soon after its introduction, it became clear that triple-drug therapy was an enormous success and saved the lives of many HIV patients [5]. One reason why therapy with three drugs works better than treatment with one or two drugs is that the rate at which resistance evolves is slower when patients are treated with three drugs [6]. It is commonly thought that resistance does not evolve in patients on triple-drug therapy because it would require a viral particle to acquire three mutations at the same time. However, in patients who are treated with triple-drug therapy, it is often observed that resistance against one of the drugs evolves, at least initially. Data from several cohort studies in different parts of the world, such as from Canada [2] and the UK (UK CHIC cohort study [3], clearly show that in most patients who fail therapy due to resistance, the virus is resistant against one of the drugs and almost never against all three. The UK study, for example, reports that out of 4306 patients who started therapy between 1996 and 2003, after two years of therapy, 13% have drug resistance. A majority of the patients with drug resistance (7%) have resistance against just one of the drugs. Less than half of the patients (6%) have resistance against more than one class of drugs and a only small number of patients (1%) have resistance against 3 classes of drugs, even though all patients of this cohort were treated with three classes of drugs. These data show that treatment can fail due to resistance against one of the drugs in a regimen. In such cases, it may be that the other two drugs cannot keep the VL completely suppressed, even though they still work. The viruses that have acquired resistance against two or three classes of drugs may have acquired these mutations at the same time or they may have acquired them one by one. For now, we will assume the latter and focus only on the probability of acquiring the first drug resistance mutation (DRM).
For many common drugs, especially reverse transcriptase inhibitors, a single mutation can confer resistance against the drug and only a small number of mutations is responsible for resistance in most patients. For example, resistance against the drug nevirapine is almost always due to one of two amino acid changes, namely K103N or Y181C in the reverse-transcriptase gene [7]. Because of the importance of a small number of mutations, several studies have investigated whether these mutations are present in untreated patients due to transmitted drug resistance or due to spontaneous mutation. Recent studies have used allele-specific PCR and related methods to determine the frequency of several important mutations in untreated patients. Low-frequency drug resistance mutations (DRMs), likely due to spontaneous mutation (and not transmitted from other patients) were detected in up to 40% of patients (see [8] for an overview). The detection of drug resistance mutations in untreated patients, together with the knowledge that a single mutation can confer resistance against a drug and allow viral escape, suggest that preexisting resistance mutations (or standing genetic variation in the population genetic jargon) may play an important role in the evolution of drug resistance in HIV.
Throughout the paper, we will assume that a single mutation can allow viral escape and we focus on the probability that such a first drug resistance mutation becomes established (i.e., it reaches such a frequency that it can be expected to become the majority variant unless treatment is stopped or changed quickly). What happens after a first mutation has become established, or how fast such an established mutation wanes in the absence of treatment are important questions, but they fall outside the scope of this study. In this paper, ''triple-drug therapy'' and ART refer to treatment with two drugs of the class NRTI plus either an NNRTI or an unboosted PI (for a list of abbreviations in the paper, see Table 2). The results are likely to be different for other drug combinations.
Starting therapy
When a patient starts therapy for the first time, one would expect that there should be a substantial probability that drug resistance evolves due to pre-existing DRMs. Indeed, recent studies have shown that the presence of drug resistance mutations at low frequency (v1%) increases the risk that treatment fails (e.g., [9], [10], [7], see [11] for a review). However, the situation is not as simple as one may hope: even if no pre-existing DRMs can be detected, resistance mutations may become established quickly, and even if DRMs are detected, treatment is still successful in the majority of patients. We will attempt to understand those observations using population genetic theory. Other authors have looked at the question of pre-existing DRMs previously (e.g., [12],
Author Summary
For HIV patients who are treated with antiretroviral drugs, treatment usually works well. However, the virus can, and sometimes does, become resistant against one or more drugs. HIV drug resistance results from the acquisition of specific and well known mutations. It is currently unknown whether drug resistance mutations usually stem from standing genetic variation, i.e., they were already present at low frequency before treatment started, or whether they tend to occur during treatment. In the current manuscript, I make use of several large datasets and evolutionary modeling to estimate the probability that drug resistance mutations are present before treatment starts and lead to viral failure. I find that for the most common type of treatment with a combination of three drugs, drug resistance evolves from pre-existing mutations in 6% of the patients. With other types of treatment, this probability varies from 0 to 39%. I conclude that there is room for improvement in preventing the evolution of drug resistance from pre-existing mutations.
[13]), however, it is worth reconsidering the topic. First of all, we now have a wealth of data available for pre-existing DRMs and the establishment of drug resistance mutations in HIV patients, and secondly, we now have a better theoretical framework to consider the role of standing genetic variation for adaptation [14].
Prevention of mother to child transmission (PMTCT)
Pregnant women in low resource settings are often treated with a single dose of the non-nucleoside reverse transcriptase inhibitor neverapine when labor starts. Single dose nevirapine (sdNVP) is the cheapest and simplest way to reduce the probability of mother-tochild-transmission, but it is shown to lead to the establishment of drug resistance mutations in the mothers and the babies. In a metaanalysis [15], found that, in 7 different studies, on average 44% of the patients treated with sdNVP had detectable NVP resistance mutation several weeks after the treatment. The presence of such mutations makes future treatment of these women harder [16]. To avoid the establishment of resistance mutations, several alternative strategies are used in combination with sdNVP. We will use the same population genetic framework as in the other two cases to try to understand why sdNVP leads to establishment of resistance mutations in so many patients, and how this can be avoided. In the current study we will only focus on the probability that NVP resistance mutations become established during treatment for PMTCT. The issue of how these mutations wane and possibly resurface when treatment is started again is important and interesting but falls outside the scope of the current paper.
Treatment interruptions
It was long suspected that treatment interruptions lead to drug resistance. Indeed, cohort studies show that treatment interruptions due to non-adherence are associated with faster accumulation of drug resistance mutations ( [17,18,19]. Clear evidence that treatment interruptions of at least a couple of weeks lead to the establishment of resistance mutations comes from clinical trials (e.g., [20,21,22] which were done in a time when it was believed that treatment interruptions may be beneficial for patients. In 2006 the SMART trial was stopped because treatment interruptions were shown to have a negative effect on patients' health [23]. However, treatment interruptions still occur, for example, when a patient is forgetful or is unable to purchase drugs due to financial or logistic barriers. It is important to understand how treatment interruptions lead to resistance and whether this effect can be avoided. The main idea that currently governs the thinking about treatment interruptions and resistance is that insufficient druglevels allow for replication and, at the same time, select for resistance (e.g., [24,25,18]. This effect is aggravated when drugs that are part of combination therapy have very different half-lifes, so that interrupting combination therapy can result in effective monotherapy. It is generally believed that this ''tail of monotherapy'' is the main reason why treatment interruptions lead to drug resistance. However, several observations are not compatible with the ''tail'' hypothesis. For example, Fox et al ( [25]) found no significant difference in the number of resistance mutations after simultaneous, ''staggered'' or ''switched'' treatment interruptions in patients from the SMART trial (a ''staggered'' stop means that the long half-life drug is interrupted several days before the other drugs and a ''switched'' stop means that before interrupting, patients switch to a regimen with only short half-life drugs). In addition, the ''tail'' hypothesis fails to explain why treatment interruptions increase the risk of resistance in patients on protease inhibitor-based (PI) regimens which do not have long half-lifes [26,20,27,28,29,30]. Another explanation is therefore needed to understand the observed patterns.
When treatment is interrupted, the viral load rapidly increases until it has reached its original level after approximately four weeks [31]. Basic population genetics tells us that such population growth also leads to an increase in the probability that DRMs are present. When treatment is started again, selection may work on such preexisting mutations, which provides a simple explanation for how treatment interruptions lead to the establishment of resistance mutations.
In this paper we will attempt to explain the observed patterns by considering selection on pre-existing variation and selection on new mutations. It is worth noting here that pre-existing does not necessarily mean old, such a mutation may have originated just a day before the start of treatment. Throughout the paper, we use a mathematical model for adaptation from standing genetic variation which we developed previously [14] and forward-intime, individual-based computer simulations. The model captures mutation, drift and selection, including changing selection pressures (due to stopping and starting of therapy) which lead to changes in population size. Because we only focus on the establishment of the first drug resistance mutation, we can ignore epistatic interactions between different drug-resistance mutations and recombination. In each of the three cases of interest, we use published data on the percentage of patients with established drug resistance mutations to estimate important parameter values (for starting ART or sdNVP) and to predict outcomes (for treatment interruptions).
Model
Model, assumptions and fixation probability of a drug resistance mutation The model we use in this paper describes the population dynamics and population genetics of a panmictic viral population in a single patient. Details of the model can be found in the supplementary material. We assume that as long as the patient is not treated, the viral population will be stable at population size N u (u for untreated). Drugs reduce the fitness of the wildtype virus to below 1 so that the population will shrink. We assume that there is a large reservoir of latently infected cells of which a fixed number (N LAT ) become activated per generation, so that the virus can not die out. Drug resistant virus can be created by mutation and is assumed to be resistant against one of the drugs in the treatment regimen. If the patient is not taking drugs, the drug resistant virus is less fit than the wildtype by a factor C rel (relative cost of the resistant virus), but if the patient is taking drugs, the resistant virus has a fitness that is higher than 1 (Fm ART w1), whereas the wildtype has a fitness lower than 1 (Fwt ART v1). In reality, there may also be resistance mutations that confer resistance against one of the drugs, but that do not lead to a fitness higher than 1. Such mutations will quickly die out and can safely be ignored in the model. Throughout the paper we focus on the the processes that allow a first major drug resistance mutation to become established in the patient. Patients are assumed to be ART-naive and have no transmitted drug resistance.
Evolutionary biologists have long known that most mutations will be lost by genetic drift even if they confer a fitness benefit [32]. This is also true for drug resistance mutations (DRMs) in patients on anti-retroviral therapy, although it is all too often ignored in drug resistance studies. The clinical relevance of this old result has recently become very clear. It was found in several studies that even though low frequency DRMs increase the risk of treatment failure and establishment of drug resistance, the majority of patients with detected low frequency DRMs will respond well to treatment [7]. This result shows that DRMs can die out, even if they have reached frequencies high enough to be detected. The reason is probably that most viral particles will not infect any new cells and produce no new viral particles, even if, on average, they produce more than 1.
The probability that a DRM becomes established in the patient depends on the number of copies that are present, the average number of offspring of the drug resistant particles and the variance in offspring number. Traditionally, fixation or establishment probabilities are calculated using the relative fitness difference between the mutant and the wildtype, but in the case of HIV it is more useful to use the fitness of the mutant virus to calculate its establishment probability. The reason is that anti-retroviral therapy works so well that wildtype fitness may be very low (much lower than 1). In such case fitness of the mutant may not be related to the fitness of the wildtype and because the wildtype cannot grow, the two types do not compete for resources. In other words, the mutant can occupy a niche that is not occupied by the wildtype. In those cases, and as long as Fm ART {1vv1, the establishment probability of the mutant will be approximately where s 2 is the variance in offspring number. In the simulations and throughout this paper, we use the variance effective population size, in which case one can assume that s 2~1 , so that Note that by setting s 2~1 , we ignore all mutations which occur in virus which is not part of the effective population size. The establishment probability of a mutation in a random viral particle (e.g., when observed in a patient) may be much lower. It is important to realize that if the establishment probability of a DRM depends on its absolute fitness, anything that reduces its fitness will reduce the establishment probability. For example, if a drug that is added to a regime reduces fitness of both wildtype and resistant virus, then it will reduce the probability that a pre-existing resistant mutant becomes established. This is true even if the effect of the added drug on wildtype and resistant virus is exactly the same. Similarly, if the immune system works well, this may also reduce the probability of establishment.
In most population genetics models, the focus is on the fixation probability, rather than the establishment probability of a mutation. And in many models, if a mutation becomes established, it will go to fixation. However, if selection pressures change, establishment does not necessarily lead to fixation. This is especially clear when we will later consider the effect of a singledose of nevirapine. A few weeks after a single dose of nevirapine, nevirapine resistance mutations can be detected in a large proportion of patients, but these mutations may never take over the whole viral population, because the treatment duration is very short and wildtype virus will quickly become a majority again (see for example, [16]). In fact, the standard results on fixation probability [32] are really results on establishment probabilities, so we can use them without problems.
Psgv vs. Pnew
For drug resistance to evolve, the viral population needs viral particles that carry drug resistance mutations. Such particles may already be present before treatment is started. To denote this possibility we use P sgv or the probability that drug resistance establishes from the standing genetic variation. If the mutation is not already present, or if was present but was subsequently lost, then the viral population has to wait for a new mutation to occur and become established. We denote this possibility as P new , or the probability that resistance evolves due to new mutations. In the latter case, we have to indicate a time window, such as per year or per generation.
The goal of this study is to understand and, albeit roughly, quantify P new and P sgv for HIV drug resistance in patients on triple-drug regimes (consisting of an NNRTI or an unboosted PI plus two NRTI's) and in patients who are treated with single dose nevirapine.
Starting standard therapy
When a patient starts anti-retroviral therapy for the first time, the viral population in that patient will move from an equilibrium without drugs to an equilibrium with drugs. At the pre-treatment equilibrium, the viral population size will at its equilibrium level (N u ), and resistance mutations are expected to be at mutationselection-drift equilibrium, where most mutations will be present at very low frequencies (see, e.g., [7]). Note that mutation-selectiondrift equilibrium is reached quickly for mutations that are very costly to the virus. So even though it may take years for neutral diversity to reach an equilibrium level in an HIV patient [33], important drug resistance mutations which are 5 or 10% less fit than the wildtype are expected to reach their (dynamic) equilibrium in weeks or months.
Standard population genetic theory predicts that the average frequency of a resistance mutation is equal to the mutation rate (m, per viral particle and per replication) divided by the relative cost (C rel ) of the resistance mutation, though drift causes actual frequencies to vary greatly between different time points and between patients (see also [34]). Even though the average frequency is independent of the population size, in larger populations, it is more likely that DRMs are present and the absolute number of drug resistant particles will, naturally, be higher. When treatment starts, resistance mutations will confer a fitness benefit to the virus and they can (but are not guaranteed to) increase in frequency and become established. The probability that this happens depends on the number of resistant particles in the population and on the establishment probability of a mutation that is present in a single particle. In [14] we derived formulas to calculate the probability that adaptation to a new environment happens from the standing genetic variation (P sgv ). We will use the approximate equation 8 in [14]: It is also possible to use the the number of resistant particles in a patient (B), if this is known, and the fitness of these copies (in the environment with drugs) to calculate the probability that a resistance mutation becomes established: where we use the probability that all copies of the resistance mutation die out to calculate the probability that at least one survives. The probability that resistance mutations become established increases with the number of copies of resistant virus and the probability that any one of these survives.
Evolution of resistance during therapy
If resistance did not evolve from standing genetic variation, it may evolve due to new mutations. The probability that this happens in a given year will depend on the number of generations (G) in a year, the mutation rate (m), the effective population size during antiretroviral treatment (N ART ) and the establishment probability of a mutation (P est ). In principle, the establishment probability during therapy may not be the same as in the very beginning of therapy, for example because the number of available cells which a particle can infect could be different. However, throughout this paper we will assume that P est depends only on the kind of therapy and not on how long a patient has been treated. Using a poisson approximation, we find that the per year probability that resistance evolves is It is debated whether during therapy, there is ongoing replication or whether a reservoir of latently infected cells is entirely responsible to residual viremia. If the reservoir reflects the composition of the viral population before treatment, then the expected frequency of the resistance mutation in the reservoir would be m C rel . If the number of latently infected cells that become activated every generation is N LAT , then the expected number of activated cells with resistant virus would be NLAT m C rel . The per year probability that resistance evolves due to activated cells from the reservoir would be It is also possible that there is ongoing replication, but that the reservoir also plays a role at the same time, so that the reality will be reflected best by a combination of equations 4 and 5. Note that N ART and N LAT are both effective population sizes, and may be much lower than the census population sizes.
Comparison with data and parameter estimation
Published data show that the rate of evolution of drug resistance is roughly constant over long times (see for example the study by [3], in which patients were followed for up to eight years). This fits with expectations if N ART and P est remain constant so that P new stays constant. However, several studies show that the probability that resistance mutations become established is higher in the first year of therapy, as compared to later years. This can be seen, for example, in a study on a large cohort in British Columbia, especially when one considers the most adherent group of patients (figure 2 in [35], see also [17]). A similar effect is seen in [11] when one considers the patients with pre-existing DRMs. This effect, that resistance is more likely to evolve in the first year of therapy as compared to later years, can be easily explained by standing genetic variation.
Under the assumption that P new is indeed constant, we can use published data to estimate both P new and P sgv . Margot et al [36] reported the number of patients in which resistance was detected in the first, second and third year after treatment initiation in a cohort of patients who were treated with NNRTI-based ART. The reported data (see table S1 in supplementary text S2) show that the probability that resistance was detected in the first year was 9.5%, whereas in the second and third year it was only 3.7% (see supplementary material for details on how this was estimated). The difference of 5.8% is likely due to standing genetic variation at the start of therapy.
We will use the estimates for P new (0.037 per year) and P sgv (0.058) from [36], in combination with other, published, estimates to get a rough estimate of the important evolutionary parameters. First of all, we will assume that the mutation rate from one nucleotide to a specific other nucleotide is 10 {5 [37], so that if there are five main resistance mutations for a given drug combination, the total mutation rate is approximately 5|10 {5 . For the remainder of the paper, we will only use this total mutation rate. If the mutation rate would be higher (lower) than our assumption, the estimated population sizes would be lower (higher) than our estimates. An overview of the parameter values we use in the paper is given in table 2.
We know that the important drug resistance mutations are at least somewhat costly for the virus. Their cost, C rel , has been estimated for several drug resistance mutations, both in vivo and in vitro (for an overview on resistance mutations in the reverse transcriptase gene see [38]). For example, [39] find that the relative cost of resistance mutation M184V is approximately 0:04{0:08. Wang et al [40] estimate a cost of 0:01{0:04 for K103N, which is the most common NNRTI resistance mutation. Other studies were not able to detect any cost of K103N, but given its low frequency in untreated patients [7], it seems likely that it is associated with a significant cost. In this paper we will use an average cost of 0:05 for all mutations.
Given the cost, the mutation rate, P sgv and P new , and using the assumption that there are 200 HIV generations in a year [41], we can find the combinations of N u , N ART and Fm ART that are compatible with the data (shown in figure 1). Estimates for the effective population size in untreated patients range from 10 3 [42] to 10 5 [43]. We know that a large proportion of untreated patients carries low frequency drug resistance mutations, but not all patients, which gives us some additional information about the population size in an untreated patient (see figure 1b). If we choose a value of N u of 2|10 3 , then we find that about half of the patients should carry pre-existing DRMs. This is somewhat higher than what is usually detected, but that can be due in part to the limits of detection of current tests [8]. An overview of the parameter estimates that were used in the simulations and for analytical predictions can be found in table 2.
Given our choice of N u , we find that Fm ART must be approximately 1:017, leading to P est &0:034. Under the assumption P est stays the same during treatment, the Margot et al data are compatible with a 18-fold reduction of the effective population size due to therapy, to an effective population size of N ART &108 . Note however, that the estimate of a 18-fold reduction depends heavily on the assumption that C rel~0 :05 . For example, had we assumed a 10% cost, then the estimated reduction would have been 37-fold , and for a 1% cost, the reduction would have been only 4-fold . The reason is that if we assume that costs are high, then we must also assume that the mutant fitness (Fm ART ) is relatively high, in order to find P sgv~0 :06, and if Fm ART is high, N ART must be low, to explain P new~0 :037.
If the evolution of resistance during therapy is not due to ongoing replication, but due to continuous activation of latent cells, then, under the assumption that C rel~0 :05, the effective number of cells (N LAT ) must be approximately 5 per generation. This means a reduction of effective population size of almost 400fold. However, it is not so clear whether in this case the word ''population size'' should still be used, because the number 5 is not an estimate of the size of the reservoir, but an estimate of the effective size of the part of the reservoir that is reactivated every generation.
The result that the frequency of resistance mutations in the reservoir depends on their fitness cost ( m C rel ), whereas the cost does not play a role for new mutations due to ongoing replication, could be harnessed to estimate the relative importance of the reservoir. If the reservoir is the most important source of resistance mutations during therapy, then the same set of mutations should be found in patients whose virus acquires resistance quickly after the start of therapy and in those who acquire mutations during therapy. However, if ongoing replication is the source of resistance mutations during therapy, then mutations with a high cost in the absence of drugs should occur relatively more often during therapy than quickly after therapy is started.
The data and the results from simulations and predictions (using equations 2 and 4) are shown in figure 2. The percentage of patients with resistance after one year is lower in the simulations than in the analytical predictions, because in the simulations, it takes time for a mutation to increase in frequency and be detected. We assume that it is detected as soon as it is more frequent than the wildtype, the result is that in the simulations (and probably in reality) P new is lower in the first year than in the other two years. It is unclear how large this effect is in reality, but it means that the 6% we find is a conservative estimate of the role of standing genetic variation. If it would take 3 months for a mutation to increase in frequency and become detected, then P new in year 1 would be 75% of its value in the later years, and P sgv would be approximately 7% in stead of 6%.
Single-dose nevirapine for prevention of mother-to-child-transmission A single dose of nevirapine (sdNVP) just before labor starts reduces the risk that a mother transmits HIV to her baby at birth, but leads to high levels of resistance in many women. Because of the long half life of nevirapine, even a single dose lasts at least a few days. However, this is a very short amount of time (only a few HIV generations) so that probably most or all detected NVP resistance mutations are due to standing genetic variation.
Because it is known that sdNVP can lead to the establishment of resistance mutations, and also to further reduce the risk that the baby becomes infected with HIV, several different treatment strategies are being used. In this study, we focus only on those strategies that include a single dose of nevirapine (and exclude, for example, pregnancy limited triple-drug therapy). Basically, sdNVP can be combined with either a short course of zidovudine monotherapy during the third trimester of pregnancy (ZDV/ sdNVP), or it can be combined with additional drugs during and after labor up to one month postpartum (sdNVP/PP). It can also be used alone (sdNVP) or combined with both (ZDV/sdNVP/PP), resulting in four possible strategies.
Under the assumption that all resistance is due to standing genetic variation, it is straightforward to predict, at least qualitatively, the effect of the four treatment options. Single dose nevirapine plus two additional drugs (sdNVP/PP) is a three drug regimen, and similar to standard antiretroviral therapy (ART), except that it only lasts a few days or weeks. We therefore expect similar levels of drug resistance due to standing genetic variation. If only NVP resistance is considered (and not resistance to the other two drugs), we expect to find somewhat lower levels than in the normal case, although the difference may not be large because resistance against NVP is more common than resistance to most other drugs. Treating with only sdNVP is different from starting ART, in that there is only one drug. The result is that the fitness of both wildtype and resistant virus will not be reduced as much as in the normal case. Specifically, NVP resistant virus will have a relatively high fitness during NVP monotherapy. This high fitness (Fm NVP ) leads to a high establishment probability (P est ) for available resistance mutations. In fact, the establishment probability may be so high that in virtually all patients that carry some NVP resistance before treatment, the resistant virus will increase in frequency during NVP treatment.
An interesting treatment option is to start with a few weeks of ZDV monotherapy before treating with a single dose of nevirapine. The ZDV treatment will reduce the population size of the virus, N u , so that the probability that NVP resistance is available and the copy number of such resistant mutants if they are available will be lower by the time the patient is treated with NVP. ZDV monotherapy ultimately leads to ZDV resistance, but the risk that resistance mutations become established during a short course is small. ZDV monotherapy reduces the viral load approximately three-fold [44]. Finally, adding ZDV treatment before labor and two additional drugs during and after labor (ZDV/sdNVP/PP) will reduce both the availability of NVP resistant virus and the establishment probability of such virus, which should lead to an even lower probability that NVP resistance mutations from standing genetic variation become established.
Comparison with data for single dose nevirapine
We identified 23 published studies that reported on NVP resistance 6 to 8 weeks after women were treated with sdNVP. Several of the studies directly compared two different treatment options. We found at least three studies for each of the four different treatment options. An overview of the studies can be found in table S2 in the supplementary text S2. For each study we recorded which of the four treatment options was used and in how many of the patients NVP resistance mutations were detected using simple Sanger (population) sequencing (we excluded studies that only recorded deep-sequencing or allele-specific PCR results, as there were too few of those to allow us to compare the treatment options). For each of the four treatment options, we also calculated the overall probability that resistance mutations were detected in a patient (simply by summing the number of patients with resistance and summing the total number of patients in the studies). We found that sdNVP leads to detectable resistance mutations in 39% of 952 patients, ZDV/sdNVP leads to detectable resistance mutations in 22% of 888 patients, adding two drugs during and after labor (sdNVP/PP) lead to detectable resistance mutations in 7.8% of 372 patients and ZDV/sdNVP/PP lead to detectable resistance mutations in none of 292 patients (see figure 3).
We now used these data, in combination with our previous parameter estimates, to estimate the fitness of a NVP resistant mutant during NVP therapy (Fm NVP ) and the reduction of the population size due to ZDV treatment (N ZDV ). We find that Fm NVP &1:54 and that ZDV reduces the effective population size approximately two-fold (table 2 and figure 4). The results show that a reduction in population size by ZDV monotherapy does reduce the probability that NVP resistance mutations become established, but adding two drugs to sdNVP helps much more. We also estimate the fitness of the mutant during therapy with nevirapine and two additional drugs and find a slightly higher value than our previous estimate (1:025 vs 1:017), though these differences are not statistically significant.
Interruption of therapy
During a treatment interruption, drugs are first removed from the body, which can take from a couple of hours to a several days or even weeks ( [45,46,47]. With some delay, depending on the half-life of the drugs, the viral population begins to grow, which is observed as an increase of viral load (see figure 5). Published data show that after treatment is stopped, viral load quickly increases in almost all patients (e.g., [2]. Davey et al [31] show that average viral load plateaus four weeks after treatment is interrupted. Garcia et al [48] and Trkola et al ( [49]) both report that a plateau is reached between four and eight weeks after treatment interruptions. An interruption is ended when treatment is started again and viral load goes down, hopefully to undetectable levels. Figure 1 shows a cartoon of the pharmacodynamics and population dynamics of a treatment interruption.
Restarting therapy
If the length of a treatment interruption is so long that the population size is back to pretreatment level and mutationselection-drift equilibrium is again reached, the probability that resistance mutations become established when therapy is started again will equal the probability that resistance mutations become established the first time a patient starts treatment, P sgv from equation 2. But if a treatment interruption is shorter than that, it is hard to calculate the exact probability that resistance will evolve upon re-initiation of therapy because neither population-dynamic, nor population-genetic equilibrium will have been reached. The absence of the population-genetic equilibrium is most problematic if resistance mutations are not very costly to the virus. However, for a costly mutation it takes only on the order of 1=C rel generations to reach mutation-selection-drift equilibrium. The absence of population-dynamic equilibrium is less problematic, because it is relatively easy to predict the population size of the virus or to measure viral load. In the simulations, we allow the population to grow exponentially until it reaches the baseline level. The resulting population size can be plugged into equation 2 to get an estimate of the probability that resistance mutations become established due to a treatment interruption.
Comparison with data for treatment interruptions
Using the parameter values from the last two sections, we can predict the risk that resistance mutations become established due to a treatment interruption of a certain length. We use the estimated fitness of the mutant virus during NVP therapy, and assume that the fitness of the mutant in absence of drugs is the same. With that value, we can calculate the fitness of the wildtype in the absence of drugs, because of the assumption that the cost of the resistance mutation is 5%. The wildtype fitness will determine how fast the virus grows in the simulations after treatment is interrupted, and therefore how long it takes before the population size is back at the pretreatment level. Specifically, we use F wt~1 :62. In the simulations, the population size plateaus after just 14 days, but P sgv reaches its expected value only after 60 days (figure 6).
We collected information from structured treatment interruption trials to test the predictions. The probability that resistance mutations become established due to a single treatment interruption was estimated for seven clinical trials with different lengths of treatment interruptions [50,22,51,52,53,54,20]. An overview of the trials can be found in table S3 in text S2 (supplementary material). We first calculated the risk under the assumption that all observed resistance was due to treatment interruptions and then subtracted the estimated probability that resistance mutations become established during therapy. The corrected values are shown in figure 6. The data show that longer treatment interruptions indeed lead to a higher risk of resistance. The risk plateaus around 37 days, which is consistent with the time it takes for viral load to reach its equilibrium level (although the simulations suggest that the risk should plateau later than the population size). The highest risk was found to be approximately 6% per interruption, just like the risk of starting therapy for the first time.
Discussion
The main aim of our study was to understand and quantify the importance of standing genetic variation for the evolution of drug resistance in HIV. We find that the probability that at least one resistance mutation becomes established due to standing genetic variation (P sgv ) depends on the kind of treatment chosen. Most clearly, it is much higher when treatment is with sdNVP (which is monotherapy) than if treatment is with triple-drug combination therapy. For standard combination therapy (ART), we use two different data sources to estimate the probability that resistance mutations from standing genetic variation become established. In the first part of this paper we used data on the number of patients in which resistance was detected in the first year of treatment versus later years. In the third part of this paper we used data from clinical trials on treatment interruptions. In both cases, we found that the probability that resistance mutations from standing genetic variation became established was approximately 6%.
The importance of new mutations as compared to pre-existing mutations could be estimated from the Margot et al ( [36]) study. We estimated that the probability that a resistance mutation becomes established during therapy (P new ) is 3.7% per year, which means that pre-existing mutations and new mutations are equally important after about one-and-a-half year of treatment. Two of the interruption studies also provided estimates for P new , which were slightly higher (4.3% and 4.8% per year) than the estimate from the Margot et al [36] study (see table S3 in text S2). It is likely that some of the patients in these studies were not perfectly adherent to treatment, so that our estimate of P new is inflated by patients who interrupted treatment. This does not affect our estimates of P sgv . However, it means that the relative importance of pre-existing mutations is highest in completely adherent patients (because new mutations are relatively unimportant for them) and lower in non-adherent patients (see [7] but see [11]).
A stochastic model was used to understand the effect of standing genetic variation on the evolution of drug resistance during HIV treatment. Four parameters are crucial to understand the role of standing genetic variation. Three of them determine the amount of genetic variation that is available (effective population size, mutation rate and cost of the resistance mutations) and one determines how likely it is that the available mutations become established (the absolute fitness of the resistant virus during treatment).
The cost and the mutation rate are parameters that are different for each specific mutation. Together, they determine the expected frequency of the mutant in an untreated patient. For example, in untreated patients the frequency of K103N was found to be lower than the frequency of Y181C [7], suggesting that m=C rel is lower for K103N. The costs for some of the most important mutations (M184V, K103N) have been estimated and are between 1 and 10% ( [38,39,40]. Throughout this paper we used a value of 5%. The effective population size in an untreated patient (N u ) determines how much variation there is in the frequency of resistant mutants between patients. If N u w1=m, the frequency in each patient will be very close to the expectation, m=C rel , but if N u v1=m, there will be a lot of variation between patients, and in many patients no resistance mutations may be available at all. Data suggest that in HIV the latter is the case (e.g., [7]), which means that, not every single point mutation is created every generation in an HIV patient. Or, more precisely, each mutation may be created, but not in a viral particle that is part of the effective population size. Mutations may even be detected in the blood stream of a patient, but may still be irrelevant if the viral particles with the mutations are eliminated before they can infect a CD4 cell. N u also determines the number of resistant viral particles in a patient with a given frequency of the mutant. With higher N u , there will be a higher number of resistant particles, and this makes it more likely that resistance mutations become established when treatment is started [11].
We find that data are compatible with an 18-fold reduction of N due to ART and a two-fold reduction of N due to ZDV monotherapy. The estimated reduction depends on the assumed cost of mutations; if we assume that mutations are twice as costly, we would find a reduction that is twice as severe. Still, the reductions we find are not nearly as severe as one may have expected based on viral load reductions. During ART, VL may be reduced 1000-fold or more (in the Margot ([36]) study from which we used the data, patients had a viral load of, on average, 8|10 4 before treatment, whereas after 48 weeks of treatment, about 80% of the patients had a viral load of less than 50, [55]). This discrepancy may be due to two effects: firstly, our estimate is an average for many patients and this average may be driven up by patients in which the drugs do not work well, or who are not adherent to therapy so that their VL does not go down as much as expected. Secondly, the relationship between effective population size and viral load may not be linear, so that a thousand-fold reduction in VL may translate in only a twenty-fold reduction in effective population size.
The fourth important parameter is the fitness of the mutant virus during treatment (Fm), which determines the establishment probability (P est ). Fm will depend on both the drugs that are used and on the specific mutation. For example, the resistance mutation K103N is more likely to become established during sdNVP than during triple-drug therapy, because additional drugs reduce Fm (Fm ART vFm NVP ). And during triple-drug therapy, K103N is more likely to become established than Y181C (even though Y181C is present at higher frequencies before treatment), likely because Fm ART is higher for K103N than for Y181C.
Starting of standard therapy
We assumed that the rate of evolution due to new mutations is constant and that the establishment of a resistance mutation from standing genetic variation leads to viral failure and is detected within one year of starting therapy. Maybe the most convincing evidence for these assumptions comes from the Li et al [11] study, where their figure 2 shows that (1) patients without detected preexisting DRMs show a constant rate of evolution of resistance and (2) patients with detected pre-existing DRMs show an increased rate compared to the patients without pre-existing DRMs, but only in the first year of treatment. We used these assumptions to estimate the probability that resistance mutations from standing genetic variation become established. However, the estimated role of standing genetic variation may be a slight underestimate, because establishment of new mutations should need some time so that P new would normally be somewhat lower in the first year of treatment. The observation that the effect of standing genetic variation only lasts a year, means that fixation of a resistance mutation must take less than a year. This limits possible values for N ART and Fm ART to such values for which the fixation time is less than 200 generations.
If resistance indeed evolves due to standing genetic variation in 6% of patients on standard ART, then there is clearly room for improvement. Note that those 6% of patients have already lost their first treatment option shortly after having started treatment. They have to switch to second-line treatment which is more expensive, usually more complicated (more pills per day) and likely has more side effects. It is therefore worth exploring ways to avoid the establishment of resistance mutations from standing genetic variation. Figure S2 suggests two options to reduce P sgv , by reducing the population size or by reducing the fitness of the resistant mutants. The first may be achieved by ZDV monotherapy, as shown in the section on PMTCT, whereas the second may be achieved by adding additional drugs to the treatment. Obviously, triple-drug combination treatment is already standard for most HIV patients, but it may be worth considering specifically which treatment options would be best to prevent the evolution of resistance from standing genetic variation. This may mean, for example, to add a fourth drug to the therapy in the first couple of weeks of treatment. Resistance to boosted PI's is very uncommon, so they may be a good choice for starting treatment, in combination with two or three other drugs.
Resistance due to sdNVP
Studying treatment with a single dose of nevirapine gives us a unique opportunity to study the effect of standing genetic variation, because treatment is so short (only a few HIV generations) that we can assume that most or all resistance mutations that are detected are from standing genetic variation. Data show that the risk that resistance mutations become established due to such treatment is very high (39%). We find that this high probability can be explained entirely by selection on pre-existing drug resistance mutations, because the fitness of NVP resistant virus is probably very high during NVP monotherapy. We estimate that its fitness is approximately 1.5. The probability that a resistance mutation becomes established can be reduced by either adding additional drugs to lower the fitness or by lowering the population size so that fewer mutants are available. A study from Zambia [56] showed that the additional drugs even help to reduce the establishment of NVP resistance mutations considerably if the additional drugs are given as a single dose (in stead of treatment for a couple of days or weeks). We did not include this study in the overview, because there was only one study that looked at this treatment option.
The results on ZDV/sdNVP/PP treatment (i.e., treatment with ZDV during pregnancy and NVP plus two other drugs during labor) are surprising in that NVP resistance mutations were not detected in any of the women who received this treatment, even though the model would predict that mutations would be detected in 4% of the women. Most of the data on this treatment option are from the Lallemant [57] paper (222 women). In this study, the authors do find some mutations that confer resistance to the NRTI's in the study (in 2.3% of the women). The same study also looked at women who were treated with ZDV/sdNVP and also in these women the percentage with resistance mutations was very low (6.4%) and much lower than the mean value for women who receive this treatment (22%). The reason for the surprisingly low values of drug resistance in this study could be that the women in the study had very low viral loads (median 2800). This probably also means that they have a low effective population size. It therefore seems unlikely that the extremely good results from the Lallemant study [57] can be replicated in other populations. However their results still show that using additional drugs to reduce the population size and to reduce the fitness of the mutant may be a good strategy to reduce the probability that resistance becomes established.
Treatment interruptions
Considering treatment interruptions, our model provides several testable predictions. 1) resistance mutations are more likely to become established after long treatment interruptions when viral loads are higher, 2) the risk that resistance mutations become established due to a treatment interruption can not be larger than the risk at the start of treatment, 3) treatment interruptions increase the risk of establishment of resistance mutations even for drugs with short half-lifes.
Data from seven clinical trials show that indeed, longer interruptions increase the probability that resistance mutations become established (figure 6). Moreover, the estimated probability appears to plateau after 37 days, which is similar to the time it takes for viral load to reach its pretreatment level. This suggests that the risk of establishment of resistance mutations is directly linked to the viral load when treatment is started again. The second prediction was also found to hold: the estimated risk that resistance mutations from standing genetic variation become established at the start of treatment was found to be similar to the risk due to a long treatment interruption (6% in both cases). The third prediction also holds, as data show that interruptions increase the risk of establishment of resistance mutations even for PI based treatment [20,51], where the ''tail of monotherapy'' cannot explain the observations. A potential problem with the data is that not only the length of the interruptions, but also the length of treatment periods between the interruptions differed between the seven studies. The trials that were compared also differed in the drugs that were used (see table S3 in text S2), which makes direct comparison difficult. Despite all these limitations, it becomes clear that longer interruptions carry a higher risk of evolution of resistance than shorter interruptions.
If interruptions lead to the establishment of resistance mutations only due to the ''tail of monotherapy'', as is usually assumed in the HIV literature [24,25,18], we would predict that: 4) treatment interruptions increase the risk that resistance mutations become established only for drugs with long half-lifes, 5) the risk that resistance mutations become established due to a treatment interruption is unrelated to the risk at the start of treatment and 6) the largest risk would be due to an interruption with a length that is exactly the time it takes for the last drug to lose its effect on the wildtype virus. All of these predictions do not hold. This is not to say that the ''tail of monotherapy'' is not important at all. But it does show that on its own, the ''tail of monotherapy'' cannot explain the risk that resistance mutations become established due to treatment interruptions. When one considers possible intervention strategies, this may be good news. If treatment interruptions are risky because of restarting rather than stopping therapy, this would give doctors a possibility to reduce the risk that resistance mutations become established even after a patient has already stopped taking his or her drugs. The establishment of resistance mutations at re-initiation of treatment may be avoided by pretreatment (such as with ZDV) to reduce the availability of mutations or by using more drugs or higher doses in the first weeks of treatment to reduced the establishment of pre-existing mutations.
General remarks
We have used a population-dynamic and population-genetic model to study several patterns of drug resistance in HIV. The model explains why resistance mutations are likely to become established in the first year of standard treatment, in women who are treated with a single dose of nevirapine and in patients who interrupt treatment. In all three cases, standing genetic variation can explain the observations.
Our results illustrate that for adaptive evolution to happen, selection and the creation of new variation need not happen at the same time, if selection can work on standing genetic variation. In the case of antiretroviral treatment, this means that insufficient drug levels (which allow for replication and selection at the same time) are not a necessary condition for the evolution of drug resistance. This result about time-heterogeneous drug levels is similar to the result on heterogeneity in space by Kepler and Perelson [58], who showed that genetic variation may be created in compartments where drugs cannot penetrate whereas selection happens in other compartments.
Our model provides a simple and quantitative explanation for why resistance is less likely to evolve when patients are treated with multiple drugs in stead of just one drug. Additional drugs reduce the fitness of a mutant that is resistant against one drug, and therefore the establishment probability of such a resistant mutant. In addition, additional drugs reduce the population size of the virus and thereby the creation of new resistance mutations. This means that there will be fewer resistance mutations with lower establishment probabilities, together leading to a strong reduction in the probability that resistance evolves. In newer therapies with boosted PIs, drug resistance has become very rare [3], which may be because boosted PIs are so strong that no single mutation can lift the virus' fitness above 1.
The model in this study may be relevant to other diseases than HIV. For example, the evolution of resistance is a problem in chronic myeloid leukemia (CML) which is a cancer of white blood cells. A recent study suggested that the probability that drug resistance evolves in CML goes down with time because the population size of the cancer goes down with time [59].
Resistance is also a problem in tuberculosis (TB), and in TB it is also known that treatment interruptions increase the risk of evolution of resistance [60]. This effect may also be due to an increased population size during the interruptions. In general, stopping treatment may be risky in cases where treatment has to be started again, which is always the case for HIV and often for TB. Each time therapy is started, resistance mutations from standing genetic variation may become established, and even if this risk is only a few percent it adds up quickly when patients interrupt treatment regularly.
Supporting Information
Text S1 Description of the model, the simulations, the calculation of the fixation probability and the data that were used for the analysis. (PDF) Text S2 Text S2 includes tables S1, S2, S3. Table S1: Number of patients with at least one resistance mutation detected by the end of the first, second and third year of NNRTI-based antiretroviral therapy. Data from [36] Table S2: Overview of clinical trials which used single dose nevirapine treatment to prevent mother-to-child transmission and which reported the number of patients with nevirapine resistance detected 6 to 8 weeks after treatment. Table S3: Overview of clinical trials with structured treatment interruptions which reported the number of patients with at least one drug resistance mutation detected. (PDF) Text S3 Computer code, written in C++, which was used to run the simulations, as used for figures 1, 3 and 6. (TXT)
==
Domain: Biology Medicine Computer Science
|
[{"added": "2011-11-26T15:10:10.000Z", "created": "2011-11-26T00:00:00.000", "id": "3189307", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Computer Science"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1002527&type=printable", "pdf_hash": "179bd879add081ae254026884719e0ddde0ca149", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:4426", "s2fieldsofstudy": ["Biology"], "sha1": "cdd82a4f14ff8fcdc14358faaa64fb9ba9bceaf7", "year": 2012}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Interaction of Motility, Directional Sensing, and Polarity Modules Recreates the Behaviors of Chemotaxing Cells
Chemotaxis involves the coordinated action of separable but interrelated processes: motility, gradient sensing, and polarization. We have hypothesized that these are mediated by separate modules that account for these processes individually and that, when combined, recreate most of the behaviors of chemotactic cells. Here, we describe a mathematical model where the modules are implemented in terms of reaction-diffusion equations. Migration and the accompanying changes in cellular morphology are demonstrated in simulations using a mechanical model of the cell cortex implemented in the level set framework. The central module is an excitable network that accounts for random migration. The response to combinations of uniform stimuli and gradients is mediated by a local excitation, global inhibition module that biases the direction in which excitability is directed. A polarization module linked to the excitable network through the cytoskeleton allows unstimulated cells to move persistently and, for cells in gradients, to gradually acquire distinct sensitivity between front and back. Finally, by varying the strengths of various feedback loops in the model we obtain cellular behaviors that mirror those of genetically altered cell lines.
Introduction
Cells have a remarkable ability to sense the direction of chemical gradients and respond by polarizing and migrating toward attractants. Chemotaxis is one of the fundamental properties of single cell organisms, such as bacteria and amoebae, as well as multicellular systems. Experiments suggest that chemotaxis involves the coordinated action of separable but interrelated processes: motility, gradient sensing, and polarization [1]. In fast moving amoeboid cells, such as the social amoeba Dictyostelium discoideum or human neutrophils, motility arises from the periodic extension of actin-rich pseudopods whose nature is quite similar in chemoattractant stimulated and unstimulated cells. Gradient sensing refers to the cell's ability to interpret extracellular gradients and to respond by directing intracellular proteins to the site of highest chemoattractant concentration. Experiments in eukaryotic cells in which motility has been impaired by inhibitors of actin polymerization, such as Latrunculin, demonstrate that gradient sensing occurs even in immobile cells, indicating that cells employ a spatial sensing mechanism that does not depend on movement. Finally, polarization is the propensity of cells to assume stable anterior and posterior edges leading to an elongated morphology. The anterior region is more sensitive to chemoattractants so that in response to a changing gradient polarized cells turn towards the new direction [2]. In contrast to gradient sensing, polarization depends on intact cytoskeleton.
The study of chemotaxis has benefitted greatly from the interplay between experimental and theoretical studies [1,[3][4][5]. A number of recent models propose that chemotaxis is a consequence of an excitable network whose activity is biased in the direction of chemoattractant stimuli [6][7][8][9][10]. Motility can be achieved if this activity directs pseudopodial protrusions [11,12]. The basis for these models is the observation that cytoskeletal and signaling pathways in cells exhibit excitable behavior, in the form of patches and waves of activity seen along the cell cortex, and that these activities coincide with the location of protrusions [10,[13][14][15][16][17][18][19][20][21].
We start with a previously described local excitation, global inhibition biased excitable network (LEGI-BEN) that captured some of the experimentally observed features of spontaneous and chemoattractant-induced signaling events, and add two modules. First, we use level set methods [22,23] and a viscoelastic mechanical model to simulate cytoskeleton-mediated cellular deformations and movements. Second, we incorporate a cytoskeleton-dependent polarity module that confers both the persistent migration seen in unstimulated cells as well as other characteristics of polarized cells. We use the model to consider a number of in silico mutants and use the resulting simulated results to consider the possible biochemical identities of elements in the model. Whereas various previous models can account for different subsets of behaviors of chemotaxing cells, we show that the complete modular framework of the polarized LEGI-BEN model presented here accounts for nearly all the reported observations.
Linking the Activity of the Excitable Network to Cellular Protrusions
Previously we and others have proposed that the spontaneous patches of signaling activity seen in motile cells could be explained by an excitable network (EN) consisting of two components: an activator (X) and an inhibitor (Y) (Fig. 1A) [7][8][9][10]15]. The activator, which involves an autocatalytic positive feedback loop, drives the inhibitor, which provides negative feedback. Experimentally, excitable behavior is observed in cells that are not stimulated by chemoattractant, indicating that the chemoattractant receptor is upstream of and not part of the EN [18]. The spontaneous nature of these activities can be recreated by including a stochastic element that triggers the EN randomly. The activity of the EN around the perimeter of a one-dimensional model of the cell can be observed by plotting the level of the activator or inhibitor as a kymograph (Fig. 1B). These patches represent localized signaling events that drive cell protrusion.
To determine whether this model could recreate cellular protrusions leading to random cell motility, we coupled the EN to a mechanical module of the cell cortex implemented in the level set framework (Fig. 1C, Methods.) (Computational methods that account for changes in cell morphology during migration are reviewed in Ref. [5]). The mechanical description of the cell was identified based on micropipette aspiration experiments using Dictyostelium cells [22]. The mechanical module incorporates several passive stresses, including the effect of cortical tension driving Laplace-like pressures on the cell and volume conservation. It also includes active stresses allowing us to test the effectiveness of the EN in driving cellular motion. In our simulations, the activity of the EN was coupled to protrusive forces, so that higher activity at one location gave stronger protrusive stress (Fig. 1C, D). We envision that the effects of the EN on the cytoskeleton mediate these protrusive forces. In these simulations, local protrusions appeared randomly around the cell, as would be expected in a cell undergoing random motility ( Fig. 1D; Video S1). Analysis of these simulations over time revealed that the activities were uniformly distributed at the population level over long time scales, though localized fluctuations do occur in shorter time scales ( Fig. 1D; inset). The localized forces caused by these heterogeneities, however, were not sufficiently persistent to propel the cells in a meaningful way. Thus, the cellular boundary extended in random directions, but the migration rate of the cell was negligible.
Incorporation of a Directional Sensing Module
To test the effect of a chemoattractant gradient we incorporated a local excitation-global inhibition (LEGI) mechanism to the excitable network, creating a LEGI-BEN system (Fig. 1E), as previously described [8,10]. Because the response regulator in the LEGI mechanism increases at the front and decreases at the back, it brings the excitable system closer or farther from the threshold at the front and back, respectively. This biases the likelihood of triggering activity in response to external chemoattractant signals. When a gradient was applied to an unstimulated cell, the activity of the cell increased everywhere around its perimeter (Video S2). Thereafter, signaling activity was found preferentially at the side of the cell experiencing the greatest chemoattractant concentration ( Fig. 1F; Video S2). The cellular response to a change in gradient was nearly immediate, and this was true for both steep (19%) and shallower (6%, not shown) gradients. When simulating cell shape changes elicited by this gradient, we found that after the application of the stimulus, the cell elongated and moved in the direction of the gradient. Analysis of the activity showed that the directional response is quite accurate and most of the activities are within the 230u to 30u region relative to the direction of gradient (Fig. 1F, inset). The magnitude of the protrusive force was chosen so that, at steady-state, the cells moved at approximately 10 mm/ min. After a shift in the direction of the gradient, the cell stopped and reversed direction nearly instantly ( Fig. 1F; Video S2).
The LEGI-BEN coupled to the mechanical module recapitulates several consistently observed cellular behaviors such as ''pseudopod splitting'' and ''cringing.'' First, cells often generate a new pseudopod by splitting an existing one [24][25][26]. In chemotaxing cells, these bifurcations appear as a series of leftright extensions. Our simulations of chemotactic cells also exhibited ''pseudopod splitting'' ( Fig. 2A). Because the midpoint of the responding area has the highest activity, negative feedback shuts down this region first, the signals propagate away in opposite directions and the pseudopod splits. Though more pronounced in stimulated cells, it was also observed during spontaneous movement (Fig. 1D, 389 s). Nascent extensions grew from localized patches of high signaling activity. These pushed the cell forward, but eventually split in two ( Fig. 2A, 20-40 s). While these extensions sometimes co-existed for a while ( Fig. 2A, 80 s), one usually won out, at which point the ''losing'' pseudopod appeared to retract into the cell ( Fig. 2A, 100-120 s). This pattern often repeated itself, giving rise to the appearance of side-to-side strokes propelling the cell.
Application of a spatially uniform dose of chemoattractant to a Dictyostelium cell results in a series of changes in cell morphology. Cells stop moving and then transiently contract (or cringe). This is followed by spreading and eventual resumption of movement. Our simulation also recreated this phenomenon (Fig. 2B). Approximately 30 s after stimulation, the cell experienced a mostly global rise in signaling activity (Fig. 2B). At this point the protrusive stresses in our model sought to push out the cell everywhere, but because of the passive constraints on cell morphology, this global
Author Summary
Chemotaxis is the movement of cells in response to spatial gradients of chemical cues. While single-celled organisms rely on sensing and responding to chemical gradients to search for nutrients, chemotaxis is also an essential component of the mammalian immune system. However, chemotaxis can also be deleterious, since chemotactic tumor cells can lead to metastasis. Due to its importance, understanding the process by which cells sense and respond to chemical gradients has attracted considerable interest. Moreover, because of the complexity of chemotactic signaling, which includes multiple feedback loops and redundant pathways, this has been a research area in which computational models have had a significant impact in understanding experimental findings. Here, we propose a modular description of the signaling network that regulates chemotaxis. The modules describe different processes that are observed in chemotactic cells. In addition to accounting for these behaviors individually, we show that the overall system recreates many features of the directed motion of migrating cells. The signaling described by our modules is implemented as a series of equations, whereas movement and the accompanying cellular deformations are simulated using a mechanical model of the cell and implemented using level set methods, a method that allows simulations of cells as they change morphology.
increase in activity had the effect of rounding up the cell. The increase of activator subsequently generates more inhibitor. Once the inhibitor prevails, it suppresses activity all around cell. Thus, the global firing was followed by an absence of signaling caused by the refractory period that follows the firing of an excitable network leading to further rounding of the cell. This was followed by spreading and eventually activities reappeared stochastically around the parameter (not shown).
Incorporation of a Polarity Module
While the simulations of Fig. 1 accurately displayed the chemotactic behavior of unpolarized cells, they lacked two important characteristics observed in stably polarized cells. First, polarized cells moving in the absence of chemoattractants travel in a persistent random walk, and this persistence is a result of having pseudopodia extend in the same direction [26][27][28][29][30][31]. Second, they have an elongated morphology with activity confined to the anterior portion of the cell. Dictyostelium cells at an early stage in their developmental program are mostly unpolarized but the degree of polarization increases as they differentiate [32]. In contrast, once activated, neutrophils are highly polarized.
To overcome these limitations and recreate more realistic cellular behavior, we introduced an additional polarity module to create a polarized-EN system. To achieve polarity we incorporated in our model a secondary set of feedback loops from the cytoskeleton indicated by an arrow linking protrusive stress to the signaling element (Fig. 3A). Positive feedback loops have been a feature of most models of polarization (reviewed in Refs. [4] and [33]) based on experimental evidence that polarization is a consequence of such loops between actin and signaling proteins (e.g. [34]). In our context, a local positive feedback loop (element Z in Fig. 3A) biases the likelihood of subsequent activity at the location of high protrusive stresses. Thus, because stresses are caused by localized increases in signaling activity, whenever high activity occurs at one location, it is more likely that subsequent bursts of activity will occur at that position again. However, only adding a positive loop is not enough to realize polarity. First, there is a lifetime to this persistence, and so the contribution of this loop is expected to subside. Second, without a counteracting negative feedback, the effect of the loop could increase over time throughout the cell, and so lead to hyperactive cells. We therefore included a global negative feedback loop that reduces the activity throughout the cell. This loop was implemented as a separate component (element W in Fig. 3A) that acts to reduce polarization. A second possibility for this inhibition would be to act against Z directly (dotted line in Fig. 3A) as might be expected if the inhibition were in the form of substrate depletion. Negative feedback loops are less common in models of polarization, though several models assume mass conservation of the polarity element, which has the same net effect [35][36][37]. Importantly, averaged over the surface of the cell, the two loops cancel each other out. However, the net effect of the two components is positive at locations of high stress and is negative elsewhere.
Simulations of the polarized-EN system in an unstimulated cell showed persistence in the location of the excitable behavior (Fig. 3). For example, in the kymographs of Fig. 3B, high activity during the period 0-500 s was centered around 290u whereas, after 600 s, it was around +90u (Fig. 3C). These kymographs show that polarity had a localized and transient biasing effect in terms of activity. However, integrating the activity of 40 simulations (each 900 s long) showed that, on average, the location of high activity was uniformly distributed around the cell as would be expected in a randomly migrating cell (Fig. 3D). To examine the temporal effect of the polarization module on the appearance of excitable behavior in any one direction, we computed the autocorrelation function for the activity at fixed angles (Fig. S1). Without the polarity module, the autocorrelation decreases to 0.3 in approximately 30 seconds and approaches zero after about two minutes. With the polarity module, it plateaus at about 0.4 after 30 seconds. The level of this plateau can be changed by varying the coefficient that controls the lifetime of the polarization element. To observe the effect of this persistence on cell motility, we used the polarized EN ( Fig. 3A) to simulate cell motility and changes in cell morphology (Fig. 3E). These simulations showed that unstimulated cells could move significant distances, though the direction and net velocity were random ( Fig. 3F; Video S3). Moreover, as the strength of the polarization increased (by varying parameter Q), the cells drifted farther away from the initial position, as measured by the meansquared displacement (Fig. 3G,H). These results follow closely observations which show that randomly migrating Dictyostelium cells 5.5 hours into development have mean-squared displacements that are approximately ten times higher than newly developed cells [31]. The length of development time also correlates with the degree of morphological polarization [32]. Integration of the Polarization Module with the LEGI-BEN Response to uniform stimuli and gradients. We next considered the effect of chemoattractant stimuli on our polarized-EN, by reintroducing the LEGI mechanism (Fig. 4A). We refer to this complete model as polarized LEGI-BEN. We first simulated the response to a spatially homogeneous stimulus. Before stimulation, the cell displayed random spontaneous activity (Fig. 4B, C). In response to the stimulus, activity increased transiently around the perimeter, lasting approximately 30 seconds. Thereafter, activity subsided throughout, before resuming their spontaneous activity (as the LEGI mechanism adapted to the uniform stimulus). In contrast to cells lacking the polarity mechanism (Fig. 4C, bottom), which displayed a strong secondary peak of elevated activity around 120 seconds after the chemoattractant stimulus, cells with the polarity mechanism do not exhibit this secondary peak (Fig. 4C, top). Consistent with our simulation results, there is widespread experimental evidence for a second peak in early-stage less polarized cells, but it is less pronounced (or nearly absent) in well-developed polarized cells [32]. Without the polarity module, the second peak appears because the LEGI module has not adapted completely, and so secondary firings of the excitable system take place. With the polarization module, the presence of an extra negative feedback loop (W) makes these less likely, effectively eliminating them.
We next tested the effect of gradients of varying steepness ( Fig. 4E-M). In all cases the activity of the cell aligned preferentially in the direction of the gradient. In 19% gradients the activity was concentrated in an arch around 630u and lateral pseudopods were rarely observed. In 6% gradients the response was still predominantly biased in the direction of the stimulus, but lateral pseudopods were observed occasionally. In 1% gradients there was alignment, but considerable more spread.
The alignment of the activity with the gradient in simulations of cells lacking the polarity mechanism also showed dependence on the gradient steepness ( Fig. 4D, F, H). In all cases, the activity in polarized cells showed better alignment with the gradient and less variability. Using the level set simulations to compare the trajectory of cells in response to these varying gradients revealed a similar gradient-dependency. Cells responded better to the steeper gradients, as evidenced by straighter trajectories (Fig. 4J-L) and greater chemotactic indices (CI) (Fig. 4M). For cells with the polarity module, these ranged between 0.1760.02 to 0.6460.21 to 0.9360.02 in 1%, 6% and 19% gradients, respectively (n = 7 in each case). These are similar to reported values in the literature. For example, CIs of 0.2, 0.6 and 0.9 were measured for cells chemotaxing in relative gradients of 1.4%, 4.8% and 10%, respectively [26]. The latter gradients were imposed by a cAMPfilled micropipette. In gradients created by microfluidics, which are closer to ours since they are not formed by a point source, CIs of 0.1-0.3, 0.15-0.4 and 0.96-0.99 have been measured in 1.25%, 2.5% [38] and13.2% gradients [39].
Response to shifts in gradient. One of the main differences between polarized and unpolarized cells is in the response to changing gradients [2,40]. In a cell with the polarity module, we first applied a 6% gradient, maintained this for 10 minutes, and then shifted the gradient 90u ( Fig. 5A; Video S4). Prior to any stimulus, the cell migrated randomly. After sensing the first gradient, the cell slowly aligned itself in the direction of the gradient and began migrating (Fig. 5A). After the direction of the gradient changed, the cell maintained its axis of activity and began a slow turning motion eventually realigning with the new gradient (Fig. 5A). A similar turning motion was observed when the direction of the 6% gradient was changed 180u (Fig. 5B).
We next repeated this simulation in a cell without the polarity mechanism ( Fig. 5C; Video S5). The response in the direction of the initial gradient was similar, although unpolarized cells lined up faster than polarized cells. Furthermore, after the change in the direction of the 6% gradient, the cell immediately shifted its activity in the new direction, no longer extending pseudopods in the old front but instead focusing its activity in the direction of the new gradient (Fig. 5C inset). Thus, the cell trajectory exhibited a nearly 90u turn. Finally, we carried out this simulation in a cell with the polarity module, but in the presence of 19% gradients ( Fig. 5D; Video S6). The response to the initial gradient was similar to the previous simulations, though the activity in response to the steeper gradient was more focused than that toward the shallower gradient (as previously observed in Fig. 4C, D) enabling the cell to move further along the gradient during the initial 900 s (compare the location of the cells at 900 s in Fig. 5A and 5D). After the change in gradient location, however, the polarized cell made a sharp 90u turn towards the new gradient. Thus, the response of a polarized cell to steep (19%) gradient changes was similar to that of an unpolarized cell to shallower (6%) gradient changes. These simulations show that polarity can be overcome by sufficiently strong gradients.
It has been observed experimentally that polarity can also be reinforced by a period of directed movement in a gradient [41]. To investigate how the time during which a cell is exposed to a gradient affects the development of polarity, we carried out simulations in which the time between application of the two gradients was altered. In Fig. 5E, F, cells migrated in response to a 12% gradient. The location of this gradient was changed 90u after either 130 ( Fig. 5E; Video S7) or 430 s ( Fig. 5F; Video S8). When the initial migration time was small, the cell made a sharp turn, displaying little polarity. However, when the cell had been migrating longer in the gradient, the cell displayed the turning behavior associated with polarized cells. These simulations show that, in our model, as in real cells, polarization is a property that develops over time, and is reinforced by the time during which the cell is exposed to a stable gradient.
Response to multiple gradients. When confronted by conflicting gradients, unpolarized, immobile cells (e.g. Latrunculin-treated) show elevated levels of signaling activity in the direction of both sources [42], a response that is recreated by the LEGI mechanism on its own [43]. Here we simulated the effect of conflicting gradients on the complete model of the cell. We started with a circular, unstimulated cell, applied two 19% gradients 180u apart and maintained these gradients no matter where the cell moved ( Fig. 6A; Video S9). At first, the cell sometimes hesitated and, in some cases, even tried to extend pseudopods in both directions (e.g. at 120 s). However, as the cell polarized, one direction won out and the cell migrated in this direction. In contrast, cells that lacked the polarity module oscillated but never settled on either source (Video S10). Thus, polarization enables cells to select between two competing sources [3,44]. We also simulated the effect of Latrunculin treatment by setting the cytoskeletal link to the mechanical module to zero irrespective of the EN behavior. The simulated Latrunculintreated cells displayed activity in both directions throughout the period of the simulation (Fig. 6B; Video S11). Interestingly, however, the stochastic component in the signaling meant that, while the activity peaks pointed towards both gradients ''on average,'' the relative strengths varied over time. This stochastic behavior was also observed in cells that were stimulated with only one gradient (Fig. 6C; Video S12). We also simulated cells that were initially moving in response to an external gradient and to which Latrunculin was added, by gradually reducing the link to the mechanical module. These cells rounded up though they continued to signal in the direction of the gradient (Fig. 6D).
Generation of ''Mutant'' Behavior by Altering Model Parameters
We next considered the effect of altering the strengths of individual loops in the signaling network. We first reduced the strength of the negative feedback loop in the polarization module by 50%. These cells could sense the gradient, however their signaling response, though still pointing on average in the direction of the chemoattractant gradient, was considerably broader (Fig. 7A, top and middle cells; Video S13). This resulted in chemotaxing cells that had multiple simultaneous protrusions which, in many cases, did not point directly towards the source. The cell morphology was quite different from the WT cells, with a broad area facing the gradient. The net movement was also slower. Cells where the negative feedback loop in the excitable network was reduced showed similar patterns of activity [10]. We also investigated the effect of diminishing the strength of the positive feedback loop (through Z) by 50%. The signaling in these cells was aligned with the external gradient. However, the overall level of activity was lower and so the cells moved only slowly in the direction of the gradient (Fig. 7A).
Lastly, we considered the role of the LEGI mechanism in enabling directional migration. We applied gradients to the cell in which the LEGI inhibitor was not regulated by receptor occupancy but is instead kept constant at the basal level. This prevents the LEGI mechanism from adapting to spatially uniform stimuli, although chemoattractant gradients are still sensed and pass on the directional signal to the excitable network. These cells could migrate in the direction of the gradient, though the effectiveness was significantly impaired (Fig. 7B; Video S14). When we raised the midpoint of the chemoattractant signal, as might be expected when cells approach a chemoattractant source, the chemotactic efficiency was further impaired compared to WT cells ( Fig. 7C; Video S15). Comparing the activity of the EN in both situations (Fig. 7D) shows that the lack of adaptation causes the level of activity to rise throughout the cell perimeter, and this has a negative effect on movement, as multiple pseudopods can occur simultaneously and in the wrong direction. This was not the case for cells with an intact LEGI mechanism, where the inhibitor ''filters out'' the mean level of chemoattractant (Fig. 7D).
Rationale for the Framework of Interacting Modules
Chemotactic cells display a variety of behaviors under various experimental conditions (Table 1). 1) Migrating cells display persistence. 2) New pseudopodia appear to split from previous ones. 3) The pseudopodia that bring about random migration coincide with patches of elevated signaling as well as cytoskeletal activity. 4) The cytoskeletal and signaling activities propagate as waves which lead to the patches of activity seen on the pseudopodia. The propagating waves suggest that these networks are excitable. 5) When exposed to spatially uniform chemotactic stimuli, cells ''freeze'' movement and then round up or ''cringe'', then spread projections in multiple directions, and finally resume normal migratory behavior. 6) These events are driven by a stereotypical kinetically complex signaling response (i.e. Ras activation or PIP 3 production) which is observed in immobilized as well as control cells. Within seconds of stimulus addition, cells produce an initial response around the whole perimeter that shuts off rapidly within 30 seconds and is followed by secondary patches lasting several minutes. 7) During persistent stimulation cells eventually adapt to the current level of stimulation but will respond again if the stimulus is increased or is reapplied after a period of recovery. 8) When cells are exposed to a gradient of chemoattractant, they produce directional responses and migrate directionally. 9) In immobilized cells, patches of response are stochastic but biased towards the high side of the gradient. 10) The directional response is amplified compared to the external gradient in the sense that it is confined to the anterior of the cell. 11) Immobilized cells exposed to two gradients produce responses on both ends, while migrating polarized cells choose one or the other sources. 12) Adaptation enables chemotactic cells to adjust their sensitivity and respond only to the steepness but not the midpoint concentration of the gradient. 13) The intrinsic polarity of cells is regulated. In Dictyostelium, for example, developed cells are more polarized than young cells. Polarity can also be enhanced by a period of migration in a gradient. 14) Polarized cells will turn when the gradient is shifted rather than creating a new front. As shown in Table 1, the modular framework of the polarized LEGI-BEN model accommodates all of these behaviors and experimental conditions whereas earlier models only account for various subsets of them.
The overall network topology has similarities with previously published models. Edelstein-Keshet and coworkers have proposed a number of models for cell polarity, motivated by the front-back appearance of Rho GTPases observed in neutrophils. Our model is similar to one of their proposed models (Model 4 in Ref. [45]). There, multiple positive feedbacks (or double negative feedback loops, for example Rac x Rho xCdc42RRac, RacRPIP 3 RRac, Rac x Rho x PIP 3 RRac) generate a bistable system. While no direct negative feedback loop is included, each of the GTPases is found in both active and inactive states, and so substrate depletion (the inactive states of the GTPases) can be considered as a negative feedback loop.
The combination of EN and Polarity modules is also comparable to a model proposed by Meinhardt [6] (who did not differentiate between the different processes) to explain chemotaxis. This model involved a positive feedback loop counterbalanced by two negative feedback loops -one local, the other global. Our combined Polarity-EN (excluding the LEGI mechanism) has two negative feedback paths, one that is local (XRY x X), and the other global (XRYRW x PRX). If we combine the two positive feedback loops of the excitable network (XRX) and the polarization model (XRYRZRPRX) then the topology of the models is similar. Neilson et al. carried out level set simulations of the Meinhardt model and generated results similar to ours for polarized cells, including pseudopod splitting, persistent random migration and turns in response to shallow gradients changing direction [12].
Models without a LEGI mechanism, or with only one positive feedback loop miss out on a number of important aspects of the overall response, however. These models do not adapt when given spatially uniform stimuli and cannot recreate the complex biphasic
14)
Turning [24,74,75] 5A-F + 2 NA 2 NA 2 + A: Biased excitability with polarity models: [6,9,12]. B: LEGI-biased excitability without polarity models: [10]; other biased ENs [11,76]; C: Excitability-only models: [7,15,17]; reviewed in Ref. [7]. D: LEGI models: [43,[77][78][79]. E: Polarization only: [45,80]; others reviewed in Ref. [3,4]. F: Stochastic with external bias models: [27,81]. G: Stochastic with persistence [27,30,82] (*these are models that fit statistics, rather than signaling models). doi:10.1371/journal.pcbi.1003122.t001 responses observed. Cells without an adaptation mechanism do not adjust sensitivity when the midpoint of the gradient is raised and hence perform less efficient chemotaxis (Fig. 7C,D). These simulations, however, show that adaptation is not absolutely required for chemotaxis. In fact, it is known that the response of migrating fibroblasts to uniform PDGF stimulation does not adapt, though these cells can only respond to gradients over a relatively narrow range of chemoattractant concentrations [46], as in our simulations of cells lacking the LEGI mechanism. In models with a single positive feedback loop, the simulated cells are always polarized. In reality, polarized and unpolarized cells can coexist in a population and cells can acquire increased polarity during a period of directed migration. The ability of the polarized LEGI-BEN to simulate cell movement under a number of varying scenarios illustrates the relative complexity and sophistication of the chemotactic signaling machinery. Experiments have demonstrated that the pathways governing chemotaxis have considerable redundancy at the biochemical level [47]. Our simulations show a similar redundancy at a systems-level, as they demonstrate that directional migration can be achieved without a LEGI mechanism (Fig. 7C, D), or without polarity (Fig. 1F). However, both mechanisms improve efficiency. As argued above, the LEGI mechanism allows the cells to respond to chemoattractant gradients over a wide range of mid-point concentrations. The polarity mechanism enables cells to migrate persistently in the absence of chemoattractant gradients and allows them to use the small directional bias obtained from the gradient to focus most activity towards the source (Fig. 4E, G).
Putative Biochemical Entities Associated with Model Elements
The modular framework of the polarized LEGI-BEN model gives rise to the entire spectrum of reported behaviors of cells but it is a conceptual model where individual biochemical entities are not assigned to specific model components or modules. An advantage of the modular approach is that, as additional data becomes available, the biochemical network within each module can be modeled in detail without altering the overall behavior of the other modules. Nevertheless, we can use several criteria to begin to assign various biochemical entities to the different modules (Table 2). First, the kinetic behaviors of certain biochemical and model components match under different conditions. Second, when levels of components and strengths of feedbacks within modules are varied, our simulated cells can ''phenocopy'' the behavior of various loss-and gain-of-function mutants.
We propose that the LEGI module incorporates the ''upstream'' components of the receptor signaling pathway (the chemoattractant-sensing GPCRs and associated G-proteins). Receptor-mediated G-protein dissociation is consistent with the local excitation process since during uniform or gradient stimulation they both rise rapidly and reach a steady-state level proportional to the level of receptor occupancy. Unfortunately, the biochemical identity of the global inhibition process that is expected to rise slowly and balance the persistent G-protein dissociation to bring about adaptation remains unknown. The receptors and G-proteins are not part of the excitable network since cells in the absence of chemoattractant or lacking G-protein function display excitability [18].
We propose that Ras and PI3K activity as well as other components traditionally viewed as elements of signal transduction pathways are part of the excitable network. Some of these, including Ras, PIP 3 , and Rac display excitable behavior such as wave propagation along the basal surface of the cell [10,17,[19][20][21]. Furthermore, constitutive Ras activity and inhibition of PIP 3 degradation cause excessive cytoskeletal activity and cellular extensions while inhibition of PI3K activity reduces this spontaneous activity [48][49][50]. We have also included many signal transduction components that either regulate, or are regulated by, Ras, PI3K, or Rac in this module. While biosensors are not available to directly test the premise, the participation of these components in propagating waves is expected since most of them behave coordinately with Ras and PI3K during uniform chemotactic stimulation. While many cytoskeletal proteins have also been shown to display excitable behavior, we have included these in the module that mediates protrusions [13][14][15]18].
The polarity module is likely to include both cytoskeletal and signaling proteins and well as ''polarity-specific'' components. Cells with elevated levels of PIP 3 or Ras activity or lacking myosin II appear to have decreased polarity [51][52][53]. However, it may be difficult to assign these components specifically to the polarity module since simulations in which the strengths of the negative feedback loops in either the polarization or excitable network modules are reduced lead to signaling levels and morphologies that are quite similar (Fig. 7A and Ref. [10]). Recently, it has been suggested that an activity akin to that achieved by W in the polarization module could be provided by membrane tension thus arguing for a role for cell mechanics [54,55]. Cells with impaired dynacortin, a global actin linking protein, are softer and also form more pseudopods that are less aligned with the gradient, reminiscent of simulations in which W is reduced.
Experimental Assessment of the Model
Here we suggest some possible experimental tests of our polarized LEGI-BEN model. 1) One assumption in the model is that the time-scale of adaptation (minutes) is longer than that of the excitable network (20-30 s). This can be tested by exposing cells to chemoattractant for several minutes, thus elevating the level of the global inhibitor in the LEGI module, and then removing the stimulus. Since excitation (E) is predicted to fall more rapidly than the inhibitor (I), the output of the LEGI module will transiently drop below its basal level. During this period of time, the spontaneous firing of the excitable network as well as its ability to be triggered by external stimulus will be decreased. 2) Currently the major evidence for excitability is observation of propagating waves. Further evidence of excitability could be obtained by testing whether cells generate all-or-none responses to supra-threshold stimuli, and whether they display a refractory period to repeated stimuli. According to our model, these hallmarks of excitable behavior should be largely independent of the actin cytoskeleton. 3) Treatment of cells with inhibitors of the cytoskeleton not only stops motility but also removes the polarity [42]. According to our model, without the mechanical or polarity module, a biased excitable network remains and activity is biased towards the gradient (Fig. 6C). We have recently found this prediction to be true when observing the dynamics of Ras and PI3K activity in cells treated with latrunculin in a steady gradient. 4) Our model hypothesizes that the persistence observed in unstimulated cells (item 1 in Table 1) is due to the same mechanism (the polarization module) that leads to polarized cells. One way to test this would be to track, in the absence of chemoattractant stimulus, the persistence of genetically modified cells that show poor polarity (e.g. cells lacking tsunami [56] or dynacortin [54]).
Signaling Network
We assume that the signaling network behaves as an excitable network [10]. It consists of two species (Fig. 1A). Component X acts as the activator: it is autocatalytic (it has strong positive feedback), and also activates the downstream component -we refer to this as the feedforward loop. The Y component provides negative feedback to X. A reaction-diffusion network, consisting of the following partial differential equations: describes the evolution of the activities of these two species. Both components in this subsystem can diffuse spatially, with diffusion coefficients D X and D Y , respectively. The signal U is the input to the excitable system, which incorporates several components: a basal level of activation (B), a stochastic component (N), contribution from the LEGI response regulator (R, described below) and the polarization component (P, also described below). The contribution of each of these is additive: The stochastic component is modeled as zero mean, white noise process with variance 1. Note that, in this context, the external gradient and the internally developed polarity compete to direct cell motion [57,58].
Receptor Signaling: Local Excitation, Global Inhibition
The LEGI mechanism involves three interacting processes (Fig. 1E). An external signal, which represents the local level of receptor occupancy (S), drives two of them: a fast, local excitation (E), and a slow, global (diffusible) inhibitor (I). These two control a response regulator, which can be active (R) or not (R T -R), where we have assumed that the total concentration (R T ) of the response regulator is constant. The system equations are given by: In the gradient simulations, the initial stimulus level is given by where r distance from each point on the cell boundary to the location of a hypothetical needle, which is either 10 mm (19% gradient) or 100 mm (6% gradient) away. This equation corresponds to the steady-state solution of the diffusion equation from a 5 mm needle in radial coordinates. The gradient is not updated as the cell moves to ensure that the gradient steepness is maintained. The gradient in our paper is defined as follows: In simulations that tested the LEGI mechanism without inhibitor ( Fig. 7B, C), the response regulator is described as follows:
Polarity Mechanism
The polarity mechanism is given by P = Z2W, where the individual components are also implemented as a local excitation, global inhibition mechanisms: For simplicity, we let D w be sufficiently high that W is spatially independent. The polarization module is activated by signal s pro , which represents actin polymerization and is proportional to Y.
Model of Cellular Deformations
To determine the effect of the model activities on the shape of a cell we used a level set framework to simulate cell shape changes as previously described [22,59]. In short, in the level set method (LSM) the cell is described as the zero-level set of a potential function Q(x,t), xMR 2 . Initially, we use a signed distance function as the potential function, defined by: if x is outside the cell, 0, otherwise: Here d(x,C) is the distance of position x to the cell boundary (initially a sphere of radius 5.1 mm). The evolution of the potential function is described by the Hamilton-Jacobi equation where v(x,t) describes the local velocity of the potential function. To obtain this velocity we apply different stresses on the cell and use a viscoelastic mechanical model of the cell to determine the local velocity. In our case we use: where s tot is the total stress applied on the cell, x m and x cor are the local displacements of the boundary and cortex, respectively, and K, D and B are viscoelastic components of the cell describing the elasticity (K) and viscosity (D) of the membrane, and the viscosity of the (B) of the cytoplasm. The velocity is given by v = dx m /dt. The total net stress (s tot ) includes the vector sum of the stresses acting on the cell. This stress includes contributions from passive components, such as surface tension, s ten = ck(x)n, where c is the local cortical tension, k is the local curvature, and n is a normal unit vector. Protrusive forces are proportional to the signal Y (using X leads to similar results) according to s pro = s 0 Y(h)n, representing actin polymerization. The conversion factor between the activity Y and the force is 35 nN/mm 2 . Based on the typical maximum activity level for Y seen in the simulations (,0.05 A. U.), this resulted in protrusive forces in the range of 1-3 nN/mm 2 , consistent with measured values of the maximum protrusive pressure due to actin polymerization (in the range of a few nN/ mm 2 [60,61]). We also include a stress that acts to ensure surface area conservation, s vol = k area (A(t)-A 0 ), where A is the surface area enclosed by the cell boundary either at time t or initially. Using these elements, we compute the total stress s tot = s pro +s ten+ s vol and use this to update the viscoelastic model parameters (x m and x cor ) above.
Model Implementation
The model and all simulations are implemented using Matlab (MathWorks, Natick, MA). Simulations were carried out in two steps. First, the PDEs for the signaling were solved around by representing the cell boundary as a one-dimensional system using periodic boundary conditions. This was discretized in space using 360 points. Spatial diffusion terms, which contain the second derivatives, are approximated by central differences in space; and by doing that, the partial differential equations are converted to ordinary differential equations. The solution of the stochastic differential equations was obtained using the SDE toolbox for Matlab [62]. The time step for simulation was set to 0.025 seconds. After solving the concentrations of all species (e.g. X, Y, W, Z), we compute the protrusive force using the concentration for Y, and use this protrusive force to update the potential function in the level set simulations, as described above. The potential function is solved on a Cartesian grid with spatial discretization of 19 points per mm. The assignment of Y activity levels to the protrusive force is done on a point-to-point pairing based on correspondence between angular positions relative to the cell centroid. The level set simulations were carried out using the Level Set Toolbox for Matlab using the first order forward Euler method [57].
Parameters
All model parameters are found in Table 3. Parameters for the LEGI and EN components were used in our previously published model [10]. The parameters of viscoelastic model were obtained using by fitting experimental measurements of aspirated Dictyostelium cells using a micropipette, as previously reported [22,59].
To choose the parameters for the polarization model we used as a benchmark that the persistence of cells is in the order of two minutes. Since the equations driving the polarity module are linear, the appearance of a pseudopod (represented by a sudden increase in s pro ) leads to an increase in Z followed by an exponential decay with rate exp(2k 2z t) (ignoring diffusion). Our nominal value for k 2Z is such that (1/k 2Z ) is approximately 1/ 66 seconds, which implies that the effect of that pseudopod is reduced to only e 22 <0.13 after two minutes. Note that during this time, more firings are possible, so that the effect of that initial pseudopod will likely be felt for longer periods. The time scale for the inhibitory element of the polarization module was chosen initially chosen in the same way. To arrive at the final values, we iterated in an ad hoc fashion, making sure that the activity of the excitable system (X and Y) showed some persistence in angle, but did not become locked in one position (sufficiently large feedback through the polarity module can lead to a bistable system). One way of measuring this is through the autocorrelation function C(t), measured at a fixed angles: where m is the mean level of signal. We computed this for unstimulated cells (where any persistence would come from the polarization module (Fig. S1) by choosing 10,000 angles and time points at random. Without the polarity module, the autocorrelation decays quickly (to less than 0.2) in approximately 30 seconds.
With the nominal parameter values of the polarization module, there is a decay (the initial correlation can be accounted by the firings of the excitable system) but the autocorrelation plateaus at approximately 0.4. Increasing the timescale of the polarization component (by making the degradation slower can increase this plateau. Increasing the degradation constant of Z eliminates any long-term correlation.
We also carried out parameter sensitivity analysis on several components of the system. Previously, we have demonstrated that the LEGI mechanism is extremely robust to parameter variations. The parameters in the mechanical model were experimentally obtained [22], and tested previously. For this reason we focused our analysis on the polarization module and the excitable network. We had previously carried out sensitivity analysis on the latter, but because the polarity module acts in feedback with this, we included it in this analysis. This consisted of varying the nominal degradation/production rates and observing the spatial distribution of X and Y, which show similar patterns. Because these drive the mechanical model in open loop, we did not do extensive tests on morphology, being constrained by the computational burden of the level set simulations.
Our results show that small changes (620%) in the rates controlling X and Y can have quite a strong effect on the excitable behavior (Fig. S2), as we noted previously [10]. These small differences in the rates of Z and W do not affect the spatial distribution of activity much or the peak levels of activity. To probe the robustness of the polarization module further, we also considered large (610 or 1/10) changes in the parameters of the polarization module (Fig. S3). Changes of this size on the rates of W change the spatial distributions of Y, but by values smaller than the change in the parameter. For example, increasing the degradation of W tenfold only increases the peak activity level by approximately 60%; decreasing this rate to 1/10 th its nominal value decreases the peak level of activity to approximately half. Moreover, the spatial distribution is largely unaffected. Changing the rates of Z has the greatest effect.
Chemotaxis Index
Chemotaxis index was computed using this following formula: After application of the gradient the cell trajectory was sampled every 5 seconds: P i . The values d i are the distances from sample points P i and P i-1 ; h i is the angle between the line connecting P i and P i-1 , and direction of the gradient. Video S3 Movement of polarized cells in the absence of a gradient. This video shows the movement of five cells with the polarized LEGI-BEN modules, but no external gradient (as in Fig. 3F). Each cell was simulated individually, and the trajectories superimposed, so was possible for different cells to overlap in the movie.
Supporting Information
(AVI) Video S4 Response of a polarized cell to a shift in the direction of a 6% gradient. The initial 6% gradient was applied at 300 s and pointed to the right. At 900 s, the direction was shifted to point to the top. This video corresponds to the simulation in Fig. 5A. This simulation uses the polarization, LEGI and EN modules. (AVI) Video S5 Response of an unpolarized cell to a shift in the direction of a 6% gradient. The initial 6% gradient was applied at 300 s and pointed to the right. At 900 s, the direction was shifted to point to the top. This video corresponds to the simulation in Fig. 5C. This simulation uses the LEGI and EN modules.
Video S6 Response of a polarized cell to a shift in the direction of a 19% gradient. The initial 19% gradient was applied at 300 s and pointed to the right. At 900 s, the direction was shifted to point to the top. This video corresponds to the simulation in Fig. 5D. This simulation uses the polarization, LEGI and EN modules. (AVI) Video S7 Development of polarity over a short exposure to a gradient. This simulation uses the polarization, LEGI and EN modules. A 12% gradient is applied at the beginning of the simulation (pointing to the right) and redirected at 130 s (pointing to the top). This video corresponds to the simulation of Fig. 5E.
Video S8 Development of polarity over a long exposure to a gradient. This simulation uses the polarization, LEGI and EN modules. A 12% gradient is applied at the beginning of the simulation (pointing to the right) and redirected at 430 s (pointing to the top). This video corresponds to the simulation of Fig. 5F. (AVI) Video S9 Response of cell to simultaneous gradients. Competing 19% gradients were applied to the cell (forming a ''V''shape with the bottom of the ''V'' at the center of the cell.) This simulation uses the polarization, LEGI and EN modules. This video corresponds to the simulation of Fig. 6A.
Video S10 Response of unpolarized cell to simultaneous gradients. Competing 19% gradients were applied to the cell. This simulation uses the LEGI and EN modules. The red line marks the track of the cell centroid. (AVI) Video S11 Response of immobilized cell to simultaneous gradients. Competing 19% gradients were applied to the cell at 180 s. This simulation uses the LEGI and EN modules but sets protrusive stresses to zero. The video corresponds to Fig. 6B. (AVI) Video S12 Response of immobilized cell to single 19% gradient. A single 19% gradient, pointing to the right, was applied to the cell at 180 s. This simulation uses the LEGI and EN modules but sets protrusive stresses to zero. The video corresponds to Fig. 6C. (AVI) Video S13 Response of cells with varying polarization modules loop strengths altered. The cells are responding to a 19% gradient pointing to the right. This video corresponds to Fig. 7A.
(AVI)
Video S14 Simulation of cells lacking adaptation. This simulation shows the response of the cell to a 19% gradient pointing to the right. The inhibitor level of the LEGI mechanism has been set constant, so the cell cannot adapt or adjust sensitivity. This video corresponds to Fig. 7B.
Video S15 Simulation of cells lacking adaptation in higher midpoint concentration. This simulation shows the response of the cell to a 19% gradient pointing to the right. The inhibitor level of the LEGI mechanism has been set constant, so the cell cannot adapt or adjust sensitivity. The midpoint chemoattractant concentration has been increased. This video corresponds to Fig. 7C.
Author Contributions
==
Domain: Biology Medicine Computer Science
|
[{"added": "2016-05-02T06:50:51.340Z", "created": "2013-07-01T00:00:00.000", "id": "1627002", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Computer Science"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1371/journal.pcbi.1003122", "pdf_hash": "f75b40f7eafdfa3a4c9f33dc77e363bf54d260c7", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:765", "s2fieldsofstudy": ["Biology"], "sha1": "f75b40f7eafdfa3a4c9f33dc77e363bf54d260c7", "year": 2013}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
An efficient algorithm for testing the compatibility of phylogenies with nested taxa
Background Semi-labeled trees generalize ordinary phylogenetic trees, allowing internal nodes to be labeled by higher-order taxa. Taxonomies are examples of semi-labeled trees. Suppose we are given collection \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathcal {P}$$\end{document}P of semi-labeled trees over various subsets of a set of taxa. The ancestral compatibility problem asks whether there is a semi-labeled tree that respects the clusterings and the ancestor/descendant relationships implied by the trees in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathcal {P}$$\end{document}P. The running time and space usage of the best previous algorithm for testing ancestral compatibility depend on the degrees of the nodes in the trees in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathcal {P}$$\end{document}P. Results We give a algorithm for the ancestral compatibility problem that runs in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$O(M_{\mathcal {P}}\log ^2 M_{\mathcal {P}})$$\end{document}O(MPlog2MP) time and uses \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$O(M_{\mathcal {P}})$$\end{document}O(MP) space, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M_{\mathcal {P}}$$\end{document}MP is the total number of nodes and edges in the trees in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathcal {P}$$\end{document}P. Conclusions Taxonomies enable researchers to expand greatly the taxonomic coverage of their phylogenetic analyses. The running time of our method does not depend on the degrees of the nodes in the trees in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathcal {P}$$\end{document}P. This characteristic is important when taxonomies—which can have nodes of high degree—are used.
Introduction
In the tree compatibility problem, we are given a collection P = {T 1 , T 2 , . . . , T k } of rooted phylogenetic trees with partially overlapping taxon sets. P is called a profile and the trees in P are the input trees. The question is whether there exists a tree T whose taxon set is the union of the taxon sets of the input trees, such that T exhibits the clusterings implied by the input trees. That is, if two taxa are together in a subtree of some input tree, then they must also be together in some subtree of T . The tree compatibility problem has been studied for over three decades [1][2][3][4].
In the original version of the tree compatibility problem, only the leaves of the input trees are labeled. Here we study a generalization, called ancestral compatibility, in which taxa may be nested. That is, the internal nodes may also be labeled; these labels represent higher-order taxa, which are, in effect, sets of taxa. Thus, for example, an input tree may contain the taxon Glycine max (soybean) nested within a subtree whose root is labeled Fabaceae (the legumes), itself nested within an Angiosperm subtree. Note that leaves themselves may be labeled by higher-order taxa. The question now is whether there is a tree T whose taxon set is the union of the taxon sets of the input trees, such that T exhibits not only the clusterings among the taxa, but also the ancestor/descendant relationships among taxa in the input trees. Our main result is a O(M P log 2 M P ) algorithm for the compatibility problem for trees with nested taxa, where M P is the total number of nodes and edges in the trees in P.
Background
The tree compatibility problem is a basic special case of the supertree problem. A supertree method is a way to synthesize a collection of phylogenetic trees with partially overlapping taxon sets into a single supertree that represents the information in the input trees. The supertree approach, proposed in the early 90s [5,6], has been used successfully to build large-scale phylogenies [7].
The original supertree methods were limited to input trees where only the leaves are labeled. Page [8] was among the first to note the need to handle phylogenies where internal nodes are labeled, and taxa are nested. A major motivation is the desire to incorporate taxonomies as input trees in large-scale supertree analyses, as way to circumvent one of the obstacles to building comprehensive phylogenies: the limited taxonomic overlap among different phylogenetic studies [9]. Taxonomies group organisms according to a system of taxonomic rank (e.g., family, genus, and species); two examples are the NCBI taxonomy [10] and the Angiosperm taxonomy [11]. Taxonomies spanning a broad range of taxa provide structure and completeness that might be hard to obtain otherwise. A recent example of the utility of taxonomies is the Open Tree of Life, a draft phylogeny for over 2.3 million species [12].
Taxonomies are not, strictly speaking, phylogenies. In particular, their internal nodes and some of their leaves are labeled with higher-order taxa. Nevertheless, taxonomies have many of the same mathematical characteristics as phylogenies. Indeed, both phylogenies and taxonomies are semi-labeled trees [13,14]. We will use this term throughout the rest of the paper to refer to trees with nested taxa.
The fastest previous algorithm for testing ancestral compatibility, based on earlier work by Daniel and Semple [15], is due to Berry and Semple [16]. Their algorithm runs in O log 2 n · τ P time using O(τ P ) space. Here, n is the number of distinct taxa in P and is the set of internal nodes of T i , for each i ∈ {1, . . . , k}, and d(v) is the degree of node v. While the algorithm is polynomial, its dependence on node degrees is problematic: semilabeled trees can be highly unresolved (i.e., contain nodes of high degree), especially if they are taxonomies.
Our contributions
As stated earlier, our main result is an algorithm to test ancestral compatibility that runs in O(M P log 2 M P ) time, using O(M P ) space. These bounds are independent of the degrees of the nodes of the input trees, a valuable characteristic for large datasets that include taxonomies. To achieve our result, we extend ideas from our recent algorithm for testing the compatibility of ordinary phylogenetic trees [2]. As in that algorithm, a central notion in the current paper is the display graph of profile P, denoted H P . This is the graph obtained from the disjoint union of the trees in P by identifying nodes that have the same label (see the section titled "Testing ancestral compatibility"). The term "display graph" was introduced by Bryant and Lagergren [17], but similar ideas have been used elsewhere. In particular, the display graph is closely related to Berry and Semple's restricted descendancy graph [16], a mixed graph whose directed edges correspond to the (undirected) edges of H P and whose undirected edges have no correspondence in H P . The second kind of edges are the major component of the τ P term in the time and space complexity of Berry and Semple's algorithm. The absence of such edges makes H P significantly smaller than the restricted descendancy graph. Display graphs also bear some relation to tree alignment graphs [18].
Here, we exploit the display graph more extensively than in our previous work. Although the display graph of a collection of semi-labeled trees is more complex than that of a collection of ordinary phylogenies, we are able to extend several of the key ideas-notably, that of a semiuniversal label-to the general setting of semi-labeled trees. As in [2], the implementation relies on a dynamic graph data structure, but it requires a more careful amortized analysis based on a weighing scheme.
Contents
This paper has five sections, in addition to this introduction. The section titled "Preliminaries" presents basic definitions regarding graphs, semi-labeled trees, and ancestral compatibility. The section titled "The display graph" introduces the display graph and discusses its properties. The section titled "Testing ancestral compatibility" presents BuildNT, our algorithm for testing ancestral compatibility. We first present the algorithm recursively, and then show how to transform it into an iterative algorithm, BuildNT N , that is easier to implement. We also give an example of the execution of BuildNT N . The "Implementation" section gives the implementation details for BuildNT N . The "Discussion" section gives some concluding remarks.
Graph notation
Let G be a graph. V(G) and E(G) denote the node and edge sets of G. The degree of a node v ∈ V (G) is the number of edges incident on v. A tree is an acyclic connected graph. In this paper, all trees are assumed to be rooted. For a tree T, r(T) denotes the root of T.
and u ≤ T v, then u is the parent of v and v is a child of u. If neither u ≤ T v nor v ≤ T u hold, then we write u T v and say that u and v are not comparable in T.
Semi-labeled trees
A semi-labeled tree is a pair T = (T , φ) where T is a tree and φ is a mapping from a set L(T ) to V(T) such that, for every node v ∈ V (T ) of degree at most two, v ∈ φ(L(T )) . L(T ) is the label set of T and φ is the labeling function of T .
For every node v ∈ V (T ), φ −1 (v) denotes the (possibly empty) subset of L(T ) whose elements map into v; these elements as the labels of v (thus, each label is a taxon). If Note that, by definition, every leaf in a semi-labeled tree is labeled. Further, any node, including the root, that has a single child must be labeled. Nodes with two or more children may be labeled or unlabeled. A semi-labeled tree T = (T , φ) is singularly labeled if every node in T has at most one label; T is fully labeled if every node in T is labeled.
Semi-labeled trees, also known as X-trees, generalize ordinary phylogenetic trees, also known as phylogenetic X-trees [14]. An ordinary phylogenetic tree is a semilabeled tree T = (T , φ) where r(T) has degree at least two and φ is a bijection from L(T ) into leaf set of T (thus, internal nodes are not labeled).
Let T = (T , φ) be a semi-labeled tree and let ℓ and ℓ ′ be two labels in L(T ). If φ(ℓ) ≤ T φ(ℓ ′ ), then we write ℓ ≤ T ℓ ′ , and say that ℓ ′ is a descendant of ℓ in T and that ℓ is an ancestor of ℓ ′ . We write , then we write ℓ � T ℓ ′ and say that ℓ and ℓ ′ are not comparable in T . If T is fully labeled and φ(ℓ) is the parent of φ(ℓ ′ ) in T, then ℓ is the parent of ℓ ′ in T and ℓ ′ is a child of ℓ in T ; two labels with the same parent are siblings.
Let T = (T , φ) be a semi-labeled tree. For each u ∈ V (T ), X(u) denotes the set of all labels in the subtree of T rooted at u; that is, X(u) = v:u≤ T v φ −1 (v). X(u) is called a cluster of T. Cl(T ) denotes the set of all clusters of T . It is well known [14, Theorem 3.5.2] that a semilabeled tree T is completely determined by Cl(T ). That is, if Cl(T ) = Cl(T ′ ) for some other semi-labeled tree T ′ , then T is isomorphic to T ′ .
Suppose A ⊆ L(T ) for a semi-labeled tree T = (T , φ). The restriction of T to A, denoted T |A, is the semi-labeled tree whose cluster set is Cl(T |A) = {X ∩ A : X ∈ Cl(T ) and X ∩ A � = ∅}. Intuitively, T |A is obtained from the minimal rooted subtree of T that connects the nodes in φ(A) by suppressing all vertices of degree two that are not in φ(A).
For a semi-labelled tree T , let us define D(T ) and N (T ) as follows.
Note that D(T ) consists of ordered pairs, while N (T ) consists of unordered pairs.
Profiles and ancestral compatibility
Throughout the rest of this paper We refer to P as a profile, and write L(P) to denote i∈[k] L(T i ), the label set of P. Figure 1 shows a profile where P is ancestrally compatible if there is a rooted semilabeled tree T that ancestrally displays each of the trees in P. If T exists, we say that T ancestrally displays P (see Fig. 2).
Given a subset X of L(P), the restriction of P to X, denoted P|X, is the profile defined as The proof of the following lemma is straightforward.
Lemma 2 Suppose P is ancestrally compatible and let T be a tree that ancestrally displays P. Then, for any X ⊆ L(P), T |X ancestrally displays P|X. For technical reasons, fully labeled trees are easier to handle than those that are not. Suppose P contains trees that are not fully labeled. We can convert P into an equivalent profile P ′ of fully-labeled trees as follows.
where ℓ is a distinct element from L ′ . We refer to P ′ as the profile obtained by adding distinct new labels to P (see Fig. 1). [15]) Let P ′ be the profile obtained by adding distinct new labels to P. Then, P is ancestrally compatible if and only if P ′ is ancestrally compatible. Further, if T is a semi-labeled phylogenetic tree that ancestrally displays P ′ , then T ancestrally displays P.
Lemma 3 (Daniel and Semple
From this point forward, we make the following assumption.
Assumption 1 For each i ∈ [k]
, T i is fully and singularly labeled.
By Lemma 3, no generality is lost in assuming that all trees in P are fully labeled. The assumption that the trees are singularly labeled is inessential; it is only for clarity. Note that, even with the latter assumption, a tree that ancestrally displays P is not necessarily singularly labeled. Figure 2 illustrates this fact.
The display graph
The display graph of a profile P, denoted H P , is the graph obtained from the disjoint union of the underlying trees T 1 , T 2 , . . . , T k by identifying nodes that have the same label. Multiple edges between the same pair of nodes are replaced by a single edge. See Fig. 3.
H P has O(M P ) nodes and edges, and can be constructed in O(M P ) time. By Assumption 1, there is a bijection between the labels in L(P) and the nodes of H P . Thus, from this point forward, we refer to the nodes of H P by their labels. It is easy to see that if H P is not connected, then P decomposes into label-disjoint subprofiles, and that P is compatible if and only if each subprofile is compatible. Thus, without loss of generality, we shall assume the following.
Assumption 2 H P is connected.
Positions
Our compatibility algorithm processes the trees in P from the top down, starting at the roots. We refer to the set of nodes in P currently being considered as a "position". The algorithm advances from the current position to the next by replacing certain nodes in the current position by their children. Formally, a position (for P) is a vector . Since labels may be shared among trees, we may have
Lemma 4 For any valid position U,
Proof By (V2), we have that T i |Desc i (U) and The lemma then follows from the definition of P|Desc P (U) .
For any valid position U, H P (U) denotes the subgraph of H P induced by Desc P (U). A valid position of special interest to us is U root , the root position, defined as follows.
That is, for each i ∈ [k], U root (i) is a singleton containing only the label of r(T i ). In Fig. 3,
Semi-universal labels
Let U be a valid position, and let ℓ be a label in U. Fig. 3, labels 1 and 2 are semi-universal in U root , but g is not, since g is in both L(T 2 ) and L( The term "semi-universal", borrowed from Pe'er et al. [19], derives from the following fact. Suppose that P is ancestrally compatible, that T is a tree that ancestrally displays P, and that ℓ is a semi-universal label for some valid position U. Then, as we shall see, ℓ must label the root u ℓ of a subtree of T that contains all the descendants of ℓ in T i , for every i such that ℓ ∈ L(T i ). The qualifier "semi" is because this subtree may also contain labels that do not descend from ℓ in any input tree, but descend instead from some other semi-universal label ℓ ′ in U. In this case, ℓ ′ also labels u ℓ . We exploit this property of semi-universal labels in our ancestral compatibility algorithm and its proof of correctness (see "Testing ancestral compatibility").
For each label ℓ ∈ L(P), let k ℓ denote the number of input trees that contain label ℓ. We can obtain k ℓ for every ℓ ∈ L(P) in O(M P ) time during the construction of H P .
Successor positions
For every i ∈ [k] and every ℓ ∈ L(T i ), let Ch i (ℓ) denote the set of children of ℓ in L(T i ). (1) In Fig. 3, the set of semi-universal labels in U root is Proof It suffices to argue that U ′ satisfies conditions (V1) and (V2). The lemma then follows from the fact that the connected components of H P (U ′ ) are label-disjoint.
U ′ must satisfy condition (V1), since U does. Suppose ℓ ∈ S. Then, for each i ∈ [k] such that ℓ ∈ L(T i ), Thus, since (V2) holds for U, it also holds for U ′ .
Overview of the algorithm
BuildNT (Algorithm 1) is our algorithm for testing compatibility of semi-labeled trees. Its argument, U, is a valid position in P such that H P (U) is connected. BuildNT relies on the fact-proved later, in Theorem 1-that if P|Desc P (U) is compatible, then U must contain a nonempty set S of semi-universal labels. If such a set S exists, the algorithm replaces U by its successor U ′ with respect to S. It then processes each connected component of H P (U ′ ) recursively, to determine if the associated subprofile is compatible. If all the recursive calls are successful, then their results are combined into a supertree for P|Desc P (U).
Page 6 of 12 Deng and Fernández-Baca Algorithms Mol Biol (2017) 12:7 In detail, BuildNT proceeds as follows. Line 1 computes the set S of semi-universal labels in U. If S is empty, then, P|Desc P (U) is incompatible, and, thus, so is P. This fact is reported in Line 3. Line 4 creates a tentative root r U , labeled by S, for the tree T U for L(U). Line 5 checks if S contains exactly one label ℓ, with no proper descendants. If so, by the connectivity assumption, ℓ must be the sole member of Desc P (U); that is, L(U) = ℓ . Therefore, Line 6 simply returns the tree with a single node, labeled by S = {ℓ}. Line 7 updates U, replacing it by its successor with respect to S. Let W 1 , W 2 , . . . , W p be the connected components of H P (U) after updating U. By Lemma 6, U |W j is a valid position, for each j ∈ [p]. Lines 8-12 recursively invoke BuildNT on U |W j for each j ∈ [p], to determine if there is a tree t j that ancestrally displays P|Desc P (U ∩ W j ). If any subproblem is incompatible, Line 12 reports that P is incompatible. Otherwise, Line 13 returns the tree obtained by making the t j s the subtrees of root r U .
Next, we argue the correctness of BuildNT.
Lemma 7 Let U be a valid position in P. If BuildNT(U ) returns a tree T U , then T U is a phylogenetic tree such that L(T U ) = L(U).
Proof We use induction on |L(U)|. The base case, where |L(U )| = 1, is handled by Lines 5-6. In this case, S = L(U) = {ℓ} and BuildNT(U ) correctly returns the tree consisting of a single node, labeled by {ℓ}. Otherwise, let W 1 , . . . , W p be the connected components of H P (U) in step 8. Since BuildNT(U ) returns tree T U , it must be the case that, for each j ∈ [p], the result t j returned by the recursive call to BuildNT(U |W j ) in Line 10 is a tree. Since |S| ≥ 1, we have |L(W j )| < |L(U )|, for each j ∈ [p].
Thus, we can assume inductively that t j is a phylogenetic tree for L(W j ). Since S ∪ j∈[p] L(W j ) = L(U ), the tree returned in Line 13 is a phylogeny with species set L(U).
Theorem 1 Let P = {T 1 , T 2 , . . . , T k } be a profile and let U root be the root position, as defined in Eq. (1). Then, BuildNT(U root ) returns either (i) a semi-labeled tree T that ancestrally displays P, if P is ancestrally compatible, or (ii) incompatible otherwise.
Proof BuildNT(U root ) either returns a tree or incompatible. We consider each case separately. . Consider any (ℓ, ℓ ′ ) ∈ D(T i ). Then, ℓ has a child ℓ ′′ in T i such that ℓ ′′ ≤ T i ℓ ′ -note that we may have ℓ ′′ = ℓ. There must be a recursive call to BuildNT(U ), for some valid position U, where ℓ is the set S of semi-universal labels obtained in Line 1. By Observation 2, label ℓ ′′ , and thus ℓ ′ , both lie in one of the connected components of the graph obtained by deleting all labels in S, including ℓ, and their incident edges from H P (U). It now follows from the construction of T that (ℓ, ℓ ′ ) ∈ D(T ). Thus, D(T i ) ⊆ D(T ). Now, consider any {ℓ, ℓ ′ } ∈ N (T i ). Let v be the lowest common ancestor of φ i (ℓ) and φ i (ℓ ′ ) in T i and let ℓ v be the label of v. Then, ℓ v has a pair of children, ℓ 1 and ℓ 2 say, in T i such that ℓ 1 ≤ T i ℓ, and ℓ 2 ≤ T i ℓ ′ . Because BuildNT(U root ) returns a tree, there are recursive calls BuildNT(U 1 ) and BuildNT(U 2 ) for valid positions U 1 and U 2 such that ℓ 1 is semi-universal for U 1 and ℓ 2 is semi-universal for U 2 . We must have U 1 � = U 2 ; other-wise, |U 1 (i)| = |U 2 (i)| ≥ 2, and, thus, neither ℓ 1 nor ℓ 2 is semi-universal, a contradiction. Further, it follows from the construction of T that we must have Desc P (U 1 ) ∩ Desc P (U 2 ) = ∅. Hence, ℓ � T ℓ ′ , and, therefore, {ℓ, ℓ ′ } ∈ N (T ). (ii) Asssume, by way of contradiction, that BuildNT(U root ) returns incompatible, but that P is ancestrally compatible. By assumption, there exists a semi-labeled tree T that ancestrally displays P. Since BuildNT(U root ) returns incompatible, there is a recursive call to BuildNT(U ) for some valid position U such that U has no semi-universal label, and the set S of Line 1 is empty. By Lemma 2, T |Desc P (U) ancestrally displays P|Desc P (U). Thus, by Lemma 4, T |Desc P (U) ancestrally displays T i |Desc i (U) , for every i ∈ [k]. Let ℓ be any label in the label set of the root of T |Desc P (U). Then, for each i ∈ [k] such that ℓ ∈ L(T i ), ℓ must be the label of the root of T i |Desc i (U). Thus, for each such i, U (i) = {ℓ}. Hence, ℓ is semi-universal in U, a contradiction.
L(U) in the supertree. The body of the loop closely follows the steps performed by a call to BuildNT(U ). Line 5 computes the set S of semi-universal labels in U. If S is empty, the algorithm reports that P is incompatible and terminates (Lines 6-7). The algorithm then creates a tentative root r U labeled by S for the tree T U for L(U), and links r U to its parent (Line 8). If S consists of exactly one element that has no proper descendants, we skip the rest of the current iteration of the while loop, and continue to the next iteration (Lines 9-10). Line 11 replaces U by its successor with respect to S. Lines 13-14 enqueue each of U |W 1 , U |W 2 , . . . , U |W p , along with r U , for processing in a subsequent iteration. If the while loop terminates without any incompatibility being detected, the algorithm returns the tree with root r U root .
Although the order in which BuildNT N processes connected components differs from that of BuildNTbreadth-first instead of depth-first-, it is straightforward
An iterative version
We now present BuildNT N (Algorithm 2), an iterative version of BuildNT, which lends itself naturally to an efficient implementation. BuildNT N performs a breadthfirst traversal of BuildNT's recursion tree, using a first-in first-out queue Q that stores pairs of the form (U, pred), where U is a valid position in P and pred is a reference to the parent of the node corresponding to U in the supertree built so far. BuildNT N simulates recursive calls in BuildNT by enqueuing pairs corresponding to subproblems. We explain this in more detail next.
BuildNT N initializes its queue to contain the starting position, U root , with a null parent. It then proceeds to the while loop of Lines 3-14. Each iteration of the loop starts by dequeuing a valid position U, along with a reference pred to the potential parent for the subtree for to see that the effect is equivalent, and the proof of correctness of BuildNT (Theorem 1) applies to BuildNT N as well. We thus state the following without proof. Theorem 2 Let P = {T 1 , T 2 , . . . , T k } be a profile. Then, BuildNT N (P) returns either (i) a semi-labeled tree T that ancestrally displays P, if P is ancestrally compatible, or (ii) incompatible otherwise.
Let Q be BuildNT N 's first-in first-out queue. In the rest of the paper, we will say that a valid position U is in Q if (U, pred) ∈ Q, for some pred. Let H Q be the subgraph of H P induced by {Desc(U) : U is in Q}. By Observation 1, H Q is obtained from H P through edge and node deletions.
Lemma 8 At the start of any iteration of BuildNT N 's while loop, the set of connected components of H Q is {V (H P (U)) : U is in Q}.
Proof The property holds at the outset, since, by Assumption 2, H P = H P (U root ) is a connected graph, and the only element of Q is (U root , null). Assume that the property holds at the beginning of iteration l. Let (U, pred) be the element dequeued from Q in Line 4. Then, H P (U) is connected. In place of (U, pred), Lines 13-14 enqueue (U|W j , r U ), for each j ∈ [p], where, by construction, H P (U|W j ) is a connected component of H P (U). Thus, the property holds at the beginning of iteration l + 1.
In other words, Lemma 8 states that each iteration of BuildNT N (P) deals with a subgraph of H P , whose connected components are in one-to-one correspondence with the valid positions stored in Q. This is illustrated by the next example. In each figure, H Q is shown on the left and the current supertree is shown on the right. For brevity, the figures only exhibit the state of H Q and the supertree after all the nodes at each level are generated. The various valid positions processed by BuildNT N (P) are denoted by U α , for different subscripts α; S α denotes the semi-universal labels in U α , and U ′ α denotes the successor of U α with respect to S α . We write L α as an abbreviation for L(U α ) The root of the tree for L α is r U α and is labeled by S α .
An example
Initially, Q = ((U root , null)). In what follows, the elements of Q are listed from front to rear.
Level 2. Refer to Fig. 6. We have S 11 = {g}, so H P (U ′ 11 ) has two components W 111 and W 112 . Let U 111 = U ′ 11 |W 111 and U 112 = U ′ 11 |W 112 . Then, The only semi-universal labels in U 12 , U 21 , and U 22 are, respectively, e, h, and i. Since none of these labels have proper descendants, each of them is a leaf in the supertree.
After level 2 is processed, Q = ((U 111 , r 11 ), (U 112 , r 11 )). The only semi-universal label in U 112 is d. Since d has no proper descendants, it becomes a leaf in the supertree.
Level 4. Refer to Fig. 8. The only semi-universal labels in U 1111 and U 1112 are, respectively, b and c. Since neither of these labels have proper descendants, each of them is a leaf in the supertree.
After level 4 is processed, Q is empty, and BuildNT N (P) terminates.
Implementation
Here we prove the following result.
Theorem 3
There is an algorithm that, given a profile P of rooted trees, runs in O(M P log 2 M P ) time, and either returns a tree that displays P, if P is compatible, or reports that is P is incompatible otherwise.
We prove this theorem by showing how to implement BuildNT N so that the algorithm runs in O(M P log 2 M P ) on any profile P.
As in the section titled "An iterative version", let H Q denote the subgraph of H P associated with the valid positions in BuildNT N 's queue. By Lemma 8, each valid position U in Q corresponds to one connected component of H Q -namely Desc(U) -and vice-versa. We use this fact in the implementation of BuildNT N : alongside each valid position U in Q, we also store a reference to the respective connected component, together with additional information, described next, to quickly identify semi-universal labels.
Let U be any valid set in Q, let Y = V (H P (U)) be the corresponding connected component of H Q , and let ℓ be any label in Y. Our implementation maintains the following data fields. (Recall that k ℓ is the number of input trees that contain ℓ.). • Y .exposed, a set consisting of all i ∈ [k] such that Y .map(i) = {ℓ} for some ℓ ∈ Y such that ℓ.count = k ℓ . • Y .weight, which equals ℓ∈Y k ℓ . This field is needed for technical reasons, to be explained later.
For the purpose of analysis, we assume that the exposed fields are represented as balanced binary search trees (BSTs), which ensures O(log k) = O(log M P ) time per access and update. The map fields are also implemented using BSTs. We store the set J U = {i ∈ [k] : U (i) � = ∅} as a BST, enabling is to determine in O(log k) time if an index i is in J U , and, if this is the case, to access Y .map(i) . The latter is also stored as a BST, allowing us to search and update Y .
Note that, in practice, hashing may be a better alternative for both exposed and map fields, as it offers expected constant time performance per operation.
The data fields listed above allow us to efficiently retrieve the set S of semi-universal labels in U, as needed in line 5 of BuildNT N (P). Indeed, suppose that U is the valid position extracted from Q at the beginning of an iteration of the while loop of Lines 3-14, and that Y = V (H P (U)). Then, by Lemma 5, we have S = {v ∈ Y .map(i) : i ∈ Y .exposed}. What remains is to devise an efficient way to update these fields for each of the connected components of H P (U) created by replacing U with its successor in Line 11.
Let U ′ be the value of U after Line 11; thus, U ′ is the successor of U. By Observation 2, H P (U ′ ) is obtained from H P (U) through edge and node deletions. We need to (a) Generate the new connected components resulting from these deletions, and (b) Produce the required map, count, and exposed data fields for the various connected components.
We accomplish (a) using the dynamic graph connectivity data structure of Holm et al. [20], which we refer to as HDT. HDT allows us to maintain the list of nodes in each component, as well as the number of these nodes so that, if we start with no edges in a graph with N nodes, the amortized cost of each update is O(log 2 N ). Since H P has O(M P ) nodes, each update takes O(log 2 M P ) time. The total number of edge and node deletions performed by BuildNT N (P) -including all deletions in the interations-is at most the total number of edges and nodes in H P , which is O(M P ). HDT allows us to maintain connectivity information throughout the entire algorithm in O(M P log 2 M P ) time, which is within the time bound claimed in Theorem 3. For part (b), we need to augment HDT in order to maintain the the required data fields for the various connected components created during edge and node deletion. In the next subsections, we describe how to do this. We begin by explaining how to initialize all the required data fields for H P = H P (U root ).
Initializing the data fields
Graph H P (U root ) has a single connected component, Y root = L(P), which is the entire vertex set of the graph. We initialize the data fields as follows.
We initialize the count fields in O(M P ) time as follows: 1. Set ℓ.count to 0 for all ℓ ∈ L(P). Once the count fields are initialized, it is easy to initialize Y root .exposed in O(k) time. Thus, we can initialize all the required fields in O(M P ) time.
Maintaining the data fields
Suppose that all data fields fields are correctly computed for every connected component that is in Q at the beginning of an iteration of the while loop in 3-14 of BuildNT N . We now show how to generate the same fields efficiently for the new connected components created by Line 11.
Computing successor positions
Let U be the valid position extracted from Q at the beginning of an iteration of BuildNT N 's while loop, and let Y = V (Desc(U)) be the associated connected component. Assume all the data fields for Y have been correctly computed. To obtain the successor of U in Line 11 of BuildNT N , we perform the following steps.
1. Identify the set S of semi-universal labels in U. As we saw, this set is given by S = {ℓ ∈ Y .map(i) : i ∈ Y .exposed}. 2. Set Y .map(i) = ∅, for every i ∈ Y .exposed. 3. Make Y .exposed = ∅. 4. For each ℓ ∈ S and each i such that ℓ ∈ L(T i ), do the following.
is a singleton set {α}, increment α.count by one. If α.count = k ℓ , add i to Y .exposed. • Otherwise, Y .map(i) is undefined. 5. For each label ℓ in S, delete the edges incident on ℓ and then ℓ itself, updating the data fields as necessary after each deletion.
The total number of operations on map and exposed fields in Steps 1-4 is O( ℓ∈S k ℓ ). Since each label becomes semi-universal at most once, the total number of operations on map fields over the entire execution of BuildNT N (P) is O( ℓ∈L(P) k ℓ ), which is O(M P ). The same bound holds for updates to count and exposed fields. Next let us focus on how to handle the deletion of a single edge in Step 5.
Deleting an edge
To delete an edge between ℓ and a child α of ℓ, we proceed as follows.
To perform Step 2, we use the well-known technique of scanning the smaller component [21]. We first consult HDT to determine which of Y 1 or Y 2 has fewer nodes. Assume, without loss of generality, that |Y 1 | ≤ |Y 2 |. We initialize Y 1 .weight to 0 and Y 2 .weight to Y .weight. We then scan the labels of Y 1 , incrementing Y 1 .weight by k ℓ for each label ℓ ∈ Y 1 . When the scan of Y 1 is complete, we make Y 2 .weight = Y 2 .weight − Y 1 .weight. We claim that any label ℓ ∈ L(P) is scanned O(log M P ) times over the entire execution of BuildNT N (P). To verify this, let N (ℓ) be the number of nodes in the connected component containing ℓ. Suppose that, initially, N (ℓ) = N . Then, the rth time we scan ℓ, N (ℓ) ≤ N /2 r . Thus, ℓ is scanned O(log N ) times. The claim follows, since N = O(M P ). Therefore, the total number of updates over all labels is O(M P log M P ).
Each execution of Step 6(a) updates each of Y 1 .map(i) and Y 2 .map(i) once.
Step 6(b) is more complex, but can also be accomplished with O(1) data field updates. We omit the (tedious) details. In summary, each execution of step 6 for some β ∈ L(P) performs O(k β ) data field updates.
Let us track the number of data field updates in Step 6 that can be attributed to some specific label β ∈ L(P) over the entire execution of BuildNT N (P). Let w r (β) be the weight of the connected component containing β at the beginning of Step 6, on the rth time that β is considered in that step. Thus, w 0 (β) ≤ ℓ∈L(P) k ℓ . We claim that w r (β) ≤ w 0 (β)/2 r . The reason is that we only consider β if (a) β is contained in one of the two components that result from deleting an edge in step 1 and (b) the component containing β has the smaller weight. Hence, the number of times β is considered in step 6 over the entire execution of BuildNT N (P) is O(log w 0 (β)) , which is O(log M P ), since w 0 (β) = O(M P ). Therefore, the total number of data field updates in Step 6, over all labels in L(P) considered throughout the entire execution of BuildNT N (P), is O(log M P · ℓ∈L(P) k ℓ ), which is O(M P log M P ).
Summary
Let us review the running times of each aspect of our implementation of BuildNT N .
• Initializing the data structures. This has two parts. • Maintaining the data structures. This also has two parts.
• Updating the HDT data structure. There are O(M P ) edge and node deletions, at an amortized cost of O(log 2 M P ) per deletion, yielding a total time of O(M P log 2 M P ). • Maintaining the relevant data fields for the connected components. We have seen that the total number of updates is O(M P log M P ). Assume, conservatively, that each update can be done in O(log M P ) time. Then, this part takes a total of O(M P log 2 M P ) over the entire execution of BuildNT N .
We conclude that the total running time of BuildNT N (P) is O(M P log 2 M P ), completing the proof of Theorem 3.
Discussion
Like our earlier algorithm for compatibility of ordinary phylogenetic trees, the more general algorithm presented here, BuildNT N , is a polylogarithmic factor away from optimality (a trivial lower bound is �(M P ), the time to read the input). BuildNT N has a linear-space implementation, using the results of Thorup [22]. A question to be investigated next is the performance of the algorithm on real data. Another important issue is integrating our algorithm into a synthesis method that deals with incompatible profiles.
==
Domain: Biology Medicine Computer Science
|
[{"added": "2017-07-17T21:43:59.887Z", "created": "2017-03-16T00:00:00.000", "id": "8548037", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Computer Science"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://almob.biomedcentral.com/track/pdf/10.1186/s13015-017-0099-7", "pdf_hash": "94ca3b3653e45461de79950c5bc125087ea3f6bb", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:6351", "s2fieldsofstudy": ["Biology"], "sha1": "94ca3b3653e45461de79950c5bc125087ea3f6bb", "year": 2017}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
The Vast, Conserved Mammalian lincRNome
We compare the sets of experimentally validated long intergenic non-coding (linc)RNAs from human and mouse and apply a maximum likelihood approach to estimate the total number of lincRNA genes as well as the size of the conserved part of the lincRNome. Under the assumption that the sets of experimentally validated lincRNAs are random samples of the lincRNomes of the corresponding species, we estimate the total lincRNome size at approximately 40,000 to 50,000 species, at least twice the number of protein-coding genes. We further estimate that the fraction of the human and mouse euchromatic genomes encoding lincRNAs is more than twofold greater than the fraction of protein-coding sequences. Although the sequences of most lincRNAs are much less strongly conserved than protein sequences, the extent of orthology between the lincRNomes is unexpectedly high, with 60 to 70% of the lincRNA genes shared between human and mouse. The orthologous mammalian lincRNAs can be predicted to perform equivalent functions; accordingly, it appears likely that thousands of evolutionarily conserved functional roles of lincRNAs remain to be characterized.
Introduction
The great majority of mammalian genome sequences are transcribed, at least occasionally, a phenomenon known as pervasive transcription [1][2][3][4]. More specifically, tiling array analyses of several human chromosomes have shown that over 90% of the bases are transcribed in at least one cell type [1,[5][6][7][8]. The analogous analysis in mouse has demonstrated transcription for over 60% of the genome [9][10][11]. Among the transcripts there are numerous long intergenic non-coding RNA (lincRNA), i.e. RNA molecules greater than 200 nucleotides in length that are encoded outside other identified genes. Some of the lincRNAs have been shown to perform various regulatory roles but the majority remain functionally uncharacterized [7,[12][13][14][15][16][17]. Furthermore, the fraction of the genome allotted to lincRNAs remains unknown.
A popular view that the vast majority of lincRNAs are byproducts of background transcription, ''simply the noise emitted by a busy machine'' [18,19], is rooted in their typically low abundance and poor evolutionary conservation compared to protein-coding sequences and small RNAs such as miRNAs and snoRNAs [20]. However, some of the lincRNAs do contain strongly conserved regions [21], and most lincRNAs show reduced substitution and insertion/deletion rates suggestive of purifying selection [12,22,23].
Given the general lack of strong sequence conservation, identification of lincRNAs on genome scale relies on expression analysis which makes comprehensive characterization of the mammalian lincRNome an elusive goal. The combination of different experimental approaches applied to transcriptomes of several species has resulted in continuous discovery of new transcripts [24], with the FANTOM project alone cataloguing more than 30,000 putative long non-coding transcripts in mouse tissues by full-length cDNA cloning [11,25]. The Support Vector Machine method has been applied to classify transcripts from the FANTOM3 project into coding and non-coding ones and accordingly estimate the number of long non-coding RNA in mouse. This analysis has led to the identification of 14,000 long non-coding RNAs and an estimate of the total number of such RNAs in the FANTOM3 data at approximately 28,000 [26].
Here we re-analyze the most reliable available sets of human and mouse lincRNAs using the latest next generation sequencing (RNAseq) data and apply a maximum likelihood approach to obtain a robust estimate of the size of the mammalian lincRNome. The results suggest that mammalian genomes are likely to encode at least twice as many lincRNAs as proteins.
Estimation of the sizes of human and mouse lincRNomes
We performed comparative analysis of the recently reported validated sets of 4662 human lincRNAs [27] and 4156 mouse lincRNAs [12,20,23] (see Methods for details) in an attempt to produce robust estimates of the human and mouse lincRNome sizes, and to measure the turnover of lincRNA genes in mammalian evolution. The validated sets consist of lincRNA species for which a specific profile of expression across tissuesand hence distinct functionality -are supported by multiple lines of evidence. Assuming that these sets of lincRNAs are random samples from human and mouse lincRNomes, comparison of the validated sets should yield robust estimates of the lincRNome size for each species. For this analysis, we deliberately chose to employ the validated sets only rather than the available larger sets of reported putative lincRNAs in order to reduce the effect of transcriptional noise and other artifacts.
A substantial fraction of the vast mammalian transcriptome, most likely the lower expressed transcripts, is expected to be nonfunctional. Therefore, to minimize the contribution of transcriptional noise, cut-off values were imposed on expression levels of lincRNA genes and their putative orthologs that were used for the lincRNome size estimation. Similarly, a series of cut-off values was applied for the fraction of indels in pairwise genomic alignments (see Methods for details).
A computational pipeline was developed to compare the sets of validated lincRNAs from human and mouse and to identify expressed orthologs by mapping the sequences to the respective counterpart genome and searching the available RNAseq data [28] (Figure 1). We then applied a maximum likelihood (ML) technique to estimate the total number of lincRNA genes in the human and mouse genomes as well as the number of orthologous lincRNA genes (see Online Methods). The following simplifying assumptions were made: 1. A lincRNA sequence in one species has at most one ortholog in the other species (that is lineage-specific duplications are disregarded). 2. The sets of experimentally validated lincRNAs are random samples from complete sets of lincRNAs (lincRNomes) for the corresponding species. 3. The experimentally validated lincRNA sets for human and mouse are uncorrelated with each other.
Let Lh and Lm be the sizes of the experimentally validated sets of lincRNAs for human and mouse, respectively. Also let Kh be the number of confirmed human lincRNAs that have an expressed orthologous sequence in mouse and Km be the corresponding number of mouse lincRNAs. Finally, Kb is the number of confirmed, expressed human lincRNAs whose orthologs in mouse are also confirmed lincRNAs. If the orthology relations between the human and mouse lincRNAs are strictly one-to-one, the number of confirmed mouse lincRNAs for which the human ortholog is also a confirmed lincRNA should be Kb as well. This is indeed the case in practice, with a few exceptions.
Given assumption (1), the lincRNAs can be partitioned into three pools: i) those present in both species, pool size Nb, ii) unique to human, Nh-Nb, and ii) unique to mouse, Nm-Nb; here Nh and Nm are the total sizes of the complete human and mouse lincRNomes, respectively. Assumption (2) allows us to compute the probability of observing a particular set of Kh, Km and Kb simply by counting the number of possible samples of Lh and Lm lincRNAs drawn at random from the respective pools of Nh and Nm that result in the given set of Kh, Km and Kb values: Maximizing the probability P in Eq. (1) with respect to Nh, Nm and Nb, we obtain (see Methods for details): To assess the robustness of the estimates, ranges of open reading frame size thresholds used to eliminate putative protein-coding genes and RPKM (reads per kilobase of exon model per million mapped reads [29]) thresholds used to gauge the expression level were employed (Tables 1 and 2). The ML estimates converged at approximately 50,000 lincRNAs encoded in the human genome and approximately 40,000 lincRNAs encoded in the mouse genome (Table 1 and Figure 2). These are conservative estimates given the use of strict thresholds on predicted open reading frame size and expression level (Table 1), so the actual numbers of lincRNAs are expected to be even greater.
Approximately two-thirds of the lincRNA genes were estimated to share orthologous relationships ( Figure 2 and Table 1). The subsets of lincRNAs with the increasing expression levels were found to be smaller and slightly but consistently more conserved (Table 2), a result that is compatible with our previous observation of positive correlation between sequence conservation and expression level among lincRNAs [23].
We next used the length distributions of human and mouse lincRNAs in the validated sets to estimate the total lengths of the lincRNomes and the fraction of the genome occupied by the lincRNA-encoding sequences, once again under the assumption that the validated sets are representative of the entire lincRNomes. Strikingly, the fraction of the human and mouse euchromatic genome sequence dedicated to encoding lincRNAs was found to be more than twofold greater than the fraction allotted to proteincoding sequences and greater even than the total fraction encoding mRNAs (including untranslated regions) ( Table 3).
Discussion
The relatively poor sequence conservation and often low expression of lincRNAs hamper robust estimation of the size of the lincRNome from expression data alone and render comparative-genomic estimation an essential complementary approach. Strikingly, the estimates obtained here by combining comparative genomic and expression analysis suggest that the mammalian lincRNome is at least twice the size of the proteome [30,31]. Given
Author Summary
Genome analysis of humans and other mammals reveals a surprisingly small number of protein-coding genes, only slightly over 20,000 (although the diversity of actual proteins is substantially augmented by alternative transcription and alternative splicing). Recent analysis of the mammalian genomes and transcriptomes, in particular, using the RNAseq technology, shows that, in addition to protein-coding genes, mammalian genomes encode many long non-coding RNAs. For some of these transcripts, various regulatory functions have been demonstrated, but on the whole the repertoire of long non-coding RNAs remains poorly characterized. We compared the identified long intergenic non-coding (linc)RNAs from human and mouse, and employed a specially developed statistical technique to estimate the size and evolutionary conservation of the human and mouse lincRNomes. The estimates show that there are at least twice as many human and mouse lincRNAs than there are protein-coding genes. Moreover, about two third of the lincRNA genes appear to be conserved between human and mouse, implying thousands of conserved but still uncharacterized functions.
that intron-encoded long-non-coding RNAs and non-coding RNAs encoded in complementary strands of protein-coding genes (long antisense RNAs) [32] are disregarded in these estimates, the total set of lncRNAs and the fraction of the genome dedicated to the lincRNA genes are likely to exceed the respective values for protein-coding genes several-fold.
In order to assess the reliability and robustness of the model with respect to parameters, we produced series of estimates of the total size of the human and mouse lincRNomes and their conserved subset with varying thresholds on expression level, extent of sequence similarity and the maximum allowed open reading frame size. Nevertheless, it is impossible to rule out some sources of bias that might have affected the estimates. For example, some orthologous lincRNA genes might remain undetected because they were not included in the UCSC genome alignments due to high divergence or synteny breaks in (for example, inversions or translocations). Such under-detection of orthologs could cause an underestimate of evolutionary conserved lincRNA genes although it has been reported that the of breakpoints is not large (,250) for the human/mouse genomic comparison [33], so this type of bias is likely to be negligible. Another, potentially more serious source of bias could be a correlation between two lists of lincRNA genes which again would result in biased estimates of evolutionary conserved lincRNA genes. However, because the human and mouse lincRNA sets were obtained using quite different approaches [12,20,23,27], there is no reason to expect that any strong correlation between the two lists would be caused by the employed experimental and/or computational procedures. An under-estimate of the number of orthologous lincRNAs as well as the total size of the mouse lincRNome also might be caused by smaller RNAseq dataset for mouse (10 tissue/cell types, see Methods for details) compared to human (16 tissue/cell types). This difference could explain the systematically smaller predicted numbers of mouse lincRNA genes (Tables 1 and 2). More generally, given that expression of a large fraction of lincRNAs appears to be tissue-specific, the availability of sufficient data for relatively small numbers of tissue/cell types could cause substantial underestimate of the size of both lincRNomes and their conserved fraction. Thus, the estimates obtained here should be regarded as highly conservative, essentially low bounds the lincRNome size and the set of orthologous lincRNA genes.
Some of the transcripts identified as lincRNAs potentially might represent fragments generated from long (alternative) 59UTRs or 39UTRs of protein-coding genes. Such transcripts could results from utilization of alternative poly(A) addition signals and/or could represent alternative splice forms separated by long introns [3,18,19,34]. If many purported lincRNAs actually are fragments of protein-coding genes, one would expect a strong correlation to exist between the expression of lincRNAs and neighboring protein-coding genes. Cabili and co-workers analyzed this correlation for the set of validated human lincRNA genes [27]. Their analysis focused on those protein-coding genes that had a lincRNA neighbor on one side and a coding neighbor on the other side, and used a paired test to compare the correlation between each protein-coding gene and its lincRNA neighbor with that between the same protein-coding gene and its protein-coding gene neighbor. This comparison showed a weak opposite trend, namely that expression of pairs of coding gene neighbors was, on average, slightly but significantly more strongly correlated than the expression of neighboring lincRNA/protein-coding gene pairs. The results of this analysis appear to be best compatible with the hypothesis that any co-expression between lincRNAs and their protein-coding neighbors results from proximal transcriptional activity in the surrounding open chromatin [35]. These findings effectively rule out the possibility that the majority of lincRNAs are fragments of neighboring protein-coding genes although there are anecdotal observations that 39UTR-derived RNAs can function not only in cis to regulate protein expression but also intrinsically and independently in trans, likely as noncoding RNAs [36].
The possibility that some lincRNA genes encode short peptides that are translated, perhaps in a tissue-specific manner, is the subject of an ongoing debate [13,[37][38][39][40]. It is extremely hard to rule out such a role for a fraction of purported lincRNAs as Table 3. The fractions of the human and mouse genomes allotted to protein-coding and lincRNA-coding sequences a . becomes obvious from the long-standing attempts to investigate potential functions of the thousands upstream open reading frames (uORFs) that are present in 59UTR of protein-coding genes in eukaryotes [41][42][43][44]. Although some of the uORFs are translated, the functions of the produced peptides if any remain unclear [45]. Even application of modern high-throughput techniques in simple eukaryotic model systems so far have failed to clarify this issue. For example, analysis of 1048 uORFs in yeast genes has supported translation of 153 uORFs [46]. Furthermore, numerous uORF translation start sites were found at non-AUG codons, the frequency of these events was even higher than for uAUG codons even though the frequency of non-AUG starting codons is extremely low for protein-coding genes [46]. Another intriguing recent discovery is the potential presence, in the yeast genome, of hundreds of transiently expressed 'proto-genes' that are suspected to reflect the process of de novo gene birth [40]. However, the functionality of these peptides remains an open question. Establishing functionality of short ORFs in mammalian genomes is an even more difficult task. For example, analysis of translation in mouse embryonic stem cells revealed thousands of currently unannotated translation products. These include amino-terminal extensions and truncations and uORFs with regulatory potential, initiated at both AUG and non-AUG codons, whose translation changes after differentiation [47]. However, contrary to these emerging indications of abundant production of short peptides, a recent genome-wide study has reported very limited translation of lincRNAs in two human cell lines [48]. In general, at present it appears virtually impossible to annotate an RNA unequivocally as protein-coding or noncoding, with overlapping protein-coding and noncoding transcripts further confounding the issue. Indeed, it has been suggested that because some transcripts can function both intrinsically at the RNA level and to encode proteins, the very dichotomy between mRNAs and ncRNAs is false [38].
Taking all these problems into account, here we adopted a simple, conservative approach by excluding from the analysis lincRNAs containing relatively long ORFs, under a series of ORF length thresholds. However, it should be noted that human and mouse lincRNAs used in this study had been previously filtered for the presence of evolutionary conserved ORFs and the presence of protein domains, and the most questionable transcripts were removed at this stage [12,20,23,27]. For example, 2305 human transcripts were excluded from the stringent human lincRNA set [27] under the coding potential criteria (the presence of a Pfam domain, a positive PhyloCSF score, or previously annotated as pseudogenes). The majority of these discarded transcripts (1533) were previously annotated as pseudogenes [27]. Similar to the stringent set of lincRNAs, these transcripts are expressed at lower and more tissue-specific patterns than bona fide protein-coding genes, suggesting that these effectively are non-coding transcripts. Nevertheless, Cabili and co-workers employed a conservative approach and excluded them from the stringent lincRNA set [27].
Questions about functional roles of lincRNAs and the fraction of the lincRNAs that are functional loom large. For a long time, the prevailing view appeared to be that, apart from a few molecular fossils such as rRNA, tRNA and snRNAs, RNAs did not play an important role in extant cells. More recently, the opposite position has become popular, namely that (almost) every detectable RNA molecule is functional. It has been repeatedly pointed out that this view is likely to be too extreme [49,50]. Although it has been shown that many lincRNA genes are evolutionarily conserved and perform various functions [7,[12][13][14][15][16][17], an unknown fraction of lincRNAs should be expected to result from functionally irrelevant background transcription [19]. In the present work, phylogenetic conservation is the principal support of functional relevance of lincRNAs. Given that neutrally evolving sequences in human and mouse genomes are effectively saturated with mutations and show no significant sequence conservation [51][52][53], expression of non-coding RNAs at orthologous genomic regions in human and mouse should be construed as strong evidence of functionality. It should be noted, however, that sequence conservation gives the low bound for the number of functional lincRNAs, and the lack of conservation is not a reliable indication of lack of function. First, the possibility exists that orthologous genes diverge to the point of being undetectable by sequence comparison, e.g. because short conserved, functionally important stretches are interspersed with longer non-conserved regions, as is the case in Xist, H19, and similar lincRNAs [54,55] [20].
The results of this work predict that, despite the fact that on average sequence conservation between orthologous lincRNAs is much lower than the conservation between protein-coding genes [12,23], 60 to 70% of the lincRNAs appear to share orthologous relationship between human and mouse, which is only slightly lower than the fraction of protein-coding genes with orthologs, approximately 80% [51]. These findings imply that, even if many of the species-specific lincRNAs are non-functional, mammalian lincRNAs perform thousands of evolutionarily conserved functional roles most of which remain to be identified.
The human and mouse validated lincRNA sets
As the human lincRNA data set, the 'stringent set' of 4662 lincRNAs, which is a subset of the over 8000 human lincRNAs described in a recent comprehensive study [27], was used. The validated set of mouse lincRNA genes was constructed by merging our previously published set of 2390 lincRNA transcripts with the set of 3051 transcripts produced by Ponting and coworkers [12]. After the merge, a unique list of 4989 GenBank transcript IDs was generated, coordinates of the newest mouse assembly, mm9, were downloaded in BED format from the UCSC Table Browser [56], and entries shorter than 200 nt were discarded. Overlapping chromosomal coordinates were merged using the mergeBed utility from BEDtools package [57], with the command line option -s (''force strandedness'', i.e. merge overlapping features only if they are on the same strand), and unique IDs were assigned to the resulting 4156 mouse lincRNA clusters. (format: mlclust_N where mlclust stands for mouse lincRNA cluster, and N is a unique integer number; see Supporting Table S1).
Expression of lincRNAs
Expression of the lincRNAs was assessed by analysis of the available RNAseq data. For human, the run files of the Illumina Human Body Map 2.0 project for adipose, adrenal, brain, breast, colon, heart, kidney, liver, lung, lymph node, ovary, prostate, skeletal muscle, testis, thyroid, white blood cells, were downloaded from The NCBI Sequence Read Archive (SRA, [URL]. nlm.nih.gov/Traces/sra; Study ERP000546; runs ERR030888 to ERR030903). For mouse, RNAseq data of the ENCODE project [58] for tissues: bone marrow, cerebellum, cortex, ES-Bruce4, heart, kidney, liver, lung, mouse embryonic fibroblast cells (MEF) and spleen, were downloaded from the UCSC . The reads were aligned with the cognate genomic sequences using TopHat [59].
The TopHat-generated alignments were analyzed using an ad hoc Python script that accepts alignments and genomic coordinates in SAM and BED formats, respectively, and uses the HTSeq Python package ( [URL]/ HTSeq) to calculate the number of aligned reads (''counts''). The RPKM (i.e. reads per kilobase of exon model per million mapped reads [29]) values were calculated from the counts values. Because we were interested to determine whether particular regions are expressed in any of the analyzed tissues, the maximum value among all tissues was assigned as the expression level of lincRNA genes and putative orthologous lincRNA genes.
Identification of open reading frames (ORFs)
An ORF was defined as a continuous stretch of codons starting from the ATG codon or beginning of the cDNA (to take into account potentially truncated cDNAs) and ending with a stop codon. The ORFs were identified by using the ATG_EVALUA-TOR program [60] combined with the ORF predictor from the GeneBuilder package [61] with relaxed parameters (the program was required to correctly predict 95% of the human and mouse cDNA training sets [61]). Control experiments with independent human and mouse cDNA data sets [61] showed a 94-98% true positive rate depending on the ORF length threshold (90, 120 or 150 nucleotides). However, a high rate of false positives is expected for such relaxed parameters. Analysis of human and mouse introns and UTRs data sets showed false positives rates of 10-20% depending on the threshold [60,61]. For the purpose of the present analysis, false positives in ORF identification represent random removal of lincRNA sequences from the samples resulting in conservative estimates of the total lincRNA number. Thus, we used the ORF cut-off values of 90, 120 or 150 nucleotides to remove putative mRNAs for short proteins separately from the human and mouse sets of lincRNAs.
Comparative genomic analysis of the lincRNA sets
To obtain the subset of human lincRNAs with expressed orthologs in mouse (Kh), human lincRNA gene coordinates of assembly hg19 were converted to mouse mm9 using the liftOver tool of the UCSC Genome Browser [62]. Out of the 4662 human lincRNAs (Lh), for 3529 putative orthologous regions were identified in the mouse genome. These sequences were checked for the evidence of expression in mouse tissues using the RNAseq data. Exon coordinates of putative lincRNAs were obtained by mapping their coordinates onto exons of all known genes of mm9 assembly of UCSC Genome Browser. The sums of exons were then used in expression level calculation to normalize for sequence length. Out of the 3369 putative lincRNAs for which the exon models could be determined, 2872 had expression level greater than zero. Similarly, the subset of mouse lincRNAs with expressed putative orthologs in human (Km) was found by converting the coordinates of initial 4156 mouse lincRNAs (Lm) from mm9 to hg19 and searching for the evidence of expression in human tissues. The exon models could be determined for 3656 of the 3677 putative lincRNAs, out of which 3157 had expression level greater than zero.
The subset of orthologous lincRNAs (Kb) was obtained by selecting those lincRNAs whose putative orthologs in another species overlap with the validated lincRNAs of that species. That is, we searched for the overlap of putative orthologs of human lincRNAs (in hg19 coordinates) with the mouse lincRNAs (in mm9 coordinates, minimal overlap 100 nucleotides). The overlap was determined using intersectBed from BEDtools package with the command line option -s (''force strandedness''). This resulted in 196 pairs of unique human and mouse lincRNAs. Approximate indel values were estimated from the sequence length differences between the lincRNAs and their orthologs, i.e. the following formula was used: where L llincRNA is the total length of lincRNA exons, and L ortholog is the total length of the exons of lincRNA ortholog. Manual examination of orthologous lincRNA alignments and putative orthologs suggested that approximately 5% of the alignments with the largest INDEL values were unreliable. Thus, all lincRNA alignments with INDEL .95% were removed from further analysis. Similarly, a cut-off was imposed on expression level of putative human and mouse orthologs of lincRNA. This cut-off was set at the lowest 5% of the expression levels of the 196 orthologous validated lincRNA genes (Supporting Table S1). All putative orthologs of lincRNA genes with lower expression values were discarded under the premise that these low values could represent experimental noise, i.e. the top 95% of the expression values EXP95% was used for all analyses (Table 1 and Supporting Table S1). In addition, EXP90%, 80%, 70%, 60%, 50%, 40%, 30%, 20%, 10% were calculated to compare subsets of lincRNAs expressed at different levels ( Table 2). We also used different sets of expression/indel filters combined with the 5 input parameters (see Results) in different experiments (Tables 1 and 2); no substantial differences between results were found (see Discussion for details). For calculating the 5 input parameters (see Results), all the collected information was stored in an SQLite database, and after applying ORF, indel and expression thresholds, final data sets were assembled (Tables 1, 2 and Supporting Table S1).
Maximum likelihood estimates
Using the experimentally validated sets of human and mouse lincRNAs and the assumptions described in the main text the probability of observing a particular set of Kh, Km and Kb for the given values of Lh and Lm is given by equation (1) in the main text. Using the Sterling's approximation for the factorial, we obtain the system of nonlinear equations for the sizes Nh and Nm of the pools and their overlap Nb that maximize the likelihood P in Eq. (1) Solving the system (3)(4)(5) for Nh, Nm and Nb we obtain Equation (2) (see main text). The confidence region around the maximum likelihood estimate Eq. (5) is an ellipsoid in the {Nh,Nm,Nb} space. The directions of its axes are given by the eigenvectors of the Jacobian matrix J of second derivatives of log P and the magnitudes of the ellipsoid's axes are given by the inverse square roots of the negatives of the eigenvalues. Computing the second derivatives of log P and evaluating them at the maximum likelihood point, we obtain We found that the confidence ellipsoid is highly elongated, and therefore the estimates for the pool sizes are strongly correlated with each other. The analytically estimated 95% confidence intervals are shown in Table 1.
In addition, a bootstrap analysis of the lincRNA numbers was performed. For this purpose, the initial sets of human and mouse lincRNAs were randomly resampled 1000 times and the calculation of the final numbers was performed using 95% indel and expression (RPKM) levels, and all ORF thresholds. The results of bootstrap analysis are given in the Supporting Table S1. The 95% confidence intervals estimated using the boostrapping procedure (Supporting Table S1) were smaller than the analytically obtained 95% confidence intervals (Table 1), thus we used the latter as conservative estimates of the 95% confidence intervals.
Supporting Information
Table S1 Comprehensive information on the human and mouse lincRNA sets. (XLS)
==
Domain: Biology Medicine Computer Science
|
[{"added": "2017-05-04T20:37:47.516Z", "created": "2013-02-01T00:00:00.000", "id": "5420817", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Computer Science"], "oa_license": "CC0", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1002917&type=printable", "pdf_hash": "71159f97bc7907306b1386982fdc73cea7cd12e1", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:999", "s2fieldsofstudy": ["Biology"], "sha1": "71159f97bc7907306b1386982fdc73cea7cd12e1", "year": 2013}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
PolyprOnline: polyproline helix II and secondary structure assignment database
The polyproline helix type II (PPII) is a regular protein secondary structure with remarkable features. Many studies have highlighted different crucial biological roles supported by this local conformation, e.g. in the interactions between biological macromolecules. Although PPII is less frequently present than regular secondary structures such as canonical alpha helices and beta strands, it corresponds to 3–10% of residues. Up to now, PPII is not assigned by most popular assignment tools, and therefore, remains insufficiently studied. PolyprOnline database is, therefore, dedicated to PPII structure assignment and analysis to facilitate the study of PPII structure and functional roles. This database is freely accessible from www.dsimb.inserm.fr/dsimb_tools/polyproline.
Introduction
Fifty percent of local protein conformations are constituted of the two regular secondary structure, i.e. a helices and b sheets, while the remaining protein structure is essentially constituted of turns that can overlap the two previous local conformation, and coil (1). Regular secondary structures are fundamental descriptor for the analysis and the understanding of the structure and function of proteins at a molecular level. As such, they are automatically used to visualize the protein 3D structures with popular software like PyMOL (2), VMD (3) or Chimera (4). Thus, the secondary structures assignment is an essential step for studying protein architecture, folding and for the prediction of 3D protein structure. Besides a helices and b sheets, a number of other regular secondary structures are often ignored, despite their importance in biological processes (5). Among other regular secondary structures the polyproline II helix (PPII) is of significant interest. PPII conformation was primarily identified in the 1950s in collagen helix by Pauling and Corey (6), and in structures containing many repeating proline amino acids (7). It was not until the end of the 1990s that this conformation has been demonstrated to occur frequently in globular protein (8), with a very high conservation ratio of 80-100% in proteins families sharing 20% sequence identity or more, a ratio close to the conservation found for a helices and b strands (9).
Page 1 of 8
This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( [URL]/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
(page number not for citation purposes) Depending on the tools used for the assignment of secondary structures, their frequency varies in the range of 3-10% of all conformations with a common core of more than 1.6% assignment shared by all tools (10). Other studies have shown similar frequency, Adzhubei and co-workers in a recent review (11) estimated about 2% of residues in Protein Databank to be in PPII-helices of length 3 and more residues. For historical reasons, this conformation had been called 'polyproline helix', although most PPIIs comprised non-proline residues and some even contain no proline at all (8).
In term of local structure conformation, Polyproline II is a left-handed helical conformation with average dihedral angle values of U ¼ À75 and W ¼ þ145 . Unlike classical regular secondary structures, PPIIs are not usually associated with conventional stabilizing internal hydrogen bonds due to this extremely extended conformation. PPII is a far more extended helix than classical a-helix (5.4 Å /turn, 3.6 residues per turn) and has a helical pitch of 9.3 Å /turn and 3 residues per turn. Thanks to this over extended conformation and high solvent exposure, residues in PPII may lead to potential interactions with other molecular partners. Thus, it was suggested that they might have an important functional role, particularly in protein-protein or protein-nucleic acid interactions and recognition (12,13). Regrettably, PPIIs are still insufficiently studied. In fact, PPII assignment is not done with the most common method of secondary structure assignment such as Dictionary of Protein Secondary Structure (DSSP; 14) and STRIDE (15), and therefore, newly solved protein structures are not assigned with PPII in Protein DataBank (16). Here we introduce a new assignment method and a dedicated webserver for PPII.
Aim and overview of database
The PolyprOnline database ( [URL]/ dsimb_tools/polyproline) contains secondary structure assignments on a large subset of the Protein Databank. It also allows to dynamically handle any new user submitted structures. Unlike other databases established for protein secondary structure analysis, PolyprOnline particularly focalize on PPII, an assignment that is rarely documented in experimentally solved structures as well as in services and tools dedicated to the analysis of protein structures. For instance, 2struc ( [URL]/) assigns protein in three secondary states using six different algorithms (17), but none of them address PPII assignment. More general tools such as PDBsum (18) give assignment by one method, PROMOTIF (19) in this case, with no details about PPII. As previously mentioned, this assignment is especially important since this conformation is the third most abundant regular secondary structure just behind a-helix and b-strand, and it is also involved in various function related to molecular interactions such as proteinprotein and protein-nucleic-acid binding. However assignment using different tools show discrepancy thus our database provide assignments with the four main methods developed so far (10).
Results and features
The data flow and processing step performed by the system are summarized in Figure 1.
Interface
Through the main interface, two types of search are possible. Both searches are detailed in text of Figure 1: simple search (analysis of one or more protein structure) and advanced search based on specific criteria to perform more complex queries. One of the most interesting features is the ability to perform secondary structure pattern query. This search is useful to look for a fragment of specified conformation contained in protein structures using a simple regular expression pattern. Pattern search uses the classical rules for regular expressions. It is possible to use conformation code letters (e.g. HHHH-PPEEE), and introduce wildcard (e.g. HHH**PP*-). It is also possible to specify the minimal and the maximal conformation length (e.g. PPPX{1,8}PP).
Outputs
The PolyprOnline webserver offers the following outputs:
A table of sortable results
Results are displayed in a table that can be sorted accordingly to the values in different columns ( Figure 2). Results in the table can also be directly downloaded in text format. All proteins in the table are identifiable by PDB code, title, size, resolution and PPII content. You can also download the assignment of each protein in classical fasta format.
Individual protein data and analysis
The PolyprOnline web server provides access to different assignment methods and allows visualization of both regular secondary structure and PPII helix ( Figure 3). We have recently underlined the discrepancies between the three different secondary structure methods able to assign PPIIs, and proposed a novel PPII assignment using the de facto standard DSSP assignment method (10,14). To better visualize the secondary structure and PPII assignments given by PROSS (21), SEGNO (22), XTLSSTR (23) and our DSSP-PPII (10,14), they are all displayed at the bottom of sequence One letter code is used to represent specific conformation. Letters are coloured accordingly to more general class of secondary structure (e.g. helix residue in red, strand in green, PPII helix in blue non-regular secondary structure in grey, coil being in dark grey colour) for a fast visualization of overall local structures. All data from protein structure analysed can be downloaded.
Ramachandran plots give the distribution of u and w torsion angles for each assignment method. The most frequent areas for a-helix and b-sheet are shown in the background of the plot (represented by a colour scale). Statistics about areas were derived from our previous study (10). Residues assigned as PPIIs are represented as white points. The image is mouse sensitive and gives additional information on residue number, nature and u and w angle values of assigned as PPII. Indeed assignments provided by the various tools can be quite different between them. Ramachandran plot lets to visually inspect u and w angle PPII value distributions and help the user to apprehend the relevance of each assignment.
Visualization and manipulation of three dimensional protein structures is allowed thanks to a JMol applet (24). It displays the assignment of secondary structures by all of the four methods and details about positions of secondary structures with a particular emphasis on PPII. This visualization can also be useful to observe difference between assignments directly in protein structure.
Protein structures dataset
A subset of the experimental protein structures extracted from the PDB was selected based on the resolution methods (RX), quality of structures (resolution lower than 3.0 Å and R-factor lower than 1.0) limited redundancy (proteins share no more than 90% of identity between each others) using webserver PISCES (20). The full list of selected structures comprised 24 761 protein chains and is available on database. The list is regularly updated. Figure 1. Data flow in PolyprOnline system. Access to the system can be done in two ways: through 'Simple query' for the analysis of one or more protein structures from their PDB code and through 'Advanced query' for performing more complex queries using different criteria such as resolution (Å ), protein length, minimal and maximal number or percentage of residues in PPII conformation assigned by a particular tool. The last type of advanced query allows local structure search on specific positions using secondary structure patterns. It is also possible to dynamically upload and process a PDB file if it is absent of the database. The query is then processed to be interpreted by our Database Management System. In the case where a PDB structure is not found in the database, a PDB file can be downloaded from the Protein Databank website and dynamically processed by the system. PolyprOnline webserver offers the following outputs to display results: Summary of all protein identified by PDB code, title, size, resolution and PPII content, printed in a sortable table according to the values in different columns ( Figure 2). From this table, individual protein data analysis can be accessed individually (Figure 3).
Assignment of PPII and other secondary structures
Currently, there is a limited number of tools for assigning PPII number. The tools available today are XTLSSTR, PROSS (version September 2004) and SEGNO (version 3.1). We have added our PPII DSSP-based program DSSP (CMBI version 2000) developed in our laboratory to this list (10). As we have previously explained, the use of multiple tools is necessary because it has been shown that PPII assignments using several methods yielded different results (10).
Secondary structures assigned by PROSS (21) are as follow: a helix (H), b turn (T), b strand (E), PPII (P), and coil (C). Assignments are based exclusively on U and W dihedral angles.
The algorithm XTLSSTR (23) uses two angles and three distances to assign secondary structure from coordinates of PDB files. It assign secondary structures: a helix (H and h), 3 10 helix (G and g), hydrogen bonded b turn (T), nonhydrogen-bonded b turn (N), Extended b strand (E and e) and PPII (P and p) SEGNO (22) uses also the U and W dihedral angles coupled with other angles to assign the secondary structures. It assign a helix (H), b-strand (E and e), isolated b-strand (B and b) 3 10 helix (G and g), p-helix (I), coil (O, coded as '-' in this database) and PPII (P and p). Sequence and analysis of secondary structures using four different protein secondary structure assignment methods are printed on a 1D alignment. One letter code is used to represent a specific conformation. Letters are coloured accordingly to more general class of secondary structure (i.e. helix residue in red, strand in green, PII helix in blue and non-regular secondary structure in grey). (B) Ramachandran plots give the distribution of u and w torsion angles of PPII amino acids for each method. The most frequent areas for a-helix and b-sheet are shown in the background of the plot (represented by a colour scale). Statistics about areas were derived from our previous study. Residues assigned as PPIIs are represented as white points. (C) Full 3D structure visualization and animation using a JMol applet of different assignment can be dynamically displayed (Ca trace only, cartoon). Local conformations are coloured with the same colour scheme as used for the 1D alignment in (A; i.e. helix residue in red, strand in green, PII helix in blue and non-regular secondary structure in grey).
DSSP-PPII is a new method for PPII assignment recently developed in our laboratory (10). It is based on the most popular secondary assignment tools: DSSP (14). DSSP assignment is based on the identification of precise hydrogen bond patterns corresponding to regular secondary structures. Assignment strategy of PPII is based on simple set of basic rules to have the highest agreement with PROSS, SEGNO and XTLSSTR methods. PPII are assigned solely in the coil region for at least two consecutive amino acids in coil with U ¼ À75 6 e and W ¼ þ145 6 e with e ¼ 29 . Basic assignment of secondary structure in DSSP defines eight types of secondary structures: a helix (H), extended b strand in parallel and or anti-parallel b-sheet conformation (E), isolated b-strand (B), 310 helix (G), Pi helix (I), bend (S) and coil (O, coded as '-' in this database). This is the basic assignment to which helix PPII (P) has been added.
Web interface and Database
Database management server used by our system is MySQL. The PolyprOnline web interface has been written mainly in PHP, Perl, R and Javascript programming languages.
Conclusion and interesting case study
To better understand structure/function and structure/ architecture relationships, the advanced search interface of PolyprOnline can be used to find proteins with a high content of PPII. Thus a query launched on the basis of PPII frequency or containing long PPII helix can highlight different properties and peculiarities. It can be noted that proteins with the highest content of PPII have an over-frequency of functions related to interaction mechanisms and/or binding, which is consistent with observations in (11). For example, Figure 4 provided some examples involved in various function such as cell adhesion (B), self binding (C) or binding to cyclin-dependent kinases (A), neurotoxicity, an effect that involved blockade of acetylcholine receptors (D) and anti-freeze effect where solvent interaction is fundamental (E). With more than 72% of residues in PPII conformation, this anti-freeze protein contains the highest percentage of PPII of our database. It can also be noted, in these examples, that the organization of these PPII present characteristics of this regular conformation: rather isolated and exposed prolines for cyclin-dependant kinase regulation subunit (A), and the characteristics of other regular secondary structures: (i) similarities with a helix motifs such as PPII-beta-beta motif in Thrombospondin (B) and Atratoxin of cobra venom (D), (ii) and analogy with both alpha and beta motif such in GTP-binding protein obg (C) and snow flea anti-freeze protein (E) where PPII arrangements appear as a six anti-parallel PPII helices bundle. All theses PPII have in common a broad exposure to the solvent as it has already been highlighted in previous studies (11). Please note that these proteins are extreme cases in term of PPII content and are provided for illustrative purposes. The largest continuous PPII helix, of 13 residues long, is found in a protein Lyase (2VK8A; 31). This quick analysis highlights the utility of PolyprOnline database for PPII study.
==
Domain: Biology Medicine Computer Science
|
[{"added": "2016-05-12T22:15:10.714Z", "created": "2014-11-06T00:00:00.000", "id": "17521069", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Computer Science"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://academic.oup.com/database/article-pdf/doi/10.1093/database/bau102/8248386/bau102.pdf", "pdf_hash": "e9ad6ef9e1645cede4af1672b76af373059b147a", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:9369", "s2fieldsofstudy": ["Biology", "Computer Science"], "sha1": "e9ad6ef9e1645cede4af1672b76af373059b147a", "year": 2014}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
riboviz 2: a flexible and robust ribosome profiling data analysis and visualization workflow
Abstract Motivation Ribosome profiling, or Ribo-seq, is the state-of-the-art method for quantifying protein synthesis in living cells. Computational analysis of Ribo-seq data remains challenging due to the complexity of the procedure, as well as variations introduced for specific organisms or specialized analyses. Results We present riboviz 2, an updated riboviz package, for the comprehensive transcript-centric analysis and visualization of Ribo-seq data. riboviz 2 includes an analysis workflow built on the Nextflow workflow management system for end-to-end processing of Ribo-seq data. riboviz 2 has been extensively tested on diverse species and library preparation strategies, including multiplexed samples. riboviz 2 is flexible and uses open, documented file formats, allowing users to integrate new analyses with the pipeline. Availability and implementation riboviz 2 is freely available at github.com/riboviz/riboviz.
Introduction
Ribo-seq quantifies the 'translatome' of actively translated RNAs in cells (Ingolia et al., 2009). Ribo-seq combines high-throughput sequencing with nuclease footprinting of ribosomes to identify the location of ribosomes across the transcriptome at codon-level resolution. Ribo-seq is often combined with RNA-seq to quantify post-transcriptional regulation and also enables quantitative mechanistic insight into the movement of ribosomes along RNA. Specialized pipelines are needed for Ribo-seq data, covering preprocessing, read mapping, gene-specific and codon-specific quantification and other downstream analyses (Li et al., 2020). We previously developed riboviz as an analysis and visualization framework for Ribo-seq data (Carja et al., 2017). Here, we present a significantly expanded and reworked version: riboviz 2.
Materials and methods
The riboviz 2 pipeline is implemented via Nextflow (Di Tommaso et al., 2017;Jackson et al., 2021) to process multiple samples from an experiment in a single command-line call. All run-specific parameters are specified by the user in a single YAML-format configuration file, documented at github.com/riboviz/riboviz. Users may also utilize a graphical user interface (GUI) to aid in the generation of this configuration file. The configuration file facilitates reproducible and transparent analyses, and allows the pipeline to run on various computing systems. riboviz 2 invokes both publicly available tools [e.g. cutadapt (Martin, 2011), HISAT2 (Kim et al., 2015), UMItools (Smith et al., 2017)], and custom Python and R scripts for data parsing and visualization.
The riboviz 2 workflow (Fig. 1A) starts with preprocessing of Ribo-seq data in FASTQ format, including adapter trimming and removing reads mapping to user-supplied contaminant sequences such as rRNA. Following preprocessing, the remaining reads are aligned to the relevant sequences as defined by user-provided FASTA and GFF3 files. Due to differences in ribosome structure and Ribo-seq protocols, the appropriate strategy for assigning reads to the codon at the ribosomal active site varies between species, e.g. Ribo-seq reads from eukaryotes and bacteria are mapped relative to
2358
This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( [URL]/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Applications Note the 5 0 -and 3 0 -end, respectively (Mohammad et al., 2019). riboviz 2 allows the user to map relative to either end of the read by specifying the displacement separately for each desired read length. riboviz 2 provides outputs typical to Ribo-seq in standard file formats, including the aligned reads in BAM format and number of read counts by read length in text format. We provide a ribogrid' intermediate data file in H5 format that contains one aligned read count matrix per transcript, organized by both 5 0 position and read length. These counts are a sufficient statistic for most downstream analyses, in that the only information used from the raw alignments is the count by both position and length. Documentation and accessor functions for this ribogrid H5 file format enable the future addition of custom analysis functions. riboviz 2 automatically outputs visualizations commonly used in publications describing Ribo-seq experiments, both for quality control to confirm that the experiment successfully recovered ribosome footprints, and as a valuable tool for analysis. These include read length distributions, proportion of reads mapping to the primary, þ1, and þ2 reading frames per gene, and metagene plots showing three-nucleotide periodicity. riboviz 2 directly visualizes the aligned read count matrix, with a heatmap of the footprint counts arranged by both 5 0 position and read length (Fig. 1B). These 'ribogrid' plots are a rich way to read out mechanistic details of Ribo-seq data such as read frame (Lareau et al., 2014). For each processed sample, the various plots output by riboviz 2 are combined into static HTML file as an overall visual summary. riboviz 2 can compare codonspecific ribosome densities to features or measures expected to correlate with elongation rates, such as tRNA gene copy numbers, and compare gene-specific features (such as codon usage metrics) to gene-level quantifications of ribosome density. In addition to these static visualizations on a per sample basis, riboviz 2 allows users to interactively visualize all of their data in an R/Shiny based web application (Fig. 2). The Shiny app is particularly useful for comparing results across control and treatment samples. Users can adjust interactive versions of the static plots already provided as well as view gene-level statistics such as read distribution along a specific gene.
New features and advantages
Flexibility across organisms riboviz 2 can be used on any organism for which a transcriptome FASTA and GFF3 file can be constructed, making it a valuable tool for users studying either model or non-model organisms. This is an advantage for riboviz 2 compared to other GUI or command-line based tools that are limited to a set of organisms or require sequence annotations to be downloaded from a specific database (Liu et al., 2020;Perkins et al., 2019;Verbruggen et al., 2019;Wang et al., 2018). The user is responsible for supplying a FASTA file appropriate to their biological question, e.g. using a published annotation to define spliced transcripts including untranslated regions, or a 'padded ORFeome' that contains fixed-width extensions to a set of open reading frames (ORFs) of interest. The user must also supply a file in GFF3 format that specifies the positions of ORFs within the transcripts. Example configuration files to run riboviz 2 on diverse datasets that span the major domains of life (Archaea, Bacteria and Eukarya), with matched transcriptome and contaminant files, are shared at (github.com/riboviz/example-datasets). These files may be used to reproduce analyses, or adapted to analyze new datasets.
Flexible end-to-end data processing workflow Another advantage of riboviz 2 is that it provides a comprehensive workflow starting from raw reads and ending with publication-quality Fig. 1. riboviz pipeline and data structures. (A) riboviz takes in user-provided sequencing, transcript annotation, and configuration files, processes the datasets and generates two major output-transcriptome-specific BAM file and a ribogrid file. These outputs are used to generate sample-specific analyses and summaries, which can be visualized as both static figures and in an interactive R/Shiny application. (B) Structure of the ribogrid file format. Ribogrid is a complete representation of transcript-specific ribosome-footprint data in an H5 file format. Each row indicates reads of a particular length and each column indicates the position of the 5 0 -end of a footprint. *Optional Fig. 2. The riboviz 2 data visualization application is powered by R and Shiny and allows users to view various aspects of their data in an interactive manner in a web browser. Shown is an example visualization using data from Guydosh and Green (2014). This dataset is also part of our example datasets repository ( [URL]. Many pipelines require input that has either already been preprocessed or aligned [see Li et al. (2020) for a summary of the functionality of other pipelines]. Instead, riboviz 2 provides comprehensive data preprocessing (e.g. adapter trimming) and read alignment by interfacing to cutadapt (Martin, 2011) and HISAT2 (Kim et al., 2015). riboviz 2 is also flexible to variations in library preparation. To the best of our knowledge, riboviz 2 is the only Ribo-seq pipeline which is prebuilt to handle multiplexed libraries or unique molecular identifiers. Following read alignment, riboviz 2 uniquely invokes an (optional) script to trim non-templated 5 0 mismatches added by some viral reverse transcriptases (Wulf et al., 2019), which otherwise leads to inaccurate quantification of read frame. riboviz 2 requires no knowledge of Python or R to take advantage of the riboviz 2 functionality, unlike many other tools (Backman and Girke, 2016;Lauria et al., 2018). As riboviz 2 is implemented as a Nextflow workflow going from raw data to visualization while requiring only a single configuration YAML file, reproducing an analysis does not require independently running various tools.
Flexible and documented data outputs
A major goal of a Ribo-seq analysis pipeline is to enable further downstream analyses of Ribo-seq data, such as differential expression analysis and identification of ribosome pausing sites. riboviz 2 consolidates the data into outputs that are suitable for downstream analysis, such as aligned read count matrices in the ribogrid H5 file. riboviz 2 aggregates raw counts per transcript into a format which can be used as input to tools such as DESeq2 (Love et al., 2014), and provides per-ORF translation values in transcripts per million (TPM).
Overall, riboviz 2 is a flexible, documented and carefully engineered open-source workflow for Ribo-seq analysis and visualization. [DBI 1936046 to P. S., DBI 1936069 to L. F. L.]; the National Institutes of Health [R35 GM124976 and subcontracts from R01 DK056645, R01 DK109714, R01 DK124369 to P. S., R01 GM132104 to L. F. L]; and start-up funds from the Human Genetics Institute of New Jersey at Rutgers University to P. S.\===
Domain: Biology Medicine Computer Science. The above document has 2 sentences that start with 'riboviz 2 is', 2 sentences that end with 'of Ribo-seq data', 2 sentences that end with 'et al., 2018)', 2 paragraphs that start with 'The riboviz 2'. It has approximately 1580 words, 82 sentences, and 15 paragraph(s).
|
[{"added": "2021-05-20T13:25:15.653Z", "created": "2021-05-17T00:00:00.000", "id": "234785713", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Computer Science"], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://academic.oup.com/bioinformatics/article-pdf/38/8/2358/43370125/btac093.pdf", "pdf_hash": "32bacd8fb34f2e24115fa8138a4843f14f47cebb", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:1166", "s2fieldsofstudy": ["Computer Science", "Biology"], "sha1": "de98a0fd9a70753856c155d0ee7842aa1b9c0bd8", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Chromonomer: A Tool Set for Repairing and Enhancing Assembled Genomes Through Integration of Genetic Maps and Conserved Synteny
The pace of the sequencing and computational assembly of novel reference genomes is accelerating. Though DNA sequencing technologies and assembly software tools continue to improve, biological features of genomes such as repetitive sequence as well as molecular artifacts that often accompany sequencing library preparation can lead to fragmented or chimeric assemblies. If left uncorrected, defects like these trammel progress on understanding genome structure and function, or worse, positively mislead this research. Fortunately, integration of additional, independent streams of information, such as a marker-dense genetic map and conserved orthologous gene order from related taxa, can be used to scaffold together unlinked, disordered fragments and to restructure a reference genome where it is incorrectly joined. We present a tool set for automating these processes, one that additionally tracks any changes to the assembly and to the genetic map, and which allows the user to scrutinize these changes with the help of web-based, graphical visualizations. Chromonomer takes a user-defined reference genome, a map of genetic markers, and, optionally, conserved synteny information to construct an improved reference genome of chromosome models: a “chromonome”. We demonstrate Chromonomer’s performance on genome assemblies and genetic maps that have disparate characteristics and levels of quality.
Researchers are generating new reference genomes at an accelerating pace. While it is now straightforward to produce enough sequence information to cover even large genomes many times over, the assembly of a realistic reference genome can still be challenging for both bioinformatic and biological reasons (Church et al. 2011;De La Torre et al. 2014;Nowoshilow et al. 2018;Ghurye and Pop 2019). A high-quality reference genome with minimized gaps and misassemblies, particularly one organized into chromosomesknown as a chromonome (Braasch et al. 2015) is a valuable research tool. Comparative genomics studies that have employed, for example, the analysis of conserved synteny of genes among distantly-related taxonomic groups have led to better understanding of how genes and genomes evolve and function (Naruse 2004;Jaillon et al. 2004;Shah et al. 2012;Lovell et al. 2014;Zhao and Schranz 2019). Likewise, understanding the population dynamics of selection and drift, as described by measures of mutation and linkage, requires chromosome-level stretches of sequence (Hohenlohe et al. 2010;Luikart et al. 2018). Reliably assembled reference genomes have aided exploration of chromosome structural conservation or rearrangement through evolutionary time (Wang et al. 2013;Jay et al. 2018), the effects of transposable element perturbation (Woronik et al. 2019), the fate of duplicated genes following divergence of organismal lineages (Brunet et al. 2006;Kassahn et al. 2009), the mechanisms of long distance regulation of genes (Kleinjan and van Heyningen 2005), and the progression of disease-resistant alleles in populations Sequencing technologies and genome assembly strategies continue to evolve, but it is still not trivial to assemble chromosome level references with highest confidence in biological accuracy for organisms with complex genomes. Since the inception of high-throughput sequencing three major assembly strategies have been employed: short-read-only assemblies, hybrid assemblies that incorporated long reads to join and gap-fill short-read assemblies, and long-read-only assemblies. While contig generation has become very robust, whether it is via the use of a de Bruijn graph in short-read and hybrid assemblies (Compeau et al. 2011), or through the use of polishing algorithms in long-read assemblies (Fu et al. 2019), most obstacles to the generation of a chromonome come from the error models of different scaffolding approaches.
Short-read assemblies rely on incorporation of "mate-pair" sequences to order and orient contigs into scaffolds (Gnerre et al. 2011;Chapman et al. 2011;Luo et al. 2012), but the approach can produce molecular chimeras during library construction or assembly chimeras during scaffolding when the short reads land in repeats. Optical maps (Pendleton et al. 2015;Howe and Wood 2015) and chromosomal conformation capture methods such as Hi-C (Lieberman-Aiden et al. 2009) have been used very effectively for scaffolding and have improved assembly quality metrics like N50 and L50. These long molecular methods are not immune from errors, however, which manifest as indels and fragment length estimation mistakes (Mukherjee et al. 2018), artifactual inversions, and occasional long-range chimeras during integration into an assembly . While all scaffolding methods remain imperfect, independent methods to explore and verify genome organization remain valuable. A genetic map is a multifunctional tool that can also serve this purpose.
Genetic map construction remains relevant for a variety of research goals; for example, comparing a genetic map with a physical genome sequence helps identify gene candidates causal for variant or mutant phenotypes (Meinke et al. 2003;Peichel and Marques 2017), and reveals variation in recombination rate across the genome (Roesti et al. 2013;Dukić et al. 2016). A map can also benefit the assembly of a reference genome by revealing points of erroneous contiguity in an assembly, by binding scaffolds into "linkage groups" that are chromosome models, and by ordering and orienting the scaffolds relative to one another. It is now relatively straightforward and rapid to genotype individuals at thousands of loci by using one of many massively parallel sequencing methods such as Restriction site-associated DNA sequencing (RADseq; (Baird et al. 2008;Davey et al. 2011;Andrews et al. 2016)). Marker-dense maps have the potential to capture a majority of the assembled genome length into linkage groups. More importantly, potentially chimeric scaffolds can be detected where the physical and genetic map relationships of markers on scaffolds conflict, such as in cases where a single scaffold's markers map to more than one linkage group. The efficacy of a genetic map for consolidating an assembly and validating its quality depends on a number of important factors, including the density and distribution of markers, the number of meiotic crossovers represented in the mapping cross progeny, the size distribution of the scaffolds, the granularity of misassembly with respect to the distance between markers, and the genotyping accuracy.
In the end, a genome assembly is a hypothesis that proposes a sequence order while the true order will always remain unknown. A useful tool should be able to automate the flagging of problematic scaffolds, resolve conflicts between the assembly and the genetic map in a rational and efficient way, and integrate additional lines of evidence that support a hypothesis of genomic structure. We present here software we call Chromonomer that corrects, orders, and orients scaffolds by integrating genetic maps and genome assemblies. Chromonomer can create chromosome-level assemblies while providing extensive documentation of how the elements of evidence fit together. To further improve assemblies, the software can integrate conserved gene synteny and raw read depth of coverage, and it provides tools to extract gene annotations from a scaffold-level assembly and translate their locations to a chromosome-level assembly (and vice versa). Earlier, prototype versions of Chromonomer have been used in a number of published genome assembly integrations (e.g., Amores et al. 2014;Fountain et al. 2016;Small et al. 2016;Kim et al. 2019;Moran et al. 2019). Here we illustrate the performance of Chromonomer with three qualitatively different teleost genome test cases representing the three major assembly strategies: 1) a high-quality, short-read-based assembly with a map made from a modestly sized genetic cross (Gulf pipefish), 2) a high-quality, long-read, optical map-scaffolded assembly with a large genetic cross (platyfish), and 3) a highly scrambled, hybrid assembly with a large genetic cross (Antarctic black rockcod).
MATERIALS AND METHODS
The primary design goal of Chromonomer is to integrate disparate information (contigs, scaffolds, and genetic maps) in a hierarchy of reliability. In cases where the first source of information is ambiguous, the software can apply additional sources. Chromonomer is designed first to trust contiguous genome assembly, in other words, the contigs, where scaffolding has not yet been inferred from other molecular information. Next, Chromonomer trusts the overall linkage map ordering, followed by the scaffolding, raw read depth of coverage, and finally, conserved gene synteny, depending on what information is available and on the user's dictate. Given this hierarchy of information, the Chromonomer algorithm 1) inserts virtual gaps into scaffolds, if depth of coverage data are supplied, 2) breaks interlinkage group scaffolds using real or virtual gaps, 3) models each linkage group as a graph with scaffolds attached to graph nodes, 4) finds a consistently ordered set of markers, 5) breaks intra-linkage group scaffolds that span non-adjacent map nodes, and optionally, 6) orders and orients any unordered scaffolds using conserved gene synteny.
The basal Chromonomer algorithm
Chromonomer requires a description of the genome assembly, which consists of an AGP (A Golden Path) file (NCBI 2019) describing the structure of the scaffolds (the set of ordered and oriented contigs and gaps), a tab-separated file describing the genetic linkage map, including the linkage group and centiMorgan (cM) position of each marker, and a SAM or BAM file (SAM/BAM Format Specification Working Group 2019) describing the alignment positions of the markers in the physical assembly. Optionally, a FASTA file containing the genome sequence can also be supplied (and with it, Chromonomer will provide a reordered FASTA file of physical sequence after the Chromonomer algorithm completes). The contig, scaffold, and marker IDs must match among the input files.
Inter-linkage group conflicts
In the first stage of the algorithm, Chromonomer resolves inter-linkage group conflicts. For each scaffold, Chromonomer collects the markers aligned to it and sorts the markers by linkage group (Figure 1A-C). Since linkage group assignment is statistically very robust, the linkage map is trusted over the physical scaffolding. So, if markers on a single scaffold belong to more than one linkage group, Chromonomer will attempt to split the scaffold ( Figure 1B). To do so, the markers must be in two or more consistently ordered sets, with an available scaffold gap (sequence of 'N' characters) between them; if multiple gaps exist, the largest gap is chosen. If such a configuration is not available, Chromonomer will discard sets of neighboring markers, starting with the smallest set, ( Figure 1C) until the scaffold can be split, or until a single, consistent set of markers remain. Split scaffolds are renamed in a user-definable way, and the details of the process are logged.
Chromonomer next determines a provisional orientation for each scaffold by calculating a linear regression between the linkage map cM positions, and scaffold-aligned basepair positions of the inclusive markers. Although not all markers are consistently ordered yet, Chromonomer will orient the scaffold in the forward direction if a positive regression results, or in the reverse direction in the case of a negative regression. This requires markers to link a scaffold to at least two cM positions in the map.
Modeling linkage groups as graphs The Chromonomer basal algorithm represents each cM position in the map as a node in a graph. Markers are used to anchor scaffolds to their respective nodes in the graph; if a scaffold spans consecutive nodes, the nodes are collapsed together providing a definitive orientation for the scaffold. If a scaffold is anchored to multiple, nonneighboring nodes, it is placed into the graph in every position where at least one of its aligned markers occurs ( Figure 1D). If multiple scaffolds collapse into the same, single graph node, their order (linear series) within the node cannot be determined from the map alone, though this cluster of scaffolds can still be ordered relative to scaffolds anchored to other nodes.
Finding a consistent set of markers Unlike how Chromonomer prioritizes map structure over scaffolding, the algorithm trusts the contiguous physical assembly over individual markers that are not corroborated by other, nearby markerssince genotyping errors can slightly change the position of a particular marker in the map. For each occurrence of a scaffold within a linkage group graph, Chromonomer will identify a maximal set of markers for the associated node that have a consistent order with respect to each other (marker base pair positions increase with map cM position, or the orders are inverted if in reverse orientation) ( Figure 1E). The markers whose order conflicts with respect to each node are logged and discarded.
Resolving intra-linkage group conflicts Chromonomer next looks at each scaffold individually within the linkage group. Scaffolds that remain in multiple nodes of the graph indicate assembly errors. Since the markers that remain are consistently ordered, Chromonomer can break scaffolds at the nearest gap between the two groups of markers from each subset of the scaffold anchored to different graph nodes (e.g., Scaffold_1 in Figure 1E and 1F), and the details of each split are logged. If an appropriate gap cannot be found to split the scaffold, the smallest set of markers at a particular graph node are discarded until the scaffold can be split across nodes, or until there is only one set of consistent markers left in a single graph node, which places the unsplit scaffold in a single location.
Finally, Chromonomer recalculates the orientation of each scaffold, again using linear regression of marker positions, and summarizes the new, chromonome-level assembly, creating a new set of sequences reflecting any scaffold splits (output in a FASTA file) and an assembly description (output in a set of AGP files) to describe the changes. An external script, translate_gtf.py, is provided to lift over a set of gene annotations from a scaffold-based assembly to a chromonome, or vice versa.
Depth of coverage and virtual gaps
Chromonomer relies on assembly gaps to break a scaffold when the genetic map indicates a misassembly. However, depending on the assembly process, a genome might be structured with very few, or without any, gaps. The alignment of raw reads back to the genome can reveal regions of anomalous depth, which are likely candidates for points of misassembly. Chromonomer can use per-base pair depth of coverage data, generated by samtools (Li et al. 2009), to identify these regions ( Figure 2). Along each scaffold, Chromonomer slides a window (user definable, default 5Kbp) and calculates the mean depth within each window. It then determines how many standard deviations any window is from the scaffold depth mean. If a window is above or below the user-definable number of deviations (default is 3), a virtual gap of zero length is inserted at the 59 end of the window into the internal AGP representation of the scaffold, which makes it available for Chromonomer's standard splitting algorithm. Any virtual gaps not used during processing will be removed before outputting modified AGP or FASTA files.
Ordering scaffolds with conserved gene synteny If a scaffold does not span more than one cM node in the linkage group, it cannot be oriented by the map. Likewise, if two or more scaffolds are anchored to a single map node, they cannot be unarbitrarily ordered within that node. For these classes of scaffolds and only these, Chromonomer can be instructed to use the order of orthologous genes from a related genome to further specify order and orientation. In other words, conserved synteny data are subordinate to map location data. The user specifies the gene annotation of a related genome (in GFF or GTF format), the annotation of the focal genome at a scaffold level, and the orthology assignment of genes between the two annotation files (using a tab-separated file).
First, for each linkage group, the overall relative orientation of the "external" chromosome (i.e., from the related genome) must be determined. To do this, Chromonomer calculates the regression of gene positions between definitively oriented scaffolds on the linkage group (that is, existing on two or more nodes in the graph) and the orthologous genes on the external chromosome. Based on this comparison, the orientation of the external chromosome is reversed if necessary.
Next, for each node in the linkage group that contains more than one scaffold (that is a node in the map that hosts unordered or unoriented scaffolds), Chromonomer tabulates the genes present on the collection of scaffolds. Any genes that fall on a non-orthologous external chromosome and any singleton, out-of-order genes are excluded, and the set of genes with a congruent order is retained. Similarly, orthologous genes that are too far away from the main set of orthologs, as determined using a trimmed mean algorithm (Bednar and Watt 1984), are discarded. The genes on the set of scaffolds are ordered according to the external chromosome position, which then allows ordering of the scaffolds in the node. Finally, the orientation of each scaffold is determined independently by calculating a regression based on the basepair positions of the genes on the scaffold and the external chromosome.
Rescaffolding based on the genetic map
The rescaffold algorithm provides a way to rationally break gapless contigs that the basal algorithm alone cannot resolve, by reprioritizing the map marker order over the assembly contig boundaries ( Figure 2). Each node in the linkage group graph will be considered the 'owner' of the sequence that 'its' markers span. Because genotyping errors can shift marker order in the map relative to the physical sequence, sets of markers from adjacent map nodes must be found that do not overlap in the physical sequence. The first step is to bucket sets of markers according to their graph node origin and to calculate a mean basepair position to represent the map node in the physical sequence. Next, the set of nodes (and their markers) are re-sorted according to these mean basepair positions. Then adjacent nodes are traversed, and the algorithm prunes any markers whose basepair positions overlap between the current and next nodes (Figure 2A). Markers are pruned in rounds, according to how far a marker is from Figure 1 The primary Chromonomer algorithm. The algorithm takes a set of scaffolds (seen here as rectangles), a set of markers (typically DNA sequence, i.e., RAD markers; represented here as shapes within the rectangles), an assembly file (AGP file), which describes how contigs and gaps are formed into scaffolds in the assembly, and a genetic map, which provides order to the markers. (A-C) Scaffolds are first evaluated to identify sets of markers mapped to different linkage groups. Those scaffolds will be split at the nearest gap (B) or pruned out (C) if a consistent set of markers cannot be found. (D) Scaffolds are anchored to their positions in the genetic map; if a scaffold appears in two locations in the genetic map, it is anchored twice. (E) A consistent ordering of markers is determined, with inconsistent markers discarded. (F) Scaffolds are oriented or split at the nearest gap, as dictated by the genetic map.
the mean basepair position for the map node. This has the effect of removing markers that are the farthest away in the physical sequence from the mean map node position first. Pruning continues until no markers overlap between the nodes. Breakpoints are required for the algorithm to proceed further, and if the sequence in the integration has no gaps, Chromonomer can insert virtual gaps via raw read depth of coverage ( Figure 2B, yellow lines). The contig is finally broken into pieces using the basal algorithms described above, and after splitting, the map nodes are re-sorted back to their cM positions ( Figure 2D); this constitutes the reordering and reorienting of the new components of the broken contig.
Chromonomer outputs
Chromonomer creates 'before' and 'after' log files for each linkage group; it creates a specific log for each modified scaffold. These logs detail marker positions relative to the genetic map and their genomic alignments. Markers dropped due to conflicts are shown and a reason for dropping them supplied. General statistics, such as integrated chromosome lengths, and the location and number of splits are logged. The software also computes a list of "promising scaffolds": scaffolds that could be integrated into the linkage groups by improving the map. Chromonomer will also produce FASTA, AGP, and GFF files that describe the newly integrated assembly.
Web-based visualization
For each run of Chromonomer, the output directory of data can be made visible to a web server (e.g., Apache, not supplied) and then served via HTTP. Chromonomer pre-computes a 'before' and 'after' JSON file (JavaScript Object Notation; [URL]/ rfc8259) for each scaffold. These JSON files are used to create graphical visualizations of the 'before' and 'after' states of the linkage groups that can be viewed in a web browser. Figures 3, 4, and 6 are based on these visualizations. Chromonomer includes JavaScript code that will also be served by the web server to render these visualizations Figure 2 Using depth of coverage to create virtual gaps, and rescaffolding the assembly. Assemblies constructed using long reads often consist purely of contigs, with no gaps. In these cases, we can input into Chromonomer depth of coverage data, generated by aligning raw reads back to the assembly, and we can identify anomalous values in depth of coverage to direct where to create virtual breakpoints in the assembly. Here, in (A) the markers clearly show a misassembly in the center region of the contig (red markers). With no gaps, the normal algorithm to split the contig will fail. (B) The scaffolding algorithm instead assumes the genetic map is the correct source of information and identifies where the contig should be broken, according to a consistent set of reordered map markers. Depth of coverage information (C) is incorporated to identify logical break points and (D) the contig is split into the respective pieces.
in the web browser using the D3 library ( [URL]/). If a web server is not available, textual versions of these visualizations are also supplied.
Genome integrations
The description of Chromonomer genomic analyses, including commands executed for Gulf pipefish, platyfish, and rockcod genomic integrations can be found in the Supplementary Methods.
Data availability
All input data were previously available in online repositories and the appropriate accession numbers are listed inline in the Supplementary Methods section. The three integrated genomes have been deposited in the Dryad Data Repository at [URL]/ 10.5061/dryad.gtht76hjm. Chromonomer is released as open source software, under the GPL v3 license. Documentation can be found, and the source code may be downloaded from [URL]:// catchenlab.life.illinois.edu/chromonomer. Chromonomer can be built on UNIX-like systems (e.g., Linux and MacOS) and has no mandatory dependencies on other software. Chromonomer can be executed on any modestly capable computer (laptop or server).
RESULTS
The basal Chromonomer algorithm: the Gulf pipefish genome Chromonomer was used to integrate the physical assembly of the Gulf pipefish (Sygnathus scovelli) with a genetic map derived from an F1 cross of 108 progeny (Small et al. 2016). This reference genome is an Illumina-based assembly following the ALLPATHS-LG assembly strategy (see Table 1 for details). The map consisted of 6593 markers on 22 linkage groups while the physical assembly was 307Mbp in total length, contained in 2104 scaffolds, and had a scaffold N50 of 640Kb and an L50 of 115. Figure 3 shows how the Chromonomer algorithm handled one linkage group. It is clear that prior to processing, marker order is inconsistent with alignment order (e.g., see Figure 3A, in which red lines are crisscrossed), but after processing their order has been corrected by discarding incongruous markers (the red lines are resolved in Figure 3B). Scaffold 76 ( Figure 3A, green) appears in the map twice, as does scaffold 12 ( Figure 3A, blue). After processing Figure 3 The Chromonomer algorithm as employed in the Gulf pipefish (Sygnathus scovelli) assembly. The figure shows all the numbered scaffolds belonging to LG5 before (A) and after (B) processing. In the diagrams, each marker in the linkage group (left) is connected by a line to its alignment position on each scaffold (right). In red in (A), scaffold_ 8 demonstrates markers with conflicting physical and map orders. In (B), the order of markers has been resolved and some conflicting markers discarded. Scaffold_76 (green) and scaffold_12 (blue), which are each anchored in two map positions, demonstrate examples of scaffolds that need to be split so a third scaffold can be inserted into the rift.
( Figure 3B), both scaffolds have been split, and in each case an additional scaffold has filled a gap (scaffolds 1860 and 1406, respectively). After Chromonomer integrated the genetic map, 266Mbp was incorporated into 22 linkage groups in the chromonomed assembly (87% of assembly length) with 550 of the 2104 scaffolds incorporated and five scaffolds having been split. Of those scaffolds not incorporated, the mean length and N50 length were 26Kbp and 4Kbp, respectively, and no markers aligned to over 92% of these 1554 relatively small scaffolds.
Incorporating conserved gene synteny into the Gulf pipefish integration
After the initial incorporation of the genetic map in Gulf pipefish, a number of scaffolds are not entirely resolved, including a large cluster of scaffolds at the top end of linkage group 14 ( Figure 4A). Using conserved gene synteny information from a congener, the greater pipefish (Sygnathus acus), Chromonomer was able to order 16 scaffolds and to orient 14 scaffolds ( Figure 4B, colored boxes). Figure 5 shows conserved synteny between S. scovelli and S. acus, before and after Chromonomer employed ortholog-based ordering. Naturally, the process has made S. scovelli look more like S. acus, which might not always be biologically correct. If the reference organism is sufficiently closely related, however, this method provides a rational hypothesis for a likely order and orientation beyond what was initially arbitrary. This rationale is supported in this case by the fact that many of the reoriented scaffolds display conserved intra-scaffold gene order, and genome-wide there is a strong pattern of conserved synteny between the two pipefish (Fig. S1). Figure 5 also shows that there remain putative true rearrangements between S. scovelli and S. acus, as demonstrated by Gulf pipefish scaffold 16 in the region at 14Mb on LG14 (22Mb in S. acus). In this case, the scaffold has high support from the genetic map for its position, while the LG14 before (A) and after (B) processing. In this example we have incorporated conserved gene synteny from the close relative Sygnathus acus to order and orient scaffolds whose position and orientation are left ambiguous by the genetic map. Colored scaffolds indicate where synteny was employed.
orthologous gene block appears in a different relative location in the S. acus genome.
Rescaffolding an assembly prior to map integration: the platyfish genome As described above, if there is a misassembly that inverts or translocates a component of the contig but does not produce scaffold gaps, the basal Chromonomer algorithm on its own will discard all inconsistent markers but the majority set, leaving the unbroken scaffold at the place in the linkage graph with the largest number of consistent markers. We can see how this would occur in the southern platyfish (Xiphophorus maculatus) assembly (Schartl et al. 2013). The assembly, prior to the application of Chromonomer, is qualitatively impressive, with 24 chromosome-length contigs, and only an additional 76 scaffolds, with an N50 of 31.5Mbp. The assembly was generated from longread data, and an optical map was used to further scaffold the assembled contigs (see Table 1 for details). However, when we compare the assembly against a high quality genetic map containing more than 22,000 markers and 267 progeny (a backcross between X. maculatus and X. helleri (Amores et al. 2014)), we find that some putative assembled chromosomes agree strongly with the genetic map (linkage group 1, Fig. S2), but others show potential misassemblies ( Figure 6A). A caveat is that, since this genetic map was produced from a hybrid cross, it is possible that some of the conflicts between the assembly and map could be due to true differences between the X. maculatus and X. helleri genomes. The map shows a large inversion relative to the assembly on LG14 between 35-53cM, but there exists no clear place for the basal Chromonomer algorithm to break the assembly. Without intervention, Chromonomer would leave the chimeric scaffold in the location with the largest set of correlated markers. We applied Chromonomer's rescaffold option (Figure 2) to prioritize the genetic map order over the contiguous assembly and resolve this situation. Figure 6B shows the result of applying this algorithm to a relatively simple case in the platyfish assembly, where marker order is not cleanly correlated between the genetic map and physical assembly. Comparison of patterns of gene synteny, before and after, relative to the medaka genome (Oryzias latipes, Figure 7) illustrates the outcome. The reordered physical assembly is now congruent with gene order in medaka. For an example of more complex corrections employing the rescaffold algorithm in the platyfish assembly, see Figs S3 and S4.
Algorithmic limits of Chromonomer: the rockcod chromonome The Antarctic bullhead notothen, or black rockcod (Notothenia coriiceps), is an extreme cold-adapted fish with an interesting karyotype.
While the ancestral haploid chromosome number in teleost fish is 24 or 25 (Naruse 2004), the black rockcod has just 11 chromosomes (V. P. Prirodina, A. V. Neyelov 1984). Using an outcrossed RADseqbased genetic map constructed from 244 progeny in an F1 pseudotestcross with 9,138 mappable markers, Amores et al. (2017) were able both to confirm this genome evolution occurred by end-to-end fusions and to identify which ancestral chromosomes became fused. The sequenced rockcod genome was also assembled using a hybrid strategy that mixed data from Illumina paired-end libraries, from 454-sequenced mate-pair libraries, and from limited PacBio RS II sequencing (see Table 1 for details). The resulting assembly was composed of 37,605 scaffolds, had a scaffold N50 of 218Kbp, and the largest scaffold was 28.8Mbp in lengtha poor result that is not atypical for predominately short-read based assemblies. We used Chromonomer to integrate the physical assembly with the genetic map for the first time. Here we found extreme discordance within the assembled scaffolds. For example, the second largest scaffold, KL668296.1 (27.5Mbp in length), contains 368 markers, but these markers were scattered in the genetic map across every one of the 11 linkage groups (Fig. S5). In fact, the four largest scaffolds map to all 11 linkage groups resulting in a remarkably disordered assembly. The pattern can be seen when gene orthologs are visualized in comparison with a related genome, in Figure 8. The x-axis shows genes from rockcod in gray (at bottom) and the corresponding orthologous genes from the blackfin icefish (Chaenocephalus aceratus) in red, located on the icefish chromosomes (y-axis), but ordered according to the rockcod. Large scaffold KL668296.1 (boxed by a dashed line in Figure 8A) spans nearly half of rockcod linkage group 1, but genes orthologous to those identified on this scaffold are found dispersed all over the icefish genome, a condition unlikely to be biologically true. A multitude of other rockcod scaffolds are also probably chimeric, most likely due to assembly errors that stem from mate-pair libraries, with error amplified by the hybrid assembly and gap closing/scaffold extension algorithms that were optimized for maximal simple statistics (like N50), but not for accuracy. After running the basal algorithm of Chromonomer, scaffold KL668296.1 was broken down into 27 coherent pieces and those were reintegrated into their respective positions according to the genetic map. The resulting increased congruence in conserved gene synteny suggests structural improvement of the assembly (Figure 8B), and importantly, the original signal of chromosome fusion is more cleanly resolved, where a region that is syntenic in rockcod is split between LG1 and LG4 in icefish. If we view the genome-wide conserved gene synteny between rockcod and the blackfin icefish, we see similarly improved synteny, but still a lot of noise. Because of the granular nature of misassembly in this genome, there are not enough markers to fully correct all of the errors, and segments containing single or small numbers of genes remain incorrectly fused to other segments (e.g., Fig. S6); such cases likely account for many of the lines crossing to non-orthologous chromosomes in Figure 9B.
DISCUSSION
In the decades since the human genome project (International Human Genome Sequencing Consortium 2001) was completed by a large consortium with massive resources (Collins 2003), sequencing technology advances have facilitated order of magnitude improvements to genome assembly though the employment of long-read, high volume sequencers and more advanced scaffolding techniques. However, even with greatly improved N50s, moving a genome from a collection of scaffolds to a chromonome (Braasch et al. 2015) with realistic long-range relationships among assembly segments remains a major impediment. Genetic maps are one of the oldest genomic resources, dating back to the beginning of the field of modern genetics (Painter 1933), but by the time of the human genome project they too required significant resources to discover and genotype markers. Short-read, massively parallel sequencing has changed genetics too, as RAD sequencing and software like Stacks made genetic mapping broadly feasible and applicable (Rochette et al. 2019). This new generation of genetic mapping provided huge numbers of markers simultaneously with the genotyping itself, and has permitted map building in a single generation. The advantages of marrying these two data streams (genome sequence and genetic map) has been demonstrated in improvement and validation of several recently released reference genomes (International Cassava Genetic Map Consortium (ICGMC) 2015; Kelley et al. 2016;Lee et al. 2019;Takehana et al. 2020;Simakov et al. 2020).
There are several pieces of software that aim to integrate genetic maps with genome assemblies. The ALLMAPS software (Tang et al. 2015) seeks to optimize the set of markers to maximize concordance between the linkage map positions of markers and their aligned genomic positions. It permutes scaffold positions in the integrated genome to minimize the distance between markers on different scaffolds. ALLMAPS then refines scaffold position and determines orientation using a genetic algorithm. Scaffolds that should be broken are flagged by ALLMAPS but breaking must be done manually. Lep-Anchor (Rastas 2020) also aims to optimize marker order, first employing a Hidden Markov Model to bin markers to different linkage groups and split inter-linkage group scaffolds, and then it uses dynamic programming to calculate the number of markers that support all particular scaffold orders. The Kermit software uses a genetic map to guide a de novo assembly (Walve et al. 2019). Kermit bins scaffolds given in an initial assembly according to their order in the genetic map, then it places raw, long reads into those bins and creates an overlap graph from that initial order. It completes the assembly based on the sequence in the overlap graph.
In the current work, we have presented Chromonomer, which broadly shares goals with tools like ALLMAPS and Lep-Anchor, however, these software treat integration as an optimization problem; by permuting the set of markers and scaffold positions to generate Figure 5 Ortholog-directed scaffold rearrangements in the Gulf pipefish (Sygnathus scovelli). Potential improvements in LG14 integrated assembly by incorporation of gene synteny between S. scovelli and S. acus. Colored scaffolds indicate where synteny was employed, and colors are consistent with Figure 4. In each panel, the S. scovelli genetic map is shown above, linking the scaffolds of the physical assembly together. Lines also connect each pair of gene orthologs between S. scovelli and S. acus. many different orders, they will discover a best order. This brute-force approach requires that the underlying objects are rationalthat is, that they can be ordered. Chromonomer is designed around a different philosophy: allow the application of different lines of evidence, and then show where the underlying components fit together and where they do not. Since each assembly and scaffolding strategy brings a particular error model along with it, the key to successfully integrating a physical assembly with a genetic map is the ability to rank the quality of different types of information and to employ the most dependable information in the most rational hierarchy. Given a high-quality assembly, both methods will result in the same, high-quality integrated genome. But given a pathological assembly, the results will be very different. Along these lines, Chromonomer is actually two distinct things: first, a tool to integrate physical and genetic assemblies, and second, a hypothesis generator to be employed during the assembly process itself. In addition, the Figure 6 Using virtual gaps and the rescaffold algorithm in platyfish (Xiphophorus maculatus). (A) The platyfish assembly shows a clear misassembly (an inverted segment between 35-53cM) when compared against the genetic map. (B) A consistent order of markers is found on the map, and depth of coverage is employed to split the CM008951.1 contig into 2 components that can then be independently reoriented.
Figure 7
Improvements in the platyfish (Xiphophorus maculatus) chromosome-level assembly. Conserved gene synteny between platyfish (Xma) and medaka (Oryzias latipes, Ola) illustrates improvements in the LG14 integrated assembly by application of the rescaffold algorithm. The top panel (A) shows synteny prior to correction; several inversions are present, including one associated with the platyfish assembly (orange, colored to match the scaffolds in Fig. 6). After correction (B), inversions and ordering are rectified.
algorithm Chromonomer appliesmodeling each linkage group as a graphis discrete and does not require optimization, which provides an execution time on the order of minutes (several orders of magnitude faster than optimization-based algorithms, described above (Rastas 2020)).
Integrating the map and the physical assembly can be used powerfully in an iterative process that leads to improvement of the final assembly via improvements in the inputs. Aided by the reporting and visualization tools in Chromonomer, a researcher can improve the genetic map when genotyping errors become obvious in the context of the physical assembly. Similarly, the researcher, when presented with the number and type of scaffold splits conducted by Chromonomer, can choose to re-examine or pare out problematic data types added of the assembly (e.g., a particular sequenced matepair library, or the output of software that hybridized different sequenced libraries together) that cause most of the artifacts. In an iterative approach, a researcher can employ knowledge of synteny from multiple species to provide evidence for a particular assembly hypothesis. For example, if a change in gene order coincides with the boundaries of a scaffold, it is likely an assembly error. One of the most Figure 8 The rockcod (Notothenia coriiceps) assembly. All of the large scaffolds in the rockcod assembly appear to be large inter-and intrachromosomal chimeras. When we examine LG1 in rockcod (A) we can see that orthologous rockcod genes are found scattered across the genome of blackfin icefish, a related taxon. The largest rockcod scaffold, KL668296.1 is highlighted by the dotted line and we can see that it is composed of sizeable pieces from all over the genome. (B) After processing with Chromonomer, the scaffold is broken up and redistributed in the assembly. We can now clearly see the conserved, two-to-one gene synteny between the icefish and rockcod.
innovative applications of Chromonomer is to use it as a tool to compare and contrast different genetic maps, perhaps from members of different populations of the same species, with the same base assembly. Here read depth and synteny information could be combined to explore the nature of structural variants within a species. Treating genomic assembly components (maps, scaffold sets, gene annotations) as independent objects, each providing a different line of evidence, provides a rich informatic framework to explore genome architecture.
Recently, chromosomal conformation capture methods, such as Hi-C, have become very useful in further scaffolding genomes, while versions of optical mapping have become more accessible as well (e.g., BioNano). These methods can be very successful in scaffolding a genome, particularly if combined with long-read-assembled contigs. Do these methods deprecate the use of genetic maps? In our opinion, no. In fact, the approaches complement each other. Chromosomal conformation capture and optical mapping are scaffolding algorithms that can improve assemblies significantly, but they also introduce errors commensurate with the quality of the data. The long-read platyfish assembly shows simple ( Figure 6) and complex (Figs. S3) errors resulting from optical map scaffolding. Both scaffolding protocols also rely on high molecular weight DNA and require nontrivial library preparations. On the other hand, genetic maps can provide one of the only independent lines of evidence (a second such line is conserved gene synteny) of the biological correctness of a scaffold, but high-quality maps can be created only in certain organisms and can come with another class of error sources. The highest quality genomes, therefore, will integrate as many of these lines of evidence as possible, including long-molecule methods and genetic maps.
Available assembly and scaffolding software still tend to operate as "black boxes", with internal algorithmic decisions opaque to the outside user. There are practical reasons for this, including the volume of data involved and difficulty in designating standardized file formats. Genome assembly as a service is also gaining in popularity, which risks further obscuring the underlying nature of a particular genome assembly. Future work in accurate reference genome construction should include software design that exports valuable internal assembly/scaffolding information in common formats, and allows practitioners to use multiple lines of evidence, properly integrated by strong underlying tools, into an evolving assembly hypothesis.
ACKNOWLEDGMENTS
We thank Clay Small and John Postlethwait for comments on the manuscript. J. Catchen was supported by NSF grant 1645087. A. Amores was supported by NIH grant R01 ODO11116 (Postlethwait, PI) and NSF grant 1543383 (Postlethwait, PI). S. Bassham was supported by NIH grant R24 RR032670 (Cresko, PI). J. Catchen and A. Amores designed Chromonomer. J. Catchen implemented the software and completed the analyses. J. Catchen, S. Bassham, and A. Amores interpreted the analyses. J. Catchen and S. Bassham wrote the manuscript. Chromonomer improves the rockcod (Notothenia coriiceps) assembly. The rockcod assembly can be chromonomed using the genetic map. (A) shows genome-wide conserved gene synteny prior to integrating the genome. (B) shows marked improvement genome-wide in the assembly after breaking down the largest scaffolds using the genetic map. However, smaller assembly errors remain.
==
Domain: Biology Medicine Computer Science
|
[{"added": "2020-02-13T09:11:13.441Z", "created": "2020-02-05T00:00:00.000", "id": "212970115", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Computer Science"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://academic.oup.com/g3journal/article-pdf/10/11/4115/37192173/g3journal4115.pdf", "pdf_hash": "e9aed5f4fd4f95fb5d65b79a32dfc3aa1f673d88", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:4783", "s2fieldsofstudy": ["Computer Science", "Biology"], "sha1": "792473a2632662625eca5538ee6b211018b9b11b", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Analysis of lifestyle and metabolic predictors of visceral obesity with Bayesian Networks
Background The aim of this study was to provide a framework for the analysis of visceral obesity and its determinants in women, where complex inter-relationships are observed among lifestyle, nutritional and metabolic predictors. Thirty-four predictors related to lifestyle, adiposity, body fat distribution, blood lipids and adipocyte sizes have been considered as potential correlates of visceral obesity in women. To properly address the difficulties in managing such interactions given our limited sample of 150 women, bootstrapped Bayesian networks were constructed based on novel constraint-based learning methods that appeared recently in the statistical learning community. Statistical significance of edge strengths was evaluated and the less reliable edges were pruned to increase the network robustness. To allow accessible interpretation and integrate biological knowledge into the final network, several undirected edges were afterwards directed with physiological expertise according to relevant literature. Results Extensive experiments on synthetic data sampled from a known Bayesian network show that the algorithm, called Recursive Hybrid Parents and Children (RHPC), outperforms state-of-the-art algorithms that appeared in the recent literature. Regarding biological plausibility, we found that the inference results obtained with the proposed method were in excellent agreement with biological knowledge. For example, these analyses indicated that visceral adipose tissue accumulation is strongly related to blood lipid alterations independent of overall obesity level. Conclusions Bayesian Networks are a useful tool for investigating and summarizing evidence when complex relationships exist among predictors, in particular, as in the case of multifactorial conditions like visceral obesity, when there is a concurrent incidence for several variables, interacting in a complex manner. The source code and the data sets used for the empirical tests are available at [URL], Bayesian networks (BN) have become a very popular tool for biological network reconstruction [1][2][3], for genotype-to-phenotype relationship studies [4] and for clinical and microarray data aggregation [5,6]. BN are directed acyclic graphs (DAG) that model the probabilistic dependencies underlying the data. These graphical models are highly attractive for their ability to describe complex probabilistic interactions between variables. They offer a coherent and intuitive representation of uncertain domains of knowledge. The graphical part of BN reflects the structure of a problem, while local interactions among neighboring variables are quantified by conditional probability distributions. Learning a BN from data requires identifying both the model structure and the corresponding set of model parameter values. Given a fixed structure, however, it is straightforward to estimate the parameter values. The task can be efficiently solved according to the maximum likelihood (ML) or maximum a posteriori (MAP) criterion under the assumption that the learning data contain no missing values [7,8]. As a result, research on the problem of learning BN from data is focused on methods for identifying the structure that best fits the data. Despite significant recent progress in algorithm development, the computational inference of network structure is currently still very much an open challenge in computational statistics [7,9]. To appreciate the complexity of learning a DAG, we note that the number of DAGs is super-exponential in the number of nodes [7].
Broadly speaking, there are two main approaches to BN structure learning. Both approaches have advantages and disadvantages. Score-and-search methods search over the space of structures (or the space of equivalence BN classes) employing a scoring function to guide the search. Another approach for learning BN structures, known as the constraint-based (CB) approach, follows more closely the definition of BN as encoders of conditional independence relationships. According to this approach, some judgments are made about the (conditional) dependencies that follow from the data and use them as constraints to construct a partially oriented DAG (PDAG for short) representative of a BN equivalence class. There are many excellent treatments of BN which surveys the learning methods [7,9]. When data sets are small, the relative benefits of the two approaches are still unclear. While none has been proven to be superior, considerable advances have been made in the past years in the design of highly scalable divide-and-conquer CB methods [10][11][12][13][14] in order to improve the network reconstruction accuracy when the number of samples is small.
In this study, we apply one of these CB algorithms, named Recursive Hybrid Parents and Children (RHPC), for representing the statistical dependencies between 34 clinical variables among 150 women with various degrees of obesity. Obesity is recognized as a disease in the U. S. and internationally by governments, health organizations, researchers and medical professionals. It is a complex multifactorial condition that needs to be studied by the means of multidisciplinary approaches involving biological expertise and new statistical and data mining tools. Features affecting obesity are of high current interest. Clinical data, such as patient history, lifestyle parameters and basic or even more elaborate laboratory analytes (e.g., adiposity, body fat distribution, blood lipid profile and adipocyte sizes) form a complex set of inter-related variables that may help better understand the pathophysiology of visceral obesity and provide guidance for its clinical management. Gregori et al. [15] performed a meta-analytic framework for the analysis of obesity and its determinants in children using Bayesian networks. Only seven lifestyle risk factors were considered as being potentially related to obesity in this population. To the best of our knowledge, our study is the first attempt to use BNs in the context of modeling the complex relationships between lifestyle and metabolic correlates of visceral obesity among women.
We use the bootstrapping method to generate more robust network structures as discussed in [6,16]. Statistical significance of edge strengths are evaluated using this approach. If an edge has a confidence above the threshold, it is included in the consensus network. Thus, if dependencies have enough support in the bootstrapping process they are captured and represented in the final consensus network. The confidence estimate assigned to each network edge is represented graphically on the final network. Such network represents a powerful computational tool for identifying putative causal interactions among variables from observational data. The consensus network graphically represents the possibly causal independence relationships that may exist in a very parsimonious manner [17]. In this study, special emphasis was placed on integrating physiological knowledge into the graph structure. Once the consensus PDAG was constructed from data, the remaining undirected edges were then directed according to our causal interpretation and additional latent variables were added to the graph for the sake of clarity, coherence and conciseness. The graphical representation provides a statistical profile of this sample of obese women, and meanwhile helps identifying the most important predictors of visceral obesity. Using the concept of a Markov Blanket we can identify all the variables that shield off the class variable from the influence of the remaining network. Therefore, BNs automatically perform feature selection by identifying the (in)dependency relationships with the class variable. We compare our findings with the results obtained using the same data and more traditional regression models.
Bayesian networks
Formally, a BN is a tuple < , P > where = <U, E > is a directed acyclic graph (DAG) with nodes representing the variables in the domain U, and edges representing direct probabilistic dependencies between them. P denotes the joint probability distribution on U. The BN structure encodes a set of conditional independence assumptions: that each node X i is conditionally independent of all of its nondescendants in given its parents Pa i . These independence assumptions, in turn, imply many other conditional independence statements, which can be extracted from the network using a simple graphical criterion called d-separation [8].
We denote by X ⊥ P Y|Z the conditional independence between X and Y given the set of variables Z where P is the underlying probability distribution. Note that an exhaustive search of Z such that X ⊥ P Y|Z is a combinatorial problem and can be intractable for high dimension data sets. We use X Y ⊥ | Z to denote the assertion that X is d-separated from Y given Z in . We denote by dSep(X, Y), a set that d-separates X from Y.
The converse does not necessarily hold. We say that < , P > satisfies the faithfulness condition if the d-separations in identify all and only the conditional independencies in P, i.e., X ⊥ P Y|Z if and only if (iff) X Y ⊥ | Z . Two graphs are said equivalent iff they encode the same set of conditional independencies via the d-separation criterion. The equivalence class of a DAG is a set of DAGs that are equivalent to . [8] established that two DAGs are equivalent iff they have the same underlying undirected graph and the same set of v-structures (i.e., uncoupled head-to-head meetings X Y Z). So we define an essential graph (also called a DAG pattern) for a Markov equivalence class to be the partially directed acyclic graph (PDAG), that has the same links as the DAGs in the equivalence class and has oriented all and only the edges common to all of the DAGs in the equivalence class. The directed links in the essential graph are called the compelled edges [7].
An important concept of BN is the Markov blanket of a variable, which is the set of variables that completely shields off this variable from the others. In other words, a Markov blanket M T of T is any set of variables such that T is conditionally independent of all the remaining variables given M T . A Markov boundary, MB T , of T is any Markov blanket such that none of its proper subsets is a Markov blanket of T. Suppose < , P > satisfies the faithfulness condition. Then, for all X, the set of parents, children of X, and parents of children of X is the unique Markov boundary of X. A proof can be found for instance in [7]. We denote by PC T , the set of parents and children of T in , and by SP T , the set of spouses of T in , i.e., the variables that have common children with T. These sets are unique for all , such that < , P > satisfies the faithfulness condition and so we will drop the superscript .
Bayesian network structure learning
Automatically learning the graph structure of a BN is a challenging topic of pattern recognition that has attracted much attention over the last few years. CB methods systematically check the data for conditional independence relationships and try to construct a partially directed graphical structure (also called a perfect map) that encodes perfectly the set of independencies. Typically, these algorithms run a c 2 independence test when the dataset is discrete and a Fisher's z test when it is continuous in order to decide on dependence or independence, that is, upon the rejection or acceptance of the null hypothesis of conditional independence. Therefore, conditional independencies that are read off from the BN structure are in total agreement with the conditional independencies that are obtained by the statistical tests. Very powerful, correct, scalable and data-efficient CB algorithms have been recently proposed [10][11][12]. They are correct (or sound) in the sense that they return the correct essential graph under the assumptions that the independence tests are reliable and that the learning database is a sample from a distribution P faithful to a DAG . The (ideal) assumption that the independence tests are reliable means that they decide (in) dependence iff the (in)dependence holds in P. In this paper we adopt one of these CB approaches [11,18]. The essential graph is obtained by running an algorithm called Recursive HPC (RHPC), where HPC stands for Hybrid Parents and Children.
Simulation experiments on artificial data
As RHPC relies on HPC to build the whole network structure, we conducted several experiments on synthetic data to assess the comparative performance of HPC, and two algorithm proposals that appeared recently in the literature, namely MMPC [12] and GetPC [10]. The source code (C++) of HPC as well as all data sets used for the empirical tests are available at [URL] authors' implementation of MMPC and GetPC can be found respectively at [URL]. edu/discover/public and [URL]. MMPC was deemed one of the best CB algorithms in [12] and GetPC was used recently in [2] for modeling gene networks. We also report the performance of our weak learner Inter-IAPC for comparison. For GetPC and MMPC, we used the softwares proposed by the respective authors (see footnote). The confidence threshold of the independence test was fixed to α = 0.05 for all algorithms. All the data sets used for the empirical experiments presented in this section were sampled from a bio-realistic network that has been previously used as benchmark for BN learning algorithms, namely Insulin (35 nodes and 52 edges). The Insulin network [19] was chosen purposely as it consists of the same number of nodes as our dataset. Four sample sizes have been considered: 200, 500, 1000 and 2000. For each sample size, 100 data sets were sampled. We do not claim that this benchmark resembles our real-world problem, however, it makes it possible to compare the outputs of the algorithms. All four algorithms were run on the target node having the largest degree (13 neighbors) in the Insulin BN to increase the difficulty of the task. The variables in the output of the algorithms were compared against the true neighbors. To evaluate the accuracy, we combined precision (i.e., the number of true positives in the output divided by the number of nodes in the output) and recall (i.e., the number of true positives divided by 13, the size of the true PC set) as ( measure the Euclidean distance from perfect precision and recall, as proposed in [10]. Figure 1 summarizes the variability of the Euclidean distance over 50 data sets in the form of quadruplets of boxplots, one for each algorithm (i. e., MMPC, GetPC, Inter-IAPC and HPC). The advantage of HPC against the other three algorithms is clearly noticeable. HPC outperforms the other algorithms in terms of Euclidean distance from perfect precision and recall.
Simulation experiments on the sample of women
The consensus PDAG obtained by running RHPC on the present sample of women is shown in Figure 2. Line thickness corresponds to the relative confidence of the edges. The edges that appeared more than 25% in the networks were included in the aggregate PDAG. The threshold was tuned on the previous Insulin benchmark samples to maximize accuracy. As may be seen, the directionality of the arrows was partially identifiable: 14 edges out of 34 were directed, indicating the presence of several robust uncoupled head-to-head meetings (T Y X).
Physiological knowledge integration into the model
Several interconnected groups of variables were identified, e.g., beer consumption, wine consumption and spirit consumption; cigarettes per day and low exercise; OM and SC fat cell sizes. In each of these densely connected subgraphs, the variables were highly interdependent and a common cause is likely to explain the observed correlations. Hence, we added some extra nodes and directed some of the links according to physiological knowledge available in the literature. The result is the partially directed acyclic graph (PDAG) that is shown in Figure 3. Dashed nodes and arrows are the latent variables that were added for sake of clarity and coherence. By definition, these latent variables are not observed, nor recorded in our data set. For example, the variable high alcohol intake was added as a common "cause" to beer consumption, wine consumption and spirit consumption; the variable unhealthy lifestyle was added as a common cause to cigarettes per day, high alcohol intake and low exercise; the latent variables fat storage and prevailing hormonal conditions were added as two distinct common causes to SC fat cell size and OM fat cell size. Almost all the undirected edges were oriented based on current literature as follows. Edges directed from the age variable were oriented based on the well-documented impact of ageing on visceral adipose tissue accumulation, blood pressure and plasma LDL-cholesterol levels [20,21]. The edge between age and tea consumption is based on the 2004 Canadian Community Health Survey, which showed a steady increase in tea consumption from 19 to more than 71 years of age [22]. The edge between tea consumption and blood pressure was oriented based on literature showing lower cardiovascular disease risk in tea consumers [23] and a direct effect of black tea consumption on peripheral blood flow and arterial stiffness [24]. The edge between age and the number of live children was attributed to the slight decrease in Canadian birth rates observed between 1961-66 and 1981-86 [25], which corresponds approximately to the period in which women of the study had their children. Accordingly, older women of the sample were more likely to have delivered slightly more children. Orientation of the edge between the number of pregnancies and the number of live children is self-explanatory.
The edge between the number of live children and OM fat cell size was derived from literature supporting that post-pregnancy weight retention is an important risk factor for obesity [26]. The finding of a specific association between the number of children and OM fat cell size was novel and warrants further investigation. The edges between OM and SC fat cell sizes and the variables obesity or visceral fat is self explanatory since the excess adipose tissue mass of obese or abdominal obese individuals is constituted of larger fat cells. Associations between fat cell size and obesity have been previously observed [27]. The edges between visceral fat or large OM fat cells and metabolic variables such as LDL-cholesterol, triglycerides and blood pressure was oriented based on the 'portal vein hypothesis', which states that visceral fat is a causal agent for metabolic disturbances [28]. However, this hypothesis has not yet been fully proven as operative and has been challenged by a number of investigators. Further studies are required to firmly establish causality. However, the fact that the association between visceral fat and metabolic disturbances is independent from overall obesity is wellaccepted [29,30]. The edges between the various components of body composition (i.e., bone density, lean body mass and obesity) were logical but it was difficult to provide causal direction between these variables. Indeed, many genetic, epigenetic, developmental and environmental factors can contribute to determine body built of a given individual. Moreover, the sizes of all compartments generally evolve in a more or less coordinated manner throughout the individual's existence [31,32]. It was expected that the variable 5-yr maximal weight would be a strong correlate of the level of obesity and lean body mass since these variables are the main components of body composition [32] and that most patients reported a stable weight in the five years preceding their inclusion in the study. The edges around the number of hours of work and the number of meals out per week were oriented based on the demonstration that increased working time was associated with food choice coping strategies [33], which we suggest is reflected by the edges to number of meals out per week, beer, wine and coffee consumption. On the other hand, the number of meals out per week was related to obesity. Accordingly, the frequency of restaurant food consumption was previously found to be positively related to body fatness [34]. Wine consumption was related directly with plasma levels of HDL-cholesterol. This edge was oriented based on epidemiological data showing a protective effect of moderate wine consumption on HDL-cholesterol levels [35]. Low leisure time physical activity was linked together with smoking habits under a latent causal variable that we termed unhealthy lifestyle. These variables were also linked with coffee and beer consumption, but had no direct link with the level of obesity. We were unable to provide orientation for these edges. Moreover, we were not able to readily explain a small number of edges. For example, the link between age at menarche, which reflects timing of puberty, and dietary supplement use is not intuitive. Further analyses and other samples will be required to clarify this apparent association.
Statistical validation
We noticed from the PDAG that OM fat cell size, visceral fat, blood pressure, tea consuption and age belonged to the triglycerides Markov boundary, though the edge between OM fat cell size and triglycerides was only moderate in strength. The influence of OM fat cell size on triglycerides was mostly mediated by visceral fat. We observed that age and triglycerides were marginally independent according to the d-separation rule. However, they became dependent conditioned on visceral fat. The PDAG was consistent with multivariate linear regression analyzes performed a posteriori on the sample (Table 1). In model 1, plasma triglyceride levels were predicted using computed tomography-measured visceral adipose tissue area (visceral fat variable) and total body fat mass (which is included in the variable obesity). Visceral fat explained 31.9% of the variance in triglyceride levels whereas overall obesity was not a significant predictor of triglyceride levels. A similar analysis in which plasma triglyceride levels were predicted by OM and SC fat cell size was also performed (Table 1, model 2). OM fat cell size explained 21.2% of the variance in triglyceride levels, whereas SC fat cell size was not a significant predictor of triglyceride levels in the model.
Discussion
The purpose of this paper was to introduce the BN methodology in the context of clinical studies, specifically obesity, and to show its effectiveness, as a component of general data mining/knowledge discovery approaches in epidemiology research. We have evaluated a consensus BN learning approach based on boot-strapping techniques on synthetic data with satisfactory results. Although our approach did not use any prior information, it was successful in uncovering biologically relevant dependencies and conditional independencies. Once the most interesting dependencies are ascertained, traditional statistical methods (e.g. linear or logistic regression, etc.) can be used to rigorously scrutinize the resulting smaller subnetworks.
In this study, special emphasis was put on integrating physiological expertise and statistical data analysis together. It is well beyond the scope and purpose of this paper to delve deeper into the problem of inferring causalities from observational data. However, the usefulness of BN stems partly from their causal interpretation. As we have seen, the graphical representation is useful as it allows tighter collaboration between the modeler and the biologist. The integration of medical knowledge into data-driven models is not only desirable, but it is also far easier and less subjective than constructing the whole BN with a priori knowledge. In this spirit, most edges were directed according to plausible causal inference although interpretation of edges as carriers of information does not necessarily imply causation.
Conclusions
Thirty-four predictors related to lifestyle, adiposity, body fat distribution, blood lipids and adipocyte sizes have been considered as potential correlates of visceral obesity in women. The analysis was performed with a novel scalable and effective constraint-based bayesian network structure learning algorithm called RHPC.
From a biological point of view, the present study confirms, among other interesting findings, that visceral fat is the predominant predictor of triglyceride levels in obese individuals. It is reassuring that an unsupervised BN analysis uncovered previously established relationships between visceral fat, blood pressure, aging and triglyceride levels. The advantage of BN method is not that it will identify the "true causes", but rather that it will perform initial data exploration to unearth new knowledge in a semi-automated and rapid fashion.
In conclusion, we suggest that BNs are valuable data mining tools for the analysis of clinical data. In addition, BNs can explicitly combine both expert knowledge from the field and information studied from the data. A need for such multi-step processes (hypothesis generation step followed by a traditional hypothesis testing step) is essential. Finally, an extension to our existing framework would be to consider Bayesian model averaging as an alternative to a single consensus model selection. This extension is currently underway.
The Recursive Hybrid Parents and Children algorithm
RHPC is based on the faithfulness assumption. As RHPC calls HPC on each node, we start discussing HPC first. HPC receives a node X and returns its adjacent nodes PC X . Under this faithfulness assumption, X and Y are not adjacent in if and only if ∃ Z U\{X, Y} such that X ⊥ Y|Z [7]. As an exhaustive search of Z is intractable for high dimension data sets. HPC perfoms a heuristic search with a severe restriction on the maximum conditioning size in order to significantly increase the reliability of the statistical independence tests. Note that other similar 'Parent and Children' learning procedures were proposed recently in the machine learning literature, namely MMPC [12] and GetPC [10]. They could be used as well. Nonetheless HPC was favored in a recent evaluation using the same conditional independence test, over a range of different networks, sample sizes and number of variables [11].
Formally, HPC can be viewed as an ensemble method for combining many weak PC learners in an attempt to produce a stronger PC learner. The algorithm was designed in order to endow the search procedure with the ability to: 1) handle efficiently data sets with thousands of variables but comparably few instances; 2) deal with datasets which present some deterministic relationships among the variables; 3) be correct under the faithfulness condition; and 4) be able to learn large neighborhoods. HPC is based on three subroutines: Data-Efficient Parents and Children Superset (DE-PCS), Data-Efficient Spouses Superset (DE-SPS), and Interleaved Incremental Association Parents and Children (Inter-IAPC), a weak PC learner based on Inter-IAMB [36] that requires little computation. HPC was shown to be correct in the sample limit under the faithfulness assumption [11,18]. For the sake of conciseness, we only discuss the main HPC routine. The algorithm details are omitted here for brevity: RHPC and its sub-routines are thoroughly described in additional file 1 for the sake of conciseness.
HPC may be thought of as a way to compensate for the large number of false negative nodes, at the output of the weak PC learner with few data cases, by performing extra computations. HPC receives a target node T, a data set and a set of variables U as input and returns an estimation of PC T . It is hybrid in that it combines the benefits of incremental and divide-and-conquer methods. The procedure starts by extracting a superset PCS T of PC T (line 1) and a superset SPS T of SP T (line 2) with a severe restriction on the maximum conditioning size (Z <= 2) in order to significantly increase the reliability of the tests. A first candidate PC set is then obtained by running the weak PC learner on PCS T ∪ SPS T (line 3). The key idea is the decentralized search at lines 4-8 that includes, in the candidate PC set, all variables in the superset PCS T ∪ SPS T that have T in their vicinity. Note that, in theory, X is in the output of Inter-IAPC(Y) if and only if Y is in the output of Inter-IAPC(X). However, in practice, this may not always be true, due to the statistical test errors that should appear, especially with few data samples. The decentralized search enables the algorithm to handle large neighborhoods while still being correct under faithfulness condition. The essential graph is obtained by running HPC on the every node and by directing the compelled edges as shown in RHPC. Note that HPC must have found dSep (X, Y ) (at line 5 of RHPC) and have cached it for later retrieval. Alternatively, HPC can be run recursively on the adjacent nodes of a target variable in order to establish a local graph without having to construct the whole BN first as discussed in [2]. RHPC applies standard techniques at lines 4-19 to identify the compelled edges. The reader is directed to [7], pp. 538, for further details. The correctness and completeness of the edge orientation in RHPC are demonstrated in [37].
Network aggregation
As discussed in the introduction, our practical goal is to extract a BN structure that encodes the conditional independencies between 34 variables given our sample of 150 women. The most common approach to discovering the structure is to use learning with model selection to provide us with a single model. However, model selection is known to be sensitive to the particular data set, especially with few instances. Had we sampled another data set of the same size from the same distribution, model selection would have learned a different model [16]. So we cannot simply accept our chosen structure as a true representation of the under-lying distribution. Averaging over the sampled structures that are generated by a sampling process produces models that are more robust, have greater confidence and place less reliance on a single dataset. Several approaches exist: generating samples of the BN structure from its marginal posterior distribution using Monte Carlo Markov chain (MCMC) [16,[38][39][40], using bootstrapping methods for computing a statistical confidence features within a BN [6,16]. In this study, we make use of the bootstrapping method to generate a more robust network structure. The 're-shuffled' dataset is generated from the original dataset (re-sampling with replacement), the graph is built from this re-shuffled set and then the procedure is repeated a sufficient number of times. Confidence in a particular edge is measured as a percentage of the number of times this edge actually appears in the set of reconstructed graphs. If an edge has a confidence above the threshold, it is included in the consensus network. Thus, if dependencies have enough support in the bootstrapping process, they are captured and represented in the final consensus network. When computing confidence estimates, we define a feature as the existence of an edge between two nodes in the PDAG. Thus, the bootstrapped network has a confidence estimate assigned to each network edge. Where directed edges are present in a PDAG, they contribute only to the confidence estimate for the edge in that direction, whereas undirected edges contribute to the confidence estimate for an edge in both directions. If an edge has a confidence above the threshold, it is included in the consensus PDAG, and if edges are found in both directions (e.g. from node X i X j and X j X i ), then the edge is undirected. Thus, if directional dependencies have enough support in the bootstrapping process, they will be captured and represented in the final PDAG.
Biological data
The sample of 150 obese women used for these analyzes consists of 34 variables related to lifestyle such as alcohol consumption, smoking habits, leisure time activity and eating patterns. Dual energy x-ray absorptiometry was used to obtain whole-body measures of body composition (bone density, lean body mass, total body fat mass). Computed tomography was used to assess body fat distribution at the abdominal level. These measures include adipose tissue areas of the abdominal fat compartments located subcutaneously and inside the abdominal cavity (visceral fat). Finally, the variables examined also include average adipocyte sizes measured both in the omental (OM) and subcutaneous (SC) adipose tissue compartments from adipose tissue samples obtained during surgery. Women included in these analyses have been the object of previous publications on other topics [41,42]. All women who participated in the protocols signed an informed consent document. The projects were approved by the ethics committee of Laval University Medical Center.
Additional material
Additional file 1: Description of the Recursive Hybrid Parents and Children algorithm. This file contains a detailed discussion of our algorithm called Recursive Hybrid Parents and Children (RHPC). RHPC takes a data set as input and returns a partially oriented DAG (PDAG for short) representative of a bayesian network equivalence class. The latter is obtained by directing the compelled edges of the skeleton. The skeleton is obtained by running an algorithm called Hybrid Parents and Children (HPC) algorithm recursively on every node. RHPC is shown to be sound in the sample limit. mathematical framework and supervised the work. AA and AT wrote the manuscript. SR and SRM critically reviewed the manuscript. All authors read and approved the final manuscript.
==
Domain: Biology Medicine Computer Science
|
[{"added": "2014-10-01T00:00:00.000Z", "created": "2010-09-28T00:00:00.000", "id": "1265128", "metadata": {"extfieldsofstudy": ["Biology", "Medicine", "Computer Science"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://bmcbioinformatics.biomedcentral.com/track/pdf/10.1186/1471-2105-11-487", "pdf_hash": "542f0d48028827b184121912181d5c57ad674cb9", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:6371", "s2fieldsofstudy": ["Medicine"], "sha1": "542f0d48028827b184121912181d5c57ad674cb9", "year": 2010}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Novel injectable calcium phosphate scaffold with human periodontal ligament stem cell encapsulation in microbeads for bone regeneration
Objectives: 1) Develop a novel construct of human periodontal ligament stem cells (hPDLSCs) encapsulated in degradable alginate microbeads (DAMB) with human platelet lysate (hPL) and injectable calcium phosphate cement (ICPC); 2) Investigate the proliferation and osteogenic differentiation of hPDLSCs in ICPC with hPL as a xeno-free supplement and animal serum replacement for bone tissue engineering applications. Methods: hPDLSCs were encapsulated in alginate-fibrin microbeads (DAMB + fibrin), alginate-hPL degradable microbeads (DAMB + hPL), or alginate-fibrin-hPL microbeads (DAMB + fibrin + hPL). The proliferation and osteogenic differentiation of hPDLSCs were investigated in culturing with the ICPC scaffold. Results: Flexural strength of ICPC was 8.4 ± 0.91 MPa, and elastic modulus was 1.56 ± 0.1 GPa, exceeding those of cancellous bone. hPDLSCs had higher viability in DAMB + fibrin + hPL group than in DAMB + fibrin. ALP was 69.97 ± 16.96 mU/mg for ICPC + DAMB + fibrin + hPL group, higher than 30.68 ± 2.86 mU/mg of ICPC + DAMB + fibrin (p < 0.05) and 4.12 ± 1.65 mU/mg of control (p < 0.01). At 7 days, osteogenic gene expressions (ALP, RUNX2, COL1, and OPN) in ICPC + DAMB + fibrin + hPL and ICPC + DAMB + fibrin were 4–11 folds that of control. At 21 days, the hPDLSC-synthesized bone mineral amounts in ICPC + DAMB + fibrin + hPL and ICPC + DAMB + fibrin were 13.2 folds and 11.1 folds that of control group, respectively. Conclusion: The novel injectable CPC scaffold encapsulating hPDLSCs and hPL is promising to protect and deliver hPDLSCs. The hPL-based medium significantly enhanced the osteogenic differentiation of hPDLSCs in ICPC + DAMB + fibrin + hPL construct, suggesting a promising xeno-free approach for bone tissue regeneration applications.
Objectives: 1) Develop a novel construct of human periodontal ligament stem cells (hPDLSCs) encapsulated in degradable alginate microbeads (DAMB) with human platelet lysate (hPL) and injectable calcium phosphate cement (ICPC); 2) Investigate the proliferation and osteogenic differentiation of hPDLSCs in ICPC with hPL as a xeno-free supplement and animal serum replacement for bone tissue engineering applications.
Introduction
Bone tissue engineering involves the use of biomaterials, cells and bioactive factors and represents a promising approach for bone regeneration. Injectable calcium phosphate cement (ICPC) is used for bone repair due to their similarity to bone minerals (Szpalski et al., 2012a;Szpalski et al., 2012b;. The first ICPC was developed in 1986 and consisted of a mixture of tetracalcium phosphate [TTCP, Ca 4 (PO 4 ) 2 O] and dicalcium phosphate anhydrous (DCPA, CaHPO 4 ). ICPC was approved by the Food and Drug Administration (FDA) for the repair of craniofacial defects due to its excellent biocompatibility, osteoconductivity, resorbability, and injectability (Friedman et al., 1998).
In stem cell therapy, MSCs need to proliferate for an optimal therapeutic dose and differentiate into multiple cell lineages for effective tissue repair and clinical treatment. Therefore, the culture medium is essential to supply nutrients for cell metabolism, proliferation, and differentiation in vitro. Currently, cell culture medium supplements are based on animal serum, mostly fetal bovine serum (FBS), which consists of a large number of low and high molecular weight biomolecules with different physiological activities (Gstraunthaler, 2003;van der Valk et al., 2018). However, it has reported the FBS could evoke the xenogeneic immunological reactions and raise the risk of transmitting bovine infections (Brunner et al., 2010). As ICPC have been widely investigated in a number of clinical trials for bone regeneration, ICPC scaffolds seeding with MSCs represent a new strategy to promote prevascularization and enhance bone tissue engineering efficacy (Lin et al., 2019). Therefore, it is urgently needed to switch the culture medium to a FBS-free, xeno-free alternative, thus enable the MSCs seeding ICPC product not cause any potential infections, allergies or malignancies and ultimately facilitate the clinical application.
Human platelet lysate (hPL) is a valuable, non-xenogenic alternative to FBS in cell culture (Bernardo et al., 2007;Rauch et al., 2011). The advantages of hPL includes that it is humanderived, serum-free, and cost-effective. An increasing number of studies demonstrated that hPL showed a growth-promoting effect on a multitude of cell lines (Fernandez-Rebollo et al., 2017). Among the various types of MSCs, periodontal ligament stem cells (hPDLSCs) play a key role in dental and periodontal tissue regeneration. Seo et al. (2004) identified hPDLSCs in the periodontal ligament (PDL) and demonstrated that hPDLSCs implanted into nude mice generated cementum/PDL-like structures that resembled the native PDL. Compared with other MSCs, hPDLSCs have a marked ability to generate multiple tissue types, including alveolar bone, cementum, and Sharpey's fibers, making them an ideal source for cell-based dental therapy. However, there has no report that investigated the effect of hPL as a cell growth supplement on the proliferation and osteogenic differentiation of hPDLSCs in culture with ICPC.
In stem cell-based bone tissue engineering, the number of functioning cells in the bone injury site appears to be of critical importance. However, it was reported that few transplanted cells survived via local administration of cell suspensions. One study demonstrated that only 50% of initial donor BMSCs remained alive 48 h after implantation, and only 5% survived after 8 weeks (Giannoni et al., 2010). Another study showed that less than 1% of human multipotent stromal cells were detectable after 30 days post-implantation (Becquart et al., 2012). Thus the preferred approach was delivering a large number of cells to the target site via a carrier scaffold. Alginate hydrogel is highly hydrated with excellent biocompatibility and has shown great potential as a carrier for cell delivery (Izeia et al., 2020;Xu et al., 2021). Previous studies have reported that alginate microbeads are an ideal vehicle to localize the cells at the target defect site for cellbased therapy (Li et al., 2012;Pal et al., 2014). In addition, it has been reported that the incorporation of hPL could improve the viability, adhesion, and proliferation of a various types of cell in the alginate hydrogel (Sandri et al., 2015;Saporito et al., 2019).
However, there has been no report that investigated the alginate hydrogel incorporation of hPL as a cell delivery system with ICPC for bone tissue regeneration.
The excellent biocompatibility of ICPC renders it suitable for cell and growth factor delivery. hPL has been used as a cell nutrient supplement for cell expansion, as well as a source of growth factors for tissue engineering applications. To date, a literature search revealed no report on the development of injectable hPDLSC-hPL-CPC-microbead constructs for bone tissue engineering. There has been no report on hPLcontaining degradable microbeads as a vehicle to deliver hPDLSCs. There has been no report of investigating hPL as a cell growth supplement for hPDLSCs in culture with CPC, using the FBS-free and xeno-free cell culture medium alternative to avoid the risk of transmitting bovine infections and the initiation of xenogeneic immunological reactions.
Therefore, the objectives of the present study were to: 1) Develop a novel construct of hPDLSCs encapsulated in hydrogel microbeads with hPL and ICPC; 2) Investigate the proliferation and osteogenic differentiation of hPDLSCs in microbeads in ICPC with hPL as the xeno-free supplement for bone regeneration. The following hypotheses were tested: 1) Encapsulating hPDLSCs in alginate microbeads would protect the hPDLSCs during the setting reaction of ICPC paste; 2) The incorporation of hPL could improve the viability of hPDLSCs in the alginate microbeads; 3) The hPDLSCs released from the degradable alginate microbeads in ICPC would be able to proliferate and undergo osteogenic differentiate when using a xeno-free growth medium with hPL.
Fabrication of injectable calcium phosphate cement scaffold
The ICPC powder consisted of a mixture of TTCP and DCPA at a TTCP: DCPA molar ratio of 1:1 because this ratio was shown to produce apatite minerals (Xu and Simon, 2005). The TTCP powder was synthesized from a solid-state reaction between CaHPO 4 and CaCO 3 (Baker Chemical Company, NJ, United States) and then ground to obtain a median particle size of 17 μm. The DCPA powder (Baker Chemical) was ground to obtain a median particle size of 1 μm. Chitosan was used as cement liquid as previous studies have indicated that the addition of chitosan to CPC enhanced the strength and durability of ICPC (Weir and Xu, 2008). Two cement liquids at 0 and 7.5% chitosan were used. The ICPC liquid consisted of chitosan malate mixed with sterile distilled water at a chitosan/(chitosan + water) mass fraction of 7.5%. The ICPC powder was mixed with the chitosan liquid at a mass ratio of 2:1. Each paste was placed into 3 mm 3 × 4 mm 3 × 25 mm 3 stainless steel molds to make flexural specimens. Each specimen was covered with a glass slide on each side, clamped, and incubated in a humidor with 100% relative humidity at 37°C for 24 h.
Load-bearing properties
The hardened specimens were demolded and immersed in distilled water for 24 h. A three-point flexural test was used to fracture the specimens in a universal Testing Machine (MTS, Eden Prairie, MN). Three-point bending strength S = 3F max L/ (2 bh 2 ), where F max is the maximum load on the loaddisplacement (F-d) curve, L is the span, b is the specimen width and h is the thickness. Elastic modulus E = (F/d) (L 3 / [4 bh 3 ]), where load F divided by displacement d is the slope (Xu et al., 2002).
hPDLSC isolation from extracted human teeth hPDLSCs were isolated from the PDL tissues on extracted human premolars collected from healthy patients. The procedures were approved by the Institutional Review Board of the University of Maryland Baltimore (HP-00052180). All patients or their guardians were informed with written consent. The PDL tissue was scraped off the middle third of the root surfaces (Chen et al., 2020). The scraped tissues were digested in a solution of 3 mg/ml collagenase I (Worthington Biochem, Freehold, NJ, United States) and 4 mg/ml dispase (Roche, Mannheim, Germany) for 1 h at 37°C in a humidified atmosphere with 5% CO 2 . Then the PDL samples were placed into culture dishes with growth medium. The medium consisted of Dulbecco's modified Eagle's medium (DMEM, GIBCO BRL, Grand Island, NY, United States) supplemented with 20% fetal bovine serum (FBS, Invitrogen, Carlsbad, CA, United States) and 1% penicillin/streptomycin (P. S, GIBCO BRL). The samples were incubated at 37°C with 5% CO 2 . Single cells were observed 3 days later, and cell colonies were formed at 7-10 days. The individual cell colonies were digested to a single cell suspension using filter paper (Whatman, TISCH Scientific, North Bend, Ohio, United States) with 0.25% Trypsin-EDTA (GIBCO BRL), and transferred to 24-well culture plates, and the cells were passaged when they reach to 80% confluency. It was shown in our previous study that the hPDLSCs were positive for STRO-1, CD146 + , and OCT4 and negative for CD34 and CD45 (Chen et al., 2020). Passage 3 cells were used in subsequent experiments.
hPDLSC encapsulation in alginate microbeads
Alginate was made degradable by oxidation at 7.5%, following a method described in a previous study Frontiers in Materials frontiersin.org 03 (Bouhadir et al., 2001). Three types of alginate-based microbeads were fabricated to encapsulate the hPDLSCs: 1) Alginate-fibrin microbeads (DAMB + fibrin); 2) Alginate-hPL microbeads (DAMB + hPL); 3) Alginate-fibrin-hPL microbeads (DAMB + fibrin + hPL). A 1.2% (mass fraction) sodium alginate solution was prepared by dissolving oxidized alginate in saline (155 mmol/L NaCl) containing 0.1% fibrinogen for preparing type-1 microbeads. For type 2 microbeads, oxidized alginate was dissolved by 50% dilution of hPL with saline. For type 3 microbeads, fibrinogen was added at a concentration of 0.1% to the alginate-hPL solution. hPDLSCs were encapsulated at a density of 100,000 cells/mL of alginate solution. The alginate-cell solution was loaded into a syringe that was connected to a bead-generating device (Var J). Nitrogen gas was fed to the gas inlet and a pressure of 8 psi was established to form a coaxial airflow to break up the alginate droplets (Zhou and Xu, 2011). For type 1 and type 3 microbeads, the droplets fell into a well of 6-well plate containing 8 ml of 200 mmol/L calcium chloride solution plus 1 NIH units/mL of thrombin (Millipore Sigma) and cross-linked for 20 min to form DAMB + fibrin and DAMB + fibrin + hPL. For DAMB + hPL, the droplets fell into a well of 200 mmol/L calcium chloride solution and crosslinked for 20 min to form microbeads. A microscope (Eclipse TE-2000S, Nikon, Melville, NY) was used to measure the size of the microbeads.
Viability of hPDLSCs inside alginate microbeads
The ICPC paste was extruded through a sterile syringe into a 12 well-plate with 0.05 ml of paste per well. The cement was set in an incubator for 30 min. The hPDLSC-encapsulating microbeads were suspended in the culture medium and added into the 24 well-plate containing CPC for co-culture. The viability and proliferation of hPDLCSs were tested in four groups: 1. ICPC + DAMB + fibrin + DMEM group (Alginate-fibrin microbeads encapsulating 1×10 5 hPDLSCs with 0.05 ml ICPC scaffold in DMEM); 2. ICPC + DAMB + fibrin + FBS group (Alginate-fibrin microbeads encapsulating 1 × 10 5 hPDLSCs with 0.05 ml ICPC scaffold in growth medium containing 10% FBS); 3. ICPC + DAMB + hPL group (Alginate-hPL microbeads encapsulating with 1 × 10 5 hPDLSCs with 0.05 ml ICPC scaffold, cultured in DMEM for the first three days and then in growth medium containing 2.5% hPL); 4. ICPC + DAMB + fibrin + hPL group (Alginate-fibrin-hPL microbeads encapsulating with 1 × 10 5 hPDLSCs with 0.05 ml ICPC scaffold, cultured in DMEM for the first three days and then in growth medium containing 2.5% hPL).
After culturing for 1, 7, and 14 days, cells were stained with a live/dead kit (Invitrogen) and observed via epifluorescence microscopy (Eclipse TE-2000S, Nikon). Two images were taken at random locations for each sample, with four samples yielding 8 images at each time point for each group. Live and dead cells were counted via Image-Pro Plus software. The percentages of live cells (P Live ) were calculated. P Live = number of live cells/(number of live cells + number of dead cells).
The proliferation rate of hPDLSCs was investigated via Cell Counting Kit-8 (CCK-8, Dojindo, Tokyo, Japan), following the manufacturer's protocol. At 1, 3, 7, and 14 days, the culture medium was carefully removed and placed with 500 µl DMEM containing 50 µl CCK-8 dye. After 2 h incubation at 37°C, a 200 µl aliquot from each well was placed in a 96-well plate and the absorbance at an optical density of 450 nm (OD450 nm) was measured with a microplate reader (SpectraMax M5, Molecular Devices, Sunnyvale, CA, United States).
To evaluate the number of hPDLSCs released from the degrading alginate microbeads, the samples were imaged via microscopy (Eclipse TE-2000S, Nikon). Two images were taken at random locations for each sample, with four samples yielding 8 images at each time point for each group. The number of the hPDLSCs with spindle-shape morphology was counted via Image-Pro Plus software. The density of the released hPDLSCs was calculated: D Release = N Release /A, where A is the area of the view field for D Release .
At 1, 7, and 14 days, cells were lysed in a 0.2% Triton X-100 (Millipore Sigma) solution. The ALP activity of the cell lysate was measured by using an ALP Assay kit (QuantiChrom, BioAssay Systems, Hayward, CA, United States) with p-Nitrophenylphosphate (pNPP) as a substrate. The ALP activity was determined by measuring the absorbance at an optical density of 405 nm using a microplate reader (SpectraMax M5). The protein of cell lysate was quantified using Protein Assay Kit (Pierce BCA, Thermo Scientific, Rockford, IL, United States) following the manufacturer's protocol. The ALP activity was normalized to the protein amount and reported as mU/mg protein (Zhao et al., 2019).
The total cellular RNA of the cells was extracted using TRIzol reagent (Invitrogen) and reverse-transcribed into cDNA using a High-capacity cDNA Reverse Transcription kit (Applied Biosystems) in a thermal cycler (GenAmp PCR 2720, Applied Biosystems). RT 2 SYBR ® Green ROX qPCR Mastermix (Qiagen, Germantown, MD, United States) was used to quantify the transcript levels of glyceraldehyde 3phosphate dehydrogenase (GAPDH), alkaline phosphatase (ALP), runt-related transcription factor-2 (RUNX2), Collagen type 1 (COL1), and Osteogenicpontin (OPN). The human-specific primers were synthesized commercially (Millipore Sigma), and the sequences of the primers are listed in Table 1. The qPCR data collection and analyses were performed using a QuantStudio ™ 3 Real-Time PCR System (Thermo Scientific). Relative expressions were calculated using the 2 −ΔΔCT method and normalized by the C t value of the housekeeping gene GAPDH. The C t value of hPDLSCs in the control group at 1 day served as the calibrator.
Bone mineral synthesis by hPDLSCs
At 1, 14, and 21 days, the hPDLSCs were fixed with 4% paraformaldehyde for 15 min and stained with ARS. After staining for 30 min, the ARS solution was removed and rinsed with distilled water to remove any loose ARS. The samples were imaged via microscopy (Eclipse TE-2000S, Nikon). For quantification, the stained mineralization was de-stained in 10% cetylpyridinium chloride (Millipore Sigma) for 30 min and the concentration was measured at an optical density of 652 nm using the microplate reader (SpectraMax M5). The ARS concentration of the control group at 1 day served as the calibrator.
Statistical analysis
One-way and two-way analyses of variance (ANOVA) were performed to detect the significant differences, followed by Tukey's test as a post hoc comparison. Statistical analyses were performed by SPSS 19.0 software (SPSS, Chicago, IL, United States) at an alpha of 0.05.
Load-bearing properties
The flexural strength and elastic modulus of the ICPC scaffolds are plotted in Figure 1. Flexural strength and elastic modulus of ICPC scaffolds was 8.4 ± 0.91 MPa and 1.56 ± 0.1 GPa, significantly higher than 6.4 ± 0.43 MPa and 1.12 ± 0.23 GPa of the control group, respectively (p < 0.05). This demonstrates that the incorporation of chitosan improved the mechanical properties of the ICPC scaffold. Both of these values are higher than those of cancellous bone reported in the literature (Damien and Parsons, 1991).
Representative live/dead staining images from 1 to 14 days are shown in Figures 2(A-F). At 1 day, there were numerous live cells (stained green) and a few dead cells (stained red) in all groups (Figures 2A-E). At 7 days, the hPDLSCs were released from the degradable microbeads in the ICPC + DAMB + fibrin + hPL, ICPC + DAMB + hPL, and ICPC + DAMB + fibrin + FBS. The cells exhibited a polygonal morphology and a relatively high viability ( Figures 2F-H). At 14 days, the degradation of microbeads accelerated and most of the cells were released ( Figures 2J-L). The PLive increased with time due to proliferation in the ICPC + DAMB + fibrin + hPL, ICPC + DAMB + hPL, and ICPC + DAMB + fibrin + FBS. At 1 day, the PLive of hPDLSCs in ICPC + DAMB + fibrin + hPL and ICPC + DAMB + hPL was slightly higher than that in the ICPC + DAMB + fibrin + FBS group, indicating that incorporation of hPL into the microbeads could better protect the hPDLSCs from the pH change and ion activities during the ICPC setting ( Figure 2M). In contrast, very few cells were observed to attach and spread in the ICPC + DAMB + fibrin + DMEM, the PLive of the hPDLSCs was lower, due to the lack of nutrition supplements.
The cell growth rate was evaluated via the CCK-8 assay ( Figure 2N). At 1 day and 3 days, the proliferation of hPDLSCs was low in the groups of ICPC + DAMB + fibrin, ICPC + DAMB + hPL, and ICPC + DAMB + fibrin + hPL. At 7 days, as more cells were released from the degradable microbeads, the cell growth was significantly enhanced. The proliferation rate in the ICPC + DAMB + fibrin and ICPC + DAMB + fibrin + hPL groups was higher than that of the ICPC + DAMB + hPL group (p < 0.05), indicating that the incorporation of fibrin accelerated the hPDLSCs migrated out from the microbeads. The absorbance of the hPDLSCs in the ICPC + DAMB + fibrin + DMEM group was lower over time, indicating that the DMEM did not support the proliferation of the hPDLSCs encapsulated in the microbeads.
FIGURE 1
Load-bearing property of scaffolds in ICPC + Chitosan group and control group: (A) flexural strength; (B) elastic modulus (mean ± sd; n = 4). Values for cancellous bone were obtained from the literature (Damien and Parsons, 1991).
Frontiers in Materials frontiersin.org 3A-F). The D Release of ICPC + DAMB + fibrin + hPL was significantly higher than that of ICPC + DAMB + hPL ( Figure 3G). This demonstrated that adding fibrin increased the degradation rate of the microbeads. As the microbeads degraded, cells were released, and the D Release of each group increased. There was no difference in D Release between each group at 14 days ( Figure 3G).
Alkaline phosphatase activity of hPDLSCs
The ALP activity of hPDLSCs encapsulated in various types of alginate-based microbeads increased with time in osteogenic medium supplemented with 10% FBS or 2.5% hPL, indicating successful differentiation towards osteoblasts (Figure 4). At 14 days, the ALP activity was 69.97 ± 16.96 mU/mg for the ICPC + DAMB + fibrin + hPL group, significantly higher than 30.68 ± 2.86 mU/mg of the ICPC + DAMB + fibrin group (p < 0.05) and 4.12 ± 1.65 mU/mg of the control group (p < 0.01). These results indicate that the hPL-based osteogenic medium significantly enhanced the osteogenic differentiation of the hPDLSCs in ICPC scaffold.
Alizarin Red staining of minerals synthesized by hPDLSCs
At 21 days of osteogenic differentiation, the hPDLSCs from the microbeads exhibited a high viability and proliferation. Figures 6A,D,G show that the hPDLSCs released from the microbeads significantly proliferated and adhered to ICPC scaffold. This indicates that the ICPC scaffold was cytocompatible.
Representative ARS staining images of bone mineral secretion by hPDLSCs in the 12-well plate with ICPC are shown in Figures 6B,E,H. The red staining of mineralized nodules formed by the hPDLSCs was denser in the ICPC + DAMB + fibrin + hPL + osteo group than that of the ICPC + DAMB + fibrin + FBS + osteo group Figures 6C,F,I. The synthesized bone mineral in ICPC + DAMB + fibrin + FBS + osteo and ICPC + DAMB + fibrin + hPL + osteo increased with culture time from 1 to 21 days. The synthesized bone mineral amount in the ICPC + DAMB + fibrin + hPL + osteo was 6.9-fold and 13.2-fold that of the control group, at 14 days and 21 days, respectively (Figure 7). These results demonstrate that the hPLbased osteogenic medium significantly enhanced the osteogenic differentiation of the hPDLSCs.
Discussion
This study represents the first report to investigate degradable alginate hydrogel incorporated with hPL inside an
Frontiers in Materials
frontiersin.org 08 injectable calcium phosphate scaffold as a cell delivery system for bone regeneration. The hPDLSCs showed excellent viability while being encapsulated in the DAMB + fibrin + hPL microbeads in ICPC scaffold. When the setting of the ICPC was complete, the hPDLSCs were gradually released from the degradable hydrogel microbeads and grew rapidly in the scaffold. Furthermore, the osteogenic differentiation of hPDLSCs was significantly enhanced by the osteogenic medium supplemented with hPL. This study is schematically shown in Figure 8. These findings demonstrate a promising and novel xeno-free approach to delivering hPDLSCs in ICPC scaffold for bone regeneration.
The ICPC paste can be filled into a bone defect with an intimate adaptation to complex defect cavities and set in situ to form bioresorbable hydroxyapatite Xu et al., 2006). However, ionic activities and pH variations during the setting of ICPC paste could exert a cytotoxic effect on the cells (Matsuya et al., 2000;Simon et al., 2004;Przekora, 2019). TTCP and DCPA dissolved in the chitosan solution as Ca 2+ , PO 4 3-, and OH − ions , which then re-precipitated to form hydroxyapatite: 2Ca 4 (PO 4 ) 2 O+2CaHPO 4 → Ca 10 (PO 4 ) 6 (OH) 2 . It was reported that the pH during the setting of ICPC could be increased to approximately 10 (Simon et al., 2004). Therefore, there was a need to protect the cells from the ICPC setting reaction.
Alginate is among the most common natural polymers for the encapsulation and delivery of cells because of its many outstanding properties such as biocompatibility, gel-forming ability, non-toxicity, and ease of process (Gasperini et al., 2014;Izeia et al., 2020;Xu et al., 2021). However, the absence of alginate degrading enzyme (alginase) in the human body limits the degradability of the alginate. As the setting reaction of ICPC paste is largely complete after 1 day, it would be desirable for the alginate microbeads to quickly degrade, thus releasing the cells from the microbeads for enhanced cell viability and proliferation. A previous study reported that alginate could be chemically modified using oxidizing agents such as sodium periodate to produce oxidized alginate that was hydrolytically degradable Frontiers in Materials frontiersin.org (Zhou and Xu, 2011). In addition, it has been reported that adding fibrin could improve the attachment and proliferation of the cells in the alginate hydrogel, as well as to accelerate the release of cells from the alginate-based microbeads (Liu et al., 2013b). Fibrinogen was converted to fibrin via the mediation of thrombin and self-assembles into fibrin mesh, providing cell binding sites for cell attachment, migration, and proliferation (Li et al., 2015). As the cells migrated out of the microbeads, the porosity of the hydrogel increased, thus accelerating the degradation of the hydrogel. A previous study showed that the cell-encapsulating-alginate-fibrin-microbeads had diameters of about 100-500 μm, and the microbeads started to degrade at day 4 and released the encapsulated hiPSCs (Wang et al., 2016). Another study encapsulated hUCMSCs into the oxidized alginate-fibrin microbeads of about 300 μm in sizes (Chen et al., 2012). The microbeads started to release the cells at 4 days and completely degraded at 21 days (Chen et al., 2012). Therefore, microbeads with sizes of several hundred microns were suitable for injection, and were quickly degraded to release the cells.
FBS is the most widely-used supplement for the cultivation and expansion of eukaryotic cells. FBS contains essential components for cell growth and maintenance such as hormones, vitamins, transport proteins, trace elements, spreading, and growth factors (Brunner et al., 2010), which is effective on most types of human and animal cells. However, FBS-expanded MSCs could evoke immune responses against xenogenic serum antigens in the human body (Spees et al., 2004;Tonti and Mannello, 2008). In addition, FBS is a potential source of microbial contaminants such as fungi, bacteria, viruses, or prions. Indeed, a 20%-50% contamination rate of the virus was reported for FBS (Even et al., 2006). Furthermore, the ingredients of FBS are not precisely defined, and lot-to-lot variability and unintended interactions with test substances can lead to unexpected or undesired outcomes in the clinical application (Jochems et al., 2002). Therefore, it is critically important and highly desirable to develop a FBS-free medium as a cell nutrition supplement for cell-based therapies.
The use of hPL for MSC expansion was first reported by Doucet et al., in 2005(Doucet et al., 2005. Since then, hPL has been proven as a viable alternative to FBS, enabling efficient propagation of MSC under animal serum-free conditions for clinical application. There are several advantages for hPL for cell growth supplement: 1) hPL can be easily obtained and produced hPDLSCs showed high proliferation in ICPC for 21 days (A,D,G). Representative ARS staining images of mineral synthesis by hPDLSCs with ICPC (B,E,H). Representative images of mineralized nodules formed by the hPDLSCs. More mineralized nodules were present in the ICPC + DAMB + fibrin + hPL + osteo group than in ICPC + DAMB + fibrin + FBS + osteo group and control group (C,F,I).
Frontiers in Materials frontiersin.org using freeze-thaw procedures (Dessels et al., 2016); 2) Because hPL is derived from humans, neither bovine viruses nor immune reactions against bovine proteins are a concern (Saury et al., 2018); 3) hPL can be used in autologous settings to minimize the risk of immunological reactions (Sánchez et al., 2003); 4) hPL is more efficient in terms of costs and proliferation rate than FBS for certain types of MSCs (Burnouf et al., 2016). In the present study, 2.5% hPL promoted the proliferation of hPDLSCs when compared with the 10% FBS. In a previous study, the use of 10% FBS showed a slightly higher proliferation of hPDLSCs as compared to 5% hPL. However, the difference in colony number was not statistically significant (Abuarqoub et al., 2019). In another study, hPDLSCs were cultured using a medium supplemented with 10% FBS, 5% PL + 5% FBS, or 10% PL. It was revealed that the media containing 5% PL + 5% FBS resulted in more significant stimulation of cell growth, when compared with those containing either 10% FBS or 10% PL. Furthermore, a tendency toward enhanced proliferation was exhibited in media containing 10% PL as compared with media containing 10% FBS (Wu et al., 2017). The differences of hPL concentration or hPDLSCs could be associated with the difference in the preparation method, the variability in the donors, and the storage conditions. Hence, the quality of hPL used by different laboratories reported in the literature may be somewhat different from each other (Christgau et al., 2006;Ogino et al., 2006;Schallmoser et al., 2007).
The implantation of ICPC scaffold with MSCs has achieved ectopic and orthotopic bone formation and critical-sized defect healing Xu et al., 2017). The seeded MSCs could directly deposit bone matrix minerals in the scaffold due to their osteogenic differentiation potential. The therapeutic benefit of the transplanted MSCs was associated with a paracrine mechanism that stimulated the recruitment of host cells. These host cells
FIGURE 7
Quantitative bone mineral synthesis by hPDLSCs. The hPDLSC-synthesized bone mineral in ICPC + DAMB + fibrin + hPL + osteo group and ICPC + DAMB + fibrin + FBS + osteo group was 13.2 folds and 11.1 folds that of the control group, respectively (mean ± sd; n = 6). Values with dissimilar letters are significantly different from each other (p < 0.05).
FIGURE 8
Schematic illustration showing the encapsulation of hPDLSCs in alginate microbeads incorporated with hPL and delivery with ICPC for bone regeneration.
Frontiers in Materials frontiersin.org included osteoblast progenitors, endothelial cells, and osteoclasts, which took over the responsibility of subsequent bone formation and remodeling (Wang et al., 2011;Wang J. et al., 2014). Interestingly, it has been revealed that MSCs could enhance bone repair by modulating the foreign body reaction to ICPC, attracting circulating monocytes, and inducing their differentiation into osteoclasts, thus favoring bone formation (Gamblin et al., 2014).
In the present study, the hPDLSCs in ICPC were gradually released from the degradable hydrogel microbeads and underwent differentiation into osteogenic lineage by the hPL-based medium, without exposure to the animal serum. Moreover, higher ALP activity, osteogenic expression, and bone mineralization were achieved in ICPC + DAMB + fibrin + hPL construct than in ICPC + DAMB + fibrin + FBS construct. Therefore, the novel hPDLSC-hPL-microbeads-ICPC construct is a highly promising xeno-free approach for bone regeneration. Future in vivo studies are needed to evaluate the bone regenerative capacity of the hPDLSC-hPL-microbeads-ICPC construct for the treatment of bone defects in animal models.
Conclusion
This study demonstrated for the first time the hPDLSCencapsulation in degradable alginate hydrogel microbeads with hPL inside an injectable calcium phosphate scaffold for bone regeneration. The ICPC scaffold was biocompatible, mechanically load-bearing, while supporting hPDLSC attachment, proliferation, and osteogenic differentiation with the hPL as xeno-free cell supplement. The microbeads incorporating with hPL protected the hPDLSCs from the setting reaction of ICPC. The encapsulated hPDLSCs in ICPC with hPL-based osteogenic medium underwent successful differentiation into the osteoblast lineage, with highly elevated ALP, RUNX2, COL1, and OPN as well as bone mineral synthesis. Therefore, the novel hPDLSC-hPLmicrobeads-ICPC construct is highly promising for bone regeneration without the risk of infection from unknown pathogens by using animal-origin serums.
Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.
Author contributions
Study conception and design: GQ, MH, JL, and HX Acquisition of data: GQ, TM, and DK Analysis and interpretation of data: GQ, TM, MW, AS, and TO Critical revision: GQ, PW, LZ, HX, and YX.
Funding
This study was supported by National Natural Science Foundation of China 31771051 (LZ) and National Institutes of Health grant R21 DE029611 (AS and HX).
==
Domain: Biology Medicine Engineering
|
[{"added": "2022-11-18T15:01:41.510Z", "created": "2022-11-16T00:00:00.000", "id": "253599070", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.frontiersin.org/articles/10.3389/fmats.2022.977853/pdf", "pdf_hash": "f07726259424b63b6d06a5ee1fa3f4a5b3a5d614", "pdf_src": "Frontier", "provenance": "peS2o-0028.json.gz:575", "s2fieldsofstudy": ["Biology", "Medicine", "Engineering"], "sha1": "f07726259424b63b6d06a5ee1fa3f4a5b3a5d614", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Effect of honeybee venom and Egyptian propolis on the honeybee ( Apis mellifera L.) health in vivo
Background: Honeybees are one of the most important pollinators in the world, and their products are nowadays included in most anticancer, antiallergic, antimicrobial drugs and are included in cosmetic treatments. In the present study, honeybee venom and Egyptian ethanolic propolis extract (EP) were focused to test their effect on health and some genes for honeybee workers (defensin2, abaecin, hymenoptaecin, vitellogenin, and juvenile hormone esterase). Results: Honeybee venom and Egyptian propolis extract (EP) were used as supplements in the nutrition with differ‑ ent concentrations in Varroa mites‑infected colonies to measure the colonies’ activities after treatment. The immune‑ related genes and antimicrobial peptides (AMPs) were evaluated by using qRT‑PCR. Treated colonies with HBV and EP showed up‑regulation of immune and immune‑related genes’ expressions and increased the life span, activities and their density of bee workers. The data illustrated that the highest gene expression fold of juvenile hormone esterase was detected in the treated colonies with Egyptian ethanolic propolis extract (EP), while the highest vitellogenin expression fold in treated colonies was with honeybee venom. The up‑regulation of antimicrobial peptides occurred in colonies with both treatments. Conclusion: The findings suggest that honeybee venom and Egyptian ethanolic propolis extract (EP) could be used as potential supplements, even at the lowest concentration to develop the immunity of worker bees to increase their efficiency and prevent loss of honeybee colonies due to several diseases closely associated with Varroa mites’ infec‑ tions that cause sudden death.
Background
Honeybees (Apis mellifera L.) are significant pollinators, and the importance of honeybee products (venom, wax, royal jelly, propolis, honey, and pollen grains) has been demonstrated in previous research that used pollen grains and royal jelly to treat colonies .
Honeybee viruses, such as deformed wing virus (DWV), kakugo virus (KV), Varroa destructor virus-1 (VDV-1), black queen cell virus (BQCV), recombinant virus (VDV-1/DWV), acute bee paralysis virus (ABPV), slow bee paralysis virus (SBPV) and others, are one of the main drivers of colony losses. Several of these viruses are vectored by mite parasites, such as Varroa destructor (De Grandi-Hoffman and Chen 2015; Beaurepaire et al. 2017), and typically affect all stages of honeybee development. Some Egyptian apiaries plague with Nosema sp. and viral infections, which reduce honeybee immunity (Abd-El-Samie et al. 2021). Honeybee viruses participated in colony collapse disorder all over the world; the viruses are detected using Page 2 of 10 Seyam et al. Egyptian Journal of Biological Pest Control (2022) 32:78 RT-PCR (Cox-Foster et al. 2007;Abd-El-Samie et al. 2021). To fight against invading microbes, insects rely on their immunity. The generation of immune effectors, antimicrobial peptides (AMPs), is a key component of humoral immunity (Lemaitre and Hoffmann 2007). AMP; abaecin is a key immune effector of honey bee involved in the response to infection by multiple parasites (Evans and Pettis 2005;Evans et al. 2006) and has been shown to have significant heritable variation in its expression (Decanini et al. 2007). Defensin comes in two varieties: Defensin I, which is thought to have a role in social immunity, and Defensin 2, which is more likely to play a role in individual immunity (Ilyasov et al. 2013). The toll and immune deficiency (IMD) pathways are two different signal transduction cascades that produce AMPs. AMP can be activated for a brief period and transported to the infection site (Lemaitre and Hoffmann 2007;Schlüns and Crozier 2007). In terms of activity potency, AMPs are comparable to antibiotics, and they can be used to build antifungal and antibacterial medications (Mahlapuu et al. 2016). The transcriptional activation of immunity-related genes (IMRGs) is not only regulated by microbial infection, but also influenced by insect hormones; hormone levels are tightly regulated by multiple internal and external factors and juvenile hormone (JH) function relates to the regulation of the yolk protein vitellogenin (Vg) (Corona et al. 2007;Pandey and Bloch 2015). The use of antifungal and antibiotic medications to treat honeybee diseases causes suppression of their immune and the appearance of pathogens that are stable to them addition pollution of honeybee products. The solution to this challenge is to improve honeybee immunity by increasing the level of AMPs expression in the honeybees themselves (Casteels et al. 1993). The expression levels of three critical immune genes encoding the antimicrobial peptides abaecin, defensin 2, and hymenoptaecin were evaluated to acquire insight into the immunological mechanisms involved in resistance to these parasites. As a result, we resorted to using the most significant natural products, honey bee venom (HBV) and Egyptian propolis ethanolic extract (EP), to raise the honeybee's immunity. The composition and the activities of honeybee venom and propolis were approved previously and characterized by advanced techniques (Tanuwidjaja et al. 2021;Ghallab et al. 2021). This study aimed to determine the effect of honeybee venom and EP as supplements in the nutrition on the regulation of honey bee AMPs gene expression, increase honeybee resistance to pathogens, reduce sudden bee death that causes colony collapse disorder, and determine the efficacy of honeybee venom and EP treatments on the secretion of juvenile hormone and vitellogenin secretion.
Egyptian ethanolic extract propolis preparation (EP)
The propolis was collected from the apiary of El-Dokki Honeybee Research Department, Plant Protection Research Institute, Agriculture Research Center, Giza, Egypt; during the spring to autumn season of 2019 after collection, propolis was kept in the freezer at − 20 °C immediately to crush easily. 100 g of crushed frozen propolis was dissolved in 1 L of ethanol 80% and left in dark place until use (Ghallab et al. 2021).
Honeybee venom collection
Honeybee colony almost density 12,000 by electrical stimulation was used to collect 1 gm of venom according to Benton protocol with some modifications (Benton et al. 1963). Thus the modifications were the electrical apparatus was powered by 18 V to honeybee sting the glass with plastic foil of apparatus for 10 s; then, the honeybee venom was collected dried at 27 °C for 35 min and then kept frozen at − 20 °C until use.
Colony selection and management
In winter season from February 2019, the apiary formed from 21 asymptomatic infested with Varroa mites' colonies (Apis mellifera L.) with poor propolis in colonies at Honeybee Research Department, Plant Protection Research Institute, was selected for the experiment. All colonies with the same strength were used in this investigation (4 frames per colony consist of one honey and pollen + 3 frames sealed and unsealed brood). Their queens were replaced with 4-month-old new Carniolan queen hybrid. The total area of a sealed and unsealed of both sides of frames was measured by inch square weekly during the experimental period using the Langstroth frame. Twenty-one honeybee colonies were divided into 3 colonies for control, and eighteen honeybee colonies were divided into 2 groups for treatments with honeybee venom and Egyptian ethanolic propolis extract; each group was divided into 3 subgroups for each concentration of treatments. The concentrations of the EP treatment were 1, 3, and 5 g/L and the concentrations of honeybee venom were 0.25, 0.5, and 1 g/L. Both treatments were added in sugar syrup only; the honey bee venom was directly dissolved in the sugar solution yielding concentrations of 0.025, 0.05, and 0.1%. After extracting the propolis with alcohol, three different concentrations of propolis were taken; 10, 30, and 50 ml of the extract, equivalent to 1, 3, and 5 g of raw propolis, respectively, was directly added to the sugar solutions, yielding concentrations of 0.1, 0.3, and 0.5%, respectively. Feeding with different concentrations of treatments continued until November 2019, and all colonies were fed once a month.
Honeybee samples collection
Adult honeybee workers' samples were collected randomly from untreated and treated colonies, and 10 honeybees' workers were collected from inside of each colony; a total of 210 bees from all colonies were kept at − 20 °C until RNA extraction.
RNA extraction
Total RNA was extracted from 5 adult workers of honeybees using BIOZOL-BIOFLUX (Catalog No. 10760055-1) following the manufacturer's instructions. RNA quality and quantity were verified using NanoDrop 2000 (Thermo Scientific, USA). The cDNAs were synthesized from the extracted RNAs.
Reverse transcription and oligonucleotides synthesis
The cDNA synthesis was performed using Applied Biosystem kit (Cat. No. 10400745). A total of 5 immune genes of honeybees (Defensin2, Abaecin, Hymenoptaecin, Juvenile hormone esterase, and Vitellogenin) were measured, and B-actin using as a reference gene for normalization (Cunha et al. 2005). RT-PCR was performed using specific primer pairs listed in (Table 1) and was synthesized by Thermo Fisher Scientific (Invitrogen).
Conventional RT-PCR
RT-PCRs were run in the thermal cycle Techne Gene Amp. (PCR system FGENO2TD), and the thermal profile used for amplification of target cDNA by using Go Taq ® Green PCR Master (2X) # A9281, according to Promega's manufacturer's. The thermal conditions began with an initial denaturation at 95 °C for 3 min, followed by 35 cycles of 95 °C for 45 s, 50 °C, and 72 °C for 1 min, then the final extension step at 72 °C for 7 min, for all genes. PCR products were analyzed by gel electrophoresis on a 1.5% agarose gel following ethidium bromide staining (Sigma-Aldrich, E7637) and visualized by UV transilluminator (CUVP upland A, USA). ExoSAP-IT (Affymetrix, USA) kit was used to purify the PCR products following the manufacturer's instruction (Cat. No. 75001.1. EA).
Sequencing of PCR products
Sequencing reactions were performed using BigDye Terminator v3.1 Cycle Sequencing (Applied Biosystems, CA, USA) and analyzed using 3130XL genetic analyzer (Applied Biosystems, CA, USA). Sequence data were collected and analyzed using BioEdit software version 7.0.0 (Hall 1999) to confirm the identification of the amplified genes.
Gene expression
The reactions were occurred by using SYBR ™ Green PCR Master Mix (2X) kit (Cat. No. 4309155) as follows: 5 µl of Master Mix, 1 µl of each primer forward and 1 µl of each primer reverse (10 µM), 1 µl of cDNA and 2 µl of nuclease water with total volume 10 µl for each reaction for the detected genes. Melting curve analysis was performed following PCR amplification. The raw threshold cycle (Ct) values of selected gene transcripts were normalized to the Ct values of β-actin. Relative expression levels were calculated using the 2 −ΔΔCt method (Livak et al. 2001). Data were presented as mean ± SE. P values were calculated by t test. Values of P > 0.05 were considered statistically significant.
Statistical analysis
Results were expressed as the mean ±standard error (SE). Statistical significance between different samples of untreated and treated honeybee colonies was analyzed using one-way ANOVA. Statistical significance was defined as P < 0.01 and P < 0.001.
Colony selection and management
Twenty-one colonies were observed weekly every month from the beginning of therapy in February 2019 until November 2019. The number of Varroa mites in treated colonies and sealed-unsealed broods reduced month after month, compared to untreated colonies. The overall average brood density of all honeybee venom treated colonies was 1860, 2644, and 3564 inches for HBV concentrations of 0.25, 0.5, and 1.0 g/L, respectively, while it was 1686, 2341, and 3182 inches for all treated colonies with Egyptian ethanolic propolis extract (EP) concentrations of 1, 3, and 5 g/L, respectively. The brood in honeybee venom treated colonies was found to be higher than in Egyptian ethanolic propolis extract (EP) treated colonies. Therefore, the hive was raised to a second round, while untreated colonies remained the same (Fig. 1).
PCR products of immune genes
The PCR products with the expected fragments of 71 base pair (bp) fragment for defensin2, 48 bp for abaecin, 112 bp for hymenoptaecin, 290 bp for juvenile hormone esterase, and 790 bp for vitellogenin were generated from asymptomatic honeybees and visualized on 1.5% agarose gel electrophoresis (Fig. 2). The obtained PCR fragments were recovered and sequenced. The nucleotide sequences of the fore-mentioned immune and immune-related genes under study were submitted to GenBank and subjected to a homology search using BLASTX which confirm their identification Additional file 1: Tables and Figs. S1-S5.
Gene expression
The studied honeybee genes were successfully amplified in all cases, and the specificity of the amplified products was confirmed by single peaks in the melting curve analysis. Expression levels of defensin, abaecin, hymenoptaecin, juvenile hormone esterase, and vitellogenin in untreated and treated honeybee colonies were compared. In general, the genes' expressions increased in both treatments in honeybee colonies than untreated colonies. In Egyptian ethanolic propolis extract (EP) treatment, nonsignificant differences were obtained for defensin2 in the treated colonies with 1.0 g/L of propolis, the hymenoptaecin was slightly down-regulated, and abaecin gene was slightly increased than untreated colonies. Defensin2 and hymenoptaecin levels increased significantly in 3 g/L EP and then declined in 5 g/L EP (Fig. 3). On the other hand, the level of vitellogenin increased as EP concentrations increased. In comparison with EP treatment concentrations, vitellogenin was up-regulated in honeybee venom (HBV) treatment concentrations, especially at 1.0 g/L. Moreover, juvenile hormone esterase increased in EP than honeybee venom treatments and it was steadily upregulated as honeybee venom treatment concentrations increased (Fig. 4a, b). The antimicrobial peptide, abaecin expression folds were the same in 0.5 and 1.0 g/L honeybee venom treatments, but the expression folds of defen-sin2 and hymenoptaecin were lower in 0.5 g/L honeybee venom treatments than in other concentrations of honeybee venom treatments (Fig. 5). The highest concentration of honeybee venom (HBV) treatment (1 g/L) caused the highest rate of gene expression folds of Vg and abaecin by 3.94, 1.79, respectively, while the lowest concentration of honeybee venom treatment (0.25 g/L) caused the highest propolis treatment's highest concentration (5 g/L) increased the expression of JHE by 3.63 folds, while propolis treatment with concentration (3 g/L) increased the gene expression folds of defensin2 and hymenoptaecin by 2.83 and 1.53 folds, respectively, but had no effect on abaecin.
Discussion
This is the first trial to use honeybee venom (HBV) and Egyptian ethanolic propolis extract (EP) in honeybee colonies as a diet with sugar syrup. In the winter, honeybee venom from winter workers did not include toxins and the allergen Api m 12, commonly known as vitellogenin, was absent (Danneels et al. 2015). In addition, colonies with queens descended from crosses between high-propolis-producing colonies showed significantly increased brood and worker bee longevity (Nicodemo et al. 2014).
In this study, the results indicated rearing of the brood density in both the honeybee venom and EP treatments, suggesting that honeybee immunity had improved and they were better able to tolerate diseases. Honeybee venom is high in proteins and enzymes (Tanuwidjaja Ghallab et al. 2021). This is in agreement with previous researchers who reported that diet is crucial to an individual's and the colony's appropriate development (Bryś et al. 2021). Precocious foraging is linked to a protein and lipid deficiency (Toth and Robinson 2005), which accelerates the aging process and leads to a decrease in the colony population. Many pathogens infect honeybees and suppress their immune systems, including the deformed wing virus (DWV), neonicotinoids and microsporidian parasites of the genus Nosema sp. (Pluta and Sokol 2021), and one of the most important parasite is Varroa destructor. Recently, Ramsey et al. (2019) suggested that Varroa were exploiting the fat body which is an essential tissue to proper immune function. According to our observations, Varroa mites decreased in both honeybee venom and EP supplemental food-treated colonies. This could be explained by some venom on the cuticle of honeybees may be due to self-grooming movement, which is thought to be an immune fortification of social insects and EP in honeybee colonies as a diet with sugar syrup. In addition, propolis' chemical compositions, which included several volatile components (Asgharpour et al. 2020; Ghallab et al. 2021), may help to resist Varroa mites permanently, as seen in propolis-treated colonies. So, in the colonies natural protein nutritional supplements should be supplemented with additional additives to improve Life especially during the winter when blooms and pollen are few because malnutrition weakens honeybees' immune systems and makes workers more susceptible to infection Bryś et al. 2021). Consequently, due to Varroa mites reduction in both honeybee venom (HBV) and Egyptian ethanolic propolis extract (EP) supplemental food-treated colonies, related virus titers may be dropped in the treated colonies, indicating that EP and honeybee venom had indirect antiviral properties, as previously reported (Baracchi et al. 2011;Borba et al. 2015;Tanuwidjaja et al. 2021). Obtained findings are consistent with Pusceddu et al. (2021) who reported that propolis high in phenols was administered to brood cells to alter reproducing parasites and that propolis in the hive can have direct impacts on brood infections (Simone-Finstrom 2017), with a good effect on honeybees and the potential to improve Varroa mite mortality. In comparison with parasitized bees from untreated cells, Damiani et al. (2010) found out that the phenotype of mite-infested bees emerging from propolis-treated cells is associated with a decreased DWV load. As a result, propolis may act as an indirect inhibitor of mite-induced viral multiplication (Annoscia et al. 2019). In addition, honeybee venom contains an apidaecin peptide which is produced by the venom glands (Van Vaerenbergh et al. 2013). The expression of apidaecin by venom apparatus epithelial cells may protect individual honey bees from infections. Alternatively, its presence in the venom could play a role in the hive's social immunity (Baracchi and Turillazzi 2010;Van Vaerenbergh et al. 2013). Moreover, honeybee venom's major component is a melittin, which inhibits the infectivity of a wide range of viruses (Memariani et al. 2020). Moreover, the present research focused on increasing honeybee immunity, which is expressed in antimicrobial peptides (defensin2, abaecin, and hymenoptaecin) and related immune genes [vitellogenin (Vg) and juvenile hormone esterase (JHE)]. JHE or JHE-like genes have not been documented to be directly involved in the immune response in insects or crustaceans; nonetheless, rising evidence has indicated the presence of hormonal immunity regulation in insects (Zhu et al. 2018).
In the present study, the over-expression of the investigated antimicrobial peptides (AMPs) defensin2, hymenoptaecin, and abaecin in treated colonies with EP and HBV was detected by RT-qPCR gene expression, the key findings of this study were immunological, and immunerelated gene expression was up-regulated in treated colonies compared to untreated colonies. According to previous studies (Abd-El-Samie et al. 2021) different honeybee viruses had spread widely among the apiaries globally and the reproduction of these viruses resulted in the reduction of vitellogenin (Dainat et al. 2012). Because this protein (Vg) has pleiotropic effects (Amdam et al. 2007;Antúnez et al. 2013), they discovered that when the titer of viral, fungal, or pathogens increased in the colonies, the rate of Vg and AMPs decreased, leading to colony collapse disorder (Antúnez et al. 2009;Chaimanee et al. 2012), although Vg damages the microorganism cell wall and has anti-microbial activity (Park et al. 2018). The level of Vg and AMPs increased with increasing treatment concentrations of honeybee venom and EP in our data; however, it was found out that even the low concentration of honeybee venom (0.25 g/L) could increase the expression folds of target genes. This is most likely due to melittin, which is associated with the insect's immune system (Baracchi et al. 2011). Additionally, propolis treatment increased the expression folds of JHE and Vg gradually with increasing the concentration of EP treatment, while the best effect of EP concentrations was 3 g/L for AMPs, but did not effect on abaecin, which is similar to Borba et al. (2015). Simone-Finstrom et al. (2009) imply that propolis activates the cellular immune system rather than the humoral immune system. Additionally, honeybees with propolis in their colonies have greater immunity and health than those without propolis (Simon-Finstrom et al. 2017). Propolis stimulates highlevel expression of the immune system response in bees challenged with microorganisms (Turcatto et al. 2018). Only 0.1 percent propolis fed in a pollen substitute diet greatly increases activation of antimicrobial peptide genes (defensin-1, abaecin, hymenoptaecin, and apidaecin) in bees infected with Escherichia coli. So, applying propolis to inside of hives allowed to a lower investment in immune function by reducing immune gene expression in uninfected adult worker bees (Simone-Finstrom et al. 2009;Borba et al. 2015). On the other hand, immune genes (Lyzozyme-2 and -3, Defensin-1), were up-regulated by pollen feeding in healthy bees (Alaux et al. 2010). The present study indicated that supplemental honeybee venom treated colonies had higher brood density and activity than propolis-treated colonies as well as untreated colonies. Honeybee venom went much beyond the traditional stereotype of predator defense, suggesting that the diverse nesting biology of honeybee species may be linked to the employment of venom in a social immunity context (Baracchi et al. 2011). Assuming that the venom gland is the only source of some cuticular peptides and that bees are unable to select which peptides spread selectively on the cuticles, differences in cuticular profiles may be influenced by the length of time these compounds can remain on the insect bodies and the frequency with which venom is applied (Baracchi et al. 2011). Indeed, bee venom is present on the cuticle of adult bees, and on comb wax, it may act as a social antiseptic device (Baracchi and Turillazzi 2010). Honeybee venom is utilized as self-defense since it contains various antibacterial toxins, including melittin, PLA2, adolpanin, dopamine, and hyaluronidase (Park et al. 2014), which can boost honeybee immunity. Obtained results revealed that JHE was gradually elevated by both supplemental treatments of honeybee venom and EP. Its titer was increased with each increase in treatment concentration, especially the highest concentration (5 g/L) of EP gave the highest expression folds of JHE. Insect metamorphosis is only triggered when JH is metabolized by juvenile JHE (Bomtorin et al. 2014;Li et al. 2021). JH promotes metamorphosis and caste differentiation in honeybee larvae and is a very significant hormone in honey bees. JH hastens development (Pandey et al. 2020), and this is what was observed in this study due to the overexpression of JHE. The JH route was found to be usually down-regulated throughout larval development, indicating that its activity is hostile to the ecdysteroid pathway. Nonetheless, in drones, the genes involved in JH synthesis had a higher expression. Providing the supplement nutrition propolis in the sugar syrup as a diet may explain the increase in JHE in the honeybee body and density due to the acceleration of larval growth (Huang et al. 2014). The ability of the juvenile hormone to impact the expression of several ecdysteroids signaling genes suggests that crosstalk between the two hormones may be significant in the bee brain and behavior regulation (Pandey and Bloch 2015). Finally, the use of a large amount of either treatment of honeybee venom and EP may cause a defect in honey bee colonies as reported previously (BenVau and Nieh 2017;Aurori et al. 2021), and it may be costly for beekeepers, so using the smallest amounts and concentrations of both treatments, especially since the low concentrations produced satisfactory results, is recommended.
Conclusions
It could be concluded that supplementing honeybee nutrition with Egyptian ethanolic propolis extract (EP) and honeybee venom as well as a treatment in honeybee colonies can improve immunological response, honeybee density, and activity. Thus, saving honeybees from death and unexpected colony collapse by employing natural therapies in the smallest quantity and at the lowest cost can be done and also protecting beekeepers from annual losses due to Varroa mites' infestation, honeybee viruses, and other diseases.
==
Domain: Biology Environmental Science Medicine
|
[{"added": "2022-06-26T13:34:45.759Z", "created": "2022-06-25T00:00:00.000", "id": "250025807", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://ejbpc.springeropen.com/track/pdf/10.1186/s41938-022-00580-0", "pdf_hash": "236c591d021f9fa766387c61a2c1cdce81e39318", "pdf_src": "Anansi", "provenance": "peS2o-0028.json.gz:6907", "s2fieldsofstudy": ["Biology", "Medicine", "Environmental Science"], "sha1": "9393d607fb16f2e032fbc1aa6ee2175709e441e8", "year": 2022}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Distinguished biomimetic dECM system facilitates early detection of metastatic breast cancer cells
Abstract Breast cancer is the most prevalent malignant tumor affecting women's health. Bone is the most common distant metastatic organ, worsening the quality of life and increasing the mortality of patients. Early detection of breast cancer bone metastasis is urgent for halting disease progression and improving tumor prognosis. Recently, extracellular matrix (ECM) with biomimetic tissue niches opened a new avenue for tumor models in vitro. Here, we developed a biomimetic decellularized ECM (dECM) system to recapitulate bone niches at different situations, bone mimetic dECM from osteoblasts (BM‐ECM) and bone tumor mimetic dECM from osteosarcoma cells (OS‐ECM). The two kinds of dECMs exhibited distinct morphology, protein composition, and distribution. Interestingly, highly metastatic breast cancer cells tended to adhere and migrate on BM‐ECM, while lowly metastatic breast cancer cells preferred the OS‐ECM niche. Epithelial‐to‐mesenchymal transition was a potential mechanism to initiate the breast cancer cell migration on different biomimetic dECMs. Importantly, in the nude mice model, the dECM system captured metastatic breast cancer cells as early as 10 days after orthotopic transplantation in mammary gland pads, with higher signal on BM‐ECM than that on OS‐ECM. Collectively, the biomimetic dECM system might be a promising tumor model to distinguish the metastatic ability of breast cancer cells in vitro and to facilitate early detection of metastatic breast cancer cells in vivo, contributing to the diagnosis of breast cancer bone metastasis.
breast cancer cells and capture the early metastatic breast cancer cells in vivo. The dECM system will promote the early diagnosis of breast cancer bone metastasis, holding great potential for clinical translation.
| INTRODUCTION
Breast cancer was the most common malignant cancer affecting women, 1,2 while bone metastasis was the most frequent complication of it. 3More than 50% of breast cancer patients showed bone metastasis with serious bone events, such as pain, pathological fractures, nerve compression syndromes, and hypercalcemia. 4Most importantly, once the bone metastasis occurred, they were virtually incurable and resulted in high rate of mortality. 4,5Therefore, early detection of metastasis before tumor cell colonization in bone should be essential for the prognosis and the survival of breast cancer patients. Recently, many efforts have been made for the early diagnosis of breast cancer bone metastasis, especially the detection of circulating tumor cells (CTCs). 6Even though the disadvantages of CTCs still limit its application in clinic, including low numbers in blood, 7 low sensitivity, and specificity of CTCs biomarkers. Moreover, CTCs might exist in blood for a long time without colonization in bone tissue. 3,8e interactions between malignant cells and surrounding microenvironment played an essential role during cancer progression and metastasis. For distant organ metastasis of cancer, Stephen Paget first proposed the "seed-and-soil" hypothesis as early as 1889, and increasing evidences demonstrated the importance of the target organ microenvironment with the selective advantage of cells to grow. 9e extracellular matrix (ECM) was the major component of tissue microenvironment with crucial ECM proteins and three-dimensional (3D) network structure. It provided biochemical and biophysical signals as the "soil" of the organ microenvironment, deciding specific tumor cell activities. 10Some researchers reported that ECM from different tissues provided distinguished microenvironment for breast cancer metastasis. 11Moreover, the alteration of ECM components and structure with distinguished tissue niches was demonstrated to affect the cancer metastasis. For example, ECM protein HAPLN1 was significantly deficient in aging skin, resulting in the alteration of the matrix's arrangement structure and reduction of the skin's flexibility, compared to the ECM from young dermal fibroblasts. 12The altered ECM in aging skin significantly promotes the migration of melanoma cells with optical tissue microenvironment. 12Cooper et al. 13 reported that paclitaxel chemotherapy increased the secretion of LOX from CD8 T cells, which remodeled the ECM in the lung and promoted breast cancer metastasis.
Conventional metastasis surrogates (Transwell migration assays, wound healing) lacked the complex ECM components and structure, which were hard to clarify the real progression of cancer metastasis in vivo.
Development of biomimetic ECM scaffolds as in vitro tumor models facilitated cancer metastasis studies with the interaction of cancer cells and matrix microenvironment.
In the past decades, decellularized ECM (dECM) from tissues or cells as bioactive materials have been extensively applied for tissue engineering and regenerative medicine. For bone tissue engineering, such biomaterials were demonstrated to mimic the complex components and structure of native tissue in vitro and promoted bone regeneration. 14Our previous studies have generated such dECM scaffolds from both tissue (small intestinal submucosa) and cells (osteoblasts or fibroblasts).6][17] Compared with tissuederived ECM, cell-derived ECM showed several advantages, including faster, easier operation, economical, and more preservation of ECM composition because of less vigorous decellularization.
9][20][21] The advantages of dECM have attracted much attention in cancer research, including structural support, suitable microenvironment, and bioactive factors. 224][25] Decellularized normal porcine livers and lungs were used to generate biomatrix scaffolds as metastatic soil and were proved to be beneficial of cancer cell invasion, colonization, and proliferation with functional readouts regarding epithelial-tomesenchymal transition (EMT) and chemoresistance. 24 Piccoli et al. reported that the pathologic dECM from colorectal cancer decreased angiogenic potential compared with healthy tissue. 25Moreover, Lv et al. developed 3D dECM scaffolds with different stiffness to mimic the microenvironment of breast cancer, including matrix stiffness, components, and structure of ECM. Lysyl oxidase (LOX) expression levels in dECM were essential for the diverse stiffness of breast cancer matrix and drug resistance. 26Therefore, dECM from bone-associated cells might serve as a potent "soil" to study breast cancer bone metastasis.
The aim of the study was to develop a biomimetic dECM system to distinguish the metastatic ability of breast cancer cells and capture the highly metastatic breast cancer cells. Our previous studies have proved that dECM from osteoblasts with dense structure and high components of core matrisome proteins provided excellent bone mimetic microenvironment, 27,28 which was also introduced here to mimic normal bone matrix. Meanwhile, dECM derived from osteosarcoma cells was introduced to mimic a bone tumor matrix. The two kinds of dECM representing different bone microenvironments were applied as a system to distinguish the metastatic activities of breast cancer cells. Cell activities of high-metastatic breast cancer cells or low-metastatic breast cancer cells on different dECM biomaterials were examined in vitro. Besides, early detection of metastatic breast cancer cells with the distinguished biomimetic ECM system was assessed in vivo. MC3T3-E1 and MG63 were used to generate bone mimetic or bone tumor mimetic dECM microenvironment. We do not know yet how using male source-derived ECM may be different from female sources. The breast cancer cells 4T1 were from female mice, while MDA-MB-231, MCF7, and HCC1937 were from female human. The purpose of the distinguished biomimetic ECM system was to detect the potential risk of bone metastasis in breast cancer mainly occurred in women.
The MC3T3-E1 and MG63 cells were seeded on tissue culture plates at a density of 1 Â 10 4 /cm 2 . The cells were cultured for 5 days in complete media and for another 7 days with ascorbic acid (50 μg/ mL). During this period, the cells were changed with fresh media every other day. The decellularization process was performed as previously reported. 29Briefly, the media were discarded, and the cells were rinsed with phosphate-buffered saline (PBS) two times, followed by the treatment of pre-warmed 0.5% TritonX-100 in PBS at 37 C for 8 min. After the wash with PBS three times, the cells were frozen at À80 C for 4 h and thawed in pre-heated PBS at 37 C for 40 min.
Triplicate freeze/thaw (À80 C/37 C) cycles were performed. After that, the dECM samples were incubated in DNase (50 U/mL)/RNase (50 μg/mL) at 37 C for 2 h and rinsed with PBS three times. The dECM generated from MC3T3-E1 was used to mimic bone microenvironment, which was defined as bone mimetic ECM (BM-ECM). Meanwhile, the dECM generated from MG-63 was used to mimic bone tumor microenvironment, which was defined as osteosarcoma ECM (OS-ECM). Collagen coating was used as a negative control.
| Residual DNA assessment
DAPI staining and residual DNA quantification were used to evaluate the residual DNA which was an important index for dECM. For DAPI staining, the samples were fixed with 4% formaldehyde for 15 min and rinsed with PBS. DAPI solution (#4083, Cell Signaling Technology) at the concentration of 1 μg/mL was added, and the samples were incubated for 5 min at room temperature avoiding light. After PBS washing, the samples were coated with anti-fade mounting solution and imaged under a fluorescence microscope (Eclipse Ti2, Nikon).
The content of residual DNA was quantified by a cell proliferation assay kit (C7026, Invitrogen) according to the manual. Native cells were washed with PBS, frozen, and stored at À80 C. The frozen cells and dECMs were thawed at room temperature and incubated in CyQUANT ® GR dye/cell-lysis buffer for 2-5 min, protected from light. The samples were measured using a fluorescence microplate reader with filters for 480 nm excitation and 520 nm emission maxima. The standard curve was prepared with λDNA (provided by the kit). The DNA content of dECM divided by the DNA content of native cells was defined as relative DNA percentage.
Briefly, dECM was scrapped mechanically from 24-well plates and transferred into a 1.5-mL Eppendorf (EP) tube. The samples were washed with GENMEDA solution, followed by centrifugation at 300 g/min for 15 min. The collected ECM was dissolved in GENMED B solution, undergone strong vortex shaking, and incubated in a water bath at 56 C for 16 h, followed by a further water bath at 96 C for 10 min. After that, 5 μL sample and 100 μL GENMED C were mixed with a vortex for 15 s and incubated at room temperature for 30 min in darkness. The samples were centrifuged at 16,000 g/min for 10 min, and the precipitation was vibrated vigorously with GENMED D. The mixed solution was transferred into a 96-well plate and measured at the absorbance of 656 nm. The GAG content was calculated according to the standard curve prepared with standard GAG samples in the kit.
9][30] Briefly, dECM or cells in 24-well plates were rinsed with PBS three times and fixed in Kahle fixative (26.7% ethanol/3.7%formaldehyde/2% glacial acetic acid in distilled water). After that, the samples were submerged in dye solution for 30 min at room temperature.
Discarded dye solution and rinsed the sample surface with distilled water until the water was clear. The collagenous staining of the samples was observed under a microscope. Subsequently, 1 mL of dye extraction buffer was added, and the collagen content of the samples was calculated using the following formula based on the absorbance at 540 nm and 605 nm according to the manufacturer's instructions.
| ECM stiffness testing
BM-ECM-and OS-ECM-coated plates were prepared for mechanical characteristics. The empty plates and COLI-coated plates were served as negative controls. Each sample was examined at least 50 different locations (each point 3 mm apart on the X or Y axis) by Piuma Nanoindenter (Optics11, The Netherlands) with a loading speed of 3 mm/s. Young's modulus was calculated accordingly by the instrument.
| Cell adhesion on biomimetic ECM
Cell morphology and adhesion ratio were performed to investigate the adhesion activities of breast cancer cells on biomimetic dECM materials. Breast cancer cells were inoculated on 24-well plates at the density of 5 Â 10 4 cells/well. At different time points (1, 2, 4, and 8 h), unattached cells were washed two times with PBS and incubated in cell counting solution (Cell proliferation kit, K1018, ApeÂBio) at 37 C for 2 h. The absorbance at 450 nm was measured. The adhesion ratio was calculated using the following formula, while the cell adhesion at 8 h was defined as 100%: To assess the cell morphology of initial adhesion, breast cancer cells (5 Â 10 4 cells/well) were inoculated in 24-well plates coated with different biomimetic dECMs. At different time points (1, 4, and 16 h), the cells were stained with FITC-phalloidin to show cytoskeletal proteins (F-actin). The experiments were performed according to the IF staining method described above. DAPI was used to show the nucleus. The cell morphology images were taken under the fluorescence microscope.
| Cell migration on biomimetic ECM
Wound-healing assay, transwell migration assay, and colony formation were performed to investigate the migration ability of breast cancer cells on the mimetic ECM stroma.
For wound-healing assay, breast cancer cells were starved in basic medium with 5% FBS for 12 h. After that, the cells were trypsinized into single cells and cultured on a 12-well plate coated with dECMs at the density of 1 Â 10 5 cells/well for another 12 h. The cells with the coated dECMs were scratched in the center of each well with a 200-μL pipette tip. The dropped cells and dECMs were washed with PBS. The migrated cells were recorded by a microscope at desired time points. The left area of the scratches without cells was quantified using ImageJ. The cell migration rate at each time point was normalized by the initial scratch area.
For transwell migration assay, MC3T3 or MG63 cells (3 Â 10 5 cells/well) were pre-inoculated on the bottom of the transwell chamber for 14 days. The posterior chamber containing biomimetic ECM at the sublayer was prepared by decellularization as described previously. Breast cancer cells (1 Â 10 5 cells/well) were seeded in the inserts. PRMI 1640 with low FBS (3%) was added in the top chamber of transwell, while PRMI 1640 with high FBS (10%) was added in the lower section. After an appropriate time of inoculation, the cells were fixed with 4% formaldehyde, and treated with 0.1% crystalline violet staining solution for 10 min. The upper surface cells were gently wiped off with sterile cotton swabs, air-dried, and observed under a microscope. The images were taken randomly, and the area of crystalline violet was quantified using ImageJ.
Clonal island formation was introduced to assess the distance of newborn cells from original cells. Breast cancer cells (1 Â 10 3 cells/ well) were singly seeded on 6-well plates coated with biomimetic ECM and cultured for 7 days with fresh medium replaced every other day. Clonal islands of breast cancer cells were formed, imaged under a microscope, and measured the size. Clonal islands with close cell numbers were selected for quantification of clonal island diameter using ImageJ software.
| RNA isolation and real-time qPCR
Cells were rinsed with PBS and lysed in Omega RNA-Solv ® Reagent (R6830-01). Total mRNA was extracted according to the manufacturer's instructions. The extracted mRNA was reverse transcribed to cDNA by One-Step gDNA Removal and cDNA Synthesis Supermix (AT311, TRANSGEN). Real-time qPCR was performed with TransStart ® Top Green qPCR SuperMix (P41014, TRANSGEN) was used for qPCR.
CXCR4, VEGF, and BSP were typical genes associated with bone metastasis of breast cancer. E-cadherin, N-cadherin, and Vimentin were used as EMT markers. The primers of these genes were synthesized by Generay Biotech Co., Ltd. The primer sequences are shown in Table 1.
| ELISA assay
Breast cancer cells (5 Â 10 5 cells/well) were inoculated into sixwell plates coated with different biomimetic ECM and cultured for 4 days. Supernatants were collected in EP tubes and centrifuged at 3000g/min for 20 min. The collected supernatant was subjected to cell-secreted VEGF content assay with VEGF ELISA assay kits (Proteintech Group, Inc.) depending on the cell species. Meanwhile, cell lysate was collected via trypsin digestion and centrifugation at 1000g/min for 5 min, followed by triplicate cycling at 20 C/À80 C.
The cell lysate was subjected to intracellular BSP content assay with BSP ELISA kit (Animalunion Biotechnology) depending on the cell species. According to the manufacturer's instructions, the samples were added to a VEGF/BSP antibody-coated microplate with horseradish peroxidase (HRP)-labeled detection antibody, and the wells were sealed with a sealing membrane for 60 min at 37 C.
Subsequently, the plate was washed five times with washing solution, and the substrate was added for 15 min at 37 C, protected from light. The absorbance was measured at 450 nm. The concentration of VEGF or BSP was calculated with a standard curve performed accordingly.
| Protein extraction and western blotting analysis
To extract ECM proteins, the dECM was mechanically separated from the tissue culture plates and placed in UA buffer
| Animal studies
The commercial small intestinal submucosa (SIS) scaffolds were cut into circular sizes with 5 mm diameter. The dried scaffolds were pretreated with PBS for at least 16 h and with α-MEM medium for another 16 h. MC3T3-E1 and MG63 cells were seeded on the scaffolds, respectively. The cells were cultured for 5 days in complete medium and for another 7 days with ascorbic acid (50 μg/mL). The ECM-coated scaffolds were prepared following decellularization as described in Section 2. The volume and size of the primary tumors in mammary glands were measured.
Then, the scaffolds were fixed in 10% neutral buffered formalin fixative (G2162, Solarbio) overnight at room temperature. RFP-labeled breast cancer cells on the scaffolds were assessed under confocal microscopy (SP8, LEICA). DAPI was used to show the nuclei.
All experimental procedures involving animals were conducted in compliance with Chinese legislation regarding the use and care of laboratory animals and were approved by the Animal Care and Use Committee, School of Medicine at Ningbo University. The approved protocol number is NBU20220099.
| Statistical analysis
Statistical analyses were performed using the GraphPad Prism 8.0a (GraphPad Software). Data were presented as the mean ± SD. Student t test was used to determine statistical differences between the two groups. When more than two groups were analyzed, oneway analysis of variance followed by a post hoc test (multi-group comparison) was used.p < 0.05 was considered statistically significant.
| Generation and characteristics of biomimetic dECMs
The ECM microenvironment was specific to cell types and cell activities. Here, the dECM from preosteoblasts (MC3T3-E1) were used to mimic normal bone microenvironment (BM-ECM) which have been applied for bone regeneration in our previous studies. 27Meanwhile, the dECM from osteosarcoma cells (MG63) was used to mimic tumor microenvironment in bone tissue (OS-ECM). The biomimetic dECMs were generated as shown in Figure 1a. Ascorbic acid was added to accumulate ECM secretion. Bright-field images, DAPI staining, DNA content quantification, and western blotting were used to assess the effect of decellularization process (Figure 1b-e). The native cells were obviously removed, and the network structure of dECM was observed (Figure 1B). Compared to OS-ECM, BM-ECM was much more compact with higher binding affinity to tissue culture plates during decellularization. Residual DNA was less than 5% in dECMs compared to that in native cells (Figure 1c). After decellularization, the cell proteins in cytomembrane (Vimentin), cytosol (GAPDH), and nucleus (Histones) were almost cleared, while the ECM proteins (FN and COLI) were reserved and even slightly enriched (Figure 1e). The retained GAGs in BM-ECM were 1.46 μg/cm 2 (56.7%), which was significantly more than that in OS-ECM (0.476 μg/cm 2 ; 40.4%; Figure 1f). Similarly, compared to the OS-ECM group, collagen content in BM-ECM was much higher (Figure 1g), with heavier staining and denser structure as stained by collagen proteins (Figure 1h). IF staining of core ECM proteins (COL1A1 and FN) further confirmed the distinguished structure of the two kinds of biomimetic dECMs (Figure 1i). Consistently, BM-ECM with compact ECM protein arrangement exhibited higher stiffness than OS-ECM as shown by Young's modulus in Figure 1j. The Young's modulus of the empty plate and COLI-coated plate was much higher than that in dECMs (Figure S4).
| Distinguished adhesion activities of breast cancer cells assessed by the biomimetic ECM system
Breast cancer cells with different abilities of metastasis were seeded on BM-ECM or OS-ECM to investigate the affinity of biomimetic dECMs on cell adhesion associated with bone metastasis (Figure 2).
The morphology of cell adhesion was assessed by F-actin staining at different time points (1, 4, and 16 h). It is interesting to find that highly metastatic breast cancer cells (4T1 and MDA-MB-231) spread better on BM-ECM than on OS-ECM as early as 1 h (Figure 2a,b). The perimeter of 4T1 cells was the highest on BM-ECM at all time points, compared to the other two groups, and statistical significance was observed between BM-ECM and OS-ECM (Figure 2e). The perimeter of MDA-MB-231 cells showed the same tendency as 4T1 cells (Figure 2f). In contrast, lowly metastatic breast cancer cells (MCF7 and HCC1937) tended to spread better with longer perimeter on OS-ECM than on BM-ECM (images: Figure 2c,d; quantification of perimeter: Figure 2g,h). Moreover, highly metastatic breast cancer cells were attached quicker on BM-ECM with higher adhesion ratio than on OS-ECM (Figure 2i,j), while the result was opposite in lowly metastatic breast cancer cells (Figure 2k,l). The results indicated highly metastatic breast cancer cells tended to adhere on bone mimetic microenvironment, while lowly metastatic breast cancer cells tended to adhere on bone tumor mimetic microenvironment.
| Distinguished migration activities of breast cancer cells assessed by the biomimetic ECM system
To determine the impact of ECM microenvironment on migration of breast cancer cells, wound healing (Figure 3) and transwell assay (Figure 4) were performed. Wound healing was used to evaluate the migration of cells on dECMs (Figure 3a), while transwell assay was used to evaluate the recruitment of cells migrated to dECM (Figure 4a-f).
The breast cancer cells with different metastasis abilities were seeded on ECM-coated plates (COLI, BM-ECM, and OS-ECM), and starved in to BM-ECM was significantly higher than that migrated to OS-ECM (Figure 4c,d). On the contrary, for MCF7 and HCC1937, the cell number migrated to OS-ECM was significantly higher than that to BM-ECM (Figure 4e,f). To further determine the effect of biomimetic dECMs on cell morphology and growth pattern, an alternate approach of microcolony formation was adopted as reported previously (Figure 4g). 31For 4T1, the cells grew tightly together on COLI and OS-ECM. However, on BM-ECM, the cells exhibited irregular cell morphology with numerous membrane protrusions (Figure 4h). Moreover, the colony size with similar cell numbers on BM-ECM was significantly higher than on OS-ECM (Figure 4i), which indicated the boosted migratory tendency of newborn cells from the original cells. 32 and HCC1937 cells were close together with round shape on BM-ECM, while the cells were much looser and more polygonal on OS-ECM (Figure 4h). Increase scattering of cells within the colony on OS-ECM of such cells was corroborated by increase in colony size without significant increase in number of cells per colony (Figure 4k,l).
The colony size of MCF7 and HCC1937 on BM-ECM with similar cell number was the lowest among the groups (Figure 4k,l).
| The expression of metastatic-associated genes on biomimetic dECMs
Besides cell activities on the two kinds of biomimetic dECMs, the molecular levels of metastatic-associated genes were further determined (Figures 5 and 6). To elaborate the metastatic activities induced by dECM niches from different perspective, bone sialoprotein (BSP), vascular endothelial growth factor (VEGF), and chemokine C-X-C motif receptor 4 (CXCR4) were chosen for assessment. BSP was an essential differentiation marker for osteogenesis. VEGF was a cytokine associated with angiogenesis, and CXCR4 was a critical chemokine receptor of CXCL12.4][35][36] Here, the mRNA levels of BSP were the highest in 4T1 cells on BM-ECM on Day 3, but no significant differences were found on Day 1 (Figure 5a).
Besides that, the mRNA levels of both VEGF and CXCR4 were the 5l). The regulation of CXCR4 mRNA levels in HCC1937 with the significant difference occurred as early as Day 1 (Figure 5l).
The protein levels of BSP, VEGF, and CXCR4 in the cells on different biomimetic dECMs were further determined (Figure 6). BSP and VEGF protein expression was quantified by ELISA kit, and CXCR4 The regulation of CXCR4 protein expression by biomimetic dECMs mainly occurred on Day 4 with the similar tendency (Figure 6i,j). For 4T1 and MDA-MB-231 cells, the highest levels of CXCR4 proteins were observed in the BM-ECM group (Figure 6i). Meanwhile, for MCF7 and HCC1937 cells, CXCR4 protein expression was the highest in the OS-ECM group (Figure 6j). Besides that, TFF1 expression was demonstrated to be associated significantly with bone relapse of breast cancer patients, with the highest-ranking gene in bone metastatic breast tumors. 37Consistently, TFF1 was significantly up-regulated in 4T1 cells on BM-ECM, while it was significantly up-regulated in MCF7 cells on OS-ECM (Figure S5).
| The promotion of cell migration by BM-ECM was initiated with EMT
The epithelial-to-mesenchymal transition (EMT) has been proposed to contribute to the metastatic spread of breast cancer cells. 38Thus, EMT might be a potential mechanism of cell migration. To identify the hypothesis, the cell morphology and typical EMT-associated markers (E-cadherin, Vimentin, and N-cadherin) were monitored (Figure 7).metastatic breast cancer cell 4T1 and low metastatic breast cancer cell MCF7 were used to assess the EMT process on biomimetic dECMs (Figure 7a). As shown in Figure 7b, 4T1 cells cultured on COLI and OS-ECM were grown as typical island with clear and smooth edges, while partial cells on BM-ECM became as long fusiform and escaped from the "island."Though the cells on BM-ECM seem significantly granular, live/dead staining confirmed the viability of MCF7 cells (Figure S6). Consistent with the cell morphology, the expression of E-cadherin was lower in 4T1 cells on BM-ECM than on OS-ECM at mRNA levels after 24 h (Figure 7c) and at protein levels on Day 2 (Figure 7i). Meanwhile, the expression of mesenchymal markers (Vimentin and N-cadherin) was significantly higher in 4T1 cells on BM-ECM than the other two groups at mRNA levels after 48 h (Figure 7d,e). Vimentin protein was also expressed highest in 4T1 cells on BM-ECM (Figure 7i). Oppositely, for MCF7 cells, partial cells on OS-ECM exhibited long shuttle shape, while the cells on BM-ECM were rounder (Figure 7b). Though the cells on BM-ECM were separated from each other which might be caused by the cell proliferation inhibition, the interaction between cells and matrix was enhanced.
E-cadherin was expressed highest in the BM-ECM group at mRNA levels after 12 h (Figure 7F), while the mesenchymal markers (Vimentin and N-cadherin) were expressed highest in the OS-ECM group at mRNA levels after 48 h (Figure 7g,h). The upregulation of E-cadherin in the BM-ECM group was observed at protein levels on Day 2, compared to the COLI and OS-ECM groups (Figure 7j). Besides the phenotype-associated genes, the essential transcription factors (TFs) (SNAI1, SLUG, and TWIST1) during EMT process were further assayed at mRNA levels (Figure S7). Consistently, the expression of the EMT-TFs was significantly higher in 4T1 cells on the BM-ECM than the other two groups at mRNA levels, while they were upregulated on the OS-ECM in MCF7 cells, compared to BM-ECM and COLI.
Taken together, BM-ECM promoted EMT of 4T1 cells, while OS-ECM promoted EMT of MCF7 cells, which might contribute to the cell migration of breast cancer cells on different biomimetic dECMs.
| Early detection of metastatic breast cancer cells in vivo
Based on the above results, we have found an interesting phenomenon. The cell activities of breast cancer cells on different biomimetic dECMs were highly associated with their metastasis abilities. Metastasis breast cancer cells were more inclined to adhere and migrate on the normal bone matrix microenvironment. However, low metastasis breast cancer cells preferred to adhere and migrate on the tumor matrix microenvironment. The association between them might be conductive to the early diagnosis of breast cancer metastasis. To demonstrate the hypothesis, we designed an in vivo model with implantation of biomimetic dECM-coated scaffolds and metastatic breast cancer cells, as illustrated in Figure 8a. Whether the biomimetic ECM system can successfully capture the early metastatic breast cancer cells was assessed later. After injection of RFP-4T1 cells for 10 days, obvious primary tumors were found in the right fourth mammary glands (Figure 8b). The tumor size was later quantified, and no significant difference was detected among different groups (Figure 8c).
Under the luminescence scan, the strongest signal was observed in the primary tumor in the right fourth mammary gland (Figure 8d, red dotted circle). Meanwhile, obvious signal could be found in the BM-ECM scaffolds, but almost no signal could be found in the other To further confirm the recruitment of RFP-labeled tumor cells, the ECM-coated scaffolds were fixed, stained with DAPI, and captured under a confocal microscope (Figure 8h). A similar trend was observed with more RFP-labeled tumor cells on the BM-ECM scaffolds than in the other two groups. Statistical significance was assessed (Figure 8i).
| DISCUSSION
The microenvironment of target organs was essential for tumor Therefore, the two kinds of dECM could be taken together as a biomimetic ECM system to be applied in breast cancer bone metastasis.
Previously, synthetic biomaterials were introduced to capture early metastatic breast cancer cells. For example, poly (lactide-co-glycolide) (PLG) scaffolds 39 and micro-porous poly (ε-caprolactone) (PCL) scaffolds 8 were fabricated and successfully for metastatic cells capture in vivo, which achieved high cell densities and reduced the tumor burden within solid organs. The scaffold-captured tumor cells were further demonstrated to be similar to those at metastatic sites. 40ough such chemosynthetic scaffolds were proved to be nontoxic, their biological activity and biomimetic performance are still limited, lacking support from biological macromolecules. In contrast, ECM was considered as a competitive biomaterial with appropriate niches.
Abundant research proved biomimetic supportive network with a plethora of extracellular signaling molecules. ECM-based biomaterials have attracted plenty of attentions as smart platforms for dynamic functionality, such as cell traction, cell-ECM interactions, and cellmediated ECM remodeling. 41Initially, ECM proteins (COLI, COLIV, and fibronectin) were used to decorate synthetic scaffolds to increase the cell adhesion and proliferation activity. Even though simply mixture of several ECM components was still hard to mimic the complex structure of ECM with abundant biomolecules as the native microenvironment to monitor the colonization process in vivo.
During the last decades, dECM was of particular interest to provide a microenvironment close to the native target tissue. 42Originally, tissue-derived dECM was widely applied for total organ or tissue replacement and regeneration with excellent biocompatibility. However, material source, vigorous decellularization process to diminish the structural stability and potential problems still limited its application.
Nowadays, cell-derived ECM has gradually entered researchers' vision with some advantages. For example, most required cells were easy to obtain and expand. The decellularization process of cells was much gentler than tissues to preserve the precise ECM structure and component distribution. Most importantly, cell-derived ECM was more selectivity and control lability since cell types and culture conditions and exhibited distinguished cell activities. Our previous studies focused on cell-derived dECM and developed a serial of bone mimetic dECMs via alteration of culture situation, including osteogenic induction, 43 mineralization by calcium, 16 and fibroblasts-osteoblasts coculture. 28 these studies, the morphology, expression, and distribution of ECM core matrisome proteins (COLI and FN) showed obvious differ- while FN, COLI, COLIII, LN, decorin, and biglycan were the major constituents in the ECM from marrow stromal cells. 45Bae et al.
found that fibroblast-derived matrix, preosteoblast-derived matrix, and chondrocyte-derived matrix exhibited respective unique compositional and structural feature with distinct microenvironment for osteogenic differentiation. 46 the present study, we generated two kinds of biomimetic ECM from preosteoblasts and osteosarcoma cells to represent the physiological status and pathological status undergoing tumor, respectively (Figure 1). Denser fibers and higher Young's modulus were found in BM-ECM than in OS-ECM. During the experiments, we also found the BM-ECM was thicker and easier to be reserved after decellularization compared to the OS-ECM. Moreover, the GAG and collagen content were increased in the BM-ECM. Osteoblasts were the main secretory cells of bone matrix. The balance between osteoblasts and osteoclasts maintained the dynamic stability of bone matrix assembly.
During bone regeneration, osteoblasts were the key active cells to generate new collagens as well as bone matrix mineralization via osteoblastic differentiation. For osteosarcoma cells, their main activities were proliferation and invasion into bone matrix, which might lead to a decrease of the core matrisome secretion and an increase of the matrisome associated proteins. Besides the basic physical and chemical characterization of the dECMs, the ECM protein components are crucial for molecular mechanism of dECM matrix function.
Reviewing the previous literatures, the composition and mechanical properties of the ECM have been considered as drivers of tumor growth, local invasion, and dissemination of cancer cells to distant metastatic sites for breast cancer. 47More and more researchers focused on the essential role of ECM as a potential therapeutic strategy. 48Some key ECM proteins during breast cancer tumorigenesis have been discovered. Collagens, the most abundant component of the tumor ECM, were reported to regulate the breast cancer initiation and metastasis, including collagen I, 49 collagen VI, 50 and collagen XII. 51Chemotherapy was further demonstrated to induce metastasis via its effects on ECM composition, which might contribute to the recurrence of breast cancer. To investigate the effect of bone mimetic dECMs on breast cancer cell migration, we designed several experiments ingeniously under different conditions (Figures 3 and 4). In most research, wound healing and transwell assay were used to analyze the cell migration under a specific factor basically the same. However, in the present study, both methods were used to clarify the different aspects of cell migration. On one hand, wound healing was used to evaluate the migration of cells on dECMs (Figure 3). For wound healing assay, the cells were cultured on dECM-coated plate with the complete confluence. On the next day after cell adhesion, the cells and dECM were removed after scratching, as illustrated in Figure 3a. The migration of the cells cultured on dECMs was activated because of the dECM stimulation. The cells on BM-ECM migrated quicker than on OS-ECM, even to the uncoated area, which was caused by the activation of dECM niche. On the other hand, the transwell assay was used to evaluate the recruitment of the breast cancer cells to the biomimetic dECMs (Figure 4a-f). For the transwell assay, the cells were cultured on the uncoated surface on the top layer of the insert membrane, when the lower layer of the insert membrane was coated with dECMs. After that, the cells migrated through the holes in the insert membrane under the attraction of the dECM. Taking together, the BM-ECM was demonstrated to promote the migration of breast cancer cells either on the dECM or to the dECM.
Increasing works showed dECM tumor models effectively recapitulated native ECM components and crucial communications between cancer cells and ECM, providing significant advantages over currently available in vitro testing platforms. 54,55Cell-derived dECMs offered highly economical platforms to detect the influence of ECM in tumor progression because of their highly controllable nature and have been recently used for the in vitro breast cancer model development. 48,56r example, Lourenco et al. generated a conditioned dECM from adipose stromal cells under the stimulation of gastric cancer cell media, which exhibited a fibrotic arrangement with the increase of aligned collagen fibers and fibronectin deposition. The generated dECM was demonstrated to promote cell proliferation, clustering, and migration of cancer cells. 57Nayak et al. engineered a biomimetic PCL scaffold coated with dECM from cancer-associated fibroblasts, which enhanced breast cancer cell attachment and viability. 58Compared to the dECM from single cell type, we developed a distinguished biomimetic dECM system with dECMs from two kinds of cells to monitor different bone matrix niches. The results showed the dECM system efficiently distinguished the low-metastatic breast cancer cells and high-metastatic breast cancer cells because the tendency of cell adhesion and migration was significantly different on the two kinds of biomimetic dECMs. The most interesting part was that the dECM system could detect early metastatic breast cancer cells in vivo with colonization on the BM-ECM scaffold. Though the accurate mechanism was unclear right now, the biomimetic dECM system provided a smart platform with native microenvironment to discover the communications between cancer cell and matrix. Most importantly, the generated dECM system can provide a promising method for the early prediction of metastatic breast cancer cells.
Even though, there are some limitations to the present studies: (1) The dECMs were derived from two different species. The normal dECM was from mice, while the osteosarcoma dECM was from a human source. Despite this, 4T1 from mice and MDA-MB-231 from human showed the same trend of cell migration on dECMs in vitro, which indicated the generated dECM system could be applied to both mice and human. One possible reason might be the conservation of key proteins in human and mouse dECM. By now, we have no idea whether the human origin of dECM is better to distinguish the metastatic ability of human breast cancer cells. The effect of dECM species on cell activities needs more investigation in the future.(2) Although the dECM system exhibits excellent clinical translation potential for early detection of metastatic breast cancer cells, the breast cancer cells applied here were immortalized cell lines. It is better to study the cell activities of primary breast cancer cells from patients for clinical application next.
| CONCLUSIONS
In summary, the distinguished biomimetic ECM system provided two different cell microenvironment, bone mimetic niche and bone tumor
(8 M urea, 150 mM Tris-HCl, pH 8.0) at 37 C for 2 h. Subsequently, the samples were sonicated (80 W, 10 s on/15 s off, 10 cycles) and placed in boiling at 100 C for 30 min. ECM protein samples were obtained by centrifuging the samples at 14,000g/min for 40 min. To extract intercellular proteins, the cells were lysed in RIPA buffer (R0010, Solarbio) containing 1 mM PMSF at 4 C for 20 min. The lysates were collected in EP tubes and centrifuged (12,000 rpm/4 C/30 min). The supernatant was transferred to a new tube for protein quantification. A standard curve was prepared with gradient concentration of BSA according to the instructions of a BCA quantification kit. The total protein amount for ECM protein (COLI and FN) assay was 50 μg per well, while the total protein amount for other protein assay was 20 μg per well. For western blotting assay, the protein samples were mixed with 5 Â loading buffer and heated for 10 min at 98 C. The proteins were separated with SDS-PAGE gel and transferred to a nitrocellulose filter membrane. The membrane was rinsed with TBS-T, followed by Fast Blocking Solution (P30500, New Cell & Molecular Biotech Co., Ltd [NCM Biotech]) for 10 min. The samples were incubated in primary antibodies at 4 C overnight. The Vimentin antibody (A19607, Abclonal) was diluted at 1:5000 in antibody dilution solution (WB500D, NCM Biotech). E-Cadherin antibody (20874-1-AP, Proteintech) was diluted at 1:1000. CXCR4 antibody (60042-1-LG, Proteintech) was diluted at 1:1000, and β-actin (AC026, Abclonal) was diluted at 1:10,000. The samples were then incubated with HRP-coupled secondary antibodies of the corresponding species (Sa00001-2, AS003, Abclonal). Original blots were provided in the Supporting Information (FiguresS1-S3). T A B L E 1 Primer sequences for realtime qPCR amplification. GeneSpecies Primers (F = forward, R = reverse) 1. Female BALB/c nude mice at the age of 8-10 weeks were purchased from Charles River and housed under specific pathogen-free conditions in the Animal Center of Ningbo University. Five days after arriving, the mice were randomly divided into three groups: (I) COLI-SIS group (n = 5); (II) BM-ECM-SIS group (n = 6); and (III) OS-ECM-SIS group (n = 6). The dECM coating SIS prepared above was surgically implanted below the scapulae of the nude mice. Briefly, the nude mice were treated with 1% pentobarbital sodium (50 mg/kg) for anesthesia and small incisions were made at the dorsal part of the scapulae bilaterally using surgical scissors. The dECM-coated SIS scaffolds were placed under the skin at the bilateral incisions, and the incisions were sewed up with sutures. The scaffolds were fixed on the skin to avoid displacement. After 2 days, breast cancer cells were transplanted orthotopically in the fat pads of the right fourth mammary glands in nude mice. A surgical incision was made around, and 1 Â 10 6 4T1-RFP cells in PBS were injected into the fat pads. After 10 days of tumor cell implantation, the nude mice were captured under IVIS Lumina III (PerkinElmer) at excitation wavelength of 580 and absorption wavelength of 620 nm. Besides in vivo evaluation, all scaffolds were stripped from the nude mice and captured together to avoid the system operation error. The signals were quantified using Living Image software later.
medium with low serum to avoid the effect of cell proliferation. The results revealed that 4T1 and MDA-MB-231 cells migrated faster on BM-ECM than on OS-ECM or COLI control (Figure3b,c), with the highest percentage of wound closure (Figure3d,e). Reversely, MCF7 and HCC1937 cells migrated slowest on BM-ECM (Figure3f,g), with the lowest percentage of wound closure (Figure3h,i). For transwell assay, the biomimetic dECMs (BM-ECM and OS-ECM) were ornamented on the lower layer of the transwell insert membrane, while the breast cancer cells were seeded on the top layer F I G U R E 1 Bone mimetic ECM (BM-ECM) exhibited higher levels of ECM components, denser structure, and higher stiffness compared to osteosarcoma mimetic ECM (OS-ECM).(a) Schematic diagram of cell-derived biomimetic ECM generation.(b) Bright-field images taken by a phase contrast microscope before (Native) and after (dECM) decellularization. After decellularization, the cells were removed and ECM structure with interweaving network was observed. The dECM structure in BM-ECM was denser than that in OS-ECM. DAPI staining (c) and residual DNA quantification (d) revealed the removement of the cells via decellularization. The relative DNA (%) was defined as the DNA content in dECMs divided by that in the native cells before decellularization.(e) Western blotting of representative proteins in cells or ECMs. Cytomembrance protein (Vimentin), cytosol protein (GAPDH), and nuclear protein (Histones) were only expressed in native cells, indicating cellular components were removed via decellularization process. ECM proteins (COLI and FN) were enriched in dECMs, indicating ECM proteins were preserved after decellularization. MC: MC3T3-E1, MG: MG63, N: native, dE: dECMs.(f) Quantification of GAG content.(G, H) The cells and dECMs were stained with Sirius Red/Fast Green Collagen Staining Kit. Collagen content was quantified according to the manufacturer's instructions of the kit (G). Collagen fibers were stained as red (H).(I) IF staining of COL1A1 (red) and OCN (green).(J) Young's modulus of biomimetic dECMs (BM-ECM and OS-ECM). At least 50 sites of each group were measured.**, p < 0.01; ***, p < 0.001. BM-ECM, bone mimetic ECM; dECM, decellularized ECM; ECM, extracellular matrix; GAG, glycosaminoglycan; IF, immunofluorescence; OS-ECM, osteosarcoma ECM. of the transwell insert membrane as illustrated in Figure 4a. Under the guidance of the lower dECMs, breast cancer cells migrated through the holes in the insert membrane to the lower layer. Consistent with wound healing assay, highly metastatic breast cancer cells (4T1 and MDA-MB-231) preferred to migrate to BM-ECM, while lowly metastatic breast cancer cells (MCF7 and HCC1937) preferred to migrate to OS-ECM (Figure 4b). Quantification of the migration rate based on the crystal violet dyeing further confirmed the tendentiousness of the cells on different biomimetic dECMs (Figure 4c-f). For 4T1 and MDA-MB-231, the cell number migrated
F I G U R E 2
Cell adhesion assessment of breast cancer cells with different metastasis abilities on biomimetic dECMs (BM-ECM and OS-ECM). Breast cancer cells with high metastasis (4T1 and MDA-MB-231) and low metastasis (MCF7 and HCC1937) were seeded on BM-ECM or OS-ECM, respectively. COLI-coated plates were introduced as a basic control, which was a major component in cell-derived dECMs.(a-d) Representative images of attached cells to show the cell spreading morphology of 4T1 (a), MDA-MB-231 (b), MCF7 (c), and HCC1937 (d). F-actin of the cells was stained by FITC-phalloidin (green), and nuclei were stained by DAPI (blue). Better spreading cells on BM-ECM at 1 h of 4T1 and MDA-MB-231 were pointed out with yellow arrows.(e-h) Quantification of cell perimeter by ImageJ software. At least 30 cells were measured for each group at each time point.(i-l) Quantification of cell adhesion ratio in different time points (1, 2, 4, and 8 h). The breast cancer cells (4T1 (i), MDA-MB-231 (j), MCF7 (k), and HCC1937 (l)) were seeded on COLI, BM-ECM, and OS-ECM for 8 h. Attached cell number was calculated with CCK kit at desired time point. Adhesion ratio was defined as the cell number at indicated times divided by the cell number at 8 h. Triplicate experiments were performed for each group.*, p < 0.05; **, p < 0.01; ***, p < 0.001. BM-ECM, bone mimetic ECM; dECM, decellularized ECM; ECM, extracellular matrix; OS-ECM, osteosarcoma ECM.
A similar trend was observed in MDA-MB-231 cells (Figure 4h,j). Oppositely, MCF7 F I G U R E 3 Cell migration assessment of breast cancer cells on biomimetic dECMs. Cells were seeded on COLI-, BM-ECM-, and OS-ECM-coated plates.(A) Schematic of wound healing assay to assess the cell migration on ECM. A scratch was made with a 200-μL pipette tip and washed with PBS. The cells with dECMs were removed from the scratched area.(b-e) Cell migration of highly metastatic breast cancer cells (4T1: b and d; MDA-MB-231: c and e).(f-i) Cell migration of lowly metastatic breast cancer cells (MCF7: f and h; HCC1937: g and i). Representative images of wound healing at different time points (b and c, f and g). Quantification of wound closure based on images by ImageJ software (d and e, h and i). The left area of scratch was displayed with yellow dotted line. The wound closure ratio was calculated with the migration area divided by the initial wound area. Triplicate experiments were performed for each group.*, p < 0.05; **, p < 0.01; ***, p < 0.001. BM-ECM, bone mimetic ECM; dECM, decellularized ECM; ECM, extracellular matrix; OS-ECM, osteosarcoma ECM; PBS, phosphate-buffered saline.
highest in 4T1 cells on BM-ECM on Days 1 and 3 (Figure5b,c). Similarly, the regulation of VEGF and CXCR4 mRNA levels by biomimetic dECMs in MDA-MB-231 was more obvious than that of BSP, with F I G U R E 4 Migration of breast cancer cells with different metastatic abilities to biomimetic dECMs via transwell assay (a-f) and microcolony assay (g-l).(a) Schematic of breast cancer cell migration from unECM microenvironment to ECM microenvironment to evaluate the cell recruitment of dECMs. MC3T3-E1 or MG63 cells were cultured on the lower layer of the transwell inserts for 10 days (5 days in complete culture medium and 5 days in the medium supplemented with 50 μg/mL ascorbic acid), followed by decellularization. Breast cancer cells were cultured on the upper layer of the filter membrane and migrated to the lower layer through the holes in the membrane under the attraction of dECMs.(b) Representative images of different breast cancer cells on dECMs. Migrated cells on the lower layer were stained with crystal violet as shown in purple.(c-f) Quantification of migrated cells on dECM-coated membrane (4T1 (c), MDA-MB-231 (d), MCF7 (e), and HCC1937 (f)). The images were taken randomly, and the area of crystalline violet was quantified using ImageJ. At least nine images were used for quantification.(g) Schematic of microcolony assay. Breast cancer cells were singly seeded on dECM-coated plates and cultured for 7 days. Clonal islands were formed.(h) Representative images of microcolonies formed by different breast cancer cells on dECMs. White dotted circles depicted the margins of the colonies.(i-l) Colony diameter and number of cells per colony of the breast cancer cells (4T1 (i), MDA-MB-231 (j), MCF7 (k), and HCC1937 (l)) on dECMs. At least five colonies were measured for each group. Triplicate experiments were performed for each group.*, p < 0.05; **, p < 0.01; ***, p < 0.001.dECM, decellularized ECM; ECM, extracellular matrix. F I G U R E 5 The effect of biomimetic dECMs on mRNA levels of metastatic-related genes in breast cancer cells with different metastatic abilities. Breast cancer cells were cultured on COLI, BM-ECM, and OS-ECM for 3 days, respectively. The mRNA expression of metastatic marker genes (BSP (a, d, g, and j), VEGF (b, e, h, and k), and CXCR4 (c, f, i, and l)) on Days 1 and 3 were assessed by quantitative real-time PCR (qRT-PCR). Both highly metastatic breast cancer cells (4T1 (a-c); MDA-MB-231 (d-f)) and lowly metastatic breast cancer cells (MCF7 (g-i) and HCC1937 (j-l)) were examined. Triplicate experiments were performed for each group.*, p < 0.05; **, p < 0.01; ***, p < 0.001. BM-ECM, bone mimetic ECM; dECM, decellularized ECM; ECM, extracellular matrix; OS-ECM, osteosarcoma ECM. the highest mRNA levels of VEGF and CXCR4 in the cells on BM-ECM on Day 3 (Figure 5d-f). Reversely, for MCF7 and HCC1937, the mRNA expression of the three genes was mainly highest in the cells on OS-ECM on Day 3 (Figure 5g-l), except CXCR4 expression in HCC1937 (Figure protein expression was measured by western blotting. In general, the protein expression tendency was basically consistent with that of mRNA. In detail, the protein levels of BSP were significantly upregulated in 4T1 and MDA-MB-231 cells on BM-ECM on Days 2 andF I G U R E 6The effect of biomimetic dECMs on protein levels of metastatic-related genes in breast cancer cells with different metastatic abilities.(a-h) The contents of BSP (a-d) and VEGF (e-h) were assayed by ELISA. Breast cancer cells with high metastasis ((4T1 (a and e), MDA-MB-231 (b and f)) and low metastasis (MCF7 (c and g), HCC1937 (d and h)) on dECMs at different time points (Days 2 and 4) were examined.(i-n) The protein levels of CXCR4 were assayed by western blotting in metastatic breast cancer cells (4T1 and MDA-MB-231) (i, k, and l) and low metastatic breast cancer cells (MCF7 and HCC1937) (j, m, and n) on dECMs at different time points (Days 2 and 4).(i and j) The WB bands were shown. Beta-actin expression was served as an internal control.(k-n) Relative protein levels of CXCR4 were assessed by the quantification of the bands in (i and j) via ImageJ. Triplicate experiments were performed for each group.*, p < 0.05; **, p < 0.01; ***, p < 0.001.dECM, decellularized extracellular matrix.4 (Figure 6a,b), which was more significant than the mRNA levels (Figure 5a,d). In contrast, BSP was highly expressed in MCF7 and HCC1937 cells on OS-ECM (Figure 6c,d). Similar to BSP, VEGF proteins were expressed highest in 4T1 and MDA-MB-231 cells on BM-ECM (Figure 6e,f), while the expression was the highest in MCF7 and HCC1937 cells on OS-ECM on Days 2 and 4 (Figure 6g,h). F I G U R E 7 Legend on next page.
Effect of biomimetic dECMs on the phenotypic transition of breast cancer cells during EMT process.(a) Schematic diagram to show EMT progression of breast cancer cells on dECMs.(b) Representative images of cell morphology. Metastatic (4T1) and low metastatic (MCF7) breast cancer cells were cultured for 48 h on COLI, BM-ECM, or OS-ECM, respectively. Cell morphology was observed under phase contrast microscopy.(c-h) The mRNA expression of EMT-related genes as measured by real-time PCR. The breast cancer cells with high metastasis (4T1 (c-e)) and low metastasis (MCF7 (f and g)) on dECMs were assayed at different time points(12, 24, and 48 h). E-cadherin (c and f), vimentin (d and g), and N-cadherin (e and h) expression were assessed.(i-n) The protein expression of EMT-related genes as measured by western blotting.(i and j) The WB bands were shown. Beta-actin expression was served as an internal control. The bands in (i and j) were quantified via ImageJ (k-n). The purpose protein value divided by the β-actin value was used to calculate the relative protein expression. E-cadherin and Vimentin expression were assayed in metastatic breast cancer cells (4T1 (i, k, and l)) and low metastatic breast cancer cells (MCF7 (j, m, and n)) on dECMs at different time points (Days 1 and 2). Triplicate experiments were performed for each group.*, p < 0.05; **, p < 0.01; ***, p < 0.001. BM-ECM, bone mimetic ECM; dECM, decellularized ECM; ECM, extracellular matrix; EMT, epithelial-to-mesenchymal transition; OS-ECM, osteosarcoma ECM.two groups (Figure8d, blue dotted circle). To make more comparable and precise, the scaffolds with different ECM coatings were stripped from the mice and scanned together (Figure8e). Consistently, the fluorescent signals were at high level in the BM-ECM group, while negative signal was observed in the COLI group and slight signals were observed in the OS-ECM group. The quantification of total signals and average signals revealed the highest level in the BM-ECM group significantly, compared to the other two groups (Figure8f,g). Moreover, no positive signal was observed in the potential metastatic organs (heart, liver, spleen, lung, kidney, and bone), while significant signals were observed in BM-ECM scaffolds at the same time (FigureS8). The results indicated that such biomimetic dECM system might be a promising tumor model for the early detection of metastatic breast cancer cells. F I G U R E 8 BM-ECM enabled early capture of metastatic breast cancer cells in nude mice, compared to OS-ECM.(a) Schematic of the in vivo model for the early detection of metastatic breast cancer cells. SIS scaffolds were coated with COLI, BM-ECM, or OS-ECM, followed by the implantation below the scapulae of the nude mice. After 2 days, RFP-labeled 4T1 cells were injected into the right fourth mammary gland. On the 10th day after injection, the nude mice were captured under IVIS Lumina.(b) The primary tumors were stripped and pictures taken together (n = 5 for COLI group; n = 6 for BM-ECM and OS-ECM groups). A steel ruler was beside to show the size of the tumors.(c) Quantification of the tumor volume.(d) Representative images of the nude mice to show the signals of primary tumors and ECM-coated scaffolds in vivo. Blue dotted circle: ECM-SIS scaffolds; red dotted circle: primary tumor in mammary gland.(e) The scaffolds were striped and captured together (n = 10 for COLI group; n = 12 for BM-ECM and OS-ECM groups). The total signals (f) and average signals (g) on the scaffolds were calculated with Living Image software.(h) The striped scaffolds were captured under a confocal microscope. DAPI was stained to show the nuclei (blue). RFP-4T1 cells were shown as red.(i) Quantification of tumor cells recruited on the ECM-coated scaffolds. At least nine images from random fields were captured for each scaffold.*, p < 0.05; ***, p < 0.001. BM-ECM, bone mimetic ECM; ECM, extracellular matrix; OS-ECM, osteosarcoma ECM; SIS, small intestinal submucosa.
metastasis via cell-matrix interactions. Bone was the major metastatic site of breast cancer, which indicated the bone niche was suitable for breast cancer cells colonization. Based on this, the bone matrix might harbor different ECM proteins under physiological or pathological conditions and further affect tumor cell activities, including cell adhesion and migration. Here, we developed two kinds of dECM to mimic bone matrix microenvironment under different situations. Decellularized ECM from normal preosteoblasts (BM-ECM) was used to mimic the normal bone matrix, while dECM from osteosarcoma cells was used to mimic the bone tumor matrix. The results demonstrated that comparative analysis of cell activities on both was helpful to distinguish the bone metastasis ability of breast cancer cells in vitro and facilitated early detection of metastatic breast cancer cells in vivo.
mimetic niche. Compared to tumor mimetic niche (OS-ECM), the structure of native bone mimetic niche (BM-ECM) was denser with more affinity to tissue plates. The ECM components in BM-ECM were detected with more collagens and GAGs than in OS-ECM. The most interesting was that the developed ECM system could efficiently distinguish the metastatic ability of breast cancer cells. Metastatic breast cancer cells tended to adhere and migrate on BM-ECM, while low metastatic breast cancer cells tended to adhere and migrate on OS-ECM. In the nude mice model with the implantation of biomimetic ECM-coated scaffolds and tumor cells, BM-ECM was demonstrated to successfully detect the early metastatic breast cancer cells before 52,52Fatherree et al.10reported chemotherapy-induced collagen IV abundant in tumor ECM and promoted invasion via activation of Src and focal adhesion kinase, while Guarin et al.52found chemotherapy-induced collagen V abundant in liver ECM and increased cancer cell invasion via α1β1 integrin and MAPK signaling. Besides the direct composition investigations, Kay try will help to discover the key ECM proteins and their functions in the process of bone metastasis of breast cancer.
==
Domain: Biology Materials Science Medicine Engineering
|
[{"added": "2023-09-03T15:31:18.954Z", "created": "2023-09-01T00:00:00.000", "id": "261483219", "metadata": {"extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1002/btm2.10597", "pdf_hash": "0397242ccd13ce6add5e60fd1a7d64d58c305ad7", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:3737", "s2fieldsofstudy": ["Biology", "Medicine", "Materials Science", "Engineering"], "sha1": "bdff9e6df7e8a31bbbda3955986d1f61ec32495c", "year": 2023}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Appraisal of Artificial Screening Techniques of Tomato to Accurately Reflect Field Performance of the Late Blight Resistance
Late blight (LB) caused by the oomycete Phytophthora infestans continues to thwart global tomato production, while only few resistant cultivars have been introduced locally. In order to gain from the released tomato germplasm with LB resistance, we compared the 5-year field performance of LB resistance in several tomato cultigens, with the results of controlled conditions testing (i.e., detached leaflet/leaf, whole plant). In case of these artificial screening techniques, the effects of plant age and inoculum concentration were additionally considered. In the field trials, LA 1033, L 3707, L 3708 displayed the highest LB resistance, and could be used for cultivar development under Polish conditions. Of the three methods using controlled conditions, the detached leaf and the whole plant tests had the highest correlation with thefield experiments. The plant age effect on LB resistance in tomato reported here, irrespective of the cultigen tested or inoculum concentration used, makes it important to standardize the test parameters when screening for resistance. Our results help show why other reports disagree on LB resistance in tomato.
Introduction
Tomato (Solanum lycopersicum L.) is the fourth most economically important crop in the world: after rice, wheat, and soybean [1,2]. Of the 200 pathogens affecting tomato production worldwide, Phytophthora infestans (Mont.) de Bary, the oomycete causing late blight (LB), is the primary cause of tomato crop loss [2]. Losses are more frequent and more severe in areas where tomato is grown near potato [3].
Research on plant age-dependent expression of P. infestans resistance in tomato is needed, especially for the method of inoculation, plant evaluation, and inoculum concentration. In contrast, numerous studies have been done on the methods for testing P. infestans resistance in potato [4][5][6][7][8][9][10][11][12][13][14]. Tomato germplasm exist with resistance against P. infestans [15][16][17][18][19][20]; however, little information is available on standardized methods for evaluating LB resistance in tomato. Information on the correlation between seedlings and mature plant resistance is also limited [21][22][23]. Moreover, the last systematic comparison of methods for testing tomato LB resistance was reported over 40 years ago [24]. Lack of standardization already causes problems. For example, the wild tomato accession S. pimpinellifolium L 3708 was reported to have LB resistance conferred by one incompletely dominant gene [15,25]. That was amended a mere seven years later, when other studies reported complex inheritance of the trait [26][27][28][29]. Similarly, the accession S. habrochaites LA 1033 was designated Ph -4 and used as a LB resistance standard by several research centers [30][31][32], despite evidence of multiple QTLs conferring the trait [17,[33][34][35]. Such discrepancies make it difficult for tomato breeders to use resistant germplasms.
The objective of this study was to (i) systematically compare the P. infestans resistance in the previously and recently reported resistant tomatoes using an array of methods, both in the field (naturally occurring infections) and with artificial inoculation (employing whole plant and detached leaf/leaflet bio-assays); (ii) determine the importance of the age-dependent expression of LB resistance; and (iii) investigate the level and stability of LB resistance of four sources of this trait in a multi-year field experiment.
Plant material
Tomato germplasm used in this study included commercial cultivars, breeding lines, landraces, and wild tomato accessions from the tomato germplasm collection at the Research Institute of Horticulture (RIH; Skierniewice, Poland), collectively referred to as 'cultigens'. Cultigens included three accessions of S. pimpinelli-folium, five of S. habrochaites, three of S. huaylasense, two of S. corneliomuelleri, three of S. peruvianum, and four of S. lycopersicum. 'Rumba' (S. lycopersicum; PNOS Oz_arów, Poland) served as the LB susceptible control, and it was readily infected by P. infestans under field conditions. The cultigens used in this study, their origin, and LB resistance background (if known) are listed in Table 1.
Plants were grown from seeds, transplanted at first true leaf stage into # / 10 cm plastic pots containing Classman Potgrond substrate (Lasland, Grady, Poland), and placed in a greenhouse. Growing plants were kept at approximately 24/18uC day/night for the tests under controlled conditions, or grown for 4 to 5 weeks in the greenhouse, and then transplanted to the field in the second half of May each year.
Field evaluations
In 2007 and 2008, all cultigens were evaluated for resistance against P. infestans under epiphytotic conditions at the Department of Genetics, Breeding, and Biotechnology experimental field area (RIH, Skierniewice, Poland). Four-to five-week old plants were transplanted to the field in a randomized complete block design, with three replications. Each plot consisted of ten S. lycopersicum plants or five plants of the wild tomato cultigens, spaced 506100 cm (within rows 6 between rows, respectively). Additionally, plants of the LB susceptible 'Rumba' were planted around the border of the experimental field to ensure high and uniform pathogen pressure. We also included a S. lycopersicum cv. New Yorker ('NY') which carries the LB resistance gene Ph-1 [36- 39]. No fungicide control was applied throughout the growing seasons in any of the field trials.
Field plots were inspected weekly throughout the season. When the susceptible control 'Rumba' foliage developed maximal area symptomatic for LB in a given year, disease ratings for each cultigen were collected. Observations were made by the same individual throughout the experiment to avoid that source of variation. LB lesions on the leaves and stems of each plant were rated using the modified scale of Zarzycka [40]: At the beginning of this study, such arithmetically biased methods of LB assessment were employed predominantly [5,8,15,19,20,24,25,31,35], and keep on being used until this day [9,16,21,26], although more accurate methods were developed [7,11,29].
Subsequently, the disease severity index (DSI) was calculated for each cultigen, respectively, as a mean of ratings for the plants, similar to other studies of this pathosystem [26,40].
In the initial field screenings performed in 2007 and 2008, four cultigens (WVa 700, L 3708, L 3707, LA 1033) exhibited high P. infestans resistance. In order to assess the stability of resistance in these cultigens, tests were conducted in 2009 to 2013, under natural LB infestations. Field trials were run in two locations: RIH, Skierniewice (Central Poland), and Boguchwała (Southern Poland), which are 300 km apart from one another. In these experiments, tomato plants were evaluated for resistance against P. infestans as described earlier.
Pathogen cultures and inoculum preparation P. infestans isolates used in this study, collected from tomato plants grown in different regions of Poland and tested in a pilot study ( Fig. 1), were deposited at RIH (Skierniewice; n = 19) or obtained from IHAR (the Plant Breeding and Acclimatization Institute -National Research Institute, Młochów, Poland; n = 27 isolates). Isolates were transferred from rye agar onto leaflets of cv. Rumba and cultured for at least two generations, each 7 to 8 days long, with incubations in darkness at 100% RH and 16uC [40]. Inoculum consisted of a sporangial suspension that was washed off the sporulating lesions on the 'Rumba' leaflets using distilled water. Sporangia counting was done with a haemocytometer, and final inoculum concentrations (inocula loads) were adjusted according to the assay protocol described below. Prior to dilution, the suspension was chilled for 2 h at 4uC, and then incubated at RT for 30 min.
The isolate IWP 13, collected in 2008 from our experimental field (RIH Skierniewice), was used in all subsequent tests. Detailed isolate characteristics are as follows: race according to Black [3,4,7,10,11], mating type A1, mtDNA haplotype Ia, and intermediate resistance to metalaxyl [41]. This isolate was chosen based on the results of three independent disease severity tests. These tests were performed with 46 isolates of P. infestans on the detached leaflets of cv. Rumba, where it induced the highest disease severity (DSI = 1.760.1). Moreover, it produced the smallest variation in symptoms among the isolates tested in a pilot study on 'Rumba' (Fig. 1).
Seedling tests
Resistance screening experiments under controlled conditions were conducted in growth chambers and in greenhouse at the RIH, Skierniewice, Poland. These assays included detached leaflet, detached leaf, and whole plant bio-assays, and are described in detail in the paragraphs that follow. In all test methods we evaluated the influence of plant developmental stage (age in weeks) and inoculum concentration on the disease severity in four tomato cultigens with various levels of P. infestans resistance/susceptibility: WVa 700, L 3708, LA 1033, and 'Rumba'. All tests were conducted in the spring and summer (April to September) of 2011 to 2013.
The detached leaflet assay
We determined the effects of the plant age (4-to 8-week old plant, in one week increments) and the inoculum concentration (5610 3 , 10 4 , 5610 4 sporangia/ml) on disease symptoms development. Each treatment combination (inoculum concentration vs. plant age) was tested in a series of three independent trials with 25 leaflets per cultigen. The third to fifth fully expanded leaves, counting from the plant's top, were collected. This criterion was established by our earlier experiments as well as by previous studies [42] on the relationship between the intensity of LB symptoms and leaf position on plants. Side leaflets were detached with scissors and immediately placed on wet cellulose wadding in a plastic tray. Sporangial suspension (40 ml) was placed on the center of the abaxial side of each leaflet. Trays with leaflets were covered with glass to maintain 90 to 100% RH and incubated at 16uC. After an initial 24 h incubation in the dark, the leaflets were turned abaxial side down and incubated at the same temperature and RH for 6 days under 12 h photoperiod and a light intensity (PPFD) of 650 mmol/m 2 /s. Symptoms of LB were assessed 7 days after inoculation, using the DSI scale described earlier.
In our preliminary tests with 20 local isolates of the pathogen, individual leaflets of the resistant cultigens displayed high variability in LB reaction. Additionally, the size of necrotic lesions did not correspond with the intensity of sporulation. Within susceptible cultigens (i.e., compatible reactions) that was not observed, and in agreement with other studies [43] the sporulation often preceded the necrosis. These differences caused difficulties in classifying cultigens using this method. Therefore, at the time of assessment by the DSI scale described above, the inoculated leaflets were tested for sporulation intensity by shaking off the sporangia (vortexing each sample for 30 sec in 1 ml of water) and counting the sporangia with a haemocytometer.
The detached leaf assay
The severity of LB symptoms on detached tomato leaves was studied in 11 stages of development (5-to 15-week old plants, in one week increments), using three concentrations of inoculum (5610 3 , 10 4 , 5610 4 sporangia/ml). The third to fifth fully expanded leaves (counting from the plant's top) were excised from greenhouse-grown plants. The petiole of each leaf was immediately placed into distilled water (100 ml) in a plastic container with a # / 1 cm hole in the lid center. The leaves in plastic containers were then placed in a plastic box and handsprayed with sporangial suspension, until the upper surface of each excised tomato leaf was completely covered. The inoculated leaves were incubated in darkness for 24 h at 18uC, and 95 to 100% RH. Following this initial incubation, samples were then incubated for 6 days under conditions similar to the detached leaflet method.
Leaves of each cultigen were evaluated using the scale described earlier (1 = 97.1-100% leaf area covered with lesions or dead; to 9 = no lesions). For each of the four cultigens undergoing evaluation, and for each treatment combination (inoculum concentration vs. plant age), a different number of leaves (12 to 24) per cultigen was used due to availability of plant materials. In addition, each treatment combination was examined in three independent sets of experiments.
The whole plant assay LB severity on whole plants was tested at six developmental stages (3-to 8-weeks of age, in one week increments) of greenhouse-grown plants, using three concentrations of inoculum (5610 3 , 10 4 , 5610 4 sporangia/ml). Plants were hand-sprayed with the sporangial suspension until complete leaf coverage and excess run-off was observed. The inoculated seedlings were incubated in the dark at growth chamber for 24 h, at 16uC and 100% RH. After this initial incubation, the inoculated seedlings were grown at 16uC with 12 h of light. The plants were rated individually seven days after inoculation. The symptomatic area of both leaves and stems were evaluated using the DSI scale described earlier. Each treatment combination (developmental stage vs. inoculum concentration) was repeated three to seven times, with each cultigen represented by 12 to 72 plants depending on the experiment.
Statistical analyses
Data from all experiments were analyzed by means of the general linear model with year, location, cultigen, plant age, inoculum concentration, and their interaction as the tested variables. Means were separated with the Tukey multiple comparison procedure at significance level of 0.05. Regression and correlation analyses were used to compare results from the different testing methods. All calculations were done with the statistical software STATISTICA 8.0 (StatSoft, Inc. 2009).
Field evaluations
Based on the results from the initial field experiments in 2007 and 2008 (Table 2), four tomato cultigens were chosen (LA 1033, L 3708, L 3707, WVa 700) from the original experimental group of 20 cultigens, to verify their P. infestans resistance in the field at two separate locations (Skierniewice, Boguchwała).
Comparative analysis of the LB resistance levels among the tested tomato cultigens indicated significant differences across the five years of study (2009 to 2013), at both locations ( Table 3). The highest and most stable level of LB resistance in the field, comparable with the baseline years 2007 and 2008, was found in LA 1033 S. habrochaites (DSI ranging from 7.2 to 9.0, depending on the year). Plants of this cultigen were either free from any LB symptoms, or developed only slight infection symptoms (classes 6 to 8). Indeed, this cultigen demonstrated high levels of resistance in the field, even under strong P. infestans pressure. For example, in 2011, both locations experienced conditions particularly conducive to LB, as confirmed by disease intensities recorded for both susceptible tested tomato cultigens ('Rumba', 'NY'), while LA 1033 showed no LB symptoms (Table 3) Intensity of LB symptoms on the S. pimpinellifolium accessions L 3708 and L 3707 which ranged from DSI = 7.4 to 8.8, was generally comparable with S. habrochaites LA 1033, and depended on the year and the location. The WVa 700 plants displayed significant differences in LB intensity levels, depending on the location and the year (Table 3). In Skierniewice, we recorded partial infection of this cultigen in 2011 and 2013 (DSI = 4.961.0 and 4.561.1, respectively). In the remaining years of the study, the WVa 700 plants exhibited very low LB intensities (DSI = 8.0 to 8.4). In the Boguchwała study, we observed consistently high levels of disease severity in this cultigen in all years of testing (DSI ranging from 4.3 to 6.1). The 'NY' (Ph-1)
The detached leaflet assay
Experimental results indicated significant effects of the plant age on the LB intensity levels across all tested cultigens. Leaflets from 4-to 5-week old plants of both S. pimpinellifolium cultigens tested (WVa 700, L 3708) exhibited higher degrees of infection than those from 7-to 8-week old plants, irrespective of the inoculum concentration used (Fig. 2). The LA 1033 leaflets showed the highest intensity of LB symptoms when inoculated with the 5610 4 sporangia/ml (DSI = 4.8 to 5.8, dependent on plant age). Comparatively, at the two lower inoculum concentrations used, the disease symptoms proved significantly decreased at all developmental stages tested.
We found a broad range of intra-cultigen variation in the lesion area of individual leaflets of resistant cultigens (WVa 700, L 3708, LA 1033) inoculated with P. infestans. This variation was observed even in the oldest specimens (8-week old) tested with the lowest inoculum load (Fig. 2). The observed high variability of this assay makes it impossible to unequivocally distinguish genotypes exhibiting specific LB reactions. Regardless of the differences in the necrotic area, the LB resistant cultigens showed a low intensity of sporulation and were not significantly different from each other. 'Rumba' displayed sporulation intensity of 23,465.3 thousands of sporangia/mm 2 , exceeding those of the resistant cultigens approximately 50-fold (WV 700: 0.760.6; L 3708: 0.560.5; LA 1033: 0.560.5 thousands of sporangia/mm 2 , respectively). Also, only in case of 'Rumba', did the size of necrotic spots correspond well with the intensity of sporulation. Finally, in contrast to the resistant cultigens tested, 'Rumba' showed consistently high and uniform disease symptoms in all detached leaflet assays (Fig. 2), regardless of plant age or inoculum load.
The detached leaf assay
Significant effects of all variables tested (cultigen, plant age, and inoculum concentration) on the LB intensity were observed. The highest variation was observed for the interaction between cultigen and plant age, and the lowest was between plant age and inoculum concentration.
'Rumba' leaves displayed the highest degree of LB infection (DSI = 1.0 to 1.1), regardless of the plant age, and lacked significant differences in their respective DSI scores (Fig. 3). Relationships between plant age and degree of disease symptoms were observed for all three remaining cultigens in the study (WVa 700, L 3708, LA 1033). Additionally, we noted a different trend in each of the three resistant cultigens. The WVa 700 leaves showed low levels of LB intensity, irrespective of plant age or inoculum concentration. The lowest LB symptoms levels displayed by L 3708 were similar to those of WVa 700, but proved comparatively more dependent on plant age and inoculum concentration (Fig. 3). In L 3708, at the lowest inoculum concentration tested, the impact of plant age on LB intensity was evident, as response dropped below the significance level in 10-week old and older plants.
Cultigen LA 1033 also showed varied reactions to P. infestans inoculation, dependent on plant age; the observed differences greatly surpassed those noted for WVa 700 or L 3708 (Fig. 3). The lowest degree of disease symptoms in LA 1033 was observed in the leaves of 15-week old plants (DSI = 8.860.4), after application of the lowest concentration of inoculum.
The whole plant assay
As in the detached leaf experiments, also in the whole plant assays we found significant effects of all variables tested (cultigen, plant age, inoculum concentration) on the intensity of LB symptoms. This necessitated an independent analysis for each variable.
In 'Rumba', we observed a significant interaction between the disease severity and the inoculum concentration (Fig. 4). Plants inoculated with the highest inoculum concentration exhibited high disease severity levels (DSI = 1.0 to 1.6, depending on plant age) in all tested stages of development (3 to 8 weeks of age). In contrast, plants treated with the lowest inoculum concentration showed lower intensity of disease symptoms (DSI = 2.6 to 5.4, depending on plant age). This wide DSI range was dependent on plant age, and for each developmental stage tested was distributed over at least two of the nine severity rating classes (reaching max. 7 classes of spread). A low degree of disease symptoms coupled with large variation in the LB susceptible 'Rumba' under 5610 3 sporangia/ ml, rendered this inoculum concentration unsuitable to correctly differentiate between the LB resistant and LB susceptible plants.
Of the resistant cultigens, WVa 700 had the lowest variation in LB symptoms at different phases of development. Indeed, WVa 700 exhibited a low degree of infection (DSI.7) at all stages of development, regardless of inoculum concentration (Fig. 4), with one exception: the 3-week old plants had increased disease severity with DSI ratings spread over all nine rating classes at the highest inoculum concentration used. We found the lowest, most uniform levels of LB intensity in 7-week old and older WVa 700 plants, regardless of inoculum concentration.
In L 3708, we observed variable levels of LB symptoms intensity, dependent on plant age and inoculum concentration (Fig. 4). Generally, younger plants (3-and 4-week old) exhibited distinctly higher levels of LB sysmptoms than older plants (7-and 8-week old). We also observed higher variation in the range of severity ratings in younger plants compared with older plants. Moreover, the range of variation in severity ratings depended on the inoculum concentration used and correspondingly increased. In addition, we observed an inverse relationship between plant age and inoculum concentration on LB intensity levels. Older plants showed a more uniform intensity of disease symptoms across inoculum loads tested, than did the younger plants. Indeed, among the L 3708 developmental stages tested, the lowest and most uniform levels of LB symptoms were observed in 7-and 8-week old plants at all inoculum concentrations tested.
LA 1033 displayed a more diverse response to P. infestans challenges than WVa 700 or L 3708 plants. This reaction, however, remained dependent on plant age at all tested inoculum concentrations (Fig. 4). A comparative analysis of DSI values for all inoculum concentrations at all developmental stages examined in this cultigen showed greater severity of LB symptoms, compared with WVa 700 or L 3708. The lowest level of LB intensity (DSI = 8.260.6) was found in 8-week old plants inoculated with the lowest concentration of inoculum. Higher inoculum concentrations caused an increase in the severity of symptoms, with only sporadic significant increases in the variation of severity ratings (Fig. 4).
In general, all control conditions assays indicated an impact of plant age on the intensity of LB symptoms at levels specific for a given tomato cultigen. Symptoms of LB tended to decrease with plant age. High variability noted for the detached leaflet assay makes this method unreliable for standard testing of the LB resistance, or requires additional assessments (e.g., sporulation intensity), but the assays may be used for efficient identification of the LB susceptible individuals, such as during the early stages of selection and breeding.
Cross-test comparison
Pooled data from the controlled condition tests (detached leaflet, detached leaf, whole plant) from all treatment combinations (cultigens, inoculum concentrations, plant ages), were individually compared with the data of field experiments. All tested laboratory techniques showed significant linear relationships with the field results. But, the detached leaf assay and whole plant assay correlated better with the field assay than did the detached leaflet assay. The determination coefficients were: 0.94, 0.83, and 0.41, respectively (Fig. 5). The stronger relationship observed for the detached leaf assay and whole plant assay with reference to the field data resulted from lower variation within the plant materials, and the domination of extreme DSI values from the field observations.
In summary, this indicates that the detached leaf and the whole plant assays may serve as reliable tools for testing tomato LB resistance. These assays could thus replace the field trials at the early stages of testing. Additionally, it is noteworthy that at high disease intensity (DSI = 1 to 3) the detached leaflet assay showed good correlation with the field results, which further supports our suggestion to use this method for initial testing in tomato LB resistance breeding.
Discussion
The choice of methods for accurate testing of genotypes for LB resistance is an important area for plant breeding programs, including the tomato-P. infestans pathosystem. Lack of agreement among published reports for LB resistance in several tomato cultigens (e.g., L 3708, LA 1033) prompted the current study. Here, our aim was to standardize and compare several methods for testing the P. infestans resistance under controlled conditions using four tomato cultigens, over several plant ages and inoculum concentrations. This would allow us to establish a uniform approach to assess the benefit of each cultigen for use in breeding for LB resistance. We also attempted to optimize the methods and to relate the results to those from natural field infection experiments.
Results of the multi-year (2008 to 2013) field experiments, from two locations (approx. 300 km apart), show that S. habrochaites LA 1033 had the lowest and most stable P. infestans infection levels of the cultigens tested. These findings are in agreement with other studies for LA 1033 reporting high resistance under natural field infection against a diverse set of P. infestans isolates, in the USA [17,33,35]. In other studies, isolates from Taiwan succeeded in infecting LA 1033 [16,31,34]. Our results of LA 1033 showing modest LB symptoms in the field could not be confirmed with detached leaf assay using P. infestans isolates derived from symptomatic LA 1033 plants (Nowakowska et al., unpublished data). The latter observation is in agreement with potato-LB studies, where P. infestans isolates derived from LB-symptomatic potato plants, carrying the Rpi-phu1 gene conferring high levels of potato LB resistance, and grown in the field [44], failed to induce the disease symptoms in the laboratory [45]. This demonstrates the complexity of the pathosystem under field conditions.
Other tomatoes with outstanding LB resistance in the field, including the S. pimpinellifolium cultigens L 3707 and L 3708 [46,47], performed insignificantly lower than LA 1033. We observed, however, their stable low LB intensities in the field at all study years and in both locations, even with high P. infestans incidence. High LB resistance has been reported in these cultigens [15,46,47], despite a lack of clear elucidation of the trait's genetic background [3,[26][27][28][29]. These cultigens have been successfully used as sources for pyramiding Ph-3 with Ph-2 LB resistance into new tomato cultivars [48][49][50][51]. Only few research groups reported rare instances, when high pathogen pressure, under highly conducive conditions, led to overcoming the resistance of L 3708, particularly in controlled condition assays [15,30,31,52].
In contrast to the aforementioned LB resistant cultigens, WVa 700 [19,38,53] exhibited varying levels of LB symptoms in the field, depending on the study year and location. These findings indicate the differences in the local pathogen populations. Consequently, such unstable expression of LB resistance in WVa 700 may occur due to a simpler genetic background for this trait, compared with those found in the other resistant cultigens. Similarly to our findings, this cultigen has failed to display stable LB resistance in other studies [19,30,31,54,55]. Finally, the Ph-1 gene present in 'NY' [36][37][38][39] provides no reliable protection against P. infestans for field tomatoes grown in Poland, as well as in other locations [2,24,30,32,54,56,57].
Collectively, the field studies indicated that under Polish conditions, LA 1033, L 3707, and L 3708 could be considered promising sources for breeding tomato for LB resistance. This has implications for Central Europe, with field production of both tomatoes and potatoes [1].
In the controlled conditions testing methods used, LA 1033 showed the largest variability in age-dependent reaction to P. infestans inoculation. Furthermore, in contrast to its consistently superior performance in the field assays, LA 1033 proved inferior to both WVa 700 and L 3708 cultigens in the detached leaf and whole plant tests. This observation underscores that the assays under controlled conditions may differ from the field tests (weather conditions, plant age, heterogenic isolate mixture, constant pathogen pressure, presence of other (a)biotic stresses). These results also suggest that full expression of LB resistance in LA 1033 occurs later than the oldest developmental stages investigated (3to 8-week old). Our observation of lower LB intensities in the detached leaves of the 13-to 15-week old plants of this cultigen further supports this hypothesis. The abundant growth of this cultigen may pose problems with generation of appropriate plant materials in the greenhouse for large scale bio-assays. This problem, however, can be solved using the detached leaf tests.
In our study, L 3708 showed higher symptoms variability relative to plant age compared with the WVa 700 cultigen. For L 3708, the trait reached stable expression in 7-to 8-week old plants in the whole plant assays, or in 10-week old leaves. This is in contrast to other studies on L 3708 [15,25], which described high levels of LB resistance when testing only 5-week old plants. Possible reasons for these discordant results include differences in the assumed methodology (isolates used, inoculum preparation and load, conditions of the assays) or in the climatic conditions related to location (e.g., intensity of sun exposure, length of day).
In the detached leaf and whole plant assays, we recorded the lowest variability for WVa 700, with low LB symptoms levels in plants of this cultigen older than 3 weeks. Moderate LB intensities were seen in whole plant assays. Turkensteen [24] reported age- dependent and progressively increasing LB resistance in the detached leaf assays of 6-to 8-week old WVa 700. Similarly, the 6week old seedlings of this cultigen exhibited high resistance against both pathogen isolates tested by Moreau et al. [19]. In contrast to our results showing high LB infection in 3-to 4-week old WVa 700 plants, under 5610 4 sporangia/ml, previous studies reported low LB intensity in this cultigen, under comparable developmental stages and inoculum concentrations [58,59]. The most likely reasons for the observed discrepancies are the pathogen isolates used or the assumed methodology, including the double isolate activation employed in our study.
Apart from the significant influence of plant developmental stage and inoculum concentration on the intensity of LB symptoms in the detached leaflet assay, we observed that LB intensity decreases with plant age. Variability of lesion size on leaflets observed using this assay, particularly in the resistant cultigens, indicated the need for improved methods for testing LB resistance (e.g., sporulation intensity). Thus, the detached leaflet assays were an inefficient testing method, although it is fast and easy. The detached leaflet assays can be used only for identification of susceptible genotypes in initial stages of breeding. Our results of tomato LB intensities are in line with the studies of LB in potato [5,9,60,61], where reported problems regarded high variability, especially for potato genotypes with moderate resistance. Although several tomato LB studies used the detached leaflet method to evaluate LB intensity [16,33,34,42,43,62], they generally included additional evaluations, such as sporulation intensity. In contrast to the detached leaflet assays, other methods used in this study (detached leaf and whole plant assays) proved more reliable for testing tomato LB resistance under controlled conditions. The observed variability among cultigens tested with these methods was significantly lower than this observed for the detached-leaflet assays. These two most effective methods, however, required different inoculum concentrations for successful separation of resistant genotypes. Both detached leaf and whole plant assays suggested an age-related LB intensity. Our results are similar to those using detached leaflet assays, except for more accurate distinction of the resistant genotypes.
Inoculation and incubation conditions influenced the reproducibility of our results. Here, infection and subsequent development of LB symptoms clearly depended on changes in temperature or RH, in accordance with the biology of the pathogen [5,8,33,57,60,63]. Standardizing the test methods (choice of isolate, preparation of inoculum, concentration of inoculum, and conditions of incubation after inoculation) resulted in greater precision in distinguishing LB resistant plants. In the light of our findings, we postulate it very useful to study the plant LB resistance in an age-dependent manner and under the controlled conditions, for each cultigen being reported, in order to better reflect field performance. This might provide an explanation for the differences in mapping of the genes or QTLs controlling the LB resistance trait [15,25,26,28,[64][65][66][67], and might be of help in detailed analyses of the emerging cultigens reported as LB resistant [16,18,20,34,50].
Of the methods studied, the detached leaf and the whole plant tests resulted in the lowest discrepancies, when compared with the field experiments. This may be due to inoculation of larger plant surface area, which may generate lower assessment errors and permit a more accurate evaluation of LB resistance. Both leaf and leaflet assays may additionally exhibit reactions to P. infestans inoculation different from those of whole plants. These may be due to differences in environmental conditions or influences on the physiological and/or biochemical processes, as a result of detachment from the plant. Thus, we propose the whole plant assays as the most reliable method of testing the tomato LB resistance under controlled conditions. This, in the case of some cultigens (notably LA 1033), may pose other challenges, to be circumvented by using alternative methods, such as the detached leaf assays. We agree with previous reports on potato-LB [4][5][6][7][8][9][10][11][12]14] and tomato-LB pathosystems [21][22][23], that the ultimate assessment of LB resistance should be performed with field tests.
Conclusions
Our five-year study under natural field infection, in two distinct Polish locations, confirmed low and stable levels of LB symptoms in LA 1033, L 3708, and L 3707 tomato cultigens. These cultigens are useful for resistance breeding programs for Central Europe. Based on field results, we consider the cultigens carrying the Ph-1 gene (e.g., 'New Yorker') an unsuitable source of LB resistance in Poland. The same is true for cultigens carrying Ph-2 (e.g., WVa 700) if they are used as the only source of LB resistance. Our comparison of three methods for assessing tomato resistance against P. infestans under controlled conditions indicated that each method may be used for different purposes in the resistance breeding. The detached leaflet assay proved useful only to separate the LB susceptible genotypes (such as within the segregating populations) and to maintain the pathogen isolates, but was unreliable for systematic screens due to high variability. Tests on detached leaves and whole plants (greenhouse) generated lower variability than those performed on the detached leaflets and were also found to be highly correlated with field tests. Our results indicate congruent trends for age-dependent expression of LB resistance in all tested tomato cultigens, irrespective of the testing method. The plant age-related LB resistance in tomato reported here, shows the need to optimize and standardize the testing parameters when reporting new sources of resistance. As documented in this study, the reliable comprehensive evaluation of a given cultigen, by means of the optimized and well-suited assays, remains crucial to maximize the benefits from the best performing crops.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2016-05-06T12:33:40.867Z", "created": "2014-10-03T00:00:00.000", "id": "17514836", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0109328&type=printable", "pdf_hash": "f50cb1823de78e2b10321186a0c79844787e3a96", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:9421", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "f50cb1823de78e2b10321186a0c79844787e3a96", "year": 2014}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Transcriptome dynamic landscape underlying the improvement of maize lodging resistance under coronatine treatment
Background Lodging is one of the important factors causing maize yield. Plant height is an important factor in determining plant architecture in maize (Zea mays L.), which is closely related to lodging resistance under high planting density. Coronatine (COR), which is a phytotoxin and produced by the pathogen Pseudomonas syringae, is a functional and structural analogue of jasmonic acid (JA). Results In this study, we found COR, as a new plant growth regulator, could effectively reduce plant height and ear height of both hybrids (ZD958 and XY335) and inbred (B73) maize by inhibiting internode growth during elongation, thus improve maize lodging resistance. To study gene expression changes in internode after COR treatment, we collected spatio-temporal transcriptome of inbred B73 internode under normal condition and COR treatment, including the three different regions of internode (fixed, meristem and elongation regions) at three different developmental stages. The gene expression levels of the three regions at normal condition were described and then compared with that upon COR treatment. In total, 8605 COR-responsive genes (COR-RGs) were found, consist of 802 genes specifically expressed in internode. For these COR-RGs, 614, 870, 2123 of which showed expression changes in only fixed, meristem and elongation region, respectively. Both the number and function were significantly changed for COR-RGs identified in different regions, indicating genes with different functions were regulated at the three regions. Besides, we found more than 80% genes of gibberellin and jasmonic acid were changed under COR treatment. Conclusions These data provide a gene expression profiling in different regions of internode development and molecular mechanism of COR affecting internode elongation. A putative schematic of the internode response to COR treatment is proposed which shows the basic process of COR affecting internode elongation. This research provides a useful resource for studying maize internode development and improves our understanding of the COR regulation mechanism based on plant height. Supplementary Information The online version contains supplementary material available at 10.1186/s12870-021-02962-2.
transcriptional differences between meristem, elongation and fixed regions is far from clear.
The genes related to many hormones have been showed to be involved in the plant height, such as genes participated in biosynthesis, transport and signaling pathways of GA and jasmonic acid (JA) [16,55,97]. Some genes affect plant height by regulating GA synthesis and transduction, such as dwarf1, dwarf3, GA20oxs, GA3oxs, CPS, dwarf plant 8 and dwarf plant 9 [7,73,81,83,101,102,108]. JA affects plant height mainly via complex phytohormone crosstalk with GA and auxin. Studies showed that JA can affect the formation and distribution of auxin by inducing the ASA1 expression and regulating the PINs and PLETHORA [98], thereby affecting cell elongation. In addition, DELLAs, GA signal reverse regulation factor, can interact with the JA pathway to coordinate normal growth and defense to biotic stresses [110]. Therefore, phytohormones are of great significance to control internode development. However, the high production cost and the instability of molecular structure in the vitro environment make direct application of phytohormones very difficult in yield. Plant growth regulators, which are compounds with similar effects to phytohormones, overcome these difficulties [105,112]. Currently, the main component of plant growth regulators used in agriculture is 1,1-dimethylpiperidinium chloride (DPC) or ethephon [70,121]. However, maize is not sensitive to DPC and the ethephon decreases grain yield of maize [52,71]. With the increase of planting density and mechanization level, a more efficient and safe new plant growth regulator is urgently needed.
Coronatine (COR), secreted by Pseudomonas syingae pathovars, is a phytotoxin [46,50,76], with similar function as JA [36,115]. It has been showed that the COR is an analog of JA [100], and is 1000 times more active than JAs [93]. The COR can lead to adverse effects for plants, such as leaf chlorosis and disease symptoms [94]. However, COR of low concentrations can increase the abiotic stress resistance [40,104,124]. At present, COR can be produced by microbial fermentation, and has the advantages of lower environmental pollution and chemical residues. Therefore, as a new environmentally friendly plant growth regulator, COR is expected to be widely used in agriculture. Previous researches have shown that COR can inhibit the elongation of maize root, hypocotyl and mesocotyls [62]. Our previous studies have showed that COR had certain effect on reducing plant height [85,99], while the molecular mechanism of COR in reducing plant height of maize is not well known.
In our study, the plant height of ZD958 and XY335, two wildly cultivated maize hybrids, could be significantly decreased under COR treatment via reducing internode length and thus improve lodging resistance. To research the underlying gene different expression that drive the responses of internode to COR, spatiotemporal transcriptome of inbred B73 internode were produced under control and COR treatment, containing the maturation, meristem and elongation regions of internode. The differences in transcription levels of the three regions at normal condition were displayed and then were compared with that upon COR treatment. In total, 8605 COR-responsive genes (COR-RGs) were reported, and internode specific genes accounted for 9.3% (802 genes). For these COR-RGs, 614, 870, 2123 of which showed expression changes in only fixed, meristem and elongation region, respectively. Gene ontology enrichment analysis indicated that different genes in the three regions control their growth. Moreover, we found that 84% of GA related gens and 80% of JA related genes were significant affected under COR treatment. In summary, the differential expression map of gene expression response in internode to COR provides a theoretical support for future study of the molecular mechanism of plant height decreased by COR.
Results
The plant height of maize is significantly decreased under COR treatment We found that the plant height of ZD958 and XY335, two wildly cultivated maize hybrids, were significantly decreased under the treatment of exogenous COR (10 μM) at the stage with nine leaves, which average decrease of about 5 cm ( Fig. 1a; Additional Fig. 1 A and Additional Data Sets 1). The grain weight per plant displayed no significant change but the yield can be increased due to lower lodging rate under COR treatment as compared with untreated controls in the field (Fig. 1b, c; Additional Fig. 2 A and Additional Data Sets 1). To explore the mechanism of decrease of plant height of maize under COR treatment, we performed the COR treatment at the ninth leaf stage for B73 inbred, which the reference genome was available [48] growing in the greenhouse. The length of 7th internode was not affected due to it was elongated completely before COR treatment, but the elongation of 9th internode was significantly inhibited in 2 days later after COR treatment ( Fig. 1d (Fig. 1g, h and i; Additional Data Sets 1). Besides, we found the fracture resistance of the 9th internode was significantly increased under COR treatment (average from 459 ± 9.63 N to 520.3 ± 11.44 N) ( Fig. 1f; Additional Data Sets 1), which might due to more lateral cell number in the internode cortex (Additional Fig. 2 B). Taken together, COR was a new plant growth regulator which could effectively reduce plant height and ear height of maize by inhibiting cell elongation during internode elongation stage, thus beneficial to improve maize lodging resistance.
The generation of spatio-temporal transcriptomes of maize internode under normal conditions and COR treatment
To explore the mechanism of plant height reduction of maize under COR treatment, we used the RNA-seq to study the transcription level of genes of the fixed region (F) of 7th internode, and the meristem region (M) and elongation region (E) of 9th internode collected in 1st, 2nd, and 4th day after COR treatment (at the stage with nine leaves) (Fig. 2a). For the convenience of subsequent description, which were named as F1_T (treatment), F2_ T, F4_T, M1_T, M2_T, M4_T and E1_T, E2_T, E4_T, respectively. Corresponding control which were collected at normal growth conditions were named as F1_C (control), F2_C, F4_C, M1_C, M2_C, M4_C and E1_C, E2_C, E4_C, respectively. In totally, 3.24 billion reads were obtained by the Illumina sequencer, and average 92.41% (Additional Table 1) of which could be uniquely mapped to the maize reference genome of B73 (RefGen_ V4) [48] by Hisat2 [53]. The normalized gene expression value was descripted by calculating the fragments per kilobase of transcript per million mapped reads (FPKM) on the strength of uniquely mapped reads. The expression level of each sample is descripted by the average FPKM values of two biological replicates because the correlation coefficient of them was high (average value of R 2 was more than 0.93, Additional Fig. 3). To reduce the error caused by transcription noise, here only genes which FPKM values were larger than 1 were defined as expressed genes. In total, the expressed genes were 24, 048 (including 1400 transcription factors (TFs)), which had expression in at least one of the 18 samples (Additional Data Sets 2).
The principal component analysis (PCA) result showed that the 18 samples were generally grouped into three categories, with each category corresponding to a specific internode region, and the COR treated and untreated samples can be separated well (Fig. 2b). In order to further increase the credibility of transcriptional results we obtained, we checked the expression patterns of 12 marker genes, which their expression regions were previously reported. ZmIncw1, ZmNAC109, ZmMYB32, and ZmIRX9 are genes involved in sugar transport, and lignin synthesis process, and were shown to be highly expressed in the fixed region [17,42,69,120,121] (Fig. 2c). ZmGSL1, ZmGRFTF1, ZmCslA1, and ZmEXPA2, related to cell division and cell wall biosynthesis, were Comparison of the grains weight per plant of ZD958 and XY335 with and without COR treatment. No significant change was observed. The data were presented as means ± SE (n = 124). Error bars indicate SE. c Comparison of the lodging rate of ZD958 and XY335 with and without COR treatment. This experiment uses four biological replicate designs for the two maize hybrids. At least 33 plants were collected in each replicate. *: 0.01 < p-value < 0.05; ***: p-value < 0.001. d Comparison of the length of 7th internode and 9th internode of B73 with and without at the three time points. The data were presented as means ± SE (n = 3). Error bars indicate SE. *: 0.01 < p-value < 0.05; **: 0.001 < p-value < 0.01. e Comparison of the finally length of 9th internode of B73 with and without COR treatment. The data were presented as means ± SE (n > 23). SE is represented by error bars. *: 0.01 < p-value < 0.05. f Comparison of the fracture resistance of 9th internode of B73 with and without COR treatment. The data were presented as means ± SE (n = 3). Error bars indicate SE. **: 0.001 < p-value < 0.01. g Gross morphologies of B73 with and without COR treatment. Scale bars, 30 cm. h Comparison of plant height of B73 with and without COR treatment. The data were presented as means ± SE (n > 15). SE is represented by error bars. **: 0.001 < p-value < 0.01. i Comparison of ear height of B73 with and without COR treatment. The data were presented as means ± SE (n > 15). SE is represented by error bars. *: 0.01 < p-value < 0.05.
shown to have highly expression in the meristem region [65,117,125] (Fig. 2d). ZmUXS (UDP-xylose synthase), ZmNST4, ZmCesA8, and ZmCesA2, involved in cell wall biosynthesis, were highly expressed in the elongation regions [2,103,117] (Fig. 2e). The preference of the expressions of these 12 marker genes in our results were consistent with previously reports, which indicated that the fixed region, meristem region and elongation region samples were collected well. The maize was treated with 10 μM COR when the 9th leaf was fully expanded. The fixed region (F, apical 1 cm of the internode) of 7th internode, and the meristem region (M, basal 0-1 cm between internode) and elongation region (E, basal 1-2 cm between internode) of 9th internode collected in 1st, 2nd, and 4th day later after COR treatment. b Principal component analysis (PCA) of 18 samples. c to e The region-specific expression genes mainly expressed in the region of fixed (c), meristem (d) and elongation (e) region. The region of fixed region, meristem region and elongation are display in light yellow, blue and green, respectively. f and g The expression of COR-induced genes. These genes are related to plant defense (f) and JA pathway (g) were showed. *: 0.01 < p-value < 0.05; **: 0.001 < p-value < 0.01; ***: p-value < 0.001 In addition, we found that ZmPAL2, ZmPAL3, ZmPAL5, and ZmPOX1, four genes related to defense processes in maize [26,82], and ZmAOS2a, ZmLOX3, ZmLOX1 and ZmOPR6, involved in JA signaling pathway [30,93], showed significant differentially expression after COR treatment (Fig. 2f, g). This was in line with that COR is not only a phytotoxin by P. syringae but also an analog of JA. In total, our spatio-temporal transcriptomes, which generated for maize internode with or without COR treatment, is high quality and accuracy.
Expression profiling of internode under normal conditions
The spatio-temporal transcriptomes generated here provided us a good opportunity to character the specific expression features of fixed, meristem and elongation regions of maize internode before exploring the effect of COR on transcription of internode.
Totally, 23,349 expressed genes were detected in internode tissues collected in normal condition, including 1357 (5.81%) TFs ( Fig. 3a; Additional Data Sets 3). These genes were classified into 14 co-expression types by the k-means clustering algorithm. The genes (9776 genes, including 472 TFs) in four modules of which were found with expression at more than one of the three regions of internode ( Fig. 3a) indicating the common functional processes in different tissue types of internode regions. Interestingly, there were 58% (13,573) of genes (belonged to eleven modules) mainly expressed at only one of the three different tissues of internode, reflecting the big difference among the fixed, meristem and elongation regions of internode.
Genes mainly expressed in the fixed region (F)
The fixed region of internode was best represented by 7777 expressed genes, including 616 TFs in the module F-I to F-V ( Fig. 3a; Additional Data Sets 3). The genes of module F-I (961 genes, 66 TFs), F-II (855 genes, 46 TFs), and F-III (855 genes, 54 TFs) were mainly expressed at 1st, 2nd, 4th day, respectively, and the genes in module F-IV (1509 genes, 188 TFs) were mainly expressed at 1st and 4th but not in 2nd day, reflecting the transcriptional dynamic during the development of fixed tissue. In addition, there were 3597 genes (including 262 TFs) in module F-V showed continuity expression at the three points of fixed tissue development. Gene annotation analysis showed that these fixed regions associated modules were mainly overrepresented with genes involved in protein kinase activity, amino acid phosphorylation, ATP binding etc. (Fig. 3b). A recent study showed that the increase of ethylene level could favor cell wall synthesis and deposition at fixed region of mature internodes [120,121]. Consistent with this, there were 17 ethylene pathway genes were highly expressed in the module F-I, including 11 ethyleneresponsive element binding protein (EREBP) transcription factors (ZmEREB23, ZmEREB54, ZmEREB97, ZmEREB104 et al.), one 1-aminocyclopropane-1-carboxylate oxidase (ZmACCO20), two ethylene receptors (ZmERS14 and ZmETR40), one gene encoding 1aminocyclopropane-1-carboxylate synthase (ZmACS6), one tasseled gene (ZmTS6) and one bHLH transcription factor (Zmpco106446).
Genes mainly expressed in the meristem region (M)
The 4124 expressed genes in module M-I to M-III, including 160 TFs, were best show the gene expression pattern of internode meristem region ( Fig. 3a; Additional Data Sets 3). The most typical characteristic of meristem region is with vigorous cell division. According to this, the module M-II genes (1670 genes including 65 TFs) were mainly involved in division related processes, including microtubule motor activity, microtubule-based movement, nucleosome assembly, nucleotide binding, helicase activity, DNA replication and repair (Fig. 3b). For example, genes encoding RAD51D and SPO11 family proteins, which were related to DNA replication process [58,72], and genes encoding cyclin family proteins (such as cyclin D1), which were related to G2 phases of cell division [41], were included in module M-II. In addition, the 591 genes of module M-I, including 24 TFs, were mainly expressed in meristem region at 1st day. ZmRAF1, a gene can increase the Rubisco content, and ZmPPD1 and ZmYCF3, two genes can increase the photosynthesis capacity of maize [74,111], were included in module M-I. This might reveal the need of large amount of organic material synthesis in meristem before entering into stage with vigorous cell division. The genes of module M-III (1863 genes, 71 TFs) were expressed in meristem at all three time points. The genes related to energy and hormone signal transduction were found in this module. Such as ZmTIDP3692 and ZmZIM20 play roles in glycolytic pathway and cell number. They play an important role in energy supply and cell division, respectively [1,79] might play an indispensable role in the meristem.
Genes mainly expressed in the elongation region (E)
The genes of module E-I and E-II represents the specific gene transcription level of the internode elongation region ( Fig. 3a; Additional Data Sets 3). Genes in module E-I (654 genes, including 41 TFs) and E-II (1018 genes, including 68 TFs), were mainly expressed in the elongation region at 1st and 4th day, respectively. ZmROP2 and ZmROP9, which are involved in early phase of directional cell expansion [28], ZmCA5P9, which is related to cell elongation [4,28,122], and ZmABI20, a B3 domaincontaining protein might associated with the stem elongation through affecting GA synthesis [38], were also found in module E-I. The module E-II are overrepresented with genes related to lipid transport, lipid binding, fatty acid biosynthetic process, transferase activity, xyloglucosyl transferase activity, cellular glucan metabolic process and copper ion binding (Fig. 3b). Three beta-expansin genes (ZmEXPB5, ZmEXPB6 and ZmEXPB7), which is associated with the synthesis of the primary wall [57], and four cellulose synthase genes (ZmCesA1, ZmCesA2, ZmCesA4 and ZmCesA9) [117] were included in module E-II. Taken together, these results suggested that the genes in module E-I and E-II were closely associated with vigorous cell elongation in the elongation region of internode. In addition, some NACs and MYBs related to cell wall biosynthesis were specifically expressed in this module, such as ZmNAC92, ZmNAC86, ZmMYB23 and ZmMYB27 [120,121].
Transcriptional disturbance of internode under COR treatment
To identify genes exhibiting responses to the COR treatment, each of nine COR treated samples were compared with their corresponding control samples without COR. Finally, a total of 8605 genes including 490 TFs were found with significantly different expression between at least one of the 9 sample pairs at the threshold of the 5% false discovery rate (FDR) and more than 2 fold changes, and were designated as COR-responsive genes (COR-RGs) (Additional Data Sets 4). For these COR-RGs, 3165, 5226 and 6664 of which were identified in the fixed, meristem and elongation regions, respectively. These genes were classified using Venn diagram (Fig. 4), which showed that there were 614, 870, 2123 COR-RGs specifically found in the fixed, meristem and elongation regions, respectively, and only 1452 (16.9% of all COR-RGs) showed different expression in all three type regions of internode. These results reflected the varied effect of COR on different tissue types. Relatively more serious influence of COR on non-fixed tissues, especially for elongation region, was consistent with the observation of plant height decrease under COR treatment.
COR-RGs specifically identified in fixed region of internode
The COR-RGs specifically identified in fixed region were categorized into two groups: up-regulated on fixed regions (F-up COR-RGs) and down-regulated on fixed regions (F-down COR-RGs), which contained 327 genes (including 15 TFs) and 287 genes (including 27 TFs), respectively ( Fig. 5a; Additional Data Sets 4). Gene ontology (GO) enrichment analysis indicated that genes involved in iron ion binding, lipid metabolic process and oxidation reduction were overrepresented in F-up COR-RGs, while genes involved in protein kinase activity and amino acid phosphorylation were overrepresented in F-down COR-RGs (Fig. 5b), including many stress tolerance related genes. Up-regulation of JA signal pathway related genes was associated with enhancement of stress tolerance in maize as reported recently [10,34]. And we found ZmLOX5, ZmLOX6, ZmLOX10, ZmAOS1, ZmAOS3, which were related to lipid metabolic process and response to JA [13,14,30] were up-regulated in fixed region after COR treatment. ZmPSEI7 is a gene encoding cysteine proteinase inhibitor, which the expression can improves the maize insect resistance [10,49,80], was also up-regulated in fixed region after COR treatment. In addition, we found the expression of ZmPOX3 and ZmCYP11, which are two genes related to tetrapyrrole pathway and their high expression is not conducive for plant resistance to biotic stress [27,37,84], were down-related in fixed region after COR treatment.
COR-RGs specifically identified in meristem region of internode
The COR-RGs specifically identified in meristem region were contained by 328 up-regulated genes (M-up COR-RGs, including 23 TFs) and 542 down-regulated genes (M-down COR-RGs, genes, including 42 TFs) ( Fig. 5a; Additional Data Sets 4). The M-up COR-RGs were mainly related to cell cycle, such as regulation of cell cycle and cell cycle checkpoint. ZmCKI4 encodes a cyclin-dependent kinase inhibitor which can inhibit the cell division [31], and ZmKRP1 is a cyclin-dependent kinase inhibitor which can inhibit the cell size number and cell division [77]. Up-regulation of these two genes suggested that the activity of cell division was generally reduced in meristem region of internode, consistent with the decrease of plant height with COR treatment. According to the reduction of activity of cell division, the genes related to cellulose biosynthetic process, membrane and transmembrane transport were downregulated in meristem region (Fig. 5b). For example, many CesA family genes, including ZmCesA1, ZmCesA4, ZmCesA6, ZmCesA7 and ZmCesA9, which are closely associated with cellulose synthesis of cell walls and can affect cell elongation, were identified as M-down COR-RGs. In addition, ZmTRPS1, a gene which can decrease cell division through altered cell wall structure [2,29,54], and ZmBR2, a green revolution gene which affects the transmembrane transporter activity and its low expression can lead to decrease of plant height [64], was also down-regulated in meristem region of internode after COR treatment. Overall, there results reflected an inhibitory effect of COR effect on cell division in meristem regions.
COR-RGs specifically identified in elongation region of internode
The COR-RGs specifically identified in elongation region of internode, which contained 1281 up-regulated genes (E-up COR-RGs, including 23 TFs) and 842 downregulated genes (E-down COR-RGs, genes, including 80 TFs) (Fig. 5a; Additional Data Sets 4), was far more than that specifically identified in fixed or meristem regions of internode. The E-up COR-RGs were enriched with genes related to the regulatory activity (e.g., catalytic activity, nucleotide and RNA binding, and RNA processing), translation (e.g., ribosome, translational elongation) and cell structure establishment (e.g., nucleosome assembly) (Fig. 5b). Previous studies showed that ZmGRAS19 and ZmGRAS58 can disturb cell elongation through affecting the formation of secondary walls, cell proliferation and cell differentiation [56], and ZmDCL101, ZmDCL104 and ZmDCL105, which encode DCL family proteins, are related to defense process and plant height [19,89], and ZmGST10, ZmGST16 and ZmGST22, which encode the glutathione transferases, are related plant defense process [21]. Here we found all these eight genes were grouped in grounded E-up COR-RGs. In addition, we found some genes which expressions were positively associated with cell elongation were downregulated in elongation region of internode. For example, ZmCesA10, ZmCesA12 and ZmCesA13, three cellulose synthase genes which the reduce of expression can inhibited cell elongation [2,24], were down-regulated. In addition, we found that genes related to response to GA, such as ZmGID1 and ZmGID2 [123], were also down-regulated after COR treatment, consistent with the inhibit of cell elongation. Overall, the indicating the defense process was activated and the vegetative growth was inhibited for elongation region of internode after COR treatment.
COR-RGs identified in more than one of the three regions of internode
Besides genes with repose specific in fixed, meristem or elongation regions of internode, there also have a lot of genes (4998, 58.1% of total COR-RGs) showed repose in more than one of the three type regions after COR treatment (Additional Fig. 4). A mainly category is genes (2447) with repose in both meristem and elongation regions but not in fixed regions, in line with the close association of meristem and elongation regions with internode length. There were 1245 genes (including 53 TFs) up-regulated in meristem and elongation regions, which mainly related to cell division, such as nucleosome assembly, helicase activity, nucleosome, DNA replication and microtubule motor activity (Additional Fig. Fig. 5 Spatio-temporal expression pattern of COR_RGs and functional enrichment analysis. a Spatio-temporal expression pattern of COR_RGs in maize internode. The FPKM values of each gene were divided by the maximum value over all samples for normalization. The COR_RGs with significant expression changes in only fixed, meristem or elongation regions were showed here. b Function classification enrichment of genes in different modules is performed using MapMan. Only items of FDR less than 0.05 are displayed 4B), and 1202 genes (including 95 TFs) down-regulated in meristem and elongation regions, which are mainly involved in protein kinase activity, lipid metabolic process, glycosyl groups transferase activity, transcription regulator activity and transferase activity, transferring acyl groups other than aminoacyl groups (Additional Fig. 4B). According to inhibit of cell divide in meristem region and cell elongation in elongation region, the expressions of ZmTHX43 and ZmIRX15 which are associated with xylan biosynthesis andZmCesA8 a constituent of secondary cellulose synthase complexes responsible for cellulose synthesis after cell expansion completion [5,9,22,25], were down-regulated. In addition, we found some genes involved in the auxinactivated signaling pathway, such as ZmARF7 and ZmIAA11 [67], were also down-regulated in meristem and elongation regions.
Internode specific genes with response after COR treatment
The spatio-temporal transcriptome data generated here gave us a good opportunity to identified internode specific genes via combined with the promulgated RNA-seq data of different maize tissues, including leaf, tassel, root, cob, silks, endosperm, pericarp, seed, ear, embryo, and anthers [18,23,59,63,96]. Totally, we identified 1376 genes (including 70 TFs) with specific expression in internode (Additional Fig. 5; Additional Data Sets 5). In these internode specific genes, 58.3% of which (802 genes, including 37 TFs) were belonged to COR-RGs (Additional Data Sets 6), significantly higher than the proportion of total expressed genes accounted by COR-RGs (35.8%), indicating the overrepresentation of internode specific genes in COR_RGs. For these COR-RGs specifically expressed in internode, 427 of which were up-regulated and 375 of which were down-regulated. Grouping according to the regions with expression change, we found 200 internode specific COR_RGs, taking 24.94% of total, were up-regulated in all the three type regions of internode. These F + M + E-up internode specific COR_RGs were enriched with genes related to triose-phosphate isomerase activity (Additional Fig. 6A, B), such as ZmTpi1 and ZmIPS1 (Inositol-3-phosphate synthase) which the expression can initiate the defense mode of plant [47,61]. In addition, ZmSDH (succinate dehydrogenase) and ZmTH1, which are related to the induction of oxidative stress [6,92] are also identified as F + M + E-up internode specific COR_RGs. These results suggested that some defensive reactions were common among the three type regions of internode after CORtreatment. In addition, we found internode specific genes ZmPGP9 which can promote inhibits auxin transport [32], ZmARR7 which the reduce of expression is benefit for improving the defense ability of maize [45], and ZmABI32 which the reduce of expression is favor for drought resistance of plants [78], were specifically down-regulated in fixed, meristem and elongation region, respectively. And the ZmIAA41 genes which related to the auxin signal was down-regulated in both fixed and elongation regions, consistent with the report that the reduce of its expressions can inhibit cell expansion and lead to plant dwarfing [109,118].
Differential expression of phytohormone-related genes under COR treatment
The plant growth and development are regulated by a complex plant hormone crosstalk, while ABA, IAA, GA and JA are critical components in these processes [35,75]. We first studied the regulation of COR on ABA, IAA, GA and JA related genes. Totally, we found 34 ABA related genes were expressed in our data and the expression of 47% (16) genes could be significantly affected by COR. In this research, 169 IAA related genes were expressed and the expression level of 52% (88) genes could be significantly affected by COR. 74 GA related genes were expressed in this research and the expression value of 84% (62) genes could be significantly affected by COR. And 35 JA related genes were expressed in this article, the expression of 80% (28) genes were significantly affected by COR. The results showed that the COR-RGs proportion of GA (84%) and JA (80%) were significantly greater than those of ABA (47%), IAA (52%) and all genes (36%) (Additional Fig. 7). Then we focused on the regulation of COR on GA and JA related genes.
Effect of COR on genes of GA pathway GA, a phytohormones of tetracyclic diterpenoid, plays essential roles during plant growth process. Among the 62 significant differentially expressed genes under COR treatment, most of the genes showed significant downregulation in the meristem region and elongation region. It consistent that reduced GA biosynthesis and suppression of GA signaling pathways lead to reduced plant height and internode shortening [3]. It's worth noting that three famous green revolution genes ZmD3, ZmGA20ox2 and ZmGA20ox3 [73,102] which are affecting GA biosynthesis were down-regulated in the elongation region after COR treatment, it consistent with that the previous researches, the mutants of these genes were observed with a dwarfing phenotype. The ZmGID1 and ZmGID2, F-box proteins modulate DELLA protein degradation, were both down-regulated in the elongation region of internode. And the ZmCPS3, ZmKS and ZmKAO, which were related to GA biosynthesis, were observed to be down-regulated in the elongation region of internode (Fig. 6a). In addition, the gibberellin stimulated-like proteins (ZmGSL1 and ZmCl22897_1a) were identified as being up-regulated by COR in the meristem region and elongating internode. Their homolog gene OsGASR3 was reported to reduce the toxicity of Xanthomonas campestris to rice and involvement in defense and affecting growth and development of rice [8]. While, we also found the DELLA protein ZmGras46 which is involved in controlling GA-induced growth and adaptability to environmental changes [44] was downregulated in the meristem region after COR treatment for 4 days. These results suggested that COR could control the expression of GA metabolic and biosynthesis genes and modulate the signal transduction for repressing internode elongation. Collectively, these regulated GA related genes might be essential for normal growth and defense processes, according to gene function annotation.
Effect of COR on genes of JA pathway
JAs are a small molecules compound derived from lipids that have core position in the transition between plant defense and normal growth. 80% of genes (28 genes) involved in JA were differentially expressed after COR treatment (Additional Data Sets 7). Unlike the gibberellin related genes which were significantly downregulated in the meristem region and elongation region, the JA related genes were most significantly upregulated in the fixed region (Fig. 6b). For example, in the oxylipin biosynthesis, the ZmLOXs, as defense signals, play important roles in inducing defense genes to work [14] most ZmLOXs were up-regulated in the fixed region after COR treatment, included ZmLOX5, ZmLOX6, ZmLOX9, ZmLOX10 and ZmLOX11. The ZmAOSs (Allene oxide synthase) which are responsible for production of JAs were up-regulated, such as ZmAOS1 and ZmAOS2 were up-regulated in the fixed region and elongation region and ZmAOS3 was mainly up-regulated in the fixed region. ZmOPR6 and ZmOPR8 encode enzymes with catalytic function that the adjacent double bond of α, β-unsaturated aldehyde or ketone can be reduced were up-regulated in the all regions, it consistent with the previous report that ZmOPR6 and ZmOPR8 are highly promoted by wound-related Fig. 6 The expression of genes in GA and JA pathways was affected by COR. a and b Effects of COR on GA (a) and JA (b) pathway genes. The degree of genes expression was indicated by log 2 (foldchange). Red represents up-regulation and blue represents down-regulation. Colour bar indicates the degree of change. The enzymatic reaction pathway is shown with solid and dashed arrows signaling molecules, such as JA and ethylene [116]. And the ZmOPR7 was up-regulated in the meristem region and elongation region. JAZ proteins as an inhibitory factor of JA signaling were also up-regulated in the fixed region, included ZmJAZ5 and ZmJAZ6 and ZmJAZ10 (Fig. 6b).
Discussion
COR can effectively reduce maize internode length, ear height and plant height The prolonged cloudy and rainy days and other environmental factors always result in severe lodging of maize. Plant height is a crucial determinants of plant architecture in maize and is closely related to lodging resistance and canopy photosynthesis at high planting density. Moderately reducing plant height is an effective strategy for improving lodging resistance in maize grown at high density and bad environment. In this study, we confirmed that COR, as a new plant growth regulator, could effectively reduce plant height and ear height of both hybrids (ZD958 and XY335) and inbred (B73) maize by inhibiting internode growth during elongation and not cause yield per plant decline (Fig. 1). These results are a further verification and complement to previous research [85,99].
Dynamic changes of genes in different regions during internode development
To understand its molecular mechanism of different internode region in response to COR treatment, we firstly analyzed the transcriptome data of the control group by 3 time points and constructed dynamic transcriptome landscape of developmental process of internode different regions (Fig. 3). The provided dynamic transcriptome data clearly demonstrated the three key regions of growing internode, including the fixed region, meristem region and elongation region, which the revealed occurrence regions are consistent with previously reported researches [117,120,121]. 2840, 5973 and 7462 genes were observed mainly expressing in the fixed region, meristem region, and elongation region, respectively, during the elongation growth of maize internode (Fig. 3). This gene bank provides a wealth of resources for future research, that will enhance our cognition of the genetic basis of internode development and also helps to understand the effect of COR on internode elongation. Especially, we detected 1376 stemcharacteristic genes (having 70 TFs), and they will become the focus in future research (Additional Fig. 5).
We found that the number of genes significantly regulated by COR in M and E regions is much higher than that in F region (Fig. 4). This showed that these two regions are most affected by COR, especially E region which is consistent with phenotypic results (Fig. 1d; Additional Fig. 1B). The most genes affected by COR in the E region are related to transcription, translation and protein metabolism. We found that the genes of secondary wall and defense process were up-regulated, which has an inhibitory effect on plant height, such as ZmGRAS19, ZmDCL101 and ZmGST10 [21,56,89] (Fig. 5). In addition, the down-regulation of some cell wall synthesis-related genes, such as ZmCesA10, ZmCesA12 and ZmCesA13, in the E region also inhibited cell elongation (Fig. 5).
COR changed pathway of GA and JA during internode elongation GAs and JAs, two important plant hormones, have a vital role in controlling plant growth and development under the different environment. GA plays essential parts during plant developmental processes, and JA as a regulator controls the response to stress. In our study, we found the most gibberellin synthetic and responsive genes were significantly inhibited in the meristem region and elongation region, it consistent that reduced GA biosynthesis and suppression of GA signaling pathways lead to reduced plant height and internode shortening [3]. We also found JA related genes were most significantly up-regulated in the fixed region, it may be related to that the lignin most is produced and stored in the secondary cell walls of fixed region and the plant defense dominated by JAs is correlated with expression of genes of lignin synthesis [20,43]. These results suggested that COR treatment mainly controlled internode growth by activating the JA pathway in fixed region and inhibiting the GA pathway in the meristem region and elongation region, thereby reducing plant height (Fig. 7).
During the growth process of plants, the balance between defense and growth is a mutual conversion process, which is a necessary condition for plants to coordinate the supply of resources according to various growth clues and environmental challenges. Notably, in our study, we found some JAZ genes, which enable plants to shut down the JAs signaling pathway in time, were up-regulated, while DELLA protein ZmGras46, known as GA signal suppressor, was down-regulated in the meristem region after 4 days of COR treatment. So that the plant can timely from the defense state to the normal growth and development state. These results may explain why COR can effectively reduce plant height, but does not affect the subsequent maize plant growth and yield per plant.
Conclusions
In summary, our transcriptome data displays a map of gene expression during internode development and a difference of gene expression after COR treatment. The biosynthesis and signal transduction of GA in cells of internode elongation region are affected by COR. At the same time, genes related to cell wall and cytoskeleton in the cells of internode elongation region are also inhibited by COR, which affect the normal expansion of internode cells. This may be one of the reasons for the shortening of maize internodes and the decrease of maize plant height after treated by COR. It provides a solid foundation for future researches of the key factors involved in regulating internode length through COR and a theoretical basis for the application of COR.
Plant materials
The hybrids of maize ZD958 and XY335 were used in our experiment which were collected from the Henan Golddoctor Seeds Co., Ltd. and Shandong Denghai Pioneer Seeds Co., Ltd., respectively. The inbred of maize B73 was used in our experiment which were collected from the National Maize Improvement Center of China. The COR was purified by the Centre for Crop Chemical Control, College of Agriculture, China Agricultural University. The ZD958 and XY335 were cultivated in Jinan (36°40′N, 117°00′E), Shandong Province, China, during the summer of 2018. The maize B73 was cultivated in the greenhouse characterized by 16 h /8 h photoperiod, 25°C /18°C day/night temperature.
The COR treatment
Coronatine was purified by the Centre for Crop Chemical Control, China Agricultural University. Coronatine purity was > 99%, measured with high performance liquid chromatography (Milford, MA, USA). COR was dissolved in 10 folds (m/v) methanol and then diluted with water before foliar spraying. The time of treating with COR is the third day after the 9th leaf is fully deployed. The concentration of COR is 10 μmol·L − 1 and the total amount of liquid is 7 ml·plant − 1 [60,85,86]. The maize treated with water, which was added the same amount of methanol as the experimental group, is control.
Determination of phenotyping
In the field experiments, we confirmed observation of the plant height, grain weight per plant and yield. We measured plant height by ruler in a separate experiment at late stage of filling, which included two treatments (COR or water) in a design of completely randomized. After harvesting, the yield and grain weight per plant are measured.
In the greenhouse, the length of 7th and 9th internode was measured at three time points (1st, 2nd, and 4th day after treatment, Fig. 2a). In addition, the length of 9th internode of B73 was measured by ruler and after measuring the height of plant and ear at late stage of filling.
And the fracture resistance was tested by stem strength tester YYD-1 (Zhejiang TOP instrument Co., Ltd., Hangzhou, China). Finally, significance analysis of these data was conducted by t test using the software Graph-Pad Prism 8 [90].
Microstructural observation of internode
On the late stage of filling after COR treatment, the middle region of the 9th internode was collected from the stem of maize. The samples were processed in Carnoy's solution (75% ethanol and 25% acetic acid mixed in equal volume) for 10 h, and then saved in 70% ethanol. Cross sections were produced from the 9th internode by double-edge razor blades and then treated with safranin. The stem microstructure was observed using Olympus BX51 microscope (Olympus China Co., Ltd., Beijing, China) basing on the methods of Xu et al. [105].
Experiment design
Our experiment comprised two factors in the completely randomized factorial design. In this experiment two replicates were designed. The specific information of experiment as follows: (a) COR factor with two levels (control of water and treating with 10 μmol·L − 1 COR) and (b) region of sampling a segment of approximately 4 mm in the top 0-1 cm region of 7th internode (F), a segment of approximately 4 mm in the base 0-1 cm region of 9th internode (M) and a segment of approximately 4 mm in the base 1-2 cm region of 9th internode (E) with three time points (1st, 2nd and 4th day after treating with 10 μmol·L − 1 COR) [119] (Additional Fig. 2). Thirty-six samples were taken for RNA extraction. Each sample After treatment with COR, the JA pathway in fixed region is activated and GA pathway in the meristem region and elongation region is inhibited so that inhibit the growth of maize plant was collected form at least three plants with the scalpel, collected in a 50 ml tube, immediately placed in liquid nitrogen, and finally stored in an ultra-low temperature refrigerator (− 80°C). Each time a sample was taken, the scalpel was rinsed with Milli Q water.
RNA extraction and preparing library
Total RNA from all the samples was extracted using the Trizol (produced by Invitrogen) basing on the manual. Then the total RNA was purified by magnetic stand (Invitrogen). The Aliquots of total RNA purified were stored in the − 80°C refrigerator. The libraries of sequencing were constructed by 5 μg total RNA using the TruSeq™ RNA sample preparation Kit (Illumina Inc., San Diego, USA) following the instructions of manufacturer. According to the protocol of library construction (Illumina), synthetic cDNA was treated with end-repair, phosphorylation and 'A' base addition. After PCR treated by NEB's Phusion DNA polymerase for 15 cycles, selection of size was performed for target fragments of cDNA on 2% Agarose of Low Range Ultra (Bio-Rad). The size of cDNA target fragments is 200-300 bp. Then the libraries were quantitated with TBS380 Picogreen (Invitrogen). All libraries of paired-end sequencing were sequenced using the HiSeq xten (2 × 150 bp read length) (Illumina Inc., San Diego, USA).
RNA-seq data analysis
In order to align the reads of paired-end and control the quality of reads, we trimmed the paired-end reads and filtered the illumina reads with the SeqPrep ( [URL]:// github.com/jstjohn/SeqPrep) and Sickle ( [URL]. com/najoshi/sickle), respectively. Then, the mapping of reads to the reference genome of maize (from the Mai-zeGDB) is performed using the Hisat2 [53]. The unique mapped reads were processed using the Cufflinks (V2.2.0) software [33]. PFKM was used to indicate the gene expression level. The R 2 between biological replicates was calculated. And correlation pictures were made through the prcomp function of R software [87] with initial settings to be convenient for graphic description of correlation among all samples with log 2 (FPKM+ 1).
The prcomp function in R software was used for PCA analysis [87] with original parameters to be easy to graphic display of relatedness among all samples. The log 2 (FPKM+ 1) of the genes were used for the analysis of PCA by R (V 3.6.1).
Gene coexpression analysis
Using the k-means algorithm of MeV (V4.9) software for the co-expression analysis for 9 different no-treatment samples [87]. The normalized expression of genes was operated by dividing their expression level at all samples with their maximum FPKM. The optimal cluster number was determined by the Figure of merit [113].
Differential expression analysis
In order to discover COR-RGs between two different samples, following the method of FPKM, each transcript's expression level of was calculated. Then the differentially expressed genes were calculated by using Cuffdiff, a part of the Cufflinks package ( [URL]. cbcb.umd.edu/) [91].
Functional enrichment analysis
Then using the function annotation module in MapMan (v3.6.0) [88] for evaluating functional category enrichment with each co-expression module. After choosing the representative protein (which was the longest protein of each gene) and running the Mercator with default settings, we conducted the MapMan annotation. Whether there are too many functional categories for a given module was tested by Fisher's exact test. The Benjamini-Hochberg correction was used to result p-values were adjusted to Q values, and 5% fault tolerance rate was applied.
Screening expression of stem-specific gene
For screening of stem-specific genes, 18 stem samples collected from our study and 19 non-stem transcriptome data [18,23,59,63,96] collected from the NCBI ( [URL]:// www.ncbi.nlm.nih.gov/) were used. We used an already reported method [11,114]. Firstly, the normalization of the expression values of all samples was performed with log 2 (FPKM+ 0.01). Secondly, the z-scores of the genes collected in different stem tissues compared with the nonstem tissues using the normalized expression value was performed. If one gene had a z-score greater than 3 in at least one of the samples of stem, this gene was determined to be stem specifically expressed. Then, combining the differentially expressed genes from the transcriptome data that we generated, we further explored the effects of COR for genes expression by performing co-expression analysis using the MeV (V4.9) software.
Additional file 1: Fig. S1 The effects of COR for ZD958, XY335 and the internode of B73. (A) Gross morphologies of ZD958 and XY335 with and
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2020-06-11T09:10:56.659Z", "created": "2020-06-04T00:00:00.000", "id": "225769786", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://bmcplantbiol.biomedcentral.com/track/pdf/10.1186/s12870-021-02962-2", "pdf_hash": "f9d0816230e5f23840c6545f2f37f202c050eae2", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:8783", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "0721dd2c2996a3cd35802778c1957d85aa74162f", "year": 2021}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Impact of inorganic salts on vase life and postharvest qualities of the cut flower of Perpetual Carnation
This study was carried out in the laboratory of Shangqiu Institute of Technology, Henan to investigate the effect of a different combination of inorganic salt on the quality and physiological characteristics of cut flowers (CFs) of Perpetual Carnation. Furthermore, to find out the best preservation solution of inorganic salt that can enhance the ornamental value of CFs of Carnation and prolong its vase life. Sucrose, 8-hydroxyquinoline, paclobutrazol, salicylic acid and different kinds of inorganic salts were added as a preservation solution. And the same amount of distilled water was used as control. The effects of these various inorganic salts on the morphological characteristics including vase life, changes in flower stems, fresh weight (FW) and water balance and the physiological characteristics including contents of malondialdehyde (MDA), cell membrane permeability and the contents of proline of carnation were investigated. The CFs placed in vase solution with inorganic salts showed significant changes in its morphology and physiological characteristics as compared to control. The changes in flower diameter (FD), FW, malondialdehyde and cell membrane permeability showed an increasing trend first and then decreasing. The value of water balance was observed with a downward trend. However, the vase life, FD, the contents of malondialdehyde, contents of proline and FW of CFs held in the preservative solution containing inorganic salts were increased than that of control. The fresh preservative solution contained sucrose 3% + 8-hydroxyquinoline (8-HQ) (200 mg·L ̅1) + paclobutrazol (100 mg·L ̅1) + salicylic acid (SA) (25 mg·L ̅1) + CaCl2 (100 mg·L ̅ 1) has the best effect on longevity (34 days), FW and FD of carnation CFs. This solution has improved the ornamental and physiological characteristics of fresh carnation CFs.
8-HQ significantly extended the post-harvest life as well as the gain of fresh weight (FW) of CFs of carnation in as compared to control (Nowak, 1990;Knee, 2000). When CFs of chrysanthemum were placed in HQ solution lead to increased its longevity, FW and lowest the water loss (Hussein, 1993). Sucrose acted as the best preservative solution and enhanced the post-harvest life of carnation by reducing ethylene production in petals (Pun et al., 2003). Addition of CaCl 2 has prolonged the flowering periods in CFs . There are also a lot of researches on how to extend the vase life of carnation CFs. These studies mainly focus on chemical reagents, nutrients, fungicides, temperature, humidity, illumination. Considerable progress has been made (Liu et al., 2009;Kazemi and Ameri, 2012;Ahmad et al., 2014). According to the results of previous studies, the growth process of plants is inseparable from glycogen and various inorganic salts. Glycogen is generally considered to be one of the important substances for maintaining regular respiratory and photosynthesis of plants, but improper use of sucrose will also affect fresh CFs. The life of the bottle insert (Liu et al., 2009), 3% sucrose proved to have the best preservation effect on fresh CFs of carnations. Inorganic salts are also crucial in maintaining the ornamental and longevity of fresh CFs (Changli, 2007;Yuping, 2009;Edrisi et al., 2012;Anwar et al., 2014). Different inorganic salt ions have different effects on the vase life of fresh CFs. The present study was conducted to evaluate the vase solution with the addition of various inorganic salts on the post-harvest life, quality and ornamental values. Therefore, it is interesting to investigate how holding solutions containg inorganic salts associate to the above mentioned factors that are involved in the process of cut flower senescence. Hence, we tried to study the role of inorganic salts in the process of flower senescence by determing the flower senescence related parameters such as MDA contents, ethylene prododuction,, proline content, water balance, cell membrane permeability, antioxidant activites during flower shelf life of CFs of carnation.
Experimental materials
The fresh CFs of the big red season were used in this experiment. These flowers were purchased from Yiyang Flower Market in Shangqiu City, Henan Province. In this experiment, fresh and healthy plants with relatively uniform appearance and without pests and mechanical damage were selected. The diameter of the flowering branches was cut at an angle of 45° before the experiment. The diameter of the flowering branches was recorded about 40 cm. The top
Introduction
Carnation (Dianthus caryophyllus L.) is a perennial herb of the genus Dianthus. It is one of the most important ornamental flowering plants and broadly used as cut flowers (CFs) (Ali et al., 2008;Onozaki, 2018) and bending plant in orchids. It has a variety of fresh CFs which are consumed in the floriculture market. It is cultivated in many countries and is widely distributed in Fujian, Hubei, Yunnan, Ningxia and other places in China, but China began to study on the fresh CFs relatively late and did not pay enough attention to the various techniques that used for the preservation of fresh CFs. After picking, it is easy to lose water results in wilting, and premature aging affects its ornamental value . In order to satisfied the consumer's demand for fresh flowers and growers profitability, it is necessary to find the best preservative solution which can prolong the vase life and improve the quality of carnation CFs.
Carnation is one of the most economically important CFs and plays a significant role in the floriculture trade. However, post-harvest senescence occurs within a few days and is a substantial limitation in the marketing of CF of carnation. Generally, early post-harvest senescence is caused by the production of ethylene synthesis. Post-harvest causes petal in rolling, and by the accumulation of bacteria on the cut stem surfaces, which produce extracellular polysaccharides that block xylem vessels and thereby increase hydraulic resistance, subsequently leading to a reduction in water uptake by the stem and premature wilting (van Doorn et al., 1995;Bowyer et al., 2003).
However, the longevity of CFs restricted by several factors such as weight loss and decay, senescence, air embolisms (Jones and Hill, 1993;Van Ieperen et al., 2001;Huang et al., 2002;Bazaid, 2004). These decays are due to bacteria, which present at a high level in preservative holding solution used by farmers, which restrict water supply to ornamental CFs and affect the post-harvest life of ornamental CFs by blocking the vascular system (van Doorn and D'hont, 1994;Loubaud and van Doorn, 2004). This blockage decreases water uptake and result in water-deficient stress, which was articulated in the form of early wilting of ornamental CFs (Put et al., 2000), led to early loss of cell turgidity and might become visible when water uptake and transpiration are out of balance during a lasting period of time, which was the result of an unrecoverable condition and the early end of CFs (Noman et al., 2017) .
The previous study showed that germicide 8-hydroxyqunoline (8-HQ) is vital preservatives used in commercial florist industry (Nowak, 1990). Application of 4 leaves were retained in flowering branches. The different chemicals such as sucrose, 8-HQ, paclobutrazol, SA, KCL, NaCl, CaCl 2 , KALSO 4 and distilled water were used in the current study.
Experimental design
This experiment consists of five treatments: A, B, C, D and E. CaCl 2 (100 mg·L‾ 1 ) was added to the basic preservation solution in treatment A. On the other hand, KCL and NaCl (200 mg·L‾ 1 ) were added to the basic preservation in treatment B. Likewise, KALSO 4 (150 mg·L‾ 1 ) was added to the basic preservation solution in treatment C; the treatment D was treated as a primary preservative, and distilled water was used as a control in treatment E. The preservation solution was poured into a 500 mL Erlenmeyer flask. Each flask was contained 250 mL of preservative solution and four carnations. The five replications were carried out in each treatment in this study. After being processed, it is placed in a light-transmitting room with ventilation and no direct light illumination at a temperature of 25 ± 2 °C and relative humidity of 60-80% (Table 1).
Observation indicators and methods of measurement
From the day of bottle insertion, the relevant indicators of the morphology and physiological characteristics of the CFs of the four seasons of carnations are regularly measured.
Determination of the shape index of fresh CFs of carnation
Bottle life observe the morphological changes of fresh CFs every day. The life of CFs is regarded as the end of the vase life by 50% petal loss or wilting (Ding et al., 2011).
Flower diameter change
Method: Using the cross measurement method, the maximum FD of each flower was measured with a ruler at 16:00 every day, repeated twice, and the average value was taken. (Note: When half of each repetition loses its ornamental value, its flower path is no longer measured.)
( )
Maximum flower stem Flower dimeter average per day change rate % 100% Initial flower diameter = ×
Fresh weight change and water balance value determination:
The weighing method is used to determine the difference between water absorption and water loss (Zhao et al., 2016). The water absorption and water loss of the fresh CF water level of the carnation were measured at 16:00 every day, and the flower weight was measured.
Determination of physiological characteristics of plants
The content of free proline in plants was determined by the ninhydrin method (Lee, 2000;Li, 2000;Shabnam et al., 2016).
The content of MDA (malondialdehyde) was determined by the thiobarbituric acid method (Draper et al., 1993;Zhao et al., 1994).
Statistical analysis
The experimental data were statistically analyzed using SPSS21.0 software, and Excel 2010 was used for chart drawing. The significane differences among th means were anlysed at P< 0.01 0r 0.05).
Vase life (days) of Carnation fresh CFs
Different treatments have different effects on the vase life of carnation fresh CFs (Figure 1). From treatments A to E, the average vase life of carnation fresh CFs was recorded as 34d, 25d, 29d, 26d and 20d, respectively. The vase life was longer in treatment A (Sucrose 3% + 8-HQ (200 mg·L‾ 1 ) + Paclobutrazol (100 mg·L‾ 1 ) + SA (25 mg·L‾ 1 ) + CaCl 2 (100 mg·L‾ 1 )) which resulted in 34d in comparison with 20d of the treatment E (control). Furthermore, it was observed that the vase life of treatment A was also significantly higher than the other two treatments (B and C) that contained holding solution with inorganic salts. It indicated that a preservative solution contained CaCl 2 is the most beneficial to extend the vase life of carnation CFs.
Changes in the flower diameter of carnation CFs
Flower path has always been one of the most important morphological indicators of fresh CFs. In this study, The preservative solutions containg inorganic salts promoted the opening of the carnation, vase life of of cut flowers, and the the FD. The trend of FD change, first increased and then decreased. However, the FDs of fresh CFs in different treatments were different. The maximum FD observed in treatment A was 7.15 cm and followed by f 7.08 cm, 7.01 and 6.76 cm in theC treatments B, C and D respectively.. The FD recored in control treatment (E) was 6.83 cm. These results indicated that the addition of CaCl 2 in the basic preservative solution increased the FD of fresh CFs as obserced in tratments (A, B and C). The FD was significantly increased in treatment A as compared to the control and basic solution without CaCl 2 ( Figure 2).
Besides, in each treatment, the added value of fresh CF diameter is also different (Figure 2). The FD in the treatment A increased by 2.87 cm compared with the initial FD. Secondly, in the B, C, and D treatments, the FD increased by 2.32 cm, 2.07 cm, and 1.39 cm, respectively, compared with the initial FD; increase by 2.33 cm. In each treatment, the FD increased the most in the A treatment, the least FD was observed in the control E, and the maximum increase in the FD was 23.2% larger than the minimum increase. It can be seen that the preservation liquid with CaCl 2 has a better effect on increasing the FD, and the increase is remarkable.
From the change of the trend of the line chart in Figure 2, it can be seen that the different treatments make the time for the four-season carnation fresh CFs to reach the maximum FD and the number of days to maintain the larger FD. In the treatment of A, the fresh CF reached the maximum FD on the 19d, and the time for maintaining the larger FD was up to 20 days. Therefore, the life of the fresh CF in the A treatment can reach 34d; the B treatment can make the fresh CF reach the maximum FD at 12d. However, the B treatment maintained a substantial flower path time of only 12d, and then the fresh CF died very quickly, and its lifespan was the only 25d; the C treatment made the fresh CF reach the maximum FD in 19d, and the larger FD time was 21d. The life of fresh CFs was 29d; the maximum FD was reached at 20d in D treatment, and the maximum FD was 14d. In the E treatment, the fresh CFs reached the maximum FD at 15d, and the larger FD was maintained for 10d. The results showed that the fresh CF had the longest life in the A treatment with calcium salt preservative solution, up to 34d. B treatment, that is, the fresh liquid added with potassium salt and sodium salt can quickly promote the bloom of fresh-cut carnation, but not very good. Maintain the life of the carnation CFs. In summary, it can be concluded that the A treatment, that is, the preservation solution of adding calcium salt can maintain the bloom and life span of the fresh CFs of the four-season carnation, maintain the preservation effect, and has the best preservation effect on the fresh CFs of carnation.
Fresh weight changes of fresh CFs of four seasons carnations
The FW of carnation fresh CFs is one of the critical indicators that affect the vase life of CFs. The change of FW in each treatment is consistent with the change of FD. With respect to preservative solution effect on maximum increase of FW of carnation CFs, date presented in Figure 3 indicated that, a significant higher FW was measured in treatment A (Sucrose 3% + 8-HQ (200 mg·L‾ 1 ) + Paclobutrazol (100 mg·L‾ 1 ) + SA (25 mg·L‾ 1 ) + CaCl 2 (100 mg·L‾ 1 )) as compared to control. FW of carnation CFs in treatments B and C was also significantly higher than those of control. The holding solution of treatments A, B and C consist of inorganic salts.
The vase life of CFs is more than 25d. The FW of the fresh CFs in treatment E was not apparent, and water loss occurring after 20d, indicating that the preservation liquid added with inorganic salts was more beneficial to prolong the vase life of the carnation fresh CFs.
It can be seen from the peak in the treatment A (Figure 3), the FW of the fresh CFs of the carnation increased, and the maximum FW increased by 16% compared with the initial FW. The maximum FW in treatment E increased by 2.4% compared with the initial FW (%) indicates that the A-preserved fresh-keeping solution, that is, the
Vase life of Carnation
Braz. J. Biol. 2020, Ahead of Print 5/9 5 addition of calcium ions in the basic preservative solution has a great influence on the FW of the fresh CFs of the four-season carnation. In the later stage of the fresh CF, the FW of the liquid containing the inorganic salt is higher than other controls. And the length of the bottle inserted is long, indicating that the fresh-keeping liquid containing inorganic salt has obvious fresh-keeping effect on the fresh CFs of the four-season carnation, and the preservation effect of the fresh-keeping liquid added with calcium is the most significant.
Changes in the value of water balance in carnation CFs
It was noticed that among the five different treatments, the value of water balance in the fresh CFs of carnation has a decreasing trend (Figure 4). In all treatments, the value of water balance in the early stage of carnation CFs decreased rapidly, positive value; the medium-term decline was slow; the late stage was up and down, which was negative. It can be concluded that in the early stage of bottle insertion, the water absorption of the CF was higher than the water loss, but with the passage of time, the water absorption decreases continuously in the middle and late stages of the bottle insertion, the water loss increases steadily, and the water balance value gradually decreases to negative. In treatments B and C, the water balance values of the fresh CFs of carnation began to appear negative at 11d and 12d after the bottle insertion and delayed by 3d compared with the control E on the 9d. In the treatment D, the water balance value of the fresh CFs of carnation was later than the negative treatment time of the three treatments A, B and C, which was earlier than the negative value of the control E. The overall effect of the treatment D was good.
In all treatments, the water balance value of carnation CFs in treatment A began to appear negative after 13d of bottle insertion, which was delayed by 4d compared with control treatment E. It can be seen that the holding solution containing CaCl 2 is most advantageous for maintaining the stability of the preservative solution.
The contents of MDA in carnation CFs
The contents of MDA in the carnation CFs were measured in this experiment. It was noticed that the contents of MDA firstly increased and then decreased in all treatments ( Figure 5). In the three treatments of A, B and C containing inorganic salts, the contents of MDA in carnation CFs were measured throughout the experiment. The preservative holding solution containing inorganic salts can reduce the contents of MDA in cutting carnation. This leads to slow down the aging of carnation CFs by increasing the resistant in CFs of the carnation. It was observed in treatment A that contents of MDA in carnation CFs were lowest, suggesting that CaCl 2 is most beneficial to prolong the vase life of CFs of the carnation.
Below the control, it indicates that the preservative solution containing inorganic salts can reduce the MDA content during the vase cutting of the carnation, increase the resistance of the carnation fresh CFs, and slow down the aging. In each measurement, the MDA content of carnation fresh CFs was the lowest in the treatment of calcium-containing A, indicating that the A-preserved fresh-keeping solution is most beneficial to prolong the vase life of carnation CFs, and the preservation effect is the best.
Effect of different inorganic salts on cell membrane permeability of carnation fresh CFs
The poor environment can cause damage to the plasma membrane of the cell, increase in cell membrane permeability, electrolyte extravasations. Extravasations of the electrolyte cause changes in the conductivity of the solution, which in turn reflects the extent of damage to the cell membrane. The higher the conductivity value, the greater the permeability of the cell membrane. The higher the degree, the more ionic extravasations in the cytoplasm and the higher the degree of aging. It was noticed that the electrolyte of the fresh CFs of the carnation generally increased first and then decreased in all treatments (Figure 6). The treatments A holding solution containing calcium salt minimizes the relative leakage of the flower electrolyte. It concluded that preservative holding solution containing CaCl 2 is the most beneficial to keep the flowers fresh.
Effect of different inorganic salts contents of proline CFs of carnation
The level of proline in plants is one of the important indicators contributing to the strength of plant stress resistance. To determine the resistance of fresh CFs of carnation in different treatments, the contents of proline in the CFs were chosen as the measurement index. It was observed in all treatments; the contents of proline in the flowers of carnation were first increased and then decreased ( Figure 7). The highest contents of proline in CFs were measured in the early, middle and late stages of bottle insertion in the treatment A. Therefore, it is concluded that the preservative solution containing calcium ions showed the vigorous resistance and is the most helpful to the accumulation of proline in CFs. It is the best preservative solution for CF of carnation.
Discussion
Fresh CFs are living organisms and play an important role in improving people's living environment and quality of life (Raza et al., 2018) Fresh CFs still have life activities after detaching from the mother plant and continue to consume nutrients. However, CFs unable to continue their normal life activities due to the loss of nutrients such as carbohydrates and inorganic salts (Noman et al., 2017). Several factors including hindering the water absorption by flower stem, dryness of petals, wilting, water loss, reducing of FW ) -dull color, fungal and bacterial infestation at cut sites of the flower branch led to the aging of the CFs. One of the big problems in post-harvest CF physiology is the blockage of the vascular system. Several factors, including air or bacterial growth, plant reaction to the actual cut might be responsible for the blockage of the vascular system. When flower stem detached from the mother plant, certain enzymes are moved to the wounded area, which may produce the chemicals that seal the wound (Loubaud and van Doorn 2004). This phenomenon minimizes water uptake. Water stresses occur due to the blockage of xylem vessels led to early wilting of flowers (Put et al., 2000). The results from this study confirm that exogenous application of different bacterial strains induced a substantial impact on improving the longevity of CFs of carnation.
The ornamental value of fresh CFs is mainly attributed by FD, FW and moisture contents. The CFs have large diameter led to the increase the ornamental value. In this study, maximum FD (7.15 cm) was obtained in holding solution supplemented with CaCl 2 followed by 7.03 cm and 7.01 cm in treatments (B and C) as compared to control (6.83 cm). In addition, significantly FD (2.37 cm) increased from initial FD in treatments A in comparison with treatments (D and E). It indicated that FD was significantly higher in treatments having inorganic salts particularly CaCl 2 . The change of FW was consistent with the change of FD, as previously reported by (Soad et al., 2011). The results of different concentrations of salicylic acid on the CF of carnations are similar (Anwar et al., 2014). FW loss is one of the most important physiological disorders of ornamental flowers after harvest, which reduced the vase life and quality. To enhance the vase life of CFs, maintain their FW is significant for commercial values (Saeed et al., 2016). Our results showed that minimum flower FW was observed in control treatment (distilled water) and maximum flower FW was observed in treatment A (Figure 3). Generally resulting in higher FW of flowers is considered good because they may result in extending vase life compared to those showing fewer ones (Soad et al., 2011). The preservative solution contained CaCl 2 with 3% sucrose decreased flower weight loss. This may be due to the effect of sucrose in delaying petal aging and flower wilting (Halevy and Mayak, 1979). These results are an agreement with (Soad et al., 2011), found that flowers placed in holding solution contained CaCl 2 showed higher FW in flowers of Gerbera. Application of inorganic salt in vase solution increased the FW of CFs of Gerbera than control (Shabanian et al., 2018). It can be seen that the holding solution containing CaCl 2 is most advantageous for maintaining the stability of the preservative solution.
Water relation is another important factor affecting the longevity of commercial CFs (van Doorn, 2012). After harvesting of CFs, water deficit and wilting of petals occur due to disturbance of water balance that affected by water loss rates and water uptake (Lü et al., 2010). Our results showed that the vase solution contains CaCl 2 significantly maintain the water balance than those of control ( Figure 4). The data of present study indicated that accumulation of proline in carnation CFs was higher in vase solution with the addition of CaCl 2 than water (control) (Figure 7). This result further supported that carnation CFs in distilled water suffered more severe water than vase solution containing inorganic salts. Proline has been broadly used as a stress indicator for evaluation levels of different stresses including water deficit (Mostofa et al., 2017;Shabanian et al., 2018). The results from our study were in agreement with the results published by previous studies conducted on ornamental CFs, which showed that sufficient water uptake is an essential factor for maintaining a favorable water balance and prolong the vase life of ornamental CFs (Perik et al., 2012;Shabanian et al., 2018).
It is evident from the Figure 1 that maximum vase life of carnation flowers was recorded by using treatment A as compared to control (distilled water). Similar results were obtained by (Soad et al., 2011) on Gerbera and by (Dineshbabu et al., 2002), who demonstrated that similar treatment extended the vase life and improved the flower quality of dendrobium.
Oxidative stress led to senescence in ornamental CFs occurs due to the presence of bacteria and fungi vase solution . Antioxidant enzymes are vital in protecting the cell from pathogens by reducing the oxidative stress and water loss (Hoque Hossain et al., 2016;Khalid et al., 2018;Khalid et al., 2019). MDA is a product of membrane lipid peroxidation during the senescence process (Zafar et al., 2019) . It maintained the cell membrane system and the plasma membrane integrity during the longevity of ornamental CFs. Its content can reflect the senescence of fresh CFs. The lower contents of MDA in CFs led to delayed aging. In all treatments, the contents of MDA in fresh CFs of carnations increased first and then decreased. Furthermore, application of inorganic salts notably CaCl 2 reduced the contents of MDA in carnation CFs as compared water, increased the vase life of carnation CFs. Similar results were observed in gerbera (Shabanian et al., 2018).
The stress resistance slows down it's aging, which is similar to the results of previous studies (Zhao et al., 2016) the effects of five treatments on the cell membrane permeability and proline content of carnation fresh-CFs and their MDA content. The performances of the three treatments (A, B and C) are better than those of the other two treatments, especially the A treatment fresh-keeping solution containing calcium salt.
Conclusion
In conclusion, longevity and quality of ornamental carnation CFs were related to their efficient ability in water uptake and higher capacity in protecting themselves against pathogen and oxidative stress. Application of inorganic salts, especially CaCl 2 as a preservative solution to carnation CFs prolong the vase life and ornamental values of CFs detached from the mother plant. Treatment A (3%+8-HQ (200 mg·L‾ 1 ) + Paclobutrazol (100 mg·L‾ 1 ) + SA (25 mg·L‾ 1 ) + CaCl 2 (100 mg·L‾ 1 ) significantly increased the vase life of carnation CFs by reducing the water deficit, MDA contents, stresses and maintaining water balance as compared to control. It may also be due to the significantly alleviated bacterial related blockage in the end stem of carnation.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2020-04-23T09:07:45.869Z", "created": "2020-04-22T00:00:00.000", "id": "216084244", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.scielo.br/j/bjb/a/5gVqPg6JYHzXjGPQgwZ449N/?format=pdf&lang=en", "pdf_hash": "5a975edbfb12d50930629dd267a578a7afcac7fb", "pdf_src": "Anansi", "provenance": "peS2o-0028.json.gz:4585", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "f41f415d7595b9835ffc05110525d50e372aab9a", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Individual Shrink Wrapping of Zucchini Fruit Improves Postharvest Chilling Tolerance Associated with a Reduction in Ethylene Production and Oxidative Stress Metabolites
We have studied the effect of individual shrink wrapping (ISW) on the postharvest performance of refrigerated fruit from two zucchini cultivars that differ in their sensitivity to cold storage: Sinatra (more sensitive) and Natura (more tolerant). The fruit was individually shrink wrapped before storing at 4°C for 0, 7 and 14 days. Quality parameters, ethylene and CO2 productions, ethylene gene expression, and oxidative stress metabolites were assessed in shrink wrapped and non-wrapped fruit after conditioning the fruit for 6 hours at 20°C. ISW decreased significantly the postharvest deterioration of chilled zucchini in both cultivars. Weight loss was reduced to less than 1%, pitting symptoms were completely absent in ISW fruit at 7 days, and were less than 25% those of control fruits at 14 days of cold storage, and firmness loss was significantly reduced in the cultivar Sinatra. These enhancements in quality of ISW fruit were associated with a significant reduction in cold-induced ethylene production, in the respiration rate, and in the level of oxidative stress metabolites such as hydrogen peroxide and malonyldialdehyde (MDA). A detailed expression analysis of ethylene biosynthesis, perception and signaling genes demonstrated a downregulation of CpACS1 and CpACO1 genes in response to ISW, two genes that are upregulated by cold storage. However, the expression patterns of six other ethylene biosynthesis genes (CpACS2 to CpACS7) and five ethylene signal transduction pathway genes (CpCTR1, CpETR1, CpERS1, CpEIN3.1 and CpEN3.2), suggest that they do not play a major role in response to cold storage and ISW packaging. In conclusion, ISW zucchini packaging resulted in improved tolerance to chilling concomitantly with a reduction in oxidative stress, respiration rate and ethylene production, as well as in the expression of ethylene biosynthesis genes, but not of those involved in ethylene perception and sensitivity.
Introduction
Zucchini (Cucurbita pepo L.) can be considered an immature fruit vegetable [1] that is produced and consumed worldwide. The harvest index of immature fruit vegetables, including zucchini, cucumber, eggplant, or green beans, is based principally on size and color, depending upon market needs. For zucchini produced in Spain and consumed in Europe, the fruits have an average length of about 20 cm, just before hardening and darkening of fruit peel, and before undesirable seed development. Immediately after harvesting, the zucchini immature fruit is subject to high rates of water loss and loss of firmness, which contributes to a rapid decrease in postharvest fruit quality and therefore considerable economic loss [1]. Although it is very important to cool the fruit as soon as possible after harvesting, one of the main deterioration symptoms associated with postharvest storage, transportation and marketing is caused by postharvest chilling injury (PCI). Cold storage of zucchini for a minimum of 2-3 days at 4°C induces the appearance of pitting in the fruit surface, and accelerates fruit dehydration and softening [2][3][4][5][6]. The fruit of most current commercial cultivars loses its commercial value when stored at 4°C for less than 7 days [5,6].
To improve the ability of zucchini fruit to withstand cold storage, we are testing different postharvest treatments that may alleviate chilling injury, and identifying different sources of cold tolerance among Cucurbita pepo germplasm, which will certainly allow us to develop new PCI tolerant commercial cultivars. The higher cold tolerance of some of the identified cultivars, as well as the positive response of sensitive cultivars to postharvest treatments that alleviate PCI, has been always found to be correlated with a reduction in oxidative stress metabolisms and cold-induced ethylene production [5][6][7]. Oxidative damage is an early response of coldsensitive zucchinis to cold exposure [8]. Reactive oxygen species (ROS) triggered by cold storage cause injury to cell membranes by breaking down the double bonds of membrane fatty acids and inducing the accumulation of malonyldialdehyde (MDA), a metabolite that is often used as an indicator of oxidative stress and the degree of structural integrity of the membranes in plants subjected to low temperatures [9,10]. We have reported that zucchini cultivars showing higher tolerance to cold storage suffer fewer PCI symptoms and consequently accumulate lower amounts of reactive oxygen species (ROS) and MDA than those which are more susceptible to cold [5]. Moreover, temperature preconditioning treatment, which induces cold tolerance in fruit, also improves their antioxidant status with lower H 2 O 2 content and induction of ascorbate peroxidase (APX) and catalase (CAT) activities [7].
Zucchini is a non-climacteric fruit that produces low ethylene at harvest and postharvest. However, postharvest fruit storage at 4°C is able to induce rapidly the production of ethylene upon transfer to 20°C, peaking at 7 days of cold storage [6]. This cold-induced ethylene does not appear to trigger PCI symptoms in zucchini, but has been found to be correlated with chilling sensitivity, less so in the fruit of those cultivars that were more tolerant to PCI, and in response to temperature conditioning treatments that alleviate PCI symptoms [6].
ISW is a passive modified atmosphere packaging (MAP) in which a polymer film with a selective permeability to CO 2 , O 2 , ethylene and water is not applied to shipping containers or retail packages, but rather to individual units of the commodity. When applied to fruit, the selective permeability of the film, and the interplay of the fruit physiology and the physical environment, produces a change in the initial atmospheric conditions to a desirable atmosphere within the package. ISW reduces postharvest losses and extends the shelf life of a number non-climacteric fruits and vegetables such as citrus, pepper and cucumber [11][12][13], but also some climacteric fruits such as apple and papaya [14,15]. In other climacteric fruits, including tomato and melon, ISW was not fully successful since it enhances undesirable flavor changes or impaired ripening [16,17]. Individual shrink wrapping adds value to fruit and vegetable products, maintaining their freshness for longer because of a reduced moisture loss, chilling injury, firmness and decay, as well as increased protection from mechanical damage during handling and transport. Rao et al. [18] studied the effect of MAP and shrink wrapping on the shelf life of cucumber and reported that shrink wrapping film can extend the shelf life of the immature cucumber fruit for up to 24 days at 10°C. However, the treatment has not yet been tested for the immature fruit of zucchini. In the present paper we have study the effects ISW on different fruit quality parameters in two zucchini cultivars stored for 14 days at 4°C. We found that maintenance of fruit weight and firmness and the reduction of PCI in ISW fruit were accompanied by a reduction in the production of cold-induced ethylene, in the expression of ethylene biosynthesis genes, and in the accumulation of oxidative stress metabolites.
Plant material and experimental design
Zucchini fruits (Cucurbita pepo L. morphotype Zucchini) were supplied by FRUTAS ESCOBY, S. L. Almería (Spain). The cultivars Natura and Sinatra were used because of their contrasting postharvest behavior during cold storage. The fruit of Natura is more tolerant to chilling injury while the fruit of Sinatra is more sensitive [6] during postharvest shelf life. In order to minimize the effects of pre-harvest management, all fruits were of the same phenological age and were grown in the same greenhouse and harvested simultaneously ( Fig 1A). The initial fruit quality and physiological parameters were assessed on a batch of both cultivars. At once the remaining fruits were divided in two groups of 60 fruits per cultivar. The fruits were randomly divided in different groups and then subjected to the postharvest treatments (i) Individual shrink wrapping (ISW) with shrink film microperforated low density polyethylene of 18 μm thickness with the following gas permeability: 7800 ml O 2 m -2 d -1 atm -1 , 42000 ml CO 2 m -2 d -1 atm -1 and 7870 ml C 2 H 4 m -2 d -1 atm -1 , and (ii) a non-wrapped treatment. Zucchini fruits were individually shrink wrapped by passing through an AWETA sorting and shrink wrapping machine at 280°C for 1,5 s.
Wrapped and unwrapped fruits were stored at 4°C for 14 days. At 7 and 14 days, four replications of 3-4 fruits per cultivar and treatment were transferred to 20°C for 6 hours before sampling and evaluation of quality parameters. Ethylene and CO 2 production, weight loss, chilling injury index and firmness were measured. Finally, two samples from the homogenated exocarps of each replication were frozen separately in liquid nitrogen and then stored at -80°C for MDA and H 2 O 2 content and gene expression analyses.
Ethylene and CO 2 production
Ethylene and CO 2 production were determined at 0, 7 and 14 days of storage. Twelve fruits were analyzed for each time, temperature and treatment, i.e. 4 replicates of 3 fruit each. Prior to analysis, the plastic film was removed from ISW fruit, which was then enclosed in sealed 10 litre containers for 6 h at 20°C. After this incubation period, gas samples were taken and ethylene content was determined three times by a gas chromatograph (Varian 3900 GC) fitted with a flame ionization detector (FID). In the same way, CO 2 was measured three times with Check mate II headspace analyzers (Dansensor). Ethylene production and respiration rates were expressed as nL g -1 6h -1 and ml of CO 2 kg -1 6h -1 , respectively.
Evaluation of weight loss, firmness and chilling injury
The percentage of weight loss during storage was assessed by weighing 12 individual fruits at 0, 7 and 14 days after harvesting. The percentage of weight loss of each fruit was calculated according to the following equation: Where Wi and Wf are initial and final fruit weight respectively. Fruit firmness was determined by a Stable Micro Systems Texture Analyzer TA. XT-Plus. A 4 mm-diameter probe was used and penetration was conducted at a speed of 1mmÁs -1 to a depth of 10 mm. Firmness was measured three times in a transversal section in the distal region of the fruit, as this part softens faster during storage.
To assess postharvest chilling injury we measured the surface and severity of PCI symptoms in each fruit after 0, 7 and 14 days of cold storage. The percentage of fruit surface affected by pitting was used to classify each fruit according to the following scale: 0 = no damage, 1 5% damage, 2 = 6-15% damage, 3 = 16-25% damage, 4 = 26-50% damage, and 5 !50% damage [6]. To assess the severity of pitting symptoms the scale was 0 = no damage, 1 = very superficial damage, 2 = superficial, 3 = moderate damage, 4 = severe damage, 5 = very severe damage. 12 fruits were analysed for each treatment and storage time. The final PCI index used was the average of both parameters. PCI was assessed at 0, 7 and 14 days of cold storage. Determination of malonyldialdehyde (MDA) and hydrogen peroxide contents MDA content was determined following the procedure described by [5] with some modifications. Exocarp was homogenized in (1:10 w/v) trichloroacetic acid (TCA) and 0.25% (w/v) 2-thiobarbituric acid (TBA) (1:10, w/v), and then heated at 95°C in a water bath for 30 minutes, followed by immediate cooling in ice and centrifugation at 4000 g for 20 minutes at 4°C. The absorbance of the supernatant was measured at 532 and 600 nm. MDA content was expressed as nmol MDA g -1 of fresh weight (FW).
The hydrogen peroxide content was determined according to the procedure described by [19]. Exocarp was homogenized with 1% (w/v) TCA (1:10, w/v), and then centrifuged at 12000 g at 4°C for 15 min. Subsequently, 0.5 mL surpernatant was mixed with 0.5 ml 100 mmol potassium phosphate buffer (pH 7) and 2 mL 1 M potassium iodide. After 1 h in the dark the absorbance was measured at 390 nm. The hydrogen peroxide content was expressed as μmol
Gene expression analysis by quantitative RT-PCR
Gene expression analysis was performed three times with three replications per time and temperature of storage, and each replication was the result of an independent extraction of total RNA from 3 different fruits. Samples were obtained from fruits stored at different storage temperatures and then rewarmed for 6 hours at 20°C. For each sample, portions of exocarp of 3-4 fruits were homogenized, frozen in liquid nitrogen, and stored at -80°C. Total RNA was extracted from each sample using a Total RNA Mini kit (Bio-Rad). The remains of DNA in RNA samples were eliminated by digestion with RQ1 RNase free DNAse (Promega). Before cDNA synthesis, we verified the absence of DNA in RNA and cDNA samples by PCR amplification with primers for CpACS4. cDNA was synthesized from 1 μg of total RNA using iScript reverse Transcription Supermix for RT-qPCR (Bio-Rad). The expression of genes was evaluated through quantitative RT-PCR by using the Rotorgene thermocycler (Qiagen) and iTaq Universal SyBR Green Supermix (Bio-Rad). S1 Table shows the different primers used for q-PCR, which were designed from the 3 0 non-coding regions of each gene by using the Primer Express v 2.0 (Applied Biosystem) software. To avoid possible cross-amplification, and before the q-PCR experiments, the size of the PCR products for each pair of primers was tested in agarose gels. Relative expression of each gene was determined by the comparative Ct (Cycle Threshold) method using C. pepo Elongation Factor 1-α (EF-1A) and Actin (ACT) genes as internal standards. To use this method, we first demonstrated that the efficiency of amplification for each amplicon was roughly equivalent, regardless of the amount of template cDNA. The absolute value of the slope of ΔCt (Ct of the target gene-Ct of the reference gene) versus serial dilutions of cDNA for a given sample must be less than 0.1. The relative expression of each gene to a calibrator sample was calculated using the formula 2-ΔΔCt, where ΔΔCt is the difference between the ΔCt of each sample and the ΔCt of the calibrator sample.
Statistical analysis
Data were treated for multiple comparisons by analysis of variance (ANOVA), followed by least significant difference test (LSD) with significance level p 0.05. Sources of variation were storage time and treatment for each cultivar. ANOVA was performed using the statistical software Statgraphic Centurion XVI (STATGRAPHICS. Statpoint Technologies, Inc.,Warrenton, VA). Normality of distribution was verified using Kolmogorov-Smirnov test, when the assumption of normality failed, the variables were transformed. Weather transformation was not possible non-parametric Kruskal-Wallis test was used to compare differences between groups, with significance level at <0.05.
Postharvest fruit quality parameters
The ISW packaging treatment greatly improved the postharvest quality and shelf life of the zucchini fruits from two different cultivars, which were subjected to cold storage for 14 consecutive days at 4°C (Fig 1A). One of the most prominent effects was on fruit weight loss. This postharvest parameter increased in the fruits of both cultivars over cold storage, although the performance of Natura fruit, with a weight loss of 4% and 6% at 7 and 14 days of storage, respectively, was much better than that of Sinatra, which lost 7% and 13% at the same storage times ( Fig 1B). As expected, ISW treatment almost abolished the dehydration of zucchini fruit and therefore the percentage of weight loss was reduced to almost 0% in both varieties (Fig 1B).
More interestingly, ISW reduced the occurrence of surface pitting of both cultivars (Fig 1C). Sinatra fruit proved more sensitive to PCI than Natura. At 7 days of storage, the Natura fruit showed an average PCI index of 2, while that of Sinatra reached a value of nearly 4 ( Fig 1C). At 14 days of storage the average PCI index of Sinatra fruit was still higher than that of Natura, but the fruit of both cultivars was so affected that it had already lost its commercial value ( Fig 1C). In ISW fruit of the two cultivars PCI symptoms were not detected at 7 days of storage, and were considerably reduced in comparison to control fruit after 14 days of storage ( Fig 1C). In fact, at 14 days ISW fruit had lower PCI symptoms than control fruits at 7 days of storage ( Fig 1C). These data demonstrate that the ISW packaging limited the advance of PCI symptoms, extending fruit shelf life for at least 7 days at 4°C.
The response of fruit firmness to ISW treatment depended on the cultivar (Fig 1D). In the more sensitive cultivar, Sinatra, this treatment considerably reduced the percentage of firmness loss at 7 and 14 days of cold storage, even leading to a slight increase in fruit firmness at 7 days. The firmness of Natura fruit was higher than that of Sinatra during the whole storage period, but it did not respond to ISW treatment ( Fig 1D).
Ethylene production and ethylene gene expression
The longer shelf life and higher quality of ISW fruit was found to be correlated with a reduction in the production of ethylene (Fig 2A). At harvest the zucchini fruit of both cultivars produced no ethylene, which demonstrates the non-climacteric nature of this immature fruit. In accordance with previous results [6], seven days of cold storage induced the production of ethylene in fruit upon rewarming at 20°C for six hours (Fig 2A), and cold induced ethylene is correlated with chilling sensitivity, being much higher in the cold sensitive cultivar Sinatra [6] (Fig 2A). The fruit of the more tolerant cultivar Natura, in fact, hardly induced the production of ethylene after 7 days of cold storage (Fig 2A). After 14 days of cold storage, although the fruit were rewarmed at 20°C, ethylene production returned to its original low value (Fig 2A). In the fruit of Sinatra, ISW significantly reduced the production of ethylene after 7 days of cold storage, and in the fruit of Natura, the production of ethylene remained very low in control and treated fruits for the complete storage period (Fig 2A).
To analyze the molecular mechanisms underlying ethylene production and response during zucchini cold storage we have studied the expression of 13 genes involved in the biosynthesis, perception and response to ethylene in control and ISW fruit of Natura and Sinatra. Genes were selected based on C. pepo transcriptome [20] and previous gene expression data in zucchini [21,22] and Figs 3 and 4 show the expression profiles of genes involved in the biosynthesis or the response to ethylene, respectively. The transcripts of three of a total of eight ethylene biosynthesis genes studied in this paper (CpACS2, CpACS3 and CpACS7) were not detected in the fruit under our postharvest conditions (data not shown). However, two of the biosynthesis genes (CpACO1 and CpACS1) were highly upregulated in the fruit of both cultivars in response to cold storage (Fig 3), which suggests that these two genes are involved in cold-induced ethylene production. In ISW fruit, the induction of CpACO1 and CpACS1 was reduced or completely suppressed in comparison with control fruits. The expression level of the other three biosynthesis genes (CpACS4, CpACS5 and CpACS6) was very low and varied slightly in response to cold and ISW treatments (Fig 3). Only in ISW fruit of Natura was there a slight increase in the accumulation of CpACS4, CpACS5 and CpACS6 at 14 days of cold storage in comparison with control fruit (Fig 3). Since at 14 days of storage the observed ethylene production was very low (Fig 2A), it is likely that this induction of ACS in ISW fruit was not accompanied by an induction of ACO genes.
The ethylene perception and signaling genes CpERS1, CpETR1, CpCTR1, CpEIN3.1 and CpEIN3.2 barely changed in response to cold storage or ISW (Fig 4), indicating that ethylene perception and signaling does not appear to play an important role in the response of zucchini Involvement of Ethylene in Postharvest Chilling Injury of Zucchini fruit to ISW packaging or cold storage in either Sinatra or Natura (Fig 4). Only the gene CpACS3.2 was slightly induced in the ISW fruit of Natura at 14 days of cold storage, but not in that of Sinatra (Fig 4).
Respiration rate and oxidative stress
Respiration rate was measured at harvest and after cold storage. In the latter case, prior to measurement the plastic film was removed from the wrapped fruits and all fruits were rewarmed at 20°C for 6 hours. At 7 days of cold storage, control fruit maintained their respiration rates during postharvest storage at 4°C in both Sinatra and Natura, but these rates fell after 14 days of cold storage (Fig 2B). In the ISW fruit of both cultivars, but especially in that of Natura, the production of CO 2 fell significantly, with a fourfold reduction in Natura fruits in comparison with control fruits (Fig 2B).
To determine ROS accumulation and membrane damage in control and ISW fruits, we measured the contents of hydrogen peroxide and MDA, the latter being a product of lipid peroxidation and membrane integrity. The results of control and ISW fruits from Sinatra and Natura stored at 4°C at 0, 7 and 14 days are shown in Fig 5. In control fruit of both cultivars, hydrogen peroxide was accumulated upon cold storage, although its content was always significantly higher in the fruit of the more sensitive cultivar Sinatra (Fig 5A). On the other hand, the hydrogen peroxide content remained stable or even fell in ISW fruit of Natura and Sinatra, respectively, with significant differences between control and ISW fruit at 7 and 14 days of cold storage in both cultivars (Fig 5A).
MDA content increased throughout postharvest zucchini fruit storage at 4°C in both control and ISW fruit of the two cultivars, but with significantly higher content in the more sensitive cultivar Sinatra at all three storage times (Fig 5B). ISW fruit reduced the accumulation of MDA with respect to control fruit in both cultivars, although the MDA content in ISW fruit of Sinatra was always higher than in those of Natura at the storage times analyzed (Fig 5B).
Discussion
ISW enhances the quality of zucchini fruit subjected to cold storage Cold storage is a postharvest technology that preserves the quality and therefore increases the shelf life of many fruits and vegetables. There are, however, a number of tropical and subtropical fruits that are very susceptible to cold during postharvest, developing PCI and accelerating loss of weight and firmness, and decay when stored at below optimum temperatures. The immature fruit of zucchini is extremely susceptible to cold storage and PCI [3,4,23], although variations for this trait have been detected among commercial and traditional cultivars of this crop [5,6]. The fruit of the two commercial hybrids analyzed in this paper, Natura and Sinatra, showed differences in their sensitivity to PCI, the former proving a more cold-tolerant genotype than the latter [6]. In correlation with PCI index, based on the percentage and severity of pitting symptoms on the fruit surface, Natura control fruit also showed a lower loss of weight and firmness (Fig 1), confirming its greater tolerance to cold storage.
Our data demonstrate that ISW packaging technology applied to zucchini was not only able to reduce weight loss resulting from the dehydration of immature fruit, but also to reduce significantly firmness loss and the percentage and severity of pitting symptoms on the fruit surface of two zucchini cultivars. The treatment also delayed the onset of PCI in the fruit of the two cultivars, retaining their harvest fresh surface appearance even after 14 days of cold storage. Under the same storage conditions, however, the control fruit of Sinatra and Natura lost their commercial value at 7 and 14 days of cold storage, respectively, essentially because of the percentage of fruit surface covered by PCI. These results indicate that ISW is able to enhance coldtolerance in zucchini, extending the shelf life of cold stored fruit for more than 7 days. This was especially noteworthy in the more sensitive cultivar Sinatra, in which ISW fruit suffered no symptoms of PCI nor loss of weight and firmness at 7 days of cold storage, while at 14 days their loss of quality parameters was less than 10% that of control fruit (Fig 1). We have found a slight increase in firmness of Natura fruit during cold storage. Whether this slight increase is a result of lignification, as has been observed in the fruit of different loquat cultivars [24,25], needs to be further evaluated. In Natura, ISW treatment also reduced weight loss and PCI, although no significant differences were detected for firmness between wrapped and unwrapped fruit throughout the storage time (Fig 1). This is the first report indicating the benefits of ISW packaging in zucchini, but similar trends in weight loss, firmness and PCI have been reported for ISW cucumbers [13,26], also an immature fruit of the Cucurbitaceae family, as well as in other fruits and vegetables [11,15,[27][28][29].
The minimum loss of freshness and firmness and the reduced PCI of shrink wrapped zucchinis may be a result of maintaining high relative humidity, produced by the limitation of water vapour diffusion and higher vapour pressure inside the package, and by the modified atmosphere (MA) around each piece of fruit. An increase in water vapor pressure around the fruit concomitant with a decrease in the transpiration rate has also been observed in other fruit and vegetable under plastic film wrapping, including broccoli [28], apricot [30], sweet cherry [31] and loquat [32].
In zucchini, treatments involving reduced concentrations of O 2 or high concentrations of CO 2 before cold storage at 2.5°C have been reported as effective in reducing PCI and in increasing polyamine levels [3], and chilling tolerance has always been found to be associated with a reduced respiration rate in refrigerated fruit [23]. Atmospheres with low O 2 (1-5%) and high CO 2 (5-10%) concentrations are able to extend the shelf life of fresh fruits and vegetables because of a reduction of the respiration rate due to substrate depletion, and a restriction of ethylene biosynthesis [33]. Temperature is also known to be one of the main external factors regulating the respiration rate of fruit [34]. Although we could not measure the evolution of O 2 and CO 2 inside individual packages, the gas selectivity and permeability of the film used (see Material and Methods) ought to be effective to reduce O 2 and to increase CO 2 within an acceptable range for zucchini conservation. In fact, in ISW fruit not only were fruit quality parameters retained, on opening the wrapping no symptoms of bad aroma or offfavours were detected, which suggests no anaerobic metabolism or accumulation of either ethanol or acetaldehyde [35]. The observed reduction in the respiration rate of ISW fruit (Fig 2) might therefore be due to a combination of low temperature of storage and modified atmosphere in the individual bags. Previous research works have also detected a reduced respiration rate in fruit and vegetables such as broccoli [28], or tomato [36] subjected to ISW, in plum, and cucumber fruits exposed to MAP [37,38], as well as in plum, peach and pomegranate protected by edible coatings [39,40]. This reduction in the respiration rates of fruits could be due to exposure to high concentrations of CO 2 and low concentrations of O 2 in the individual bags [41,42]. In the present work, Natura fruits showed a higher reduction in respiration rate than Sinatra ones, also indicating that in zucchini this regulation is genotype dependent and has a far greater effect on ISW fruit of the more cold tolerant cultivar. These differential responses may be linked to differences in low temperature response between the two cultivars.
ISW inhibits the production of ethylene and the expression of ethylene genes
In a previous paper we reported that the storage of zucchini fruit at low temperatures induces the production of ethylene once the fruit is rewarmed for a few hours at 20°C. As shown in Fig 2, this chilling-induced ethylene peaks at 7 days of cold storage plus 6 h of rewarming, and is higher in Sinatra, the cultivar showing higher sensitivity to PCI. Our results agree with previous data showing that chilling-induced ethylene is correlated with chilling injury and chilling sensitivity in zucchini [6]. In accordance with temperature preconditioning treatments, which also alleviate PCI in zucchini, the ISW treatment significantly inhibited chilling-induced ethylene in the more sensitive cultivar. In the more tolerant genotype Natura, this reduction was not obvious due to the very low level of ethylene production in the refrigerated fruit of this cultivar (Fig 2). This inhibition of ethylene production promoted by ISW was previously reported in the climacteric papaya [15], but also in the non-climacteric fruit of pepper [29], and the immature fruit of cucumber [13]. MAP has been also found to reduce the production of ethylene in diverse fruit such as plum and apple [37,43].
The role of ethylene in the onset of PCI is unclear. Since many cold sensitive fruits produce ethylene concomitantly with the development of PCI [44], it has been speculated that this ethylene could be involved in the development of PCI [45]. Nevertheless this ethylene could also be the consequence of cold stress and result from PCI in cold sensitive fruits. Indeed, in zucchini we have recently reported that this chilling-induced ethylene was not necessary for the onset of PCI, but rather a cold induced response [6]. Despite this, the inhibition of ethylene production observed in ISW fruit again demonstrates that there is a correlation between chilling-induced ethylene and chilling sensitivity in zucchini, and that the reduction of ethylene in cold stored fruit was always associated with a reduction in PCI and therefore with a higher tolerance to cold storage. As proposed previously, this chilling induced ethylene may therefore be used as an earlier marker of the tissue that may show the secondary, downstream visual symptoms of PCI [6]. The gene expression data also support this conclusion, as many of the upstream regulatory genes in the ethylene signal transduction pathway did not change under cold, but those catalyzing steps in ethylene production did.
Even less is known about the molecular mechanisms associated with ethylene production and response as a result of the postharvest storage of zucchini fruit at low temperatures. The biosynthesis of ethylene comprises two steps: S-adenosymethionine is converted to 1-aminocyclopropane-1.carboxylate (ACC) by the limiting enzyme ACC synthase (ACS), and ACC is then converted to ethylene by ACC oxidase (ACO), and the genes coding for these two enzymes belong to multigenic families in plant genomes [33]. In this paper we have studied the expression profiles of one ACO and seven ACS genes during the storage of zucchini fruit at 4°C for 14 days and in response to ISW packaging. The results indicate that the promotion of ethylene biosynthesis in chilled fruit depends mainly on two of the analyzed genes, CpACS1 and CpACO1, whose expressions were highly induced in control fruit under low temperature storage, and were downregulated in response to ISW. The transcripts of other ACS genes such as CpACS2, CpACS3 and CpACS7 were not detected in the fruit during postharvest, and those of the genes CpACS4, CpACS5 and CpACS6 were not regulated by either cold or ISW packaging in the cold sensitive cultivar Sinatra, although they were slightly upregulated in the ISW fruit of the more tolerant cultivar Natura (Fig 3). The regulation of ACS and ACO genes in response to chilling is not constant among different fruits. Thus ACS and ACO genes have been found to be either upregulated or downregulated in response to cold storage in papaya [46], citrus [47], banana [48] or peach [49]. We have found that CpACS1 and CpACO1 are more sensitive to low temperature, and that the reduction of ethylene production promoted by ISW treatment was correlated with a downregulation of these two genes. The genes CpACO1 and CpACS1 would therefore be the perfect biotechnological targets to improve cold tolerance. In fact, we have already started the identification of loss-of-function mutants for these two genes in an EMS mutant collection of zucchini.
None of the five ethylene perception and response genes analyzed in this paper was found to be significantly modulated by either cold storage or ISW packaging during the postharvest storage of zucchini fruit. The perception genes CpETR1 and CpER1, and the signal transduction genes CpCTR1 and CpEIN3.1, did not change their expression in response to the treatment. Only the gene CpEIN3.2 was slightly upregulated in the ISW fruit of the cultivar Natura (but not in those of Sinatra) after 14 days of cold storage (Fig 4). In other systems, including both climacteric and non-climacteric fruits, the expression of ethylene receptors and ethylene signal transduction genes in response to cold storage varied greatly and depended on the isoform analyzed. In the climacteric papaya, the expression of ETR1, ERS1 and CTR1 was not induced by cold storage, but the expression of CTR2, EIN3.1a and EIN3.1b was induced after 20 days of cold storage [46]. Ethylene perception and signaling genes were also differentially regulated in the climacteric fruits tomato, kiwi and pear [50][51][52][53][54][55]. In non-climacteric fruit, such as that of grapefruit and loquat, ethylene receptors and response genes were also differentially upregulated by cold storage [56,57], although in these cases ethylene is not required for fruit ripening. Taken together these results indicate that chilling is able to induce some isoforms of ethylene biosynthesis, perception and response genes in both climacteric and non-climacteric fruit. In non-climacteric fruit the induction of these genes is not associated with fruit ripening but should rather be associated with fruit response to cold stress. The regulation of specific ethylene perception and response genes such as EjETR1 and EjEIL1 in loquat, has suggested that they may be involved in PCI development [57]. Our results in zucchini suggest that certain ethylene biosynthesis genes, but not all of them, participated in the fruit response to chilling temperature and ISW treatment, but that ethylene perception and signal transduction pathway genes do not appear to be involved in PCI development of control fruit, nor in the reduction of PCI symptoms as a result of ISW packaging.
ISW inhibits oxidative stress and oxidative damage
It has been widely reported that the storage of fruits at low temperatures is able to induce oxidative stress, leading to an increase in the production of reactive oxygen species (ROS) causing progressive oxidative damage, such as lipid peroxidation of both cellular and organelle membrane that aggravates oxidative stress through the production of lipid-derived radicals [8][9][10]58]. The accumulation of MDA, one of the final products of the peroxidation of unsaturated fatty acids in phospholipids, is frequently used as a marker of membrane oxidative damage. To combat this oxidative stress, plants trigger various enzymatic and non-enzymatic antioxidant defense mechanisms. Oxidative damage to the tissues will therefore depend on a delicate equilibrium between ROS production and the antioxidant defense mechanisms that operate in the cells scavenging for ROS overproduction [58].
In zucchini, it has been reported that the storage of fruit at low temperatures induces changes in both the accumulation of ROS and in the activity of antioxidant enzymes such as catalases, ascorbate peroxidases and superoxide dismutases [5,7,8,59]. Postharvest treatments that have been shown to induce chilling tolerance in zucchini, including temperature preconditioning treatments [7,8] and external applications of polyamines [59], were able to induce the enzymatic oxidative defense mechanisms while reducing the content of H 2 O 2 and MDA. Changes in carbohydrate content in zucchini fruit under low temperature have been related not only with the importance of soluble carbohydrate as osmoprotectants and stabilizers of cell membranes but also with their role as ROS scavengers [60]. Our results in this paper also indicate that the success of ISW, the most effective current treatment to induce cold tolerance in zucchini, is also concomitant with a reduction in oxidative stress induced by chilling. The treatment not only reduced the production of H 2 O 2 throughout the storage period in both Natura and Sinatra, but also diminished the accumulation of MDA, which manifested itself in a reduction of cell damage. The manipulation of this oxidative stress and oxidative damage is therefore essential to reduce PCI symptoms and to increase the postharvest quality of zucchini fruit under cold storage.
In conclusion, individual shrink wrapping of zucchini fruit with a selective film during postharvest was able to induce cold tolerance in two cultivars showing differences in chilling sensitivity, by considerably reducing chilling injury and the loss of weight and firmness during the storage period at 4°C. This improvement in fruit quality parameters was associated with a reduction in the production of ethylene and a downregulation of ethylene biosynthesis genes CpACS1 and CpACO1, together with a reduction in the respiration rate of fruit and the inhibition of oxidative stress and oxidative damage processes.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2017-07-21T04:09:32.712Z", "created": "2015-07-15T00:00:00.000", "id": "13653422", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0133058&type=printable", "pdf_hash": "66c5bc8df9236d77c991bef04ac19c650263fdc9", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:812", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "0219d48e44c8acfac7eac51cdfa9ef782eab9b0d", "year": 2015}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
A new allele of flower color gene W1 encoding flavonoid 3'5'-hydroxylase is responsible for light purple flowers in wild soybean Glycine soja
Background Glycine soja is a wild relative of soybean that has purple flowers. No flower color variant of Glycine soja has been found in the natural habitat. Results B09121, an accession with light purple flowers, was discovered in southern Japan. Genetic analysis revealed that the gene responsible for the light purple flowers was allelic to the W1 locus encoding flavonoid 3'5'-hydroxylase (F3'5'H). The new allele was designated as w1-lp. The dominance relationship of the locus was W1 >w1-lp >w1. One F2 plant and four F3 plants with purple flowers were generated in the cross between B09121 and a Clark near-isogenic line with w1 allele. Flower petals of B09121 contained lower amounts of four major anthocyanins (malvidin 3,5-di-O-glucoside, petunidin 3,5-di-O-glucoside, delphinidin 3,5-di-O-glucoside and delphinidin 3-O-glucoside) common in purple flowers and contained small amounts of the 5'-unsubstituted versions of the above anthocyanins, peonidin 3,5-di-O-glucoside, cyanidin 3,5-di-O-glucoside and cyanidin 3-O-glucoside, suggesting that F3'5'H activity was reduced and flavonoid 3'-hydroxylase activity was increased. F3'5'H cDNAs were cloned from Clark and B09121 by RT-PCR. The cDNA of B09121 had a unique base substitution resulting in the substitution of valine with methionine at amino acid position 210. The base substitution was ascertained by dCAPS analysis. The polymorphism associated with the dCAPS markers co-segregated with flower color in the F2 population. F3 progeny test, and dCAPS and indel analyses suggested that the plants with purple flowers might be due to intragenic recombination and that the 65 bp insertion responsible for gene dysfunction might have been eliminated in such plants. Conclusions B09121 may be the first example of a flower color variant found in nature. The light purple flower was controlled by a new allele of the W1 locus encoding F3'5'H. The flower petals contained unique anthocyanins not found in soybean and G. soja. B09121 may be a useful tool for studies of the structural and functional properties of F3'5'H genes as well as investigations on the role of flower color in relation to adaptation of G. soja to natural habitats.
Background
Soybean (Glycine max (L.) Merr.) is believed to have been domesticated in north-eastern China from its wild relative, Glycine soja Sieb. & Zucc. [1]. Glycine soja is native throughout China, the adjacent area of Russia, Korea, Japan and Taiwan [1]. Flower color of G. soja is almost exclusively purple; by contrast, 33% (5,544 out of 16,855) of the soybean accessions in the USDA Soybean Germplasm Collections have white flowers (Dr. R. L. Nelson, personal communication 2006). The reason why G. soja almost lacks flower color variants is uncertain [2]. A few white-flowered G. soja accessions were reported in a Chinese germplasm collection, but these had high 100-seed weight, strongly suggesting recent outcrossing with G. max [2]. One white-flowered plant (PI 424008C) was found in 1998 among the progeny of a purple-flowered G. soja accession (PI 424008A) that was originally introduced from South Korea in 1976 [2]. Genetic analysis indicated that the white flower was caused by a recessive allele at the W1 locus similar to the white-flowered soybeans [2].
In soybean, six genes (W1, W2, W3, W4, Wm and Wp) primarily control flower color and two genes (T and Td) control pubescence color [3,4]. The hydroxylation pattern of B-ring in flavonoids plays an important role in the coloration of seed coats, flower and pubescence of soybeans. The B-ring of flavonoids can be hydroxylated at either the 3' position leading to the production of cyanidin-based pigments, or at both the 3' and 5' positions to produce delphinidin-based pigments.
Two key enzymes involved in this pathway are flavonoid 3'-hydroxylase (F3'H) and flavonoid 3'5'-hydroxylase (F3'5'H) which are both microsomal cytochrome P450 dependent monooxygenases that require NADPH as a co-factor [5]. Chromatographic experiments suggested that T and W1 loci are responsible for the formation of flavonoids with 3', 4' and 3', 4', 5' B-ring hydroxylation patterns, respectively [6][7][8]. Hence, T and W1 are presumed to encode F3'H and F3'5'H, respectively. The F3'H cDNA was cloned and characterized from a pair of near-isogenic lines (NILs) for the T locus, To7B (TT, tawny pubescence) and To7G (tt, gray pubescence) [9]. Sequence analysis revealed that they differed by a singlebase deletion of C in the coding region of To7G. The deletion generated a truncated polypeptide lacking the GGEK consensus sequence of F3'H gene and the hemebinding domain, resulting in non-functional protein.
The W1 gene has a pleiotropic effect on flower and hypocotyl color: soybean cultivars with purple/white flowers have purple/green hypocotyls. The soybean F3'5'H gene was cloned from NILs for W1 and confirmed that W1 encodes F3'5'H and that the gene of white-flowered NILs contained a 65 bp insertion in the coding region [10]. In addition to the F3'5'H protein, a cytochrome b5 is required for full activity of F3'5'H in petunia, and mutation in cytochrome b5 results in a reduction in F3'5'H activity and alteration of anthocyanin amount and composition [11].
Yasuda discovered B09121, a flower color variant of G. soja, at a slope between a paddy field and a ditch in Karatsu, Saga Prefecture (southern Japan) in 2002 (unpublished result) ( Figure 1). Its banner petals have a pale pinkish hue and a pronounced light purple pigmentation that originates from the base of the petals and spreads in streaks towards the petal margins. We designated the flower color as light purple. Considering its growth habit, small seed size and unique flower color, it is unlikely that the flower color of B09121 was derived from outcrossing with soybean. To our knowledge, B09121 is the first example of a flower color variant of G. soja found in nature. The first objective of this study was to determine the genetic basis of purple flower color by crossing experiments. The second objective was to analyze the flavonoids in flower petals of G. soja accessions. The third objective was to clone and characterize a gene responsible for light purple flowers.
Genetic analysis
A Canadian soybean cultivar Harosoy (W1W1 W2W2 w3w3 W4W4 WmWm WpWp tt) with purple flowers and gray pubescence and a NIL of a US soybean cultivar Clark for W1 gene, Clark-w1 (L63-2373, w1w1 W2W2 w3w3 W4W4 WmWm WpWp TT) with white flowers and tawny pubescence were crossed with B09121 having light purple flowers and tawny pubescence in 2005. Flowers of Harosoy and Clark-w1 were emasculated one day before opening and pollinated with B09121. Hybridity of the F 1 plants was ascertained by tawny pubescence color in crossing with Harosoy and by spindly growth habit in crossing with Clark-w1. Seeds of L63-2373 were provided by the USDA Soybean Germplasm Collection. The NIL was produced by crossing Clark with T139 and backcrossing the progeny to Clark up to BC6 [14].
A total of 120 to 130 F 2 seeds derived from the crossing with Harosoy and Clark-w1 were planted in field (lowhumic andosols) on June 13 in 2007 at the National Institute of Crop Science, Tsukuba, Japan (36°06'N, 140°05'E). A bulk of 30 seeds each of 36 F 3 families derived from the cross with Clark-w1 were planted at the same location on June 8 in 2008. N, P and K were applied at 3.0, 4.4 and 8.3 g m -2 , respectively. Flower color was scored in individual F 2 and F 3 plants. Banner petals were collected with forceps at the day of opening. Two 200 mg samples of banner petals were collected in 2 ml of MeOH containing 0.1% (v/v) HCl for anthocyanin analysis. Two 200 mg samples in 2 ml of absolute MeOH were also collected for the determination of flavonol and dihydroflavonol. High performance liquid chromatography (HPLC) of anthocyanins, flavonols and dihydroflavonol was performed following previously described protocols [12]. The 2 ml extracts were filtered through disposable filtration units (Maishoridisc H-13-5, Tosoh, Japan) and 10 μl from each sample was subjected to HPLC analysis. The amount of flavonoids was estimated from the pertinent peak area in the HPLC chromatogram (detection wavelength of anthocyanins = 530 nm; flavonols= 351 nm; dihydroflavonols = 290 nm). The peak area was subjected to analysis of variance using Statistica software (StatSoft, Inc. Tulsa, OK).
RNA extraction and cDNA cloning
Total RNA was extracted from banner flower petals (100 mg) of Clark, Clark-w1 and B09121 using the TRIZOL Reagent (Invitrogen) according to the manufacturer's instructions. cDNA was synthesized by reverse transcription of 5 μg of total RNA using the Superscript III First-Strand Synthesis System (Invitrogen) and an oligo(dT) primer according to the manufacturer's instructions. The full-length cDNA was cloned from Clark and B09121, using a pair of PCR primers (5'-AACTAGCAAATTAAT-TAGCTT and 5'-CAACCCAAACATTACTTAT) and end-to-end PCR. The PCR mixture contained 0.5 μg of cDNA, 10 pmol of each primer, 10 pmol of nucleotides and 1 unit of ExTaq in 1 × ExTaq Buffer supplied by the manufacturer (Takara) in a total volume of 50 μl. A 5 min denaturation at 94°C was followed by 30 cycles of 30 sec denaturation at 94°C, 1 min annealing at 58°C and 1 min extension at 72°C. A final 7 min extension at 72°C completed the program. The PCR was performed in an Applied Biosystems 9700 thermal cycler. The~1.8 kbp PCR product was cloned into pCR 2.1 vector (Invitrogen) and sequenced.
dCAPS and indel analyses
Genomic DNA of Clark, Clark-w1, B09121 and 36 F 2 plants that were used for F 3 progeny tests was isolated from trifoliolate leaves by CTAB [19]. A pair of PCR primers (5'-GTCTAACGAGTTCAAGGCCAT, 5'-CAACTTGGCCAAAAAGGGTAT) was designed to detect a single-base substitution at nucleotide number 653 that is unique to B09121. The first primer contains a nucleotide C that is mismatched with its target DNA to artificially create a restriction site of NcoI (CCATGG) in Clark ( Figure 2). The base substitution within the restriction site would result in presence/ absence of the restriction site in the amplified product to generate a polymorphism. The PCR mixture contained 30 ng of genomic DNA, 5 pmol of each primer, 10 pmol of nucleotides and 1 unit of ExTaq in 1 × ExTaq Buffer supplied by the manufacturer (Takara) in a total volume of 25 μl. After an initial 30 sec denaturation at 94°C, there were 30 cycles of 30 sec denaturation at 94°C, 1 min annealing at 56°C and 1 min extension at 72°C. A final 7 min extension at 72°C completed the program. The amplified products were digested with NcoI, and the digests were separated on an 8% nondenaturing polyacrylamide gel in 1 × TBE buffer (90 mM Tris-borate, 2 mM EDTA, pH 8.0). After electrophoresis, the gel was stained with ethidium bromide and the DNA fragments were visualized under UV light.
A pair of indel PCR primers (5'-TTTTGAGCTTATTC-CATTTGG, 5'-TGAATATTCGAACCCAACCA) was designed to identify the 65 bp insertion in F3'5'H gene of soybean lines with w1 allele based on the previous report [10]. The PCR profile and electrophoresis conditions were identical with the dCAPS analysis except that annealing was conducted at 59°C.
Semi-quantitative RT-PCR analysis
Semi-quantitative RT-PCR was conducted by reversetranscription of 5 μg of total RNA using the Superscript III First-Strand Synthesis System and an oligo d(T) primer according to the manufacturer's instruction. To test the transcription level of the F3'5'H gene, PCR reactions were carried out in a volume of 25 μl, using 125 ng of cDNA. The initial 30 sec denaturation at 94°C was followed by 26 cycles of 30 sec denaturation at 94°C, 1 min annealing at 59°C and 1 min extension at 72°C. A final 7 min extension at 72°C completed the program. The primers were 5'-GACGCTGAGGATATTCAACC and 5'-AGAAATCTGTGAGGTCACGA. A soybean actin gene was used as a control. The initial 30 sec denaturation at 94°C was followed by 20 cycles of 30 sec denaturation at 94°C, 1 min annealing at 56°C and 1 min extension at 72°C. A final 7 min extension at 72°C completed the program. The primers were 5'-CTGGGGATGGTGTCAGCCACAC and 5'-CACC-GAACTTTCTCTCGGAAGGTG. PCR products were loaded on a 1.2% agarose gel, stained by ethidium bromide and visualized under UV light.
Accession Numbers
Sequence data from this article have been deposited with the DDBJ Data Libraries under accession nos. AB540111 (Clark) and AB540112 (B09121). Table 1). Results of the F 3 progeny tests supported the hypothesis that light purple flower was controlled by a new allele at the W1 locus that was dominant to the w1 allele. We designated the allele as w1-lp (light purple). The gene symbol was approved by the Soybean Genetics Committee. Dominance relationship of the locus was W1 >w1-lp >w1.
In contrast, flower petals of white-flowered line PI 424008C contained no anthocyanins. Further, the flower petals of B09121 contained about half the amount of A1 to A4 compared with Clark. The HPLC chromatogram of B09121 exhibited three additional anthocyanin peaks, A5 to A7, that were not found in soybeans or other G. soja lines (Figure 3). Based on the comparison of retention time with authentic specimens, peonin, cyanin and chrysanthemin, A5, A6 and A7 were estimated as peonidin 3,5-di-O-glucoside, cyanidin 3,5-di-O-glucoside and cyanidin 3-O-glucoside, respectively.
Eight peaks, F1 to F8, corresponding to flavonol glycosides, were detected by HPLC analysis: F1 (kaempferol Table 1 Segregation of flower color in F 1 plants and F 2 population derived from a cross between a soybean cultivar Harosoy with purple flowers and B09121, a Glycine soja accession with light purple flowers, and segregation of flower color in F 1 plants, and F 2 and F 3 populations derived from a cross between a soybean near-isogenic line Clark-w1 with white flowers and B09121 in Tsukuba, Japan. The amounts of flavonol glycosides estimated by peak areas in HPLC analysis are presented in Table 4. PI 424008A and PI 424008C contained all eight of the flavonol glycosides found in Clark. COL/AOMORI/1983/ NASU-2 lacked F4 and Kokaigawa-1 lacked F8. B09121 was devoid of F4 and F7. However, the total amount of flavonol glycosides was not significantly different among the Clark and G. soja lines included in this report. One peak (F9) corresponding to dihydroflavonol (aromadendrin 3-O-glucoside) was found by HPLC analysis in Clark and all of the G. soja lines. The amount of aromadendrin 3-O-glucoside estimated by peak area in HPLC analysis is presented in Table 5. The G. soja lines had 33 to 155% more aromadendrin 3-O-glucoside than Clark. position 210) and from glutamic acid to valine (amino acid position 475), respectively ( Figure 2). The latter amino acid substitution was also found in Chin-Ren-Woo-Dou, whereas the former was unique to B09121.
dCAPS and indel analysis
The PCR reaction for dCAPS analysis produced bands with expected size of about 100 bp in Clark, Clark-w1 and B09121 (Figure 4). NcoI digested the bands of Clark and Clark-w1 and shortened the bands by 18 bp. In contrast, the band from B09121 was largely undigested. In addition, a faint band with approximately the same size was also observed among the digested bands in B09121. Similar results were obtained in dCAPS analysis using a different set of PCR primers and a different restriction enzyme (HphI) (data not shown). The PCR reaction for indel analysis produced shorter bands (255 bp) in Clark and B09121, and a longer band in Clark-w1 (310 bp) due to the 65 bp insertion (Figure 4). In addition, a faint band was also observed in Clark-w1 that was approximately the same size as the shorter bands in Clark and B09121.
In dCAPS analysis of the F 2 population, plants with w1w1 genotype (white flower) had only shorter bands, whereas plants with w1-lpw1-lp genotype (light purple flower) had longer bands with faint bands similar to B09121 ( Figure 5). Heterozygous plants with the w1-lpw1 genotype (light purple flower) had both bands at similar band intensity. In indel analysis, plants with w1-lpw1-lp genotype had only shorter bands, whereas plants with w1w1 genotype had longer bands and faint shorter bands. Heterozygous plants with w1-lpw1 genotype had both bands at similar band intensity. Thus, dCAPS and indel markers co-segregated in plants with white and light purple flowers. In contrast, the F 2 plant with purple flowers had the two bands with similar intensities in dCAPS analysis and only a shorter band in indel analysis.
Semi-quantitative RT-PCR analysis
Results of semi-quantitative RT-PCR suggested that the transcript level of the F3'5'H gene was not substantially different among lines (data not shown).
Discussion
Flower color of G. soja is almost exclusively purple, whereas about 30% of soybean cultivars have white flowers. The reason why G. soja almost lacks flower color variants is uncertain [2]. In 1998, researchers found a white-flowered variant of PI 424008A, a USDA accession of G. soja with purple flowers that was originally introduced from South Korea in 1976 [2]. The mutation may have occurred during propagation at USDA. To our knowledge, B09121 is the first example of flower color variant found in the natural habitat. Genetic analysis suggested that light purple flower is controlled by a new allele at the W1 locus, w1-lp. Dominance relationship of the locus was W1 >w1-lp >w1. Interestingly, one F 2 plant and four F 3 plants with purple flowers were generated in the cross with Clark-w1. Considering the fact that purple-flowered plants were produced from heterozygous plants (w1-lp w1) in both F 2 and F 3 generations and that frequency of purple-flowered plants was similar (about 1%) across generations, the purple flower color may have been produced by a crossover in the W1 gene instead of seed contamination or out-crossing. The purple-flowered F 2 plant produced F 3 plants with purple and light purple flowers at a 3:1 ratio; this suggests that the region including the 65 bp insertion was eliminated from the genome. The dCAPS and indel analyses indicated that the base substitution was heterozygous but the indel region was homozygous without the 65 bp insertion in the purple-flowered F 2 plant. The results further supported elimination of the insertion that is responsible for gene dysfunction from the plant by intragenic recombination. It remains to be investigated whether the existence of tandem repeats derived from the insertion is responsible for the high frequency of intragenic recombination. Sequence analysis of F3'5'H cDNA from Clark and B09121 indicated that two amino acids (amino acid numbers 210 and 475) were substituted. The former substitution has not been observed in soybean cultivars examined to date. It may be responsible for light purple flower and unique anthocyanin composition. However, no catalytic domains have been assigned to the region. The spontaneous mutation leading to flower color change may not have affected amount of F3'5'H gene transcripts, based on the results from semi-quantitative RT-PCR analyses.
310
Flavonoids in flower petals of G. soja with purple flowers were generally similar to those of soybean cultivars with purple flowers. Flower petals of PI 424008C with white flowers had no anthocyanin but contained comparable amounts of flavonol glycosides and dihydroflavonols. It is consistent with the result from Clark-w1, a soybean NIL with white flowers, whose flower petals contained no anthocyanin although it had levels of flavonol glycosides and dihydroflavonol similar to Clark [12]. The present results confirmed that W1 solely controls anthocyanin biosynthesis in G. soja.
Purple flowers of soybean and G. soja contain four major anthocyanins with 3'4'5'-substituted form, malvidin 3,5-di-O-glucoside, petunidin 3,5-di-O-glucoside, delphinidin 3,5-di-O-glucoside and delphinidin 3-O-glucoside. In contrast, flower petals of B09121 contained lower amounts of the four major anthocyanins. In addition, they contained small amounts of the 5'-unsubstituted versions of the above anthocyanins, peonidin 3,5-di-O-glucoside, cyanidin 3,5-di-O-glucoside and cyanidin 3-O-glucoside. It appears that F3'5'H activity was reduced and F3'H activity was increased in flower petals of B09121. Flower petals of soybean and G. soja contain large amounts of kaempferols and very small amounts of quercetins [12,13], suggesting that F3'H activity may be very low. Further, alleles at the T locus encoding F3'H did not affect 3'-hydroxylation of flavonols in flower petals [12,13]. Therefore, it is unlikely that the F3'H gene might be responsible for the anthocyanin alteration. Alternatively, mutation in the F3'5'H gene possibly led to a reduction in F3'5'H activity and an increase in F3'H activity. In petunia, a cytochrome b5 is required for full activity of F3'5'H. A mutation in cytochrome b5 reduced 3'4'5'-substituted anthocyanins and increased 3'4'-substituted anthocyanins [11]. It is possible that the amino acid substitution might directly affect the amount and composition of anthocyanins or interact with cytochrome b5. Functional analysis using yeast recombinant assays may be necessary to identify the amino acid substitution that led to light purple flower and unique anthocyanin composition, to investigate the association with cytochrome b5, and to verify whether the amino acid substitution generated F3'H activity. Transformation experiments using a soybean line with w1 allele may be necessary to verify the function of the F3'5'H gene from B09121. B09121 may be the first example of a flower color variant of G. soja found in the natural habitat. It may be a useful tool for studies of the structural and functional properties of F3'5'H genes as well as investigations on the role of flower color in relation to adaptation of G. soja to natural habitats.
Conclusions
This study is the first report of a flower color variant of wild soybean G. soja discovered in nature. Genetic analysis revealed that light purple flower of the accession B09121 was controlled by a new allele of W1 locus encoding F3'5'H. The new allele was designated as w1lp. The dominance relationship of the locus was W1 >w1-lp >w1. In crossing experiments, purple-flowered plants were generated in the cross between B09121 and a soybean near-isogenic line with w1 allele. F 3 progeny test, and dCAPS and indel analyses suggested that the plants with purple flowers might be due to intragenic recombination. Flower petals of B09121 contained lower amounts of four major anthocyanins common in purple flowers and contained small amounts of the 5'-unsubstituted versions of the above anthocyanins that are absent in soybeans and other G. soja accessions. The results suggested that F3'5'H activity was reduced and flavonoid 3'-hydroxylase activity was increased in the flower petals. The cDNA of B09121 had a unique base substitution resulting in the substitution of valine with methionine. B09121 may be a useful tool for studies of the structural and functional properties of F3'5'H genes as well as investigations on the role of flower color in relation to adaptation of G. soja to natural habitats.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2017-06-20T22:37:16.855Z", "created": "2010-07-28T00:00:00.000", "id": "5756800", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://bmcplantbiol.biomedcentral.com/track/pdf/10.1186/1471-2229-10-155", "pdf_hash": "21546ee13b1f5ef3d0ca5e6a8f48b2ee3ded7ec7", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:3990", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "e424ca99228345de6fcf7c5982804a54000ceab6", "year": 2010}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Localization and dynamic change of saponins in Cyclocarya paliurus (Batal.) Iljinskaja
Cyclocarya paliurus is a unique tree species of that grows in southern China. The tree contains distinctive saponins in the leaf that has hypoglycemic and hypolipidemic effects. It was aimed to detect localization of saponins and suitable time of harvest for medicinal uses. Histochemical, cytochemical localization and UV-spectrophotometry were carried out in C. paliurus plant. We found that in all organs, the saponins were primarily located in the parenchyma cells and the highest saponins accumulation was in the palisade tissue in leaves. Cytochemical localization results indicated that saponins were mainly distributed in the chloroplast, vesicle, and plasmalemma. On average, the total saponins content in leaves (20.57 mg·g-1) was two and three times greater than in root (10.19 mg·g-1) and shoot (6.20 mg·g-1), respectively. Moreover, the saponins content in the leaf and root exhibited fluctuations, which were highest in September. Considering saponins levels and biomass, we conclude that harvesting all leaves in September is an economical and effective strategy for medicinal use in C. paliurus.
Introduction
Cyclocarya paliurus is widely distributed in the southern region of China and is commonly known as Qing-Qian-Liu for its fruit clusters that look like a string of old Chinese copper coins [1]. Previous studies have found that C. paliurus contains a variety of bioactive substances, among which are saponins. Saponins are known to have a wide range of pharmacological properties, including anti-inflammatory, anti-tumor, anti-hyperglycemic, and antihyperlipidemic effects [2][3][4][5]. Saponins have also been found to boost the immune system and reduce high blood pressure [2,6]. There are several distinctive saponins that have been isolated and identified from C. paliurus leaves, including cyclocarioside A and cyclocarioside I, which are 200 and 250 times as sweet as sucrose, respectively [7,8]. To date, most studies involving saponins from C. paliurus have mainly focused on the pharmacological function of saponins on diabetes [9][10][11] and hyperlipidemia [12]. Studies have also investigated the separation and purification of active compounds [13] and the extraction and identification of new compounds [14]. However, little is known about the storage and localization of saponins in the C. paliurus plant. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 In recent years, histochemical localization has been used to fix accumulating position of secondary metabolites in Chinese medicinal materials. According to types of composition, different methods have been used to determine cellular localization of alkaloids [15], saccharides [16], saponins [17], quinones, and flavonoids [16,18]. Research investigating the locations of saponins in Panax ginseng, Bupleurum chinense, and Polygala tenuifolia by vanillin-HAc and perchloric acid staining has found that saponins accumulated primarily in the parenchyma cells in the secondary phloem, cortex, and periderm of the root and shoot, as well as the epidermal and mesophyll cells in leaves [17,19,20]. However, differences in saponins accumulation among organs indicated that the distribution and types of saponins largely depend on species and tissues [21]. The initial saponins extraction from leaves is typically conducted by adding ammonium sulfate, lead acetate or another neutral salt to an aqueous solution of acid saponins [22]. Then Cai et al. (2009) proposed an approach that used lead acetate precipitation to localize the saikosaponin of Bupleurum scorzonerifolium at the subcellular level, through which, they found that saikosaponin mainly located in cytoplasm of epidermal cells, especially in vacuole, in addition, saponins were also found in the plasmodesma and golgi apparatus [23].
As a medicinal woody plant, harvest season and harvest organ are vital for C. paliurus. However, which organ (leaf, stem or root) is the main location for saponins synthetic? And what season is the optimum for saponins accumulation? To determine the suitable harvest strategy, histochemical and cytochemical analyses were used to detect saponins content in various vegetative organs across growing seasons. Combined with previous work on the biomass of C. paliurus, we aim to provide supports to determine harvest strategy for medicinal use of C. paliurus. Meanwhile, the study also can make foundation to understand the pathways of synthesis and transport of saponins; more importantly, it will provide a basis for developing a strategy of cultivation for medicinal use.
Plant materials
No specific permissions were required for the location. The field studies did not involve endangered or protected species.
Samples for histochemical and saponins content determinations were collected from 6-year-old C. paliurus trees that had previously been established as a trial field in Zhenjiang, Jiangsu Province, China (119˚32'E, 32˚16'N). Current year shoot, leaflets, and mature lateral roots were collected at the end of April, May, June, July, and September of 2016. The four individuals in every replicates selected in this study were average sample trees determined by tree height and breast diameter, and there were three replications. Root samples were rinsed with running deionized water.
From a separate and second source of C. paliurus, samples were collected for cytochemical localization. These samples were collected from yearling that had been cultivated in a greenhouse (25 ± 2˚C, relative humidity at 60-70%) in Nanjing Forestry University, Nanjing, Jiangsu Province, China (118˚48'E,32˚04'N). Tender leaves were collected in April, and fully developed were collected in May. Sample came from one individual was regarded as one replicate, and there were three replications.
Histochemical localization
Histochemical localization was carried out according to methods outlined by Teng et al. (2009) and Tan et al. (2008) [17,20]. Sections of fresh tissue 25-35 μm in length were cut using a freezing microtome Leica-CM1850 (Leica, Beijing, China), treated with 5% lead acetate for 10 min to precipitate saponins from the tissue, and then stained with a mixed solution of 5% vanillin-HAc and perchloric acid (v:v = 1:1) for 5-10 min. Treated sections were then observed and photographed under a light microscope Olympus CX-41 (Olympus, Tokyo, Japan). The control samples were fixed in formalin-aceticacid-70% alcohol (FAA) for 30 days to remove saponins. Sections were treated and stained as frozen sections. Ten sections were randomly selected from each sample for observation.
Cytochemical localization
Leaflets were treated following published protocols [23]. Leaflet samples were cut into 1 mm 3 fragments and fixed with 2% glutaraldehyde (prepared with 3% lead acetate in 0.1 M sodium cacodylate buffer at a pH of 7.2) for 4 h at 4˚C. Samples were then washed four times for 30 min each time with 3% lead acetate, following which, the cuts were incubated with 1% osmic acid (prepared with 3% lead acetate in 0.1 M sodium cacodylate buffer at a pH of 7.2) overnight at 4˚C, then washed twice (each for 30 min) with 0.1 M sodium cacodylate buffer at a pH of 7.2. The fragments were then gradually dehydrated with a grade series of ethanol of 30%, 50%, 75%, 85%, 95%, and 100%. Each ethanol step was exposed to the sample for 30 min. Next, samples were incubated with epoxy propane transition and embedded with Epon812. The fragments of the leaflet were cut into sections 50-70 nm long using an ultramicrotome LKB-V (LKB, Sweden). After samples were stained with uranyl acetate, the sections were subsequently observed and photographed using a transmission electron microscope HITACHI 7650 (HITACHI, Tokyo, Japan). The control sample was treated with the same procedures as described above, but the fixing step with lead acetate was excluded. Ten sections were randomly taken from each sample for observation.
Determination of total saponins
A total of 0.500 g dried powder of tissue (leaflets, shoots, roots) of C. paliurus was degreased with petroleum ether twice at 80˚C, 4 h for each time [24]. The defatted retained residue was air-dried at ambient temperature. Total saponins in the residue were extracted using an ultrasonic-assisted method [25]. Briefly, the residue was soaked with 10 ml ethanol solvent (80%, v/ v) for 12 h, and sonicated (KQ250B, Kunshan Ultrasonic Instruments Co., Ltd., Kunshan, China) at 59 kHz for 40 min at 70˚C, then filtered through 0.45 μm microporous membrane. The above process was repeated. Two extracts were combined, condensed under low pressure, and adjusted to 10 ml with 100% ethanol.
The total saponins content was determined by using UV spectrophotometry UV-4802H (Unico, American) at a wavelength of 550 nm in accordance with the protocol described by Deng et al. [26]. A 5% vanillin-glacial acetic acid-perchloric acid solution stained reagents, calculated by the standard oleanolic acid (Nanjing Zelang Medical Technology Co. Ltd., Nanjing, China) curve, expressed as milligrams of oleanolic acid equivalent per gram of dry weight (mg�g-1).
Statistical analysis
Data were presented as mean ± standard deviation. One-way analysis of variance (ANOVA) were conducted to test the significance of total saponins accumulation by Bonferroni test. All statistical analyses were performed at a 95% confidence level. Calculations were conducted using SPSS 20.0 software (SPSS Inc., Chicago, IL, USA).
Histochemical localization of saponins in vegetative organs
Saponins can exhibit characteristic colors from light red to dark red when they react with a 5% vanillin-glacial acetic acid-perchloric acid solution [27]. Thus, the distribution and the accumulation of saponins can be judged according to the change of color. In our study, the distribution patterns of saponins in mature lateral roots were quite different between new and old roots (Fig 1A and 1B). In the new root, secondary phloem contained massive saponins (dark red). In the phelloderm of the periderm, parenchyma cells in the central and vascular cambium were light red, indicating small quantities of saponins. No color appeared in phellem layer and secondary xylem (Fig 1A), suggesting there were no saponins present in these locations. In the old roots, saponins found in phelloderm were concentrated, and those found in the secondary phloem were scattered. Meanwhile, red also occurred in the xylem ray located in the secondary xylem (Fig 1B). The distribution pattern in other parts was similar to that in the new root ( Fig 1A).
In young shoot (April), dark red continuous bands were detected in the epidermis, and dispersal particles were found in the cortex parenchyma. Pale red was identified in the secondary phloem and xylem ray, and no color was evident in other parts of tender cuttings (Fig 1D). In the mature shoot (September), a purple-red color was primarily distributed in phelloderm, cortex, and secondary phloem. Light red was evident in the xylem ray, and no color emerged in other section of the shoot (Fig 1E).
In leaves, deep red primarily occurred in palisade tissue of both the tender and mature leaflets, indicating that a significant quantity of saponins accumulated in these locations (Fig 1G and 1H). The thickness and density of the palisade tissue in the mature leaflets were greater than tender leaves (Fig 1G), predicating that more saponins are accumulated in mature leaflets. However, few saponins were found in the spongy tissue of both leaves. Saponins were also found in the primary vein in the mature leaflet. The secondary phloem, epidermis, and cortex were deep red, and the secondary xylem of the mature leaflet was pale red (Fig 1K). Dark red only occurred in undifferentiated phloem of the main vein in tender leaflets, and no color was found in other parts (Fig 1J). Controls of root, shoot, leaf, and main vein (Fig 1C, 1F, 1I and 1L) did not produce any characteristic red when reacting with the chromogenic agent.
Cytochemical localization of saponins in leaves
In mesophyll cells of tender leaflets, the quantity of black granules (a proxy for saponins) [23,28] was observed in cytoplasm, and the undeveloped chloroplast also contained a limited number of saponins (Fig 2A and 2B). However, in mature mesophyll cells, saponins mainly accumulated in the chloroplast alongside starch granules (Fig 2D). In the epidermal cells of both the tender and mature leaflets, saponins were located in the cytoplasm adjacent to the cell wall (Fig 2C and 2F). However, no saponins were found in the cell wall. Saponins on the endoplasmic reticulum vesicles were found in mature leaflets (Fig 2F), and many black granules occurred outside the cell wall (Fig 2G). Black granules also occurred in the nucleus (Fig 2A and 2E). When leaves matured from April to September, the quantity of black granules and starch grains increased in the chloroplast and decreased in the nucleus.
Dynamics of total saponins content in organs during the growing months
Lateral roots, shoots, and leaflets were taken at the end of each month from April to September. The content of total saponins in the three organs was significantly different (P < 0.05). Overall, the average saponins content in leaves was 20.57 mg�g -1 , which was two times the content in the roots (10.19 mg�g -1 ) and three times the content in the shoots (6.20 mg�g -1 ). The maximum saponins content was 24.45 mg�g -1 in leaves, >13.57 mg�g -1 in roots, and > 6.71 mg�g -1 in shoots.
Dynamic trends of total saponins during the growing months were observed in all three vegetative organs (Fig 3). The content of saponins fluctuated greatly in the leaf, varying from period (from June to July). Interestingly, the second peak in all samples appeared during the initial growth in May.
Histochemical localization of saponins in vegetative organs
Histochemical localization of saponins has been reported in perennial medicinal plants, such as Panax ginseng, Gynostemma pentaphyllum, Bupleurum chinense, and Polygala tenuifolia [17,19,20,27]. The study showed that distribution and accumulation of saponins in C. paliurus had tissue specificity and saponins accumulated mainly in the parenchyma of vegetative organs of C. paliurus, which was consistent with the reports provided by Liu et al. on Gynostemma pentaphyllum [27] and Teng et al. on Polygala tenuifolia [20]. We also found that the distribution of saponins was almost in the peripheral structure of vegetative organs, suggesting that saponins could be responsible for protecting plants against some pathogens and insects [29]. The localization of saponins in shoot showed that the transport of saponins in different organs could be accomplished by phloem. The high content of saponins in the main veins of leaves indicates that the main veins of leaves should be crushed during the processing of products in order to facilitate the release of active ingredients.
Cytochemical localization of saponins in leaves
Saponins reacted with lead acetate and produced black complexes under electron microscope [23]. Based on the histochemical results in our study, the leaf of C. paliurus is an efficient material for the production of saponins. Thus, we conducted cytochemical localization on the leaf only. Localization of secondary metabolites at the cellular level has been the primary focus of studies that assess biosynthesis and physiological functions [30]. Our results of cytochemical localization in leaf of C. paliurus indicated the synthesis site of saponins in the cells and routes of saponins transfer in adjacent cells. The content of saponins and starch grains located in chloroplast increased gradually with the maturation of leaf. We speculate that saponins are mainly synthesized from photosynthate in chloroplasts, and then transport via plasma membrane system like endoplasmic reticulum vesicles. Chloroplasts, as a site of saponins synthesis and accumulation, has been reported in P. ginseng by Yokota [31], who found that Rb1 (a ginsenoside) was localized to the chloroplasts, peroxisomes and cytoplasm of leaf parenchymal cells. Furthermore, Yao et al. presented that terpenoid synthetase activity was remarkably increased on chloroplast membrane in the needles of Pinus massoniana seedlings when treated with MeJA [32]. Comparing the distribution of saponins in tender and mature leaves, the accumulation and synthesis of saponins are related to the structure and function development of chloroplast. Therefore, the photosynthetic efficiency should be considered to increase the accumulation of saponins and yield during the directional cultivation of C. paliurus. For black granules in the nucleus (Fig 2A and 2E), we are not sure whether they are saponins or other substances, and no publication declared the presence of saponins in nucleus. Further verification is needed to support this phenomenon.
Dynamics of total saponins content in organs during the growing months
Determination of harvesting time for medicinal plants depends on the contents and yield of active ingredients in organs at different developmental stages [20,33]. As dynamic patterns presented (Fig 3), significant variations in the content of saponins among organs and growing months were found. Results showed that the maximum accumulation occurred in September, which was consistent with our findings from the histochemical localization. Usually, most of photosynthesis products are used as the source of primary growth (vegetative growth), leading to less metabolites accumulation in earlier growing stage. Interestingly, different from other plants [17,20], the content of saponins in the C. paliurus leaves maintained at a relatively high level, and kept increasing during the blossom period (late of April to May). The expression patterns of key genes related to the synthesis of saponins in yearling seedlings (data unpublished) supported this phenomenon to some extent, but more validation needs to be done on six-yearold plants. The increase of saponins content in earlier growing stage is supposed to be a stress reaction, owing to its sweet taste attracting more predators, and more secondary metabolites can play a role as chemical defense compounds against herbivores or microorganisms [30]. Subsequently, attributing to consumption of more photosynthate in developing fruits, a downward trend of saponins content in leaf occurred in June and July. With the leaf maturation and aging, an upward trend of saponins content appeared in August and September, suggesting that more photosynthetic products conversed into metabolites before dormancy. These results were consistent with the findings in B. chinense [17]. The high content of saponins in initial growth stage also provides guidance for leaf harvesting. Proper harvest in this period could prolong the time of Cyclocarya paliurus tea market and increase economic benefits.
Compared with roots and shoots, higher content of saponins in leaves indicated that leaves would be the main organ of saponins synthesis and storage. Consistent few content in shoot suggested that shoot might be the channel for saponins transportation.
High yield of saponins depends not only on the contents, but also on the biomass of the tree. As we know, for perennial deciduous woody plants as C. paliurus, more biomass of leaves is prone to be attained than shoots and roots. In common, the biomass of leaf in C. paliurus is increasing along with season from April to September. Owing to the largest leaf area and the lowest moisture content before leaf aging, the leaf biomass reaches the maximum in September. Moreover, leaf harvesting in September has little effect on tree growth in the following year. Comprehensively, we recommend that harvesting leaf in September is the efficient option of medicinal use for C. paliurus.
Supporting information S1 File. Raw data for content of saponins. The data used for plotting after analysis is shown in the file. (XLSX)
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2019-10-17T08:56:31.915Z", "created": "2019-10-15T00:00:00.000", "id": "204740004", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0223421&type=printable", "pdf_hash": "6cf0ab032dcb0a5331a62e5e5ee58ff878a8d47c", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:8520", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "ce394b40c7215a3c86515fdf2295b9c5b2011d8c", "year": 2019}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Towards a tailored indoor horticulture: a functional genomics guided phenotypic approach
As indoor horticulture gathers momentum, electric (also termed artificial) lighting systems with the ability to generate specific and tunable wavelengths have been developed and applied. While the effects of light quality on plant growth and development have been studied, authoritative and reliable sets of light formulae tailored for the cultivation of economically important plants and plant traits are lacking as light qualities employed across laboratories are inconsistent. This is due, at least in part, to the lack of molecular data for plants examined under electric lights in indoor environments. It has hampered progress in the field of indoor horticulture, in particular, the transition from small-scale indoor farming to commercial plant factories. Here, we review the effects of light quality on model and crop plants studied from a physiological, physical and biochemical perspective, and explain how functional genomics can be employed in tandem to generate a wealth of molecular data specific for plants cultivated under indoor lighting. We also review the current state of lighting technologies in indoor horticulture specifically discussing how recent narrow-bandwidth lighting technologies can be tailored to cultivate economically valuable plant species and traits. Knowledge gained from a complementary phenotypic and functional genomics approach can be harvested not only for economical gains but also for sustainable food production. We believe that this review serves as a platform that guides future light-related plant research.
By 2050, the world population is projected to reach 9 billion and this will demand at least 70% more food to feed the growing population 1 . In an effort to fight hunger and malnutrition, people around the world face common interrelated challenges of population growth, resource availability, and environmental changes. As more cropland is devoted to non-food purposes such as fuel extraction, cotton farming, housing and other infrastructures, cropland expansion is no longer an attractive nor a feasible option 2 . Furthermore, environmental change including but not limited to the melting of glaciers, soil erosion, and desertification, threaten to reduce land productivity 3,4 . Thus, it is necessary to find sustainable means for growing plants especially food crops. One solution is indoor horticulture where electric light sources are employed in place of sunlight to grow plants in highly controlled closed environments where parameters (e.g., light, temperature, nutrient, carbon dioxide level, and humidity) essential for plant growth can be optimized and maintained throughout its development by a technology-driven approach. This approach enables careful management of waste emission and allows for the application of ''vertical farming.'' Vertical farming refers to the growing of crops in layers or inclined surfaces stacked vertically either on its own or integrated into structures with other primary functions such as skyscrapers and warehouses. This form of farming generates higher yield per unit area of land, thus providing attractive features especially for urban horticulture 5,6 . Resources such as water, light, and nutrients can be tightly regulated to optimize growth parameters yield within a clean enclosed area of farming that reduces the use of pesticides 7 . As such, indoor horticulture presents a sustainable approach to feed the growing population while minimizing the negative impact of agriculture on the environment. For instance, vertical farming has been previously shown to increase lettuce yield per unit area compared to conventional horizontal hydroponics 8 .
In recent years, the use of electric lighting for horticulture has become an increasingly attractive option for plant cultivation either as supplementary or sole light source [9][10][11][12][13][14] . The application of light-emitting diodes (LEDs) lighting for plant growth was first documented in lettuce (Lactuca sativa L.), in which its development under a 16 h photoperiod of monochromatic red LED (660 nm peak) supplemented with a 30 μmol m −2 s −1 photosynthetic photon flux density (PPFD) of blue fluorescent lamps (400-500 nm range) at a total PPF of 325 ± 10 μmol m −2 s −1 , was comparable to those grown under cool-white fluorescent and incandescent lamps 15 . Thereafter and concomitant with the advancement of LED technology, is a surge in the application of LED lighting in horticultural activities (Fig. 1). In general, public interests in this area as reflected by popular Google search terms ''food security'' and ''horticulture science'', has remained relatively consistent over the past decades, but interests in ''LED horticulture'' and ''LED grow lights'' have recorded marked increases in the last 5 years (Fig. 1b). These areas attracted comparable but increasing academic interests in the same period when searched against PubMed database using the same keywords (Fig. 1a). Notably, commercial ''LED grow lights'' were "in-trend" since 2008 presumably for small-scale ornamental plants in home-and office-based applications but the use of LED lighting technology for industrial-scale indoor horticulture (search term: ''LED horticulture'') only gained attention in the last 5 years as reflected by a sharp increase since August 2013 (Fig. 1b) thus implying a maturation of this technology for indoor horticulture applications. In recent years, indoor light sources have achieved significant progress in terms of energy-and costefficiencies. Concomitant with such technological advances is their application as electric lighting for indoor farming. In this regard, its application especially at an industrial scale, has been hampered at least in part by (1) the differential growth, morphology, and developmental behavior of plants cultivated under electric lightings, and (2) the lack of knowledge of how different light wavelengths influence plant performance in general and specific plant traits in particular. The former has been observed especially when using narrow waveband lights such as LEDs and lasers, which yield plants with variable traits and visual appearances [16][17][18] . Here, we pool together recent reports of electric lights for indoor farming and propose how a complementary phenotypic and functional genomics approach can be used to determine a set of light qualities and catalog a wealth of phenotype-specific molecular signatures that are tailored specially for indoorgrown plants.
A recent account of indoor light qualities on plant growth and development
It is well-documented that light qualities (i.e., wavelengths and ratios) and their quantities can be used to manipulate plant characteristics as increasingly advanced lighting systems such as the narrow waveband LED lights and single-wavelength lasers have presented unparalleled technical and economical advantages [19][20][21][22][23][24][25] . Since the photoreceptors of plants absorb light corresponding to the regions of red and blue, a combination of appropriate ratios of LED lights are, therefore, required to achieve normal growth and photosynthetic capabilities such as that observed in a recent study on one Crassulacean acid metabolism (CAM) plant, the ice plant (Mesembryanthemum crystallinum) as well as in ornamental plants such as cabbage tree (Cordyline australis), weeping fig (Ficus benjamina), and gloxinia (Sinningia speciose) [26][27][28] . In the former for instance, red (670 nm peak) and blue (465 nm peak) LEDs provided at a ratio of 9:1 at a PPFD of 350 μmol m −2 s −1 and 16 h photoperiod, yielded the highest shoot and root biomass and shoot/root ratio in the succulent M. crystallinum plants compared to just red or blue LED lights alone 27 .
Using electric lights to cultivate ornamental plants
Since it has been established that a combination of light wavelengths provided at suitable ratios can be used to cultivate plants in indoor environments, LED lights have been tested on many ornamental plants including but not limited to perilla, petunia, carnation, sunflower, rose, and chrysanthemum (Supplementary Table S1). Ornamental plants serve as good case studies for the testing of narrowbandwidth electric lights such as LEDs, because they are in general smaller in size, consume fewer resources, have easily amenable features of broad esthetic values and are already accustomed to growth under indoor (home-or office-spaces) conditions compared to crop plants. In the case of Anthurium (Anthurium andreanum) and moth orchids (Phalaenopsisis), they not only displayed positive development but also yielded higher biochemical content when grown under LED lights (white light at 460 and 560 nm peaks; red at 660 nm peak and blue at 460 nm peak) compared to those grown under fluorescent lights (545-610 nm) (FL) 29 . In this study, the PPFD was maintained at 25 μmol m −2 s −1 under a 16 h photoperiod. Specifically, Anthurium showed the greatest plantlet length and number of leaves when treated with white, blue, and the combination of white, blue and red LEDs. In addition, more roots were observed in cultures treated with FLs and blue LEDs with 6.6 and 6.0 roots, respectively, than in cultures treated red LEDs (1.5 roots). Chlorophyll a, b, and total chlorophyll content were significantly higher in the blue LED treatment (0.692 mg g −1 fresh weight), while the lowest total chlorophyll content was found in the red LED and FL treatments yielding 0.327 and 0.375 mg g −1 fresh weights. Meanwhile, moth orchids displayed the greatest plantlet length and number of leaves when treated with FLs, white, and a combination of blue and red LEDs. Chlorophyll a content was significantly higher in the blue LED treatment (0.2813 mg g −1 fresh weight), while chlorophyll b content was higher in blue and the combination of blue and red LED treatments, yielding 0.1368 and 0.1468 mg g −1 fresh weights, respectively. Total chlorophyll (0.421875 mg g −1 fresh weight) was highest under blue LED while the lowest total chlorophyll content was found in FL treatments and white LEDs yielding 0.1810 and 0.2500 mg g −1 fresh weights, respectively 29 . We have summarized the effects of electric lights on the growth and development of other ornamentals studied in recent years in Supplementary Table S1. Based on the studies performed on different ornamentals, it is clear that narrow-bandwidth lights provided at suitable quality and quantity can induce esthetically valuable features.
Using electric lights to cultivate crop plants
Unlike ornamentals, the use of electric lights on crop plants have a more specialized purpose, i.e., to improve the yield, biomass and/or nutrients for human consumption. Indoor horticulture has great potential to address food security since this form of crop farming is independent of geographical factors and weather conditions. It is, therefore, not surprising that narrowbandwidth electric lights, in particular LEDs, have been broadly employed in studies involving crop plants especially green crops such as lettuce, cabbage, and artichoke (Supplementary Table S1). For instance, artichoke (Cynara scolymus) seedlings grown under red LED lights with a higher PPFDs than blue and white LEDs but only a third of natural light, yielded 60-100% more shoot dry weight and were 67-115% taller than those grown in greenhouse under natural light at a 16 h photoperiod. Nonetheless, seedlings grown under blue or white lights yielded 67-76% less biomass compared to greenhousegrown seedlings 30 . In another report, cucumber (Cucumis sativus) plants grown under monochromatic LED lights (purple at 394.6 nm peak, blue at 452.5 nm peak, green at 522.5 nm peak, yellow at 594.5 nm peak and/or red at 628.6 nm peak at 350 μmol m −2 s −1 PPFD) for 12 h per day have reduced growth, CO 2 assimilation rate, and quantum yield of photosystem II (PSII) electron transport compared to plants grown under white light control provided by natural light supplemented with incandescent reflector lamps 31 . On the contrary, red and blue LEDs improved the growth of lettuce (Lactuca sativa) in another study 32 . Another crop plant, Chinese cabbage (Brassica campestris) also showed improved vegetative and reproductive growth as determined by physical measurements and biochemical contents when treated with weak (80 μmol m −2 s −1 PPFD) blue, blue + red at 1:8 ratio and red LED lights in comparison to fluorescent lamps and sunlight at a 12 h photoperiod 33 . Conversely, an earlier study reported mostly lower growth rates for LED-illuminated cabbage 34 . While it is clear that electric lights have the potential to improve the yield and traits of crop plants in indoor farming, the effects are, however, inconsistent as they vary considerably across different studies. This inconsistency is further highlighted in Supplementary Table S1, which summarizes recent findings involving the use of electric lights on green crop plants.
Using electric lights to cultivate fruits, herbs, and other economically valuable plants Electric lights have also been employed on fruits such as tomato and strawberry as well as herbs such as mint, basil, and dill (Supplementary Table S1). For instance, electric lights can enhance fruit set and yield of strawberry plants although the fruit color appears less saturated as well as inducing greater biomass and phenolic content in basil (Ocimum basilicum) 16,18 . Different combinations of red and blue LED lights were able to increase oil yield in mints (Mentha piperita, M. spicata and M. longifolia), and improve growth and flower buds formation in basil (Ocimum basilicum) 35 (see Supplementary Table S1 for details). Furthermore, electric lights have also been applied on plants that have other economic values such as medicinal plants and assisting reforestation goals. In the latter, narrow-bandwidth lights enable the selection of desirable traits of seedlings pre-cultivated in indoor environments prior to outdoor planting. In one particular example, the seedling quality traits of oak tree (Quercus ithaburensis) can be improved under blue, red, and farred LED lights 36 (see Supplementary Table S1 for details).
Electric lights affect water usage efficiency of plants
One benefit of indoor horticulture is the reduction in water consumption. It was estimated that indoor plant cultivation can reduce water usage up to 90% compared to that used in open field farming. However, recent studies have suggested that the employment of electric lights negatively affected the water usage efficiency of crop and ornamental plants. For instance, the use of red and blue LED lights has resulted in reduced water usage efficiency in tomato (Solanum lycopersicum) and lisianthus (Eustoma grandiflorum) compared to those grown under high pressured sodium (HPS) lamps under a treatment of 100 ± 25 μmol m −2 s −1 photosynthetically active radiation (PAR) at a 16 h photoperiod. In both tomato and lisianthus, whole plant water usage efficiency decreased by 31% under the red and blue LED treatment compared to the HPS treatment. However, under red and white LED lights, whole plant water usage efficiency decreased by 25% for tomato and 15% for lisianthus in comparison to HPS treatment 37 . Therefore, in arid regions, it is necessary to consider the water usage efficiency of plants grown under electric lights in addition to trait improvements.
Using electric lights to cultivate model plants and the relevance of such studies
Meanwhile, electric lights have also been tested on the model plant Arabidopsis thaliana albeit to a lesser extend compared to other economically valuable plants. While Arabidopsis can not represent crop plants and ornamentals, it however presents a fast and convenient way to study the effects of various parameters of electric lights including but not limited to wavelengths, ratios, and intensity. The knowledge gained from studies on Arabidopsis can in turn, serve as authoritative guides for the employment of electric lights on economically valuable plants. In one study, Arabidopsis thaliana plantlets grown under LED lights provided by two LED systems: type L18SP673 L, Valoya and the custom-made LED system (Roschwege) that combined LEDs with the light color warm white 3000K with 660 nm LEDs of moderate and 730 nm wavelength LEDs of low intensity, achieved higher rosette dry weight, seed mass, and developed faster compared to those grown under fluorescent lights (Osram L36W/830 und L36W/840, Osram) under a 180 μmol m −2 s −1 PPFD and at a 16 h photoperiod 38 . However, a recent study employing single-wavelength laser lights revealed a delay in the emergence of new leaves in Arabidopsis seedlings as well as their bolting and flowering times 17 . In this study, the single-wavelength laser beams were generated from a laser illumination system consisting of two diode-pumped solid-state lasers (Laserglow Technologies, Toronto, Canada), adjusted to a ratio of 9:1 of red (671 nm): blue (473 nm) and giving an average of total photon flux density of 90-100 μmol m −2 s −1 . The same study also reported that leaves of laser-illuminated Arabidopsis plants have a lower total chlorophyll content and dry weight.
Currently, studies on Arabidopsis thaliana are limited because the outcome of such studies is often deemed not practical for ''real-world'' applications. There is currently a lack of genomics data to explain the inconsistencies and sometimes contradictory phenotypes in various crops grown under electric lights. This has hampered progress in the field of indoor farming especially in the transition from small-scale plant chambers or home-and officespaces to large indoor plant factories. As such, a functional genomics approach that will be discussed in the following sections, can at least in part bridge this knowledge gap and explain these varying reports. However, in order to ascertain clear co-relations between the phenotypes and their corresponding genomic signatures in response to electric light qualities, which is the core message of this review, a well-characterized model plant such as Arabidopsis is required. Arabidopsis is by far the best characterized with regards to their genetics and is also easily amenable since there is already a large collection of light-related mutants. Cataloging functional genomics data that co-relates with the specific phenotypes of Arabidopsis can, therefore, form a basis that guides optimization works in other economically important crop and ornamental plants, hence the relevance of using Arabidopsis as a reference point. 16,17 ) and such inconsistencies can perhaps be reconciled through answers from a molecular perspective.
Using functional genomics for economic gains
While various methods ranging from phenotypic, physical, and physiological measurements to biochemical and biophysical characterizations have been employed to study the effects of electric lights on the growth and development of plants, a molecular approach that employs a system-wide profiling of gene expression and function, is however under-utilized. Here, we explain citing recent examples, how functional genomics can be used to reveal crucial molecular signatures that can directly link the observed traits of plants grown to their light regimes. Functional genomics present a powerful molecular basis to gain insights on cellular biological processes and act as an inference point for biotechnological manipulations or indoor horticultural ambitions. Modern molecular tools in particular functional genomics, have generated a wealth of molecular data, which enabled mapping of previously elusive biochemical pathways governing plant responses to environmental cues including light perception and adaptation 40 . On the other hand, high throughput and increasingly sensitive instruments have contributed to the identification of missing components and assigning new functions to uncharacterized proteins 40 . Thus far, the major focal functional genomics tools that can be utilized to gain a systems-view on the influence of various light regimes on plant growth and development include transcriptomics, proteomics and metabolomics. Besides microarray technology, for transcriptome-wide studies, RNA-sequencing (RNA-seq) has also become a method of choice. RNA-seq technology provides a wide dynamic identification range of low-abundant transcripts and genetic variants thus permitting detection of more highly confident differentially expressed genes. Above all, functional genomics approach has been hastened by an increase in the number of available sequenced plant genomes. In the following sections, we discuss the employment of functional genomics in uncovering, at systems level, the physiological, biochemical and biophysical changes imposed by alterations in light regimes.
Functional indicators diagnostic of light-induced regimes
Cellular processes predict the phenotype we observe and the efficiency of light absorption and utilization through the analysis of marker genes such as photosystem ii reaction center protein a (psbA), ascorbate peroxidase 1 (apx1), and light-harvesting chlorophyll a/b-binding protein 1 (LHCB1). This system-wide approach is diagnostic for functional and/or structural changes when induced by different light regimes. Distinct wavelengths of light impose different effects on plant physiology and morphology due to the differential sensitivity of photoreceptors. Various light responses are facilitated by a coordinated action of at least one photoreceptor 41 . Very few plant growth and development functional studies monitoring the biological role of different light routines and in particular LED lights, have been reported. Previously, microarray studies identified genes differentially regulated by green light and low red:far-red light ratio 42,43 . In the former study, etiolated Arabidopsis seedlings were treated with a short, single 100 µmol m −2 pulse of green light. The green light treatment induced expression of phytochrome A-regulated, nuclear encoded genes corroborating proper function of the sensitive phytochrome system. This is associated with a robust increase in stem elongation. On the contrary, in tobacco (Nicotiana tabacum), with the same temporal and fluence-response kinetics, plastid-encoded transcripts decreased in accumulation 42 . Taken together, the increase in stem growth rate and a decrease in plastid transcripts denotes a mechanism that influences progression of early commitment to light environment, assisting adaptation of seedling development during the critical process of early establishment 42 .
Light regimes affect hormonal-related gene expression
It has been observed that by modifying low red:far-red ratio, light signaling and hormonal-related genes, particularly in the abscisic acid pathway were affected 43 . Using RNA-seq technology to analyze the molecular mechanisms by which various light qualities control Norway spruce seedling growth and phytohormone levels, a study showed that red light regulates the biosynthesis of gibberellic acid and thus, promotes stem elongation 44 . Furthermore, the authors show that blue light led to an increase in genes associated with secondary metabolites biosynthesis and potentially enhancing plant defenses 44 . Recently, RNA-seq technology was applied to study the effects of various wavebands of light on plant responses at transcriptional level 45 . In this study, grapes (Vitis vinifera) plantlets grown in vitro under white fluorescent lamp (FL40D-EX/38, Huadian CO., China), blue LEDs (peak at 440 nm), green LEDs (520 nm), and red LEDs (peak at 630 nm) lights exhibited over 600 differentially expressed genes with respect to white light. Here, taking advantage of the gene expression, phenotypic, and physiological data, the authors could link plant responses to different light spectrum and the expression patterns of particular sets of genes. For example, exposure to red and green light principally triggered responses associated with shadeavoidance syndrome (SAS) like enhanced stem elongation and decreased chlorophyll levels accompanied by the increased expression of genes encoding histones (H1, H2A, H2B, H3, H4), auxin-repressed protein, xyloglucan hydrolase, early light-induced protein (ELIP) and microtubule proteins 45,46 .
In addition, specific light treatments were observed to induce differential expression of many genes associated with diverse cellular functions, including those involved in ribosome pathway and primary metabolism such as starch and sucrose metabolic pathways, which is also supported by an increase of these metabolites in the plants. Notably, the authors highlighted a potential negative impact of adding sucrose to the culture media where it could at least in part, contribute to the observed increase in root growth and the upregulation of defense genes associated with SAS after exposure to red and green light. Unlike in the red and green light exposures, blue light-induced expression of genes associated with microtubules, chlorophyll biosynthesis, and sugar degradation. However, in the blue light accumulation of genes associated with auxin-repressed proteins and defense-related genes decreased. Generally, the observed effects from blue light exposure may explain the detected increase in leaf growth, chlorophyll synthesis, and chloroplast development as well as increase in chlorophyll a/chlorophyll b ratio in the leaves 45 . In this study, red light seems to promote high carbohydrate to protein ratio, whereas blue light has been observed to induce a low carbohydrate to protein ratio in plants 47 . Overall, plantlets grown under blue light have comparable growth to that observed under white light while exposure to red and green lights seem to cause shade stress on the plantlets 45 .
Light quality has also been shown to influence plantmediated effects on herbivores and beneficial anthropods. Ultraviolet-B radiation (UV-B, 280-315 nm) treated plants attracted parasitoids. This can be attributed to UV-B light treatments promoting upregulation of oxylipin biosynthesis genes, which are also involved in jasmonic acid synthesis 48 . The increased transcription of these oxylipin biosynthesis genes and oxylipin biosynthesis could account for an increased emission of parasitoid-attracting volatiles from UV-B light treated plants. Previously, in sweet basil (Ocimum basilicum), an increase in the composition of volatile organic compounds was detected in essential oils following UV-B light treatment 49 . However, this observation could not be consistently reproduced in other plants suggesting that the indirect effects of UV light depletion for higher plants may be species specific 21 .
All in all, changes in light regimes or exposure to plants influence hormonal associated pathways, which impacts on gene or protein expression and the growth and development of the plants.
Light regimes influence photosynthetic apparatusassociated genes
On another note, most genes related studies target either a single or a few genes that are linked to the photosynthetic apparatus or other light associated responses. For example, blue light with maximum intensity at 452.5 nm has been shown to induce expression of ten genes including fructose-1,6-bisphosphate, fructose-1,6-bisphosphate aldolase, ribulose-5-phosphate kinase, rubisco large subunit, rubisco small subunit, rubisco activase, triose-3-phosphate isomerase and ribulose phosphate epimerase all of which encode key enzymes in the Calvin cycle 32 . In addition, blue light acts as a catalytic wavelength in acquiring high quantum yields of photosynthesis and activating respiration 21 . On the contrary, green (with maximum intensity at 522.5 nm), yellow (594.5 nm), and red (628.6 nm) lights cause downregulation of these key Calvin cycle enzymes 32 .
Another study examining the impact of singlewavelength laser light (adjusted to a ratio of 9:1 of red (671 nm) : blue (473 nm) giving an average total photon flux density of 90-100 µmol m −2 s −1 ) on plant growth and development looked at the expression levels of six photosynthetic marker genes, each representing a main component of the photosynthetic system 17 . Four of the six genes (photosystem I p700 chlorophyll a apoprotein A1, ferredoxin 2, psbA, and LHCB1) showed lower expression in laser-grown plants than cool-white fluorescent grown plants. The low expression of psbA, a gene that is generally associated with photo-inhibition and the photo-damaged of PSII particularly when the light absorption exceeds consumption [50][51][52] , suggests that the laser-grown plants had reduced photo-inhibition. In addition, the expression of two light-stress marker genes, ascorbate peroxidase 1 and glutathione s-transferase was observed to decrease in plants grown under laser light compared to white light signifying that the laser illumination conditions induce less stress than the white fluorescent light (for review see ref. 17 ).
Influence of light regimes on the protein abundance
Just like the transcriptional gene regulation studies, proteomics studies reflecting the fate of translational changes influenced by different light regimes are yet to be fully explored. Similar to influencing transcriptional changes, different light wavelengths induce changes in the global proteome of cells. Strong blue light has been shown to activate the incorporation of carbon in amino acids thereby inhibiting the biosynthesis of starch in leaf chloroplasts while increasing the biosynthesis of proteins 47 . However, this increase of protein biosynthesis induced by exposure to blue light can be abolished by prolonged exposure of plants to red light 53 . Recently, a global proteomics comparative study was performed on laser-and white light-grown plants to evaluate the impact of variable light exposure on the proteome of the plants as well as relate this molecular data to the physiological data collected from the same plants 17 . In this study, plants grown under laser illumination have lower expression of proteins indicative of light and radiation stress responses. There were 115 differentially regulated proteins of which only 17 proteins were upregulated. The majority, 98 proteins, were downregulated and of these 43 were annotated as localized in the chloroplast and 12 of the 43 proteins are involved in photosynthesis. Of important to note is that seven light-harvesting chlorophyll-protein (LHC) complexes including LHCB1.4 (At2g34430) and LHCB3 (At5g54270) were among the most downregulated proteins in the laser-illuminated plants. This corroborates well with the observed changes at the transcripts level. The LHC proteins play an important role in fine-tuning the amount of light energy to be channeled to the reaction centers, a process that enables plants to adapt to a wide spectrum of light environments to drive photosynthesis 54 . Since the LHC family of proteins is light-stress induced, a reduction in its abundance is indicative of reduced photo-oxidative stress under laser-illuminated plants. Furthermore, 16 proteins enriched in gene ontology category "response to light stress" were downregulated in the laser-illuminated plants, lending more support to the concept that laserlight regime induces less stress on plants than white fluorescent light. Phenotypic and biochemical data characterization further supported these observations at molecular levels, including gene expression patterns of marker genes. For example, phenotypically, the average leaf length and area of the first two leaf pairs were observed to be higher in laser-illuminated plants than the white fluorescent light (control) exposed plants but bolting and flowering times were slightly delayed. The authors argued that the delay in flowering time was most likely due to the absence of far-red light that has been shown to promote flowering 55,56 . From a biochemical point of view, laser-grown plants leaves had lower total chlorophyll content compared to control plants. Similarly, it has previously been reported that chlorophyll levels in spinach (Spinacea oleracea) are reduced under a 9:1 ratio of red LED (660 nm) and blue fluorescent lamp at a total PPFD of 282 µmol m −2 s −1 as compared to coolwhite fluorescent light 57 . Additionally, absence of elevation of light-stress response proteins in the proteomics data is indicative that the laser-light regime, which was employed is suitable and sufficient for plant growth 17 .
A highly tailored approach to indoor plant cultivation
Single-wavelength lights (e.g., lasers) and their tunability can conceivably provide a highly tailored approach to indoor plant cultivation, allowing for more flexibility than narrow waveband LEDs [58][59][60] . A recent report on the model plant Arabidopsis thaliana, showed that plants can complete a full cycle of growth and development under singlewavelength lights comprising of adjusted red (671 nm) : blue (473 nm) ratio of 9:1 17 . This has paved the way for more comprehensive light formulae containing well-defined mixtures of light wavelengths, intensities and ratios, to tailor for the cultivation of different plants and to obtain specific plant traits for indoor horticulture activities 59 .
We, therefore, propose the use of a complementary phenotypic and functional genomics approach to determine the optimal light conditions for plant growth in indoor environments using Arabidopsis thaliana as the model plant since the Arabidopsis plant has a short life span, a large collection of light-related mutants and a fully sequenced genome and proteome, and importantly, the findings can be easily translated to crop plants and where necessary further optimize light conditions with model crop plants like rice (Oryza sativa) or tomato 61 . As illustrated in Fig. 2, we propose that a preliminary broad screening that document extensively the phenotypes of model plants grown under different light regimes (i.e., different wavelength combinations, intensities and ratios) should be conducted using narrow waveband and/or single-wavelength lights to determine the optimal light conditions that yield economically important traits. Subsequently, a functional genomics approach involving transcriptomics and proteomic analyses of plants displaying economically beneficial traits under the respective light regimes can be conducted, to ascertain the molecular signatures governing the regulation of genes involved in expression of these desirable phenotypes 62 . A catalog of phenotype-specific molecular signatures can act as an authoritative guide to determine the optimal light qualities for crop plants, for cultivating different traits and for potential biotechnological innovations that are specific for indoor horticulture applications 40,63 . Given that the knowledge gained from a complementary phenotypic and functional genomics study can be harvested for economical gains, this highly tailored approach to indoor horticulture can, therefore, contribute to sustainable food production.
(see figure on previous page) Fig. 2 An illustration of a highly tailored indoor horticulture approach. a Light regimes comprising of well-defined mixtures of singlewavelength lights in their pre-determined optimal ratios and intensities can be used to grow different species of plants, cultivate economically important plant traits, and optimize the growth and development stages of plants in highly controlled indoor environments. b Plants displaying favorable traits and growth parameters under the optimized light regimes are subjected to functional genomics where their underlying molecular signatures can be harvested for biotechnological innovations to produce plant traits and yield that are economically attractive especially when grown under light regimes of indoor environments
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2018-11-02T16:03:32.631Z", "created": "2018-11-01T00:00:00.000", "id": "53190713", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41438-018-0065-7.pdf", "pdf_hash": "2baeb9cde248b5bff5ce6015556d1f7b71685d32", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:7068", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "2baeb9cde248b5bff5ce6015556d1f7b71685d32", "year": 2018}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Flow system for automated analysis of maize pollen.
Pollen grains are haploid gametes of uniform shape and size, and can be obtained in large quantity. If appropriate traits are used, they can be an excellent material for investigation of rare but important biological events like intracistronic recombinations or mutations induced by very low level of mutagens. This advantage will be further improved, if the laborious counting and examination can be made automatically. For automation of pollen analysis, techniques of flow analysis and image analysis would be applicable. Flow analysis with a optical detector was tested using maize pollen. Pollen grains were transported by gentle suction through a glass capillary which was placed under a microscope. Interruptions of the light path by pollen grains were detected by a silicon photocell after optical magnification and converted into electric pulses. The frequency distribution of pulse height was examined by a multichannel pulse height analyzer. 10(6) pollen grains would be counted and classified within about 30 min for a pollen suspension dilute enough for separation of each pulse. The flow system tested seems promising for detection of Wx mutant pollen in a wx pollen population after iodine staining if illumination of sample particles is improved. ImagesFIGURE 1.
Introduction
Pollens of higher plants are haploid and can be handled in large numbers. If appropriate genetic markers can be used, they seem to be good material for investigation of rare but important biological events like intracistronic recombination (1-3) or mutation (4)(5)(6). Since pollen grains are uniform particles in size and shape, analysis by electronic instruments would seem possible. The advantages of use of pollen would be further improved if the laborious and time-consuming procedure of scoring of pollen could be replaced by automated analytical instruments. Such instrumental analysis would be useful both in counting very large numbers of pollen and in objective classification of characters of pollen grains.
Two types of measurement can be expected, i.e., flow system and image analysis. Both systems may be used as total examination or as sampling analysis. In the flow system, however, basically all the particles are counted and examined, one by one, when they pass through the detector. The simplest * Department of Induced Mutation, National Institute of Genetics, 1,111, Yata, Misima, Sizuoka, Japan 411. January 1981 instrumentation may be a combination of particle detector and counter. If the characteristics of the particle can be classified into groups by the nature of the signals which were generated at the detector when the particle passes, addition of counters and signal analyzer will complete a basic measuring system. One such system was assembled to examine pollen grains of maize (Fig. 1). The present paper reports the preliminary results of an automated flow analysis system and makes practical evaluation for further improvements.
Materials and Methods
System Set-up The flow system used is shown schematically in Figure 2. Pollen samples were suspended in water and agitated by a small stirring blade. They were transported through a glass capillary (about 0.8 mm diameter) by gentle suction of air using a microtube pump. The capillary was fixed between a slide glass and cover slip, and the spaces around the capillary were filled with balsam. The capillary was positioned precisely in the light path of a microscope. Interruptions of the light path by pollen grains 165 FIGURE 1. Pollen of waxy maize stained with iodine. In this material, the darkly stained pollen grain at the center is a wild type (Wx) recombinant. Phenotypic revertant towardl non-waxy may be comparable to this dark pollen grain.
were detected by a silicon photocell (Hamamatsu TV, S876-16BR) after optical magnification of about x 10. Electric pulses from the silicon photocell were amplified and sent to a multichannel pulse height analyzer (Canberra, Series 30). The multichannel analyzer (MCA) used was capable of storage of data up to 106_1 counts in each of its 1024 channels. It also had regions of interest (ROI) function with integrated data readout. Each input pulse was assorted to corresponding channel according to its pulse height. The number of assorted pulses was recorded in counters of each channel. Counts of neighboring channels could be pooled by ROI function by designating the region of interest. The frequency distribution of pulse height could be displayed on a CRT screen of the MCA while collecting and analyzing the data. This display could be transferred onto paper by use of an X-Y plotter. Numerical data could be printed out by a line printer. The X-Y plotter and line printer complete the instrumental set up. Extra care was taken to stabilize the light source to illuminate the capillary. Mechanical fixation, soldered leader wires, and an electronic voltage regulator were used.
Sample Preparation
The present experiments were intended to test the flow type automatic analysis for detection of phenotypic reverse mutation of waxy (wx) pollen to non-waxy (Wx) pollen. For this purpose, mature maize pollen was used to test the flow system and to obtain optimal flow rate and other operational data. 166 Mature maize pollen had been harvested in the field and stored in 70% ethanol. Some younger pollen usually used in fine structure analyses of wx locus, was also tested after iodine staining (Fig. 1). In both cases, pollen grains were washed clean with 70% ethanol a few times to eliminate small debris in the sample suspension.
Pulse Shape Analysis
Shapes of the electric pulses generated by interruptions of light by pollen grains were brought to a standstill for observation on a CRT screen of a dual channel synchroscope (Iwatsu Electric, Synchroscope SS-5050) using digitized memory (Kawasaki Electronica, Transient Memory TM-1410). The Transient Memory had outputs for an X-Y plotter to make a paper copy of pulse shape. To the second channel of the Synchroscope, either calibration voltage for pulse height or timing pulses for pulse width could be applied for measurements.
Measuring Procedure
After a 1 hr warmup to stabilize the instruments, the sample vial and exhaust vial were set in position. The agitator of the sample vial and microtube pump for transportation were turned on. When the grains start moving through the capillary, electric pulses were confirmed on a monitor Environmental Health Perspectives synchroscope (Matsushita Communication Industrial, VP-5102A). MCA was activated for data collection by turning on the "collect" switch. When most of the sample suspension was transported, and before air bubbles came up, data collection by MCA was terminated by turning off the "collect" switch. ROI was set after data collection to include the major part of the peak of frequency distribution of pulse height. The X-Y plotter and digital printer recorded results of analysis. January 1981
Flow System
Although the capillary used for maize pollen grain was quite large, particles flowed at the center of the capillary when suspension was transported at the rate of 2 ml/min or 4 m/min. This centering of pollen grains in flow was an unexpected advantage of the present system and made focusing of pollen images on the photocell easy. Clogging in the flow system could be avoided by use of a thick capillary and cleaned pollen suspension. However, the cleaning of the pollen suspension eliminated both small debris and empty shells of abortive pollen. The latter fraction would be biologically important in some experiments, but in the present experiments only fully grown pollen grains were analyzed.
Pulse Shape
The silicon photocell used had a narrow lightsensitive area of 1.2 mm x 6 mm, and was placed to transverse the image of the capillary. This gave a sharp pulse and high resolution. Pulse shapes were uniform with little variation in pulse height when a diluted pollen suspension was used. Examples of the pulse shapes are shown in Figure 3. The count rate for this concentration was 500 pulses/sec or 3 x 104 pulses/min. Clustering or clumping of the particles occurred when the concentration of suspension was higher. An example of such piled up pulses is shown in Figure 4.
Frequency Distribution of Pulse Height
Electric pulses were generated by interruption of the light path by pollen grains. The height of each pulse reflects the characteristics of the pollen, the optical density, which might relate to size and color of pollen grain. The size of pollen grain was also expressed as width of each pulse. In the present experiments, the size of pollen grains did not vary very much by visual examination. The frequency distribution of pulse height was analyzed by MCA. An example of the frequency distribution of pulse height of unstained mature pollen of maize is shown in Figure 5 together with ROI readout data. In case of piled up pulses like those in Figure 4, the peak of the frequency distribution of pulse height was broad. A mixture of younger pollen was also examined after iodine staining, but, as discussed later, pulse heights did not differentiate between Wx and wx as far as transmitting light was used.
Discussion
Automation of the pollen analysis may be worthwhile both for quick and accurate counting of the total number of pollen grains and in objective classification of the character of pollen. If well cleaned maize pollen were to be used, counting of the total number of pollen grains would be easy with the flow system reported here and a low cost electronic counter. Counting of pulses which exceeded in their height the predetermined threshold level could be made by using a discriminator or comparator circuit. Recent advancements in the electronic industry have developed a compact and relatively low cost multichannel pulse height analyzer (MCA). The MCA used here had 1024 counters, each capable of 106_1 counts, and was able to sort pulse to the corresponding counter according to pulse height. This could be used as main analyzer of the flow system.
Theoretically, four classes of pulses, each differing in height, would be expected in reversion experiments of waxy maize pollen if optical detection of particles was used. They would be, in order of increasing height, (1) small debris and noises inherent to the elements used, (2) abortive pollen, (3) normal pollen, and (4) mutant Wx pollen. As described before, the height of the pulse might reflect optical thickness and color of the pollen grain. The size of pollen grains did not vary significantly, but as maize pollen grains were eggshaped and the directions of their axes might vary randomly in flow, the optical thickness as sensed by the photodetector might deviate considerably. However, the difference in color of typical Wx and wx pollen after iodine staining was unmistakably clear. Wx pollen stains dark blue-black and wx light brown. For objective classification of pollen character, use of an appropriate color filter will help to differentiate mutant Wx pollen from parental wx pollen.
Illumination of pollen grains was important. In the case of wx-Wx pollen experiments, Wx pollen can be detected best with reflecting light after iodine staining. However, if only reflecting illumi-nation were used with a dark background, darkly stained Wx pollen would produce lower electric pulses than pulses produced by light colored wx pollen. This would make distinguishing between darkly stained mutant Wx pollen and empty shells of abortive pollen difficult. Improvement is possible by adopting both types of illuminations. Transmitting illumination would produce an electric pulse which corresponds to each pollen grain, and reflecting illumination would permit subtraction of the height according to lightness of color. This would leave pulses from dark Wx pollen less affected and maintaining their height highest among the four classes of pulses described before. An additional red color filter would help in differentiating mutant Wx pollen pulses from background wx pollen pulses.
It should be mentioned that in the present set-up, sample and exhaust vials were exchangeable and pollen was intact after examination. Pollen could be counted or examined repeatedly for the total number and presence of mutant pollen grains or could be spread on a large slide glass for visual examination.\===
Domain: Biology Medicine Agricultural And Food Sciences. The above document has
* 2 sentences that start with 'Interruptions of the light path',
* 3 sentences that start with 'The frequency distribution of pulse',
* 2 sentences that start with 'The capillary was',
* 2 sentences that start with 'The X-Y plotter and',
* 2 sentences that start with 'An example of',
* 2 sentences that start with 'The size of pollen',
* 2 sentences that end with 'of each pulse',
* 2 sentences that end with 'the "collect" switch'. It has approximately 1975 words, 107 sentences, and 25 paragraph(s).
|
[{"added": "2014-10-01T00:00:00.000Z", "created": "1981-01-01T00:00:00.000", "id": "7155708", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "pd", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1289/ehp.8137165", "pdf_hash": "64696ce5a5a74e8ea75e19644946b2cb1ab6aacd", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:3111", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "64696ce5a5a74e8ea75e19644946b2cb1ab6aacd", "year": 1981}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
This below document has 2 sentences that start with 'Calves were individually', 2 sentences that start with 'Faecal egg counts were determined', 2 sentences that start with 'Composite faecal cultures for each', 2 sentences that end with 'four dairy farms in Co', 2 sentences that end with 'day 0 and day 14', 2 sentences that end with 'on the four farms, respectively'. It has approximately 1767 words, 66 sentences, and 19 paragraph(s).
===
Ivermectin treatment failure on four Irish dairy farms
We report on the use of the faecal egg count reduction test to evaluate the performance of ivermectin in treating gastrointestinal nematode infections in first grazing season (FGS) calves on four dairy farms in Co. Kilkenny, Ireland. On each farm, FGS calves were injected subcutaneously with ivermectin in accordance with their live weight (day 0). Calves were individually faecal sampled on both day 0 and day 14. Faecal egg counts were determined using the Mini-FLOTAC technique. Composite faecal cultures for each farm were performed on each sampling occasion. The faecal egg count reductions (mode) ranged from 17.3–80.2% with the lower 95% confidence limit ranging from 3.1–72.3% on the four farms, respectively. Ivermectin-resistant nematodes were detected on all farms, with evidence of Ostertagia resistance on one farm. This study highlights the urgent need for Irish producers to reappraise their parasite control practices. Electronic supplementary material The online version of this article (10.1186/s13620-019-0142-8) contains supplementary material, which is available to authorized users.
Introduction
Given the increase in the number of reports of the detection of anthelmintic-resistant nematodes in cattle [1][2][3], there is a clear need to establish the efficacy of commonly used anthelmintics on Irish cattle farms so as to ensure that neither animal welfare nor performance is compromised.
Of the anthelmintics available to beef and dairy producers, macrocyclic lactones (Group 3 -MLs) such as ivermectin are a popular choice [4,5] and their popularity with producers can be readily explained by both their ease of use (e.g. available as pour-ons, low dose volume) and their persistence of activity against gastrointestinal nematodes [6,7]. However, concerns exist over their long term sustainability and resistance to them has already been reported on two Irish beef research farms [8].
Although the controlled efficacy test is regarded as the gold standard for determining anthelmintic efficacy [9], the faecal egg count reduction test (FECRT) is more widely employed as it does not involve the slaughter of potentially expensive animals. The guidelines for conducting the FECRT were devised chiefly for sheep [10] but they can also be used for other species such as cattle, horse and pigs. Using this test, anthelmintic resistance (AR) is classified as occurring when the FECR is less than 95% with the lower 95% confidence limit being less than 90%. Resistance is only suspected when just one of these conditions are met.
In this study, we report on the use of the FECRT to evaluate the performance of ivermectin in treating gastrointestinal nematode (GIN) infections in first grazing season (FGS) calves on four dairy farms in Co. Kilkenny, Ireland.
Materials and methods
On each of four dairy farms, a minimum of 12 FGS Holstein Friesian female calves which had not received any prior anthelmintic treatment and were on pasture for a minimum of eight weeks, were randomly selected and then injected subcutaneously with ivermectin (Ivomec Classic Injection for Cattle and Sheep 10 mg/ml, Boehringer Ingelheim Vetmedica GmbH) by a private veterinary practitioner using an injection syringe. Calves were treated in accordance with their live weight (day 0) as determined with the aid of a weigh tape at a dosage rate of 1 ml of the product per 50 kg of live weight. All dosage volumes were rounded up to the nearest millilitre. Calves were individually rectal faecal sampled on both day 0 and day 14. Faecal egg counts were determined using the Mini-FLOTAC technique [11] (limit of detection of five eggs per gram of faeces (EPG)). Data were analysed using the 'shiny-eggCounts' package ( [URL]:// shiny.math.uzh.ch/user/furrer/shinyas/shiny-eggCounts/ ). Composite faecal cultures for each farm were performed (2 g of faeces per calf ) on each sampling occasion to determine the composition of nematode genera in each treatment group. Cultures were incubated at 27°C for eight days and 100 L 3 larvae per culture were identified to genus level using standard identification keys [12] on recovery. All L 3 larvae were identified when counts were less than 100. A questionnaire (see Additional file 1) was also prepared so that details of parasite control practices and grazing management strategies for FSG calves on each farm could be recorded.
Results and discussion
Across all farms, the mean (arithmetic) FEC on day 0 did not exceed 100 EPG ( Table 1). The FEC reductions (mode) ranged from 17.3-80.2% with the lower 95% confidence limit ranging from 3.1-72.3% on the four farms, respectively. Cooperia L 3 larvae were detected in all four post-treatment faecal cultures (Table 2), with both Ostertagia and Bunostomum L 3 larvae also being identified in the post-treatment faecal culture from Farm 2.
With regard to the survey conducted, none of the farmers had previously used either faecal egg counting or calf growth rates as a guide to determine the need for anthelmintic treatment. All four farmers reported using the 'dose and move' system, with all calves in the group being treated at the same time with anthelmintics prior to moving to silage aftermath. All anthelmintic treatments given to FGS calves in the previous year were for preventive purposes only using an avermectin-based product, with the first treatment on two of the farms given at six weeks post turnout. Three of the farmers reported that they treat FGS calves a minimum of three times in the first year, while three also use the same parcel of grazing land each year for FGS calves. On two farms, calves were turned out to pasture in May while on the other two farms the month of turnout was March and April, respectively.
Using standardised criteria for defining the occurrence of AR [10], we report the presence of ivermectin resistant nematodes on all four study farms. None of these farms had previously reported issues of anthelmintic treatment failure. Despite the reporting of resistance by Cooperia to MLs now being a relatively common occurrence [3], cases of Ostertagia resistance to MLs as reported here, are much less frequent [13,14]. Although Cooperia infections can potentially affect animal performance [15,16], Ostertagia is still a much more significant parasite of FGS cattle, and any decline in the treatment efficacy of an anthelmintic in treating this parasitic infection can lead to significant penalties with regard to animal health, welfare and performance. It is difficult to determine what level of importance should be ascribed to the presence of Bunostomum spp. in the post-treatment culture of Farm 2 given that only one L 3 larvae was detected , while no L 3 larvae of this genus were observed in the pre-treatment faecal culture. Furthermore, it is important to state that caution always needs to be exercised with regard to the interpretation of faecal larval culture results given that they may not accurately reflect the composition of the worm burden of the host animal [17]. This may be as a result of both the differences in the fecundity of the worm genera and the rates of larval mortality occurring during culture [18].
It should be recognised that a number of factors such as sample size, the detection limit of the method used to determine FEC, the pre-treatment FEC values, the level of FEC aggregation within the treated group and the method used to generate confidence intervals can influence both the detection and interpretation of treatment efficacy/inefficacy [19,20]. In an effort to mitigate against the influence of some of these factors on both test sensitivity and specificity, 15 calves were randomly selecting for sampling on day 0. This is based on [21]. It was therefore decided to use the Mini-FLOTAC technique for FEC determination as this method has a considerably lower detection limit compared to the standard McMaster method (limit of detection of 50 EPG) and this would help to offset the negative influence of the low pre-treatment FEC values in determining the true efficacy of ivermectin in this study. A number of the parasite control practices and grazing management strategies employed by the farmers in this study favoured the development of AR. The 'dose and move' system [22] which was previously advocated as a parasite avoidance strategy may actually accelerate the development of AR as the only surviving nematodes that will seed the new pasture with eggs will be resistant types [23]. In addition, the use of anthelmintics on a preventive basis only and the failure to use common markers of parasitic infection such as FEC determination or measuring calf performance may result in potential overuse of anthelmintics. Indeed, an overuse of anthelmintics may hasten the development of AR by reducing the population of nematodes in refugia. The in refugia population refers to that portion of the nematode population not exposed to anthelmintic treatment [24].
The ultimate challenge in controlling nematode challenge in calves is to strike a balance between calf performance and maintaining the size of the population in refugia. This involves the regular monitoring of livestock throughout the grazing season for evidence of parasitism with commonly used markers such as FEC. Although FEC in general are not a reliable guide of the parasite burden of a calf as faecal egg output conforms to a stereotypic excretion pattern independent of the nature of the infection [25], whereby an initial increase in egg output is followed by a subsequent decrease which occurs logarithmically [26]. This is as a result of the fecundity of female nematodes being governed by a density-dependent mechanism which appears to involve the host animal [27]. However, it has subsequently been determined that the control of egg output in female nematodes by density-dependent mechanisms in the early stages of the grazing season appears to be minimal [28] and FEC do accurately reflect the level of challenge experienced by calves in the first two months of the grazing season. As a result, FEC measured two months post turnout are a useful tool in predicting the level of parasitic challenge in the latter half of the grazing season [28] and may potentially be used as a guide as to whether clinical parasitism may arise later in the season [29]. This can be used as a basis for determining the need for anthelmintic treatment.
Conclusions
The detection of the presence of ML-resistant nematodes on all four farms, and in particular Ostertagia resistance to ivermectin on one farm, should serve as a timely reminder that greater efforts need to be made to delay the development of further resistance to commonly used anthelmintics on Irish farms. With this in mind, a more targeted approach to the control of GIN infections is advocated, providing producers are aware of the risk of dictyocaulosis occurring under these grazing conditions.
Additional file
Additional file 1: Questionnaire used to generate the survey data. (DOCX 16 kb)
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2019-05-25T13:03:01.837Z", "created": "2019-05-15T00:00:00.000", "id": "162183095", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://irishvetjournal.biomedcentral.com/track/pdf/10.1186/s13620-019-0142-8", "pdf_hash": "fd72e86a0c1f6084d5651184e5eb9980049039c8", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:5959", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "fd72e86a0c1f6084d5651184e5eb9980049039c8", "year": 2019}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Epizootic Spread of Schmallenberg Virus among Wild Cervids, Belgium, Fall 2011
Schmallenberg virus was detected in cattle and sheep in northwestern Europe in 2011. To determine whether wild ruminants are also susceptible, we measured antibody seroprevalence in cervids (roe deer and red deer) in Belgium in 2010 and 2011. Findings indicated rapid spread among these deer since virus emergence ≈250 km away.
Schmallenberg virus was detected in cattle and sheep in northwestern Europe in 2011. To determine whether wild ruminants are also susceptible, we measured antibody seroprevalence in cervids (roe deer and red deer) in Belgium in 2010 and 2011. Findings indicated rapid spread among these deer since virus emergence ≈250 km away. D uring summer and fall of 2011, a nonspecifi c febrile syndrome among adult dairy cows in northwestern Europe was reported. During November 2011, an enzootic outbreak causing fetal death or neurologic signs in newborn lambs, kids, and calves emerged throughout several countries in Europe. Both syndromes were associated with the genome of a new Shamonda/Sathuperi-like orthobunyavirus named Schmallenberg virus (SBV) in the blood (adults) or central nervous system (newborns) (1,2). Susceptibility of wild ruminants can be expected on the basis of the behavior of related viruses of the Simbu serogroup. Therefore, we measured seroprevalence of antibodies against SBV in wild red deer (Cervus elaphus) and roe deer (Capreolus capreolus) and looked for the viral genome in fetuses from pregnant deer found dead.
The Study
Blood samples were collected during postmortem examination of 313 red deer and 211 roe deer shot during the 2010 and 2011 hunting seasons. The 524 samples were randomly collected during October-December from 35 hunting estates in 4 of the 5 provinces in southern Belgium ( Figure 1). The animals' sex; age; body condition; and macroscopic aspects of hooves, mucosae, and internal organs were recorded. IgG against the recombinant nucleoprotein of the emerging SBV was detected by using an ELISA kit (ID Screen Schmallenberg Virus Indirect, version 1; ID.vet Innovative Diagnostics, Montpellier, France). Results are expressed as percentages of the reference signal yielded by the positive control serum; serologic status is defi ned as negative (<60%), doubtful (60%-70%), or positive (>70%). Neutralizing antibodies against SBV were sought as described (3) in subsets of roe deer serum (IgG-negative and IgG-positive according to ELISA), and a linear relationship between percentages and reciprocal neutralizing titers was found. In addition, necropsies were performed on 22 fetuses and 5 newborn red deer fawns; brain samples were tested for SBV genomic RNA and cellular β-actin transcripts by reverse transcription quantitative PCR (3). Contingency tables were analyzed by using χ 2 analysis to detect associations between seroconversion and species, sex, age, sampling location, and sampling date. Signifi cance level was p<0.05.
No gross lesions compatible with any disease were found in any deer. All 299 serum samples collected during the fall of 2010 were negative for IgG against SBV. However, among the 225 samples from deer killed in 2011, seroprevalence was 43.1% (95% CI 36.6%-49.6%). No signifi cant association was found between species and seroconversion: 40.5% (95% CI 31.6%-49.5%) among red deer and 45.9% (95% CI 36.5%-55.2%) among roe deer (p = 0.42). Acquired immunity against SBV was thus already high, suggesting that SBV had quickly spread since its emergence ≈250 km northeast during late summer 2011.
A signifi cant association between month of sampling and seroconversion was detected for both deer species (Figure 2). This late circulation of virus might be surprising because biting midges of the genus Culicoides, which reportedly transmit SBV (4), are not usually active during cold months. However, during fall 2011, temperatures in the region were substantially higher than normal (5) and thus compatible with persistent wild-ruminant exposure to biting midges until mid-December. No association was found between seroconversion and sex of the deer (p = 0.71 and 0.85 for red and roe deer, respectively), age (p = 0.99 and 0.24), and location of sampling (p = 0.47 and 0.23). These results suggest a similar level of exposure to infected vectors and a similar degree of susceptibility to infection among all animals in the study area (13,058 km 2 ).
In most animals that had been found dead, gross lesions were consistent with trauma (e.g., fractures, hematomas, hemoperitoneum/thorax, ruptured spleen) suggestive of impact against a vehicle. No fetus or newborn showed morphologic alterations of the neck, trunk, or limbs suggestive of arthrogryposis. No macroscopic abnormalities were seen in the cerebral cortex, cerebellum, and spinal cord. All β-actin-positive samples of these 27 fetuses and newborns remained negative for SBV RNA. Unfortunately, postmortem decay rendered fetal serum not suitable for analysis.
Conclusions
SBV infects wild cervid populations, and infected insect vectors were homogeneously distributed over southern Belgium in the fall of 2011. Emergence probably took place in 2011. However, because seroprevalence was already 20% in red deer and 34% in roe deer during October and because our results show that the proportion of the infected population increased exponentially during October-December, we suggest that the virus began circulating months earlier than the currently believed August/September (3). We recently showed that among the fetuses of pregnant cows that were infected after the establishment of the fi rst placentome, 28% were infected and that an arthrogryposis/ hydranencephaly syndrome follows if transplacental virus transmission occurs before fetuses are immunocompetent (6). For this study, no feedback from forest rangers, no macroscopic observations, and no PCR results suggested transplacental contamination. However, aborted fetuses and stillborn and distorted nonviable newborn fawns are almost impossible to collect in the wild (quickly eaten by scavengers), and the absence of SBV-specifi c genetic material or morphologic alterations at necropsy are not evidence of noninfection. Therefore, no objective facts confi rm or refute transplacental transfer.
Because the virus can infect the fetus only after the fi rst placentome has developed and because roe deer embryos remain in diapause until January (7), it is unlikely that SBV has contaminated many roe deer fetuses. Because 90% of roe deer were already SBV positive in mid-December and because circulating antibodies prevent transplacental passage of the closest phylogenetic relatives of the virus (8), we suggest that roe deer fetuses were probably not infected. On the contrary, red deer mate in September, and the fi rst functional placentome is established by the end of October (9); thus, 80% of pregnant red deer were exposed to the emerging virus when placental transfer was possible. Furthermore, 35% of pregnant red deer were infected in November and December, i.e., after establishment of the fi rst placentome and before the fetus was immunocompetent. By extrapolating the rate of transplacental infection among cattle (6), we determined that 28% of these pregnancies resulted in contamination of the fetus, i.e., 10%, of expected pregnancies. Because unrestricted replication of Simbu-like viruses occurs in the central nervous system of immunologically incompetent ruminant fetuses (1), which can lead to a typical arthrogryposis/hydranencephaly syndrome, a 10% loss among fawns can be expected in 2012.
In the same geographic area, 5 years apart, 2 arboviruses have emerged: as vectors and infect sheep, goats, cattle, and red deer. Although most (>50%) red deer seroconverted against BTV-8, only a few (<3%) roe deer sampled in the same places and at the same time were BTV-8-positive (10), which sharply contrasts with the SBV seroconversion rates reported here. This fi nding invalidates the assumption that less exposure of roe deer to infected midge bites explains the almost complete absence of seroconversion against BTV-8 in this species. The emergence of SBV thus reveals the existence of roe deer-specifi c anti-BTV-8 host factors, posing a fascinating question.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2017-06-18T04:00:24.952Z", "created": "2012-12-01T00:00:00.000", "id": "6033606", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.3201/eid1812.121067", "pdf_hash": "c787556bbd60d178e9ab79d4f8ba1d122c84e2dd", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:1008", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "c6e42df978da200c3eef17c3697f557a894a9dcd", "year": 2012}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Flowering time control in European winter wheat
Flowering time is an important trait in wheat breeding as it affects adaptation and yield potential. The aim of this study was to investigate the genetic architecture of flowering time in European winter bread wheat cultivars. To this end a population of 410 winter wheat varieties was evaluated in multi-location field trials and genotyped by a genotyping-by-sequencing approach and candidate gene markers. Our analyses revealed that the photoperiod regulator Ppd-D1 is the major factor affecting flowering time in this germplasm set, explaining 58% of the genotypic variance. Copy number variation at the Ppd-B1 locus was present but explains only 3.2% and thus a comparably small proportion of genotypic variance. By contrast, the plant height loci Rht-B1 and Rht-D1 had no effect on flowering time. The genome-wide scan identified six QTL which each explain only a small proportion of genotypic variance and in addition we identified a number of epistatic QTL, also with small effects. Taken together, our results show that flowering time in European winter bread wheat cultivars is mainly controlled by Ppd-D1 while the fine tuning to local climatic conditions is achieved through Ppd-B1 copy number variation and a larger number of QTL with small effects.
INTRODUCTION
Flowering time is one of the most important phenological stages in crop development, as it is key to adaptation, yield potential and yield stability Mühleisen et al., 2013). Wheat (Triticum aestivum L.) covers more of the world's surface than any other food crop and has the largest production volume of all staple crops in Europe (FAO, 2013). This worldwide expansion and success were possible because of the adaptability of wheat flowering time to different environmental conditions as facilitated by the vast natural variation provided by the hexaploid wheat genome . Understanding the genetic control of flowering time may increasingly gain importance as extreme weather conditions will be expected to occur more frequently already in the near future (Beniston et al., 2007) with potentially negative effects on yield. For example, tailoring flowering time of wheat to local climatic conditions facilitates avoiding high temperatures and drought stress during anthesis and grain filling (Bennett et al., 2012;Bentley et al., 2013). The prediction of flowering time thus plays a key role in adaptation breeding as well as to efficiently transfer promising genotypes into regions with different climatic conditions. Furthermore, in order to increase wheat yield potential, an initiative has recently been started to launch hybrid breeding in wheat (Longin et al., 2012;Whitford et al., 2013). Especially for hybrid seed production, the timing of flowering between male and female parental lines has to be synchronized and in addition, the vulnerability during anthesis can result in decreased pollen production (Pickett, 1993;Longin et al., 2013;Langer et al., 2014).
Flowering time in cereals is controlled by three different signaling pathways: the vernalization (Vrn), photoperiod (Ppd) and earliness per se (Eps) pathway (for review see Snape et al., 2001;Distelfeld et al., 2009;Kamran et al., 2014). The group of Vrn genes regulates the transition from the vegetative to the generative phase in response to temperature (Distelfeld et al., 2009;Allard et al., 2012) and thus determines winter and spring growth habit. These genes play, however, only a minor role for winter wheat flowering time provided that the vernalization requirement is fulfilled (Kamran et al., 2014). Wheat is a photoperiod sensitive crop and thus flowers only after a critical day length has been reached. However, photoperiod insensitivity has been selected by plant breeders for several decades to enhance yield in certain climatic conditions. Photoperiod (Ppd) loci genetically control the response to photoperiod, with photoperiod insensitive alleles inducing flowering irrespective of day length. Ppd-1 encodes a pseudo-response regulator (PRR) (Turner et al., 2005) and the Ppd-1 homeoloci are located on group 2 chromosomes. Eps summarizes all other loci that affect flowering time independently of vernalization and photoperiod response (Worland, 1996;Kamran et al., 2014;Zikhali et al., 2014). Furthermore, different studies reported a moderate but significant correlation between heading time and plant height and Wilhelm et al. (2013) reported a significant effect of Reduced height (Rht)-B1 on heading suggesting that genes controlling plant height might also affect flowering time.
For a long time single nucleotide polymorphisms (SNPs) and small insertions-deletions (INDELs) were assumed to be the major types of DNA polymorphisms underlying genotypic variation. However, during the last decade, copy number variation (CNV) was found to be abundant in the human genome (Iafrate et al., 2004;Sebat et al., 2004), affecting the human phenotype and often linked to diseases. By contrast, the extent to which CNVs affect genotypic variation in plants is largely unknown. Copy number variation refers to genomic rearrangements of sequences typically larger than 1 kb, resulting in the gain or loss of DNA segments (Zmienko et al., 2014). Notably, in polyploid plants copy number variation refers to the number of copies per haploid genome. CNVs mainly occur in intergenic regions but can also encompass protein-coding genes or sequences containing regulatory elements. Such CNVs changing the number of functional copies or regulatory elements can affect the expression level of genes. The effects of CNVs have remained undetected in classical QTL mapping experiments because the CNVs are generally not detectable by the commonly used marker systems (e.g., SSRs, DArTs, SNPs). A still small but growing number of reports suggest that copy number variations also contribute to the genotypic variation of important traits in plants, including flowering time (Díaz et al., 2012;Zmienko et al., 2014).
QTL for heading and flowering time in wheat have been identified in several linkage mapping and association mapping studies, mostly based on biparental collections or collections of rather diverse germplasm (Hanocq et al., 2004;Griffiths et al., 2009;Reif et al., 2011b;Rousset et al., 2011;Bennett et al., 2012;Le Gouis et al., 2012;Kamran et al., 2014). The aim of this study was therefore, to employ a candidate gene approach and genotypingby-sequencing to generate high-density marker data to dissect the genetic architecture of flowering time in European winter bread wheat cultivars. In particular, the objectives of our study were to (1) employ high-density genome-wide association mapping to identify main effect QTL for flowering time based on a population of European winter bread wheat with 410 genotypes evaluated in multi-location field trials, (2) assess the frequency of alleles at Ppd, Vrn, and Rht candidate genes as well as copy number variation at the Ppd-B1 locus and to evaluate their effects on flowering time, (3) assess the contribution of epistasis to the genetic architecture of flowering time, and (4) to draw conclusions for plant breeding.
PLANT MATERIALS, FIELD EXPERIMENTS AND METEOROLOGICAL DATA
A total of 410 winter bread wheat (T. aestivum L.) lines were used for this study. Genotypes were European varieties released during the past decades mainly in Austria, Czech Republic, Denmark, Eastern Europe, France, Germany, Poland, Russia, Turkey, and the United Kingdom. The genotypes are referred to as elite germplasm to distinguish it from genotypes not derived from breeding programs. Experiments were conducted in 2012 at three locations in partially replicated designs with a replication rate of 1.27 per location (Williams et al., 2011). Locations were Hohenheim (48 • 42 50 N, 9 • 12 58 E, 400 m above sea level (asl), growing season mean temperature 9.6 • C and mean precipitation 790 mm, soil type silty loam), Ihinger Hof (48 • 44 50 N, 8 • 55 18 E, 493 m asl, growing season mean temperature 8.7 • C and mean precipitation 923 mm, soil type silty clay) and Oberer Lindenhof (48 • 28 26 N, 9 • 18 12 E, 700 m asl, growing season mean temperature 7.4 • C and mean precipitation 1115 mm, soil type silty loam; later referred to as Lindenhof). Entries were sown in observation plots of two rows and 1.25 m length.
Meteorological data were obtained from weather stations directly located at the fields where the experiments were grown ( Figure S1A). Hourly air temperature values, measured at 2 m height, were available for Hohenheim and Ihinger Hof and temperatures at 01.00, 07.00, 13.00, and 19.00 h for the station Lindenhof. In order to demonstrate vernalization conditions for each location, cumulative vernalized day degrees were calculated for days from germination employing the method described by Weir et al. (1984) using, however, all available temperature values. Vernalization is best obtained between 3 and 10 • C while temperatures between −4 and 3 • C as well as between 10 and 17 • C result in slower vernalization (Weir et al., 1984;Eagles et al., 2010). For winter wheat, the vernalization requirement can be assumed to be fulfilled when the sum of the accumulated vernalized day degrees has reached 33 vernal days (Weir et al., 1984;Eagles et al., 2010). Days of full vernalization and sum of accumulated vernalized day degrees at January 1st for each location are given in Figure S1B.
Heading date, representing flowering time, was recorded as days after January 1st when 75% of the spikes of an observation plot had emerged to 75% from the flag leaf sheath. Thermal time, often used to more consistently describe the phenological development of plants (Eagles et al., 2010;Rousset et al., 2011;Allard et al., 2012;Cane et al., 2013), was calculated for this time period as the sum of accumulated degree days ( • Cd) which are a function of the daily mean temperature and the base temperature of 0 • C (Weir et al., 1984).
MOLECULAR DATA ANALYSIS AND CANDIDATE GENES
All lines were genotyped by genotyping-by-sequencing (GBS) at Diversity Arrays Technology (Yarralumla, Australia) using the Wheat GBS 1.0 assay. Markers with more than 25% missing values and those with a minor allele frequency smaller 0.05 were removed resulting in a total of 23,371 markers for which a map position was available and that were used for the analyses. Associations among the 410 genotypes were analyzed by applying principal coordinate analysis (Gower, 1966) based on the Rogers' distances of the individuals (Wright, 1978) and was done with the software package Plabsoft (Maurer et al., 2008).
PHENOTYPIC DATA ANALYSIS
The phenotypic data were analyzed based on the following statistical model: y ijko = μ + g i + l j + gl ij + r jk + b jko + e ijko , where y ijko was the phenotypic observation of the ith wheat line at the jth location in the oth incomplete block of the kth replication, μ was an intercept term, g i the genetic effect of the ith genotype, l j the effect of the jth location, gl ij the genotype-by-location interaction, r jk the effect of the kth replication at the jth location, b jko the effect of the oth incomplete block of the kth replication at the jth location, and e ijko was the residual. Error variances were assumed to be heterogeneous among locations. Variance components were determined by the restricted maximum likelihood (REML) method assuming a random model. Significance of variance component estimates was tested by model comparison with likelihood ratio tests. Best linear unbiased estimates (BLUEs) were estimated across locations assuming fixed effects for the genotype. Heritability (h 2 ) on an entry-mean basis was calculated as the ratio of genotypic to phenotypic variance according to Melchinger et al. (1998). All statistical analyses were performed using ASReml 3.0 (Gilmour et al., 2009).
ASSOCIATION MAPPING
For association mapping an additive genetic model was chosen and mapping was done with a mixed model incorporating a kinship matrix as described previously (Yu et al., 2005;Würschum and Kraft, 2014). In brief, the model was: y ijp = μ + a p + g i + l j + e ijp , where y ijp is the adjusted entry mean of the ith wheat line at the jth location carrying allele p, μ the intercept term, a p the allele substitution effect of allele p, g i the genetic effect of the ith wheat line, l j the effect of the jth location, and e ijp the residual including the genotype-by-location interaction effect. The allele substitution effect a p was modeled as fixed effect whereas g i and l j were regarded as random effects. The variance of the random genetic effect was assumed to be Var(g) = 2Kσ 2 G , where σ 2 G refers to the genetic variance estimated by REML and K was a 410 × 410 matrix of kinship coefficients that define the degree of genetic covariance between all pairs of entries. We followed the suggestion of Bernardo (1993) and calculated the kinship coefficient K ij between inbreds i and j on the basis of marker data as described by Würschum et al. (2011Würschum et al. ( , 2012. For the detection of main effect QTL, a genome-wide scan for marker-trait associations was conducted. To control for multiple testing, we followed the suggestion of Kraakman et al. (2004) and tested at a false discovery rate (FDR) of 0.20 (Benjamini and Hochberg, 1995).
The two-dimensional epistasis scan was done based on 2594 equally spaced markers by extending the above model to markermarker interactions including the subordinated main effects. For the significance level for the epistatic QTL we used an α-level of 0.01 and followed the suggestion of Holland et al. (2002) dividing the α-level by the number of possible independent pairwise interactions between chromosome regions, assuming two separate regions per chromosome (P < 1.2e-5). The circular plots illustrating the epistatic interactions were created with Circos (Krzywinski et al., 2009). The total proportion of genotypic variance (p G ) explained by the detected QTL was calculated by fitting all QTL and the segregating candidate genes simultaneously in a linear model to obtain the adjusted R 2 R 2 adj which corrects for the number of parameters in the linear model. The ratio p G = R 2 adj / h 2 , where h 2 refers to the heritability of the trait, yielded the proportion of genotypic variance (Utz et al., 2000). The p G values of individual QTL were accordingly derived from the sums of squares of the QTL (SS QTL ) in this linear model.
RESULTS
The 410 winter bread wheat genotypes were evaluated at three locations where the vernalization saturation for winter wheat was reached after 38, 39, and 44 days from germination ( Figures S1A,B). The sum of accumulated vernalized day degrees at January 1 st was 53.5, 47.0, and 53.9 ( Figure S1B). We recorded heading date as days after January 1st and also calculated thermal time to heading by taking into account the temperature per location. The genotypic variance as well as the genotype-by-location interaction variance were significantly larger than zero (P < 0.01) for both traits. The heritability estimates were high with 0.93 for heading date and 0.94 for thermal time to heading ( Table 1). Heading date BLUEs across locations showed a wide range of 27 days between the earliest and latest variety. The BLUEs per location revealed a strong effect of the location on heading date with Hohenheim being the earliest location and Lindenhof the latest (Figure 1). While the temperature profiles at the three locations ran in parallel, they were highest for Hohenheim and lowest for Lindenhof ( Figure S1A). The BLUEs per location for thermal time to heading, i.e., taking the different temperatures at each location into account, largely eliminated the differences between the three locations (Figure 1).
The candidate gene approach revealed no differences between the 410 genotypes for the Ppd-A1, Ppd-B1, Vrn-A1, Vrn-B1, Vrn-B3, and Vrn-D1 genes. All varieties carried the alleles Ppd-A1b.1, Ppd-B1b.1, Vrn-A1c, vrn-B1, vrn-B3, and vrn-D1. By contrast, locus was also present in this panel of cultivars and the majority of the genotypes had one copy (382), 21 had two copies and 5 had three copies of Ppd-B1. Consistent with Cane et al. (2013) we also observed one genotype ('Naridana') for which the Ppd-B1 gene appears to be absent. Genotypes carrying the photoperiod insensitive Ppd-D1a allele were substantially earlier flowering than the genotypes with the photoperiod sensitive Ppd-D1b allele (Figure 2A). Ppd-D1 consequently also explained 58.2% of the genotypic variance for thermal time to heading ( Table 2). For Ppd-B1 copy number variation, the genotypes carrying two copies of the gene flowered earlier that the ones with only one copy while the few genotypes with three copies appear to be even earlier ( Figure 2B). Increasing copy number at Ppd-B1 resulted in earlier flowering in a Ppd-D1b background while it may have no or only a small effect in plants homozygous for Ppd-D1a. While there was no difference in thermal time to heading for the two alleles at the Rht-D1 locus, we found that the plants carrying the semi-dwarfing Rht-B1b allele flowered earlier than the tall Rht-B1a plants (Figure 2A). This however, was due to different frequencies of Ppd-D1a alleles in both groups. Within plants carrying the Rht-B1a allele only 12.5% carried the photoperiod insensitive Ppd-D1a allele whereas within the plants carrying the Rht-B1b allele 55.3% carried Ppd-D1a. After a grouping of the plants according to their allele status at Ppd-D1, no differences in thermal time to heading were observed any more between the two Rht-B1 alleles. Consistently, both Rht-B1 and Rht-D1 did not explain any genotypic variation of thermal time to heading ( Table 2). By contrast, the height-reducing allele at the Rht8 locus (192bp allele) reduced thermal time to heading and may even have a small effect in a Ppd-D1a background ( Figure S4). However, overall the effect of Rht8 was found to be small as it explained only 1.5% of the genotypic variance ( Table 2).
The 410 individuals of the mapping population were genotyped by a genotyping-by-sequencing approach which after quality checks yielded 23,371 polymorphic markers with known map position that were used for further analyses. In accordance with previous studies using elite European winter bread wheat (Reif et al., 2011a;Würschum et al., 2013b), the principal coordinate analysis revealed only a slight population structure and the first two principal coordinates explained 19.2 and 8.2% of the variance, respectively ( Figure S2). The genomewide scan for marker-trait associations identified 5 significantly associated markers on chromosomes 5B, 1D, and 6D (Figure 3) which explained between 0.1 and 2.9% of the genotypic variance ( Table 2). In addition, we performed a genome-wide scan with Ppd-D1 as cofactor in the model which revealed another QTL on chromosome 4B explaining 1.5% of the genotypic variance. Together, the five candidate genes and the six identified QTL explained 72.2% of the genotypic variance. The epistasis scan identified 30 significant epistatic interactions (Figure 4) which explained between 0.1 and 2.2% of the genotypic variance (Table S1).
For 379 varieties the country of origin was known and we used this information to analyze the frequency of Ppd-D1, Ppd-B1 CNV, and QTL alleles dependent on the geographic origin ( Figure 5). For Ppd-D1 the photoperiod insensitive allele Ppd-D1a was rare in the UK, Denmark, Germany, Poland, the Czech Republic and in Austria. By contrast, about a third of the French lines carry this allele and in Eastern Europe and Russia it is the predominant allele. For Ppd-B1 copy number, the one-copy allele is the prevalent allele in all regions and the two-copy allele was mainly found in varieties from the more southern countries. In addition, we analyzed the frequencies of the alleles causing earlier flowering for the detected QTL in the same geographic regions and observed a similar picture. With the exception of one QTL (CloneID 1089381), the allele causing earlier flowering is the minor allele in the first group of countries and occurs at a higher frequency in Eastern Europe and Russia.
PHENOTYPIC EVALUATION OF FLOWERING TIME IN EUROPEAN WINTER BREAD WHEAT CULTIVARS
Flowering time is of importance in plant breeding as it is central for the adaptation of wheat to different climatic regions and consequently also affects yield potential. The observed genotypic variance of heading date and thermal time to heading was several times larger than the genotype-by-location interaction variance which is in accordance with previous studies in European elite wheat germplasm (Reif et al., 2011b;Langer et al., 2014). All three test locations were located in the south of Germany but we observed strong differences between them with regard to flowering time of the wheat varieties (Figure 1). As vernalization requirement was fulfilled ( Figure S1B) and photoperiod was similar for all three locations, temperature remained as a likely cause for the observed differences between the locations which was confirmed by the calculation of thermal time to heading. Taking the temperature at the locations into account eliminated the differences between them observed for heading date. This illustrates the strong effect of temperature on flowering of wheat. However, the low genotype-by-location interaction variance as compared to the genotypic variance indicates that all genotypes responded similarly to the different temperature regimes. Consequently, there appear to be no major QTL for flowering time in response to temperature segregating in European winter bread wheat that affected the trait under the growth conditions represented by our three locations. Nevertheless, a different set of test locations with more diverse temperature profiles may reveal such QTL and the observed genotype-by-location interaction variance may well be caused by medium or small effect temperature response QTL. The large range in heading date of 28 days (Table 1) can be explained by the different European origins of the varieties included in this study as varieties from southern European countries tended to flower earlier than the more northerly originating ones ( Figure S3). Especially in view of the expected climate change, plant breeders need germplasm which allows a flexible response to different climatic conditions. Responding to early summer drought and heat stress, early flowering genotypes can be advantageous provided late frost can be avoided. As earliness is often associated with reduced height and potentially reduced resource capture (Addisu et al., 2009;Bentley et al., 2013) could be a trade-off regarding yield exploitation. The challenge for future wheat breeding is therefore, to modify flowering time to suit local climatic conditions while maintaining or even increasing yield potential. In order to efficiently exploit the variation and to transfer genotypes between regions, e.g., from southern France to Germany or vice versa, selection tools like marker-assisted selection may be advantageous for future wheat breeding. We, therefore, investigated the genetic control underlying variation of flowering time in the panel of European winter bread wheat varieties by a candidate gene approach and by genome-wide association mapping.
CANDIDATE GENES IN EUROPEAN WINTER BREAD WHEAT CULTIVARS AND THEIR EFFECT ON THERMAL TIME TO HEADING
A number of genes involved in the different pathways affecting flowering time were identified so far in wheat and their effects were studied by candidate gene approaches (Eagles et al., 2009(Eagles et al., , 2010Rousset et al., 2011;Bentley et al., 2013). In this study we assessed the frequency of some of the known flowering time genes and their contribution to the genetic architecture of thermal time to heading in the panel of European winter bread wheat varieties. The common spring allele Vrn-A1a (Yan et al., 2004) was not detected and the whole panel did not segregate for either of the tested vernalization loci (vrn-B1, vrn-B3 and vrn-D1). This strict winter habit and thus a consequent vernalization requirement is not surprising, as only winter wheat varieties were included in this study. Genes inducing reduced plant height, predominantly the gibberellic acid insensitive alleles of Rht-B1 and Rht-D1, were important components of the "green revolution" (Hedden, 2003). Plant height and its possible effect on flowering time is often discussed as reported associations between flowering time related traits and plant height vary strongly. Langer et al. (2014) observed no correlation in European elite germplasm and neither did Cane et al. (2013) for Australian wheat while other studies did find significant correlations between the two traits (Bordes et al., 2008;Longin et al., 2013;Wilhelm et al., 2013). Furthermore, Wilhelm et al. (2013) reported a significant effect of Rht-B1 on days to heading. In our study we also observed a significant correlation between plant height and heading time as the short genotypes tended to flower earlier than the tall ones. This effect was mainly due to the Rht-B1b allele (Figure 3) and likewise Rht8 was found to explain a small proportion of the genotypic variance of thermal time to heading ( Table 2). This suggested a significant effect of Rht-B1 on heading date in wheat. However, a more detailed analysis revealed that the differences in thermal time to heading between plants carrying either of the two Rht-B1 alleles was due to different frequencies of the photoperiod insensitive Ppd-D1a allele in these two groups. Within either of the two Ppd-D1 alleles, Rht-B1 had no effect on thermal time to heading. Consistently, both Rht-B1 and Rht-D1 did not contribute to the genotypic variance when the effect of Ppd-D1 was taken into account ( Table 2). This indicates that the observed correlations between flowering time and plant height are due to different frequencies of Ppd-D1a in genotypes with different Rht alleles but not to a direct effect of the Rht loci on flowering time.
Photoperiodism in wheat is mainly controlled by the Ppd-D1 locus located on chromosome 2D which greatly influences flowering time by the separation of genotypes into photoperiod sensitive and insensitive ones and accelerates flowering by www.frontiersin.org October 2014 | Volume 5 | Article 537 | 7 several days in European environments (Worland, 1996;Beales et al., 2007;Kamran et al., 2014). The major photoperiod insensitive Ppd-D1a allele has a 2089bp deletion upstream the coding region which increases expression of the gene and is associated with upregulation of the floral activator TaFT1 (Beales et al., 2007;Shaw et al., 2012;Bentley et al., 2013). The photoperiod insensitive Ppd-D1a allele was introduced into European material in the early Twentieth century when Italian breeders used Japanese germplasm as source (Worland, 1996). Nishida et al. (2013) recently identified alleles of the Ppd-A1 and Ppd-B1 loci which accelerate heading in Japanese germplasm and Bentley et al. (2013) showed that photoperiod insensitive alleles at Ppd-A1 or Ppd-B1 can have effects as strong as that of Ppd-D1 when introgressed into hexaploid wheat. However, unlike Ppd-D1a neither of these alleles appears to be present in the current European winter wheat material. By contrast, Ppd-D1a is present in European winter wheat varieties and genotypes carrying this allele showed a strongly decreased thermal time to heading (Figure 2A). Consistently, this locus also explained the by far largest proportion of genotypic variance ( Table 2). This illustrates that not only in biparental populations but also in our panel of European winter wheat varieties Ppd-D1 is the major determinant of flowering time under fully vernalized conditions. The Ppd-D1a allele is mainly present in Eastern and Southern European and Eurasian varieties (Figure 5), as up to the beginning of the twenty-first century photoperiod insensitivity has been introduced into most wheat cultivars grown below 48 • latitude (Rajaram and van Ginkel, 2001). The main reason for this is that photoperiod insensitivity is beneficial for crops grown in regions with high summer temperatures in order to avoid heat or drought stress during premature developmental stages (Bentley et al., 2011). In addition, earlier flowering increases yield especially in Southern Europe. The prevalence within Eastern European and Russian varieties may be due to the incorporation of Italian material as reduced photoperiodic sensitivity is advantageous for winter wheat plants adapted to dry land environments, prevailing in the continental climate regions of Russia and the former Soviet Union (Worland et al., 1994;Litvinenko et al., 2001). By contrast, the low proportion of Ppd-D1a alleles in varieties originating from northern parts of Europe is due to the considerably lower yield potential of photoperiod insensitive as compared to sensitive varieties in these regions, owing to the shortened vegetative phase. One of the prime examples for copy number variation in plants is flowering time in wheat which has recently been shown to be affected by CNVs (Díaz et al., 2012). Díaz et al. (2012) showed that Ppd1 on the B genome (Ppd-B1) can be present in different copy numbers. Wheat genotypes with only one copy are photoperiod sensitive whereas an increased copy number (2-4 copies) results in a day-neutral, early flowering phenotype. These experiments were conducted on a rather limited number of lines and with phenotypic data from controlled greenhouse conditions. Cane et al. (2013) found all four alleles to be present in southern Australian wheat and reported that the three-copy allele (termed Ppd-B1a) and the four-copy allele (Ppd-B1c) reduced days to heading as compared to the one-copy allele (Ppd-B1b) whereas the two-copy allele (Ppd-B1d) increased days to heading. In our collection of European winter wheats, the majority of the cultivars carried the one-copy allele and only few the two-or three copy alleles. The allele with four copies which is characteristic for Chinese spring was not present in our collection, in contrast to Cane et al. (2013) who found this allele in a number of modern Australian cultivars and current breeding lines. In contrast to the findings of Cane et al. (2013), our results indicate that both the two-and the three-copy alleles reduce the time to heading in our panel and under the growth conditions present at our test locations (Figure 2B). The proportion of genotypic variance explained by Ppd-B1 copy number variation under field conditions was small compared with Ppd-D1, however still larger than that of any of the detected QTL ( Table 2).
Taken together, our candidate gene analyses revealed that the Reduced height loci appear to have no effect on flowering time in European winter bread wheat varieties and confirmed Photoperiod-D1 as the major source of variation of flowering time. In addition to the major effect of Ppd-D1, copy number variation at the Ppd-B1 locus may add in the fine-tuning of adaptation of European winter wheat to local climatic conditions.
QTL AND EPISTATIC QTL FOR THERMAL TIME TO HEADING
Even though Ppd-D1 explained 58% of the genotypic variance, there is still variation for flowering time in European winter bread wheat not explained by this major regulator. In order to identify additional components of the genetic architecture underlying the trait we performed a genome-wide association mapping in our panel of varieties. This identified six small effect QTL (Figure 3, Table 2) indicating that beside Ppd-D1 there are no other major QTL affecting flowering time in European winter bread wheat. Rather, the fine tuning to local climatic conditions appears to be controlled by many small effect QTL. With regard to a knowledgebased breeding, such small effect QTL that escape detection in QTL mapping approaches could probably be better captured by genome-wide prediction approaches which may be an option for the future and warrants further research. While it is not possible to unambiguously assign the detected QTL to one of the flowering pathways, the QTL on chromosome 1D (CloneID 1218093) might be part of the Eps signaling pathway as Zikhali et al. (2014) recently reported an Eps QTL on this chromosome.
The geographic pattern of distribution of alleles of the six QTL resembled the distribution of the two Ppd-D1 alleles. The allele conferring earlier flowering was more prominent in the varieties originating from southern regions than in the northern regions suggesting that these QTL contribute to the adaptation of winter wheat varieties to different European climatic conditions. The total proportion of genotypic variance explained by Ppd-D1 and the detected QTL was 72.2% suggesting that there are other QTL with effects too small to be detected or with medium or large effects which remained undetected. Surprisingly, we did not identify a QTL on chromosome 2D in our genome-wide scan, despite Ppd-D1 being located on this chromosome. We did, however, identify an unmapped marker (CloneID 3940868) that of all tested markers was most strongly associated with the trait and which is located in the promotor region of Ppd-D1. The linkage disequilibrium (LD) between this marker and any of the mapped markers on chromosome 2D was low suggesting a rapid decay of LD around this locus in this germplasm. Bentley et al. (2013) have recently attempted to develop markers flanking Ppd but found none of them to be polymorphic on their lines suggesting that the region around Ppd may be largely monomorphic. Our result illustrates that despite the comparably high number of markers obtained by the genotyping-by-sequencing approach there are still chromosomal regions that are not covered with this marker density and consequently regions where QTL may have remained undetected.
Another potential source for the variance not accounted for by Ppd-D1 and the detected QTL is epistasis. Epistasis refers to interactions between two or more loci in the genome (Carlborg and Haley, 2004) and has recently been shown to contribute to the genetic architecture of complex traits in different crops including maize, wheat and rapeseed (Buckler et al., 2009;Reif et al., 2011b;Liu et al., 2012a;Steinhoff et al., 2012;Würschum et al., 2013a). Reif et al. (2011b) showed the contribution of epistasis to the genetic architecture of flowering time in elite winter wheat and in addition, Bentley et al. (2013) have recently shown a dependency of the effect of Ppd-D1a alleles on the genetic background. Consistent with these previous findings, we also identified epistatic QTL for thermal time to heading. While the proportion of genotypic variance explained by individual epistatic QTL was small, combined they could contribute a substantial proportion to the observed variation for flowering time.
In conclusion, the genome-wide scan identified only six QTL suggesting that in addition to the major regulator Ppd-D1 the genetic architecture of flowering time in European winter bread wheat is controlled by many QTL with only small effects and potentially by epistasis.
CONCLUSIONS
In this study we employed a large panel of European winter bread wheat varieties to unravel the genetic architecture underlying flowering time in this germplasm set. Using a candidate gene approach and genome-wide association mapping, we show that in fully vernalized winter wheat more than half of the genotypic variation is attributable to the major photoperiod regulator Ppd-D1. The remaining variation appears to be due to copy number variation at the Ppd-B1 locus, other small effect QTL and epistatic QTL. With regard to a knowledge-based breeding of wheat these results suggest that only Ppd-D1 is worth to be included in marker-assisted selection programs whereas the subsequent fine tuning to local conditions is better done based on phenotypic selection in the field.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2016-06-17T23:43:56.281Z", "created": "2014-10-09T00:00:00.000", "id": "5844406", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.frontiersin.org/articles/10.3389/fpls.2014.00537/pdf", "pdf_hash": "30a455f1c72ac82698268cbd631a332442e5b3c2", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:8292", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "30a455f1c72ac82698268cbd631a332442e5b3c2", "year": 2014}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Allelic Analysis of Sheath Blight Resistance with Association Mapping in Rice
Sheath blight (ShB) caused by the soil-borne pathogen Rhizoctonia solani is one of the most devastating diseases in rice world-wide. Global attention has focused on examining individual mapping populations for quantitative trait loci (QTLs) for ShB resistance, but to date no study has taken advantage of association mapping to examine hundreds of lines for potentially novel QTLs. Our objective was to identify ShB QTLs via association mapping in rice using 217 sub-core entries from the USDA rice core collection, which were phenotyped with a micro-chamber screening method and genotyped with 155 genome-wide markers. Structure analysis divided the mapping panel into five groups, and model comparison revealed that PCA5 with genomic control was the best model for association mapping of ShB. Ten marker loci on seven chromosomes were significantly associated with response to the ShB pathogen. Among multiple alleles in each identified loci, the allele contributing the greatest effect to ShB resistance was named the putative resistant allele. Among 217 entries, entry GSOR 310389 contained the most putative resistant alleles, eight out of ten. The number of putative resistant alleles presented in an entry was highly and significantly correlated with the decrease of ShB rating (r = −0.535) or the increase of ShB resistance. Majority of the resistant entries that contained a large number of the putative resistant alleles belonged to indica, which is consistent with a general observation that most ShB resistant accessions are of indica origin. These findings demonstrate the potential to improve breeding efficiency by using marker-assisted selection to pyramid putative resistant alleles from various loci in a cultivar for enhanced ShB resistance in rice.
Introduction
Rice (Oryza sativa L.) feeds more than half of the world's population [1] and genetic improvement of this food crop can serve as a major component of sustainable food production. Rice sheath blight (ShB), caused by the soil-borne fungal pathogen Rhizoctonia solani Kühn, is a major disease of rice that greatly reduces yield and grain quality worldwide [2]. Due to the high cost of cultural practices and the phytotoxic influence associated with the application of fungicides, the use of ShB resistant cultivars is considered the most economical and environmentally sound strategy in managing this disease. Understandings of genetic control will facilitate cultivar improvement for this disease and secure global food production.
The necrotrophic ShB pathogen has a broad host range and no complete resistance has been identified in either commercial rice cultivars or wild related species [3,4]. However, substantial differences in susceptibility to ShB among rice cultivars have been observed under field conditions [5,6]. Differential levels of resistance and the associated resistance genes have been studied among rice germplasm accessions [7]. Rice ShB resistance is believed to be controlled by multiple genes or quantitative trait loci (QTLs) [8]. Since Li et al. [9] first identified ShB QTLs using restricted fragment length polymorphism (RFLP) markers under field conditions, over 30 resistant ShB QTLs have been reported using various mapping populations, such as F 2 s [10][11][12][13][14], double haploid (DH) lines [15], recombinant inbred lines (RILs) [8,[16][17][18], near-isogenic introgression lines (NIL) [19] and backcross populations [20][21][22][23]. 'Teqing' and 'Jasmine 85' have been repeatedly involved in these studies as the ShB resistant parents. We are the first to map rice ShB QTLs using association mapping strategy in a global germplasm collection.
In association mapping, each identified marker usually has multiple alleles in the mapping panel and each allele in a marker locus contributes differently to the associated trait. Agrama and Yan [42] reported that three alleles at each of three associated loci (allele 87 of RM490, 105 of RM413 and 122 of RM277) and two alleles at another locus (182 and 183 of RM263) had significantly greater contribution to straighthead resistance than other counterparts. Li et al. [47] determined that allele 126 bp had the greatest effect on increasing grain yield, plant weight and grains/panicle branch among eight alleles of RM471. There is no study on allelic distribution for associated loci in a global rice germplasm collection.
Linkage disequilibrium (LD), defined as the non-random association of alleles at separate loci located on the same chromosome [24], is a prerequisite for association mapping. The distance at which LD declines with genetic or physical distance determines the marker density needed for achieving a reasonable mapping resolution. The extent of LD may vary among different genomic regions [48]. Numerous studies on global germplasm collections indicate 25 cM as a reasonable resolution for association mapping in rice [41,47,49]. In our study, we used 154 simple sequence repeat (SSR) markers plus an indel to provide coverage of 10 cM across the rice genome for sufficient mapping resolution.
Accurate phenotyping is essential for mapping, especially when the target trait is controlled by multiple genes or QTLs such as ShB resistance. All the previous studies phenotyped ShB resistance under field conditions with only one exception, Liu et al. (2009) [18], where a micro-chamber method (MCM) was adapted. The MCM described by Jia et al. (2007) [6] has proven to effectively minimize the confounding effects of environmental and morphological factors, thus generating more reliable data. Furthermore, numeric measurement of ShB in the MCM should be more accurate than the traditional visual scoring under field conditions. Because of these advantages, the MCM has been widely applied in studies of ShB resistance [18,50,51].
Using association mapping, our objectives were to 1) map QTLs associated with ShB resistance phenotyped with the MCM, 2) identify putative resistant alleles in a global germplasm collection, and 3) explore the use of ShB putative resistant alleles in a breeding program.
Variation of ShB severity ratings
The 217 sub-core entries in the mapping panel originated from fifteen geographic regions including 77 countries worldwide. India had the most entries (6.5%), followed by China (5.5%), Indonesia (4.1%), Japan (4.1%) and Taiwan (4.1%). Their name, origin, ShB severity rating, structure group and entry number in the Genetic Stocks Oryza (GSOR) collection ( [URL]/ Main/docs.htm?docid = 8318) are presented in (Table S1). The ShB severity ratings among the 217 entries were distributed normally, ranging from 0.25660.111 to 0.90960.096 with an average of 0.52160.008 (Fig. 1). The resistant check Jasmine 85 was rated 0.47260.021 and susceptible check Lemont was rated 0.94660.080. Twenty-four entries (11.1%) were significantly more resistant to ShB than Jasmine 85 at the 5% level of probability while 54 others (24.9%) had similar resistance. Population structure Structure analysis from Q1 to Q10 across twenty runs for the 217 sub-core entries genotyped with 155 genome-wide DNA markers using STRUCTURE demonstrated that when Q reached five, the Pr(Q) became more-or-less plateaued, so Q5 captured the major structure in our data. Thus, the mapping panel was divided into five subgroups and each entry was classified to an appropriate subgroup using STRUCTURE. Inferred by reference cultivars recommended by Agrama et al. [52,53], the five subgroups were denoted as temperate japonica (TEJ), aus (AUS), aromatic (ARO), indica (IND), and tropical japonica (TRJ) ( Fig. 2A). A similar structural pattern was seen with the PCA analysis with the first two axes explaining 75.03% of variation (Fig. 2B). Furthermore, the genetic distance based on cluster analysis also divided the mapping panel into five major clusters (Fig. 2C). All three approaches led to the same conclusion: a five-group structure could clearly and sufficiently explain the existing genetic diversity in the mapping panel. In the mapping panel, IND had the most entries (86), followed by TRJ (49), AUS (39), TEJ (36), and ARO (7). Among 24 entries having greater resistance to ShB than the resistant check, Jasmine 85, 20 belonged to IND, two to AUS and one each to TRJ and admix (TRJ-AUS-IND).
Determination of the best fit model
From dimension 1 to 10 in the PCA and structure Q1 to Q10, PCA5 had the smallest Bayesian Information Criterion (BIC) value, indicating that PCA5 should be the best fit model to map ShB QTLs (Table 1). Hence, we tested each of 155 molecular markers for association with ShB resistance using PCA5, and plotted the observed versus expected -Log10(P) before and after correction using the genomic control (GC). The plots of the PCA5+GC were distributed more uniformly and was much closer to the expected -Log10(P) than PCA5 alone (Fig. 3). In other words, the PCA5+GC model showed better control for Type I errors. Therefore, the GC approach was applied to correct the biased estimation. The P values generated from the PCA5 model after the GC correction were used to present the significance level of each marker.
Marker loci and their alleles associated with sheath blight
Ten marker loci were identified to be significantly associated with ShB resistance at the probability level of 5% or lower, three on chromosome (Chr) 11, two on Chr1, and one each on Chr2, 4, 5, 6 and 8 ( Table 2, Fig. 4). RM237 on Chr1 at 27.1 Mb had the highest significance rating for ShB at the 0.002 level of probability. RM11229 on the long arm of Chr1 explained the most phenotypic variation (9.5%) with significance at the 0.044 level of probability. RM11229 and 1233 each had six alleles, the most among the 217 sub-core entries, followed by RM341 and 254 (five alleles), RM237, 8217,146 and 408 (four), RM133 (three) and RM7203 (two) ( Table 2).
Among the six alleles of RM11229, allele 158 was present in 18 entries that had the lowest average ShB rating (0.414), and thus, it was designated as the 'putative resistant allele' of this marker locus. Accordingly, ten alleles, one each from the ten associated marker loci, were noted as the putative resistant allele in Table 2 because they had the greatest effect to decrease ShB among all the alleles for their respective loci ( Table 2). ShB rating was the smallest for putative resistant allele 158 of RM11229 among the ten putative resistant alleles. Of the other five putative resistant alleles, 139 of RM341 (present in 17 entries), 340 of RM146 (28 entries), 88 of RM7203 (120 entries), 169 of RM254 (12 entries) and 177 of RM1233 (35 entries), had lower ShB means ranging 0.447-0.470 than the resistant check Jasmine 85 (0.472), suggesting a stronger effect for resistance to ShB than Jasmine 85. The remaining four putative resistant alleles had similar ShB ratings with Jasmine 85, suggesting a similar effect for the level of ShB control.
Number of putative resistant alleles and sheath blight resistance
As the number of putative resistant alleles in the germplasm increased, so did germplasm resistance to ShB (Table S1). GSOR 310389 from Korea contained the most putative resistant alleles, eight out of ten, and had a ShB rating of 0.351 which was significantly more resistant than the resistant check Jasmine 85 which contained three putative resistant alleles and had a ShB rating of 0.472. Among seven entries containing six putative resistant alleles with a mean of 0.386 ShB, GSOR 310475 and 311475 were more resistant than Jasmine 85 and had ShB ratings of 0.324 and 0.336, respectively. Among 28 entries having five putative resistant alleles with a mean ShB rating of 0.444, seven were significantly more resistant than Jasmine 85. Seven, out of 35 entries which carried four putative resistant alleles and had a mean ShB 0.466, were identified to be significantly more resistant than Jasmine 85. The mean ShB ratings for entries containing three, two, one and zero putative resistant alleles were 0.483, 0.535, Table 1. Comparative analysis of different subgroups using structure (Q model) and different dimensions in principal components analysis (PCA) for association mapping of sheath blight resistance using 217 entries genotyped with 155 molecular markers. 0.582 and 0.598, respectively. There was a strong and negative correlation between the ShB severity rating and number of putative resistant alleles (r = 20.535, p,0.0001).
Our mapping results showed that most entries containing a large number of putative resistant alleles were IND (Fig. 5 and Table S1). All entries with six or more putative resistant alleles were IND with only one exception of AUS.
Discussion
Pyramiding putative resistant alleles for cultivar improvement R. solani is a soil-borne necrotrophic fungus and its group AG1-IA has a broad host range including rice, maize, wheat, sorghum, bean (Phaseolus spp.) and soybean [Glycine max (L.) Merr.] [54]. The pathogen's ability to persist in soil and on crop residues allows it to, survive in multiple ways and makes disease management difficult. There is no complete resistance to ShB in rice because the resistance is quantitatively controlled by numerous genes or quantitative trait loci (QTLs). More than 30 QTLs responsible for ShB have been reported in rice [8,[10][11][12][13][14][15][16][17][18][19][20][21][22][23]. However, all of these studies have been limited to conventional mapping populations from a small number of parents, which limits the alleles in the progeny to those present in the parental lines.
From our diverse mapping panel including 217 sub-core entries, we identified ten marker loci significantly associated with ShB resistance, each locus had numerous alleles, each allele contributed differently to ShB resistance, and the putative resistant allele of each locus contributed the most (Table 2). Highly significant correlation demonstrated that as more putative resistant alleles pyramided in a germplasm entry, the entry had greater resistance to ShB. Among 24 entries that were significantly more resistant than the resistant check Jasmine 85 that had three putative resistant alleles, 17 (71%) contained four or more putative resistant alleles. Among 54 entries that had similar resistance with Jasmine 85, 29 (54%) possessed three or more putative resistant alleles.
These findings suggest that marker-assisted breeding for ShB resistance can be conducted on an allelic level by pyramiding putative resistant alleles in a cultivar. This can be accomplished in a two pronged approach by combining parental lines based upon their combination of number of different ShB resistant loci and possessing the putative resistant alleles at these loci. For example, resistant entry GSOR 310389 that has eight putative resistant alleles could be crossed with a commercial cultivar having fewer putative resistant alleles, and the progeny would be selected based on those having the most highly putative resistant alleles at the most loci. The breeding program would end up with the selection of progeny containing the most putative resistant alleles, potentially having greater resistance to ShB than either parental line.
Pyramiding responsible genes has been successfully applied in rice breeding for disease resistance including bacterial blight (Xanthomonas oryzae pv.) and blast (Magnaporthe oryzae). [57] combined three major genes, Pi-d, Pi-z and Pi-k h , and bred blast resistant cultivar Jefferson. All the successful applications of gene pyramiding have been at the gene loci level. Our findings in this study will help enhance the application to allelic level in crop breeding. The allelic application will improve breeding efficiency, increase cultivar resistance to sheath blight in rice and ultimately secure food production worldwide.
Allelic analysis can only be applied in association mapping where large number of diversified genotypes are used and multiple alleles are involved at each associated marker locus in the mapping panel. Using this method, germplasm accessions that are identified in the association mapping strategy to possess multiple putative resistant alleles can be crossed with other accessions that have a different complement of putative resistant alleles. The selection of progeny possessing the most putative resistant alleles should be more effective than it is for resistant loci. In this regard, association mapping offers advantages for identifying parental material and specific alleles that can enhance breeding.
Putative resistant alleles and ancestry background for sheath blight
Jia et al. [58] reported 52 entries that are significantly more resistant to ShB than Jasmine 85. The resistant entries were identified from 1,794 entries of the USDA rice core collection that has 35% indica, 27% temperate japonica, 24% tropical japonica, 10% aus and 4% aromatic genotypes [52]. Based on the ancestry classification, there are 621 indica entries in the core and 45 of them are included in the resistant list, making a resistance frequency of 7.2% for indica germplasm. Accordingly, the resistance frequency is 2.8% for aromatic, 1.7% for aus, and 0.2% each for temperate japonica and tropical japonica. In a study conducted by Zuo et al. [23], japonica cultivars showed higher sheath blight severity than indica cultivars. They describe a general observation that japonica rice is more susceptible than indica rice. Furthermore, Jasmine 85, Tetep and Teqing, used as parents in many studies on mapping ShB resistance, all belong to indica.
This study demonstrated that: 1) a majority of the ShB putative resistant alleles existed in indica germplasm, 2) most of the resistant entries with a large number of putative resistant alleles were indica, conversely 3) only a very small portion of putative resistant alleles existed in japonica, and 4) the most susceptible entries with very few or no putative resistant alleles were japonica (Fig. 5 and Table S1). Entry GSOR 310389 is an example which had eight out of ten putative resistant alleles, showed a high level of resistance to ShB, and is indica. The results from association mapping match well with the phenotypic observation that most resistant genotypes are indica and resistant germplasm is rare in japonica.
ShB associated markers and QTL identification
Our genome-wide search found ten marker loci that were significantly associated with sheath blight resistance (Fig. 4). Both RM11229 (Chr1 at 22.7 Mb) and RM7203 (Chr11 at 1.1 Mb) are novel QTLs that have not been previously reported. RM11229 is approximately 5.0 Mb away from a ShB QTL reported by Channamallikarjuna et al. [16] and RM7203 resides about 3.3 Mb away from one QTL identified by Li et al. [9]. The remaining eight QTLs identified in this study were either quite near (less than 1.4 Mb distant) or within the interval of previously identified QTLs. The ten associated markers identified in this study were located on seven chromosomes (Chr1, 2, 4, 5, 6, 8 and 11) (Fig. 4).
On Chr1 our study identified RM11229 and RM 237, which occur within 4.4 Mb of each other, as markers associated with ShB resistance. RM11229 explained the most phenotypic variation (9.5%) and its putative resistant allele 158 bp had the smallest average ShB score (0.414 in Table 2), indicating the greatest resistance among the ten putative resistant alleles. RM237 was the most significant marker for ShB resistance (p = 0.002). Therefore, the 4.4 Mb gap between RM11229 and RM237 on Chr1 should be a target area for fine-mapping ShB resistant genes in rice. RM237 at 26.8 Mb is near the ShB QTL region spanning 27.6 to 34 Mb found by Charnnamallikarjuna et al. [16].
On Chr4, three reports uniformly indicated ShB QTLs on the long arm at 29.8 to 33.6 Mb. This is a very narrow region of 3.8 Mb on the physical map, and corresponds to a small estimated cM distance in the mapping population developed from susceptible Lemont and resistant Teqing parents [8,9,59]. Our identified marker RM8217 at 32.6 Mb was within the qSB-4-2 (30.6,33.6 Mb) by Pinson et al. [8] and Qsbr4a (31.7,33.6 Mb) by Li et al. [9], and in close proximity (within 0.6 Mb) to QRlh4 (29.8,32.0 Mb) by Xie et al. [59]. This small area confirmed by multiple studies strongly suggests a reliable location harboring ShB QTL.
On Chr5, 6 and 8, we identified one ShB associated marker from each chromosome. RM146 on Chr5 at 18.1 Mb is close to the Rsb 1 reported by Che et al. [10] between RM164 320 at 19.1 Mb and RM39 300 at 20.7 Mb near the centromere. RM133 (Chr6 at 0.2 Mb) and RM408 (Chr8 at 0.1 Mb) were close to the qShB6 (0.5,1.8 Mb) described by Liu et al. [18] and the QSbr8a (1.5,2.1 Mp) by Li et al. [9], respectively. The putative resistant alleles of both RM133 (allele 230 bp) and RM408 (allele 119 bp) were common with more than 45% of entries among the 217 subcore entries in the mapping panel. RM7203 on the short arm of Chr11 at 1.1 Mb was a novel ShB QTL that has not been reported. The putative resistant allele 88 of RM7203 existed in 55% of the 217 entries, so was the most common allele and had a ShB mean (0.470) similar to Jasmine 85 (0.472).
At the bottom of Chr11, two markers RM254 (at 23.7 Mb) and RM1233 (at 26.5 Mb) were identified, explaining relatively high phenotypic variation among ten markers, 5.3% and 5.1%, respectively. The RM1233 was one of the flanking markers for qSBR11-1 (26.5,27.2 Mb) reported by Channamallikarjuna et al. [16]. The putative resistant alleles, 169 bp of RM254 and 177 bp of RM1233, were in 12 and 35 entries, respectively. Their ShB ratings were lower than Jasmine 85 in average, indicating a stronger resistance.
Above comparisons confirm eight out of ten marker loci identified in our association mapping with two novel QTLs, RM11229 and RM7203, for sheath blight resistance. The confirmation of previously identified QTLs provides validation for the accuracy of QTLs identified in our study. Furthermore, the comparisons demonstrate that association mapping can locate many QTLs over the entire genome since the mapping panel includes a large number of diversified entries of germplasm. In biparental linkage mapping studies, fewer QTLs are typically identified and can only be located in a limited area in the genome where the two parents differ.
Germplasm panel
The rice mini-core collection of United States Department of Agriculture (USDA) contains 217 entries [53] derived from 1,794 entries of a core collection [60]. The core collection has been shown to be representative of the genetic diversity found in more than 18,000 accessions of the USDA rice whole collection [60]. The mini-core has proven to be an efficient platform for association mapping and has been successfully applied to mapping QTLs for improving grain yield [47]. We excluded fourteen entries of wild species to minimize interference due to different genetic structure [61] and replaced them with fourteen core entries known to have greater resistance to ShB than Jasmine 85, a common resistant check in the comprehensive evaluation of the core collection [58]. The replacement aimed to enhance detection of QTLs by increasing the frequency of putative resistant alleles in the panel.
Phenotyping
A complete set of 1,794 entries in the USDA rice core collection was evaluated in 2008, using the MCM with three replications, three plants in each replication following a randomized incomplete block design over time [58]. Rice cultivars, Lemont (susceptible) and Jasmine 85 (resistant), were included as repeated checks in each replicate to serve as standards for evaluation. Both the check cultivars have been used as standard checks in many other studies regarding ShB resistance [11,12,14,18]. In 2009, the 217 entries of the mini-core collection, plus those core entries that showed significantly more resistance than Jasmine 85, were re-evaluated using the same protocol. LSmean of ShB severity from six replications including 18 observations of each entry was used for association mapping.
The isolate RR0140-1 of R. solani was selected from 102 isolates collected state-wide from Arkansas rice fields due to its slow growing phenotype [51]. Slow growing isolates cause relatively consistent disease reaction and differentiate susceptible cultivars from moderately resistant ones better than fast growing isolates [51]. Field evaluations showed no differences in disease reactions between the slow growing isolates and the fast ones [51]. Further, the RR0140-1 isolate have been adapted by numerous studies [6,18,50]. Pathogen inoculum of RR0140-1 were grown by placing sclerotia in the centre of potato dextrose agar (PDA) plates (Sigma-Aldrich, St. Louis) containing 0.005% (wt/vol) tetracycline, and then transferred to a fresh PDA medium for 5-6 days at 27uC under darkness. Mycelium discs (7 mm in diameter) were excised from the outer growing area in the culture plate where the outer mycelia were mostly active. Rice seedlings were inoculated at the three-leaf stage.
In the greenhouse, each 12612 cm pot was filled with presterilized soil to ensure that the study was not confounded by the presence of soil borne R. solani inoculum. Pots with drainage holes were placed in flats filled with shallow water (,5 cm). Five seeds of each accession were planted in each pot and thinned to three uniform plants before pathogen inoculation. The three remaining plants in a given pot were referred to as one experimental unit or replicate. Each of the three seedlings in a pot was individually inoculated with a round mycelium disc of RR0140-1 pathogen as described by Jia et al. [6] with modification. Each disk was pressed up to the base of the seedling stem, assuring that the mycelium was in contact with the plant. After inoculation, each pot was immediately covered with a 2-liter soft drink bottle with the bottom and cap removed. Relative humidity was maintained over 80% in the bottle, which favoured growth of the sheath blight pathogen on the plants. The greenhouse temperatures were set for day/night at 30/22uC, respectively with a 12 h photoperiod.
Plant response to the sheath blight pathogen was measured using the ratio between the height of the pathogen growing up the plant and the height of the leaf collar on the last emerged leaf. Because mature plant height varied from 70 to 202 cm in this collection [60], the ratio excluded possible interference of plant height in scoring disease response. Therefore, the smaller the ratio, the greater the resistance was for an entry. Measurements were taken when the ratio reached 1.0 for 75% of the susceptible check plants, Lemont, so that the maximum susceptibility was scored 1.0.
ShB rating data were analyzed using the GLIMMIX procedure in SAS version 9.1.3 [62]. The experimental design of randomized incomplete block formed the basis of the statistical model, where the accession is a fixed effect and block is treated as random effect. The LSMEANS option was used to calculate the least-square means (LSMs) of each entry and the LSMs were used for the association mapping. The statistical differences of the accession to each check (Jasmine 85 and Lemont) were determined by a Dunnett's multiple comparison test, using the diff = control option.
Genotyping DNA was extracted from leaf tissue of five plants for each of the 217 entries using a rapid alkali extraction procedure [63] and genotyped with 154 SSR markers plus an indel. The 155 molecular markers covered the entire rice genome with an average genetic distance of 10 cM, described by Li et al. [47]. PCR amplifications were performed according to Agrama et al. (2007) [42]. For each marker, forward primers were labeled with either 6FAM, NED or Hex (Applied Biosystems, Foster City, CA, USA or Integrated DNA Technologies, Coralville, IA, USA). The amplifications were performed using MJ Research Tetrad thermal cyclers (Bio-Rad, Hercules, CA, USA). PCR products were pooled based on color and size range of amplified fragments (typically three markers per run along with ROX-labeled size standard), and the DNA was denatured by heating samples at 94uC for 5 min. The samples were separated on an ABI Prism 3730 DNA Analyzer according to the manufacturer's instructions (Applied Biosystems). Data were analyzed using GeneMapper v. 3.7 software (Applied Biosystems).
Population Structure Analysis
Analysis of population structure in the mapping panel was performed using STRUCTURE software [64,65]. Rare alleles, with frequency of less than 5% in the panel, were treated as missing data for structure analysis, principal components analysis (PCA), cluster analysis and association mapping. We implemented a model-based clustering method for inferring population structure using distinctive allele frequencies and assigning individuals into Q clusters. Twenty independent runs were performed for each value of Q, ranging from one to ten, using the admixture model with a burn-in of 50,000 iterations followed by 100,000 iterations during analysis. Subgroups were determined on the basis of the following criteria: (1) likelihood plot of these models; (2) stability of grouping patterns across twenty runs; and (3) germplasm information about the materials under study. To validate the population structure and compare the different models, PCA was conducted to obtain eigenvectors for further model testing and association analysis. Genetic distance was calculated with PowerMarker [66] using Nei's method [67]. The resulting unweighted pair-group method with arithmetic mean (UPGMA) tree was viewed using MEGA 4.0 [68].
Model comparisons and association analysis
The flexible mixed model [69] was used to control population structure. For the purpose of model comparisons, the phenotypic vector is modeled asy~X bzQvzZuze, where b is vector of marker effects to be estimated. The term Qv contains the coordinates of the individuals of p dimensions in Q matrix generated by STRUCTURE [64,65] and PCA matrix by NTSYSpc version 2.1 [70]; X and Z are the incidence matrices of 1 s and 0 s that relate y to b and u, respectively. u is a vector of polygene background effects; and e is a vector of residual. The phenotypic covariance matrix was assumed to have the form V~Z(2Ks 2 g )Z T zIs 2 e , where K is the K matrix including relative kinship coefficients defining the degree of genetic covariance between a pair of individuals [71], I is an n6n identity matrix, s 2 g is the genetic variance attributable to genome wide effects, and s 2 e is the residual variance.
When model K or Q+K or PCA+K were tested for a fit, we found little convergence because of the low level of relatedness among entries in the panel. Thus, a simple model (ignoring the effect of population structure), possible linear models of Q2-Q10 (considering different number of subgroups from two to ten) and PCA1-PCA10 (PCA matrix with different number of dimensions from one to ten) were compared for the best fit to ShB determined by Bayesian information criterion (BIC). In order to control false positive rates, the genomic control (GC) method [72] was further used for correcting population structure. The association mapping was conducted using the best fit model with TASSEL v.2.1 [73], followed by the GC. The associated markers with ShB resistance were claimed at the probability level of 0.05. The ShB severity ratings of germplasm entries that carried the same allele in an associated marker locus were averaged to estimate allelic effect on the ShB rating. Among the alleles of each associated marker locus, the allele with the lowest ShB mean was indicative of the strongest effect and was designated as the 'putative resistant allele'.
Supporting Information
Table S1 Accession number in the Genetic Stocks Oryza (GSOR) collection, sheath blight (ShB) mean, cultivar name, country of origin, structural group, and number of putative resistant alleles present for 217 entries from the USDA rice core collection. (DOC)
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2017-04-20T09:30:36.572Z", "created": "2012-03-12T00:00:00.000", "id": "18397215", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CC0", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0032703&type=printable", "pdf_hash": "640adc27fea05cb9e772c6fac6b9f92be5532fce", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:39", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "527618d5d4da7965e521f05650890c3b2083612b", "year": 2012}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Altered chloroplast development and delayed fruit ripening caused by mutations in a zinc metalloprotease at the lutescent2 locus of tomato.
The chloroplast is the site of photosynthesis in higher plants but also functions as the center of synthesis for primary and specialized metabolites including amino acids, fatty acids, starch, and diverse isoprenoids. Mutants that disrupt aspects of chloroplast function represent valuable tools for defining structural and biochemical regulation of the chloroplast and its interplay with whole-plant structure and function. The lutescent1 (l1) and l2 mutants of tomato (Solanum lycopersicum) possess a range of chlorophyll-deficient phenotypes including reduced rates of chlorophyll synthesis during deetiolation and enhanced rates of chlorophyll loss in leaves and fruits as they age, particularly in response to high-light stress and darkness. In addition, the onset of fruit ripening is delayed in lutescent mutants by approximately 1 week although once ripening is initiated they ripen at a normal rate and accumulation of carotenoids is not impaired. The l2 locus was mapped to the long arm of chromosome 10 and positional cloning revealed the existence of a premature stop codon in a chloroplast-targeted zinc metalloprotease of the M50 family that is homologous to the Arabidopsis (Arabidopsis thaliana) gene ETHYLENE-DEPENDENT GRAVITROPISM DEFICIENT AND YELLOW-GREEN1. Screening of tomato germplasm identified two additional l2 mutant alleles. This study suggests a role for the chloroplast in mediating the onset of fruit ripening in tomato and indicates that chromoplast development in fruit does not depend on functional chloroplasts.
The chloroplast is a defining organelle of plant cells, serving as the site of the biochemical reactions associated with photosynthesis as well as the synthesis of amino acids, fatty acids, carotenoids, vitamins, and a range of specialized metabolites (Armbruster et al., 2011). The chloroplast is derived from an ancient cyanobacterial endosymbiont with many of the ancestral genes of this organism incorporated into the host genome (Timmis et al., 2004). Current estimates suggest that there are between 2,000 and 3,000 chloroplastlocalized proteins, the vast majority of which are encoded by the host genome (Armbruster et al., 2011). Therefore, chloroplast homeostasis requires coordination of the nuclear and plastid genomes together with the transport of these nuclear-encoded proteins into the chloroplast, their processing, folding, and assembly into functional protein complexes and their subsequent degradation. These processes must be intricately regulated to reflect changes in plastid function during development, the metabolic status of the cell, and be responsive to environmental conditions that perturb chloroplast function. Developing a systematic understanding of the function of chloroplast-localized proteins has broad implications for defining and manipulating plant metabolism and considerable progress has been made in this area, particularly in model organisms (Waters and Langdale, 2009;Ajjawi et al., 2010;Armbruster et al., 2011). Investigating the function of chloroplast-localized proteins in diverse species with novel physiology or specialized metabolism will provide additional and important insight into the functions of this unique organelle.
The onset of fruit ripening is marked by a reprogramming of cellular metabolism that often includes cell wall modifications together with associated fruit softening, the conversion of chloroplasts into chromoplasts and the synthesis of brightly colored pigments such as carotenoids, the conversion of starch into simple sugars, and the synthesis of volatile compounds that influence taste and aroma. These biochemical processes are highly regulated and contribute to the transformation of the fruit from an unpalatable and often toxic organ into one that is attractive and nutritious for seed-dispersing fauna (Barry, 2010;Klee and Giovannoni, 2011). The factors that regulate the ripening transition are not fully understood, although specific transcription factors belonging to the MADSbox and SQUAMOSA PROMOTER BINDING PRO-TEIN families are necessary for ripening and affect multiple aspects of the ripening process, including ethylene synthesis, softening, color change, and aroma volatile production (Herner and Sink, 1973;Sink et al., 1974;Vrebalov et al., 2002Vrebalov et al., , 2009Manning et al., 2006;Itkin et al., 2009;Kovács et al., 2009;Chung et al., 2010;Jaakola et al., 2010;Karlova et al., 2011;Martel et al., 2011;Seymour et al., 2011;Lee et al., 2012). In climacteric fruit such as tomato (Solanum lycopersicum), these transcription factors act upstream of ethylene biosynthesis and ethylene is required for full development of the ripe phenotype (Barry and Giovannoni, 2007;Klee and Giovannoni, 2011;Martel et al., 2011). However, recent research has also implicated additional hormones in mediating the onset of fruit ripening, including brassinosteroids and abscisic acid (Nakano et al., 2003;Lisso et al., 2006;Symons et al., 2006;Zhang et al., 2009;Chai et al., 2011;Jia et al., 2011). Together, these data suggest a complex interplay of transcriptional regulation and hormonal signals influence the ripening of fleshy fruits.
The chloroplast and subsequent development of the carotenoid-rich chromoplast in ripening fruits greatly impacts overall fruit quality with plastid-derived primary and secondary metabolites influencing multiple aspects of the ripe phenotype, including nutrient content, color, and aroma (Klee, 2010). Consequently, many of the enzymes involved in these ripening-induced metabolic changes are localized or associated with either the chloroplast or the chromoplast (Chen et al., 2004;Klee, 2010). Additionally, several mutants that alter different aspects of fruit quality do so through changes in either chloroplast or chromoplast biochemistry. For example, the high-pigment1 (hp-1), hp-2, and hp-3 loci possess altered chloroplast number and ultrastructure, contributing to fruits with elevated levels of chlorophyll, carotenoids, and flavonoids, together with altered patterns of aroma volatile production (Yen et al., 1997;Bino et al., 2005;Kolotilin et al., 2007;Galpaz et al., 2008;Kovács et al., 2009). Similarly, inhibition of chlorophyll degradation due to mutations in tomato and pepper (Capsicum annuum) homologs of the chloroplast-targeted STAY-GREEN protein of rice (Oryza sativa), are responsible for the green-flesh and chlorophyll retainer mutations of tomato and pepper that influence fruit color (Barry et al., 2008;Borovsky and Paran, 2008). Recently, multiple aspects of the ripening process, including the ethylene climacteric and carotenogenesis, were found to be disrupted in the Orange ripening (Orr DS ) mutant that encodes the M subunit of the plastidial NADH dehydrogenase complex (Nashilevitz et al., 2010). These studies illustrate the important role of the chloroplast in influencing fruit ripening and the quality of fleshy fruit.
The lutescent mutants of tomato, lutescent1 (l1) and l2, are nonallelic monogenic mutants that have identical phenotypes resulting in an early and progressive loss of chlorophyll from leaves and fruits. In fruit tissue, the loss of chlorophyll during development renders the fruit a whitish-yellow color prior to the onset of ripening (Kerr, 1956). Here we report that l1 and l2 mutants exhibit pleiotropic chlorophyll-deficient phenotypes including hypersensitivity to high-light stress and delayed deetiolation. In addition, mutation at the l1 and l2 loci leads to a delay in the onset of fruit ripening, suggesting the possibility of a chloroplastderived signal that stimulates fruit ripening. Concomitant with the diverse chlorophyll-deficient phenotypes, positional cloning of the l2 locus revealed the presence of a point mutation that introduces a premature stop codon in a tomato homolog of ETHYLENE-DEPENDENT GRAVITROPISM DEFICIENT AND YELLOW-GREEN1 (EGY1), which encodes a chloroplast-targeted zinc metalloprotease required for chloroplast development in Arabidopsis (Arabidopsis thaliana; Chen et al., 2005).
lutescent Mutants Display Pleiotropic Chlorophyll-Deficient Phenotypes
Developing fruits of the l1 and l2 mutants of tomato are characterized by reduced chlorophyll content, leading to mature fruits that are whitish yellow rather than the typical green color associated with wild-type Ailsa Craig (AC) fruit ( Fig. 1A). At the cellular level mature green AC fruits viewed using confocal laserscanning microscopy show significant amounts of chlorophyll autofluorescence and this is almost completely absent in mature fruits of l1 and l2 mutants (Fig. 1B). The chlorophyll loss observed in l1 and l2 fruits is progressive with increasing age but the average chloroplast number per cell in young developing fruits at 20 d post anthesis was considerably lower in l1 and l2 compared to AC (Fig. 1C). Growing l1 and l2 mutants over several years it was noted that the chlorophyll loss in fruits is typically more exaggerated in plants grown during the summer when light intensities are higher and photoperiods longer and was often accompanied by accumulation of purplish epidermal pigments that are likely to be anthocyanins (Fig. 1D). A lack of chlorophyll accumulation is also observed in developing pistils of l1 and l2 flowers (Fig. 1E). Progressive and early chlorophyll loss is also a feature of the leaves of l1 and l2 (Fig. 1F). This phenomenon is exaggerated in plants grown under highlight conditions where substantial chlorophyll loss was observed even in young expanding leaves but is also apparent in l2 plants held in the dark for 2 weeks (Fig. 1, G and H). Leaves of a second l2 mutant allele, designated l2 3779, which was identified in an ethane methylsulfonate (EMS) mutant screen in the M82 cultivar (Menda et al., 2004; see below for details of this allele) had 13% less chlorophyll than leaves of wild-type cv M82 when grown under light conditions. However, following 2 weeks in darkness, l2 e3779 lost 95% of the chlorophyll content whereas M82 leaves lost only 37% (Fig. 1H).
lutescent Mutants Exhibit Reduced Rates of Deetiolation and Thylakoid Membrane Formation
The conversion of etioplasts into chloroplasts and the subsequent development of photosynthetic competency is regulated by light (Fankhauser and Chory, 1997). To establish whether the l1 and l2 loci impact photomorphogenesis, the rate of deetiolation was determined in the cotyledons of dark-grown seedlings exposed to light. Dark-grown seedlings of l1 and l2 displayed the typical etiolated phenotype observed in AC seedlings (data not shown). However, upon exposure to light, a delay in chlorophyll accumulation was observed in both mutants ( Fig. 2A). Cotyledons of wild-type AC seedlings initiated chlorophyll synthesis within 2 h of transfer to the light and this steadily increased over the 24 h time frame of the experiment. In contrast, both l1 and l2 accumulated chlorophyll at a slower rate and achieved less than 60% of the levels attained by AC seedlings after 24 h of light exposure. Concomitant with the reduced rates of chlorophyll accumulation during deetiolation, examination of plastid ultrastructure revealed impaired development of the thylakoid membranes in the chloroplasts of l1 and l2 mutants compared to those of AC plastids (Fig. 2B). Although chlorophyll accumulates more slowly during deetiolation, the chlorophyll content of fully expanded leaves is similar in AC, l1, and l2 before the onset of early chlorophyll loss (Fig. 2C).
Photosynthesis and Yield in lutescent Fruit
Developing tomato fruit are green due to the presence of chloroplasts that are mainly localized in the outer pericarp wall. Chlorophyll concentration in the outer pericarp tissue of fruit of cv M82 varies between fruit and with the stage of development and on Data are presented as the mean 6 SE of at least 103 cells. Means followed by different letters indicate statistical significance at a = 0.05. D, Image of a mature l1 fruit displaying elevated anthocyanin accumulation in the fruit peel. E, Chlorophyll-deficient phenotypes in the pistils of l1 and l2. F, Fully expanded leaflets of AC, l1, and l2 highlighting increased susceptibility to chlorophyll loss in each mutant. G, Phenotype of 3-week-old AC, l1, and l2 plants grown under high-light conditions. H, Chlorophyll breakdown in the dark in leaves of l2 e3779. Chlorophyll concentration was determined in leaves of plants of mutant l2 e3779 and its isogenic wild-type cv M82 grown for either 7 weeks in the light or 5 weeks in the light followed by 14 d under complete darkness. Data are presented as the mean 6 SE (n = 6).
average is 50-to 100-fold lower than in leaves (data not shown). The photosynthetic activity of M82 and l2 e3779 fruit at the mature green stage of development was determined by chlorophyll fluorescence (Fig. 3A). The concentration of chlorophyll was 22.2 6 1.2 mg g 21 fresh weight in M82 and 2.8 6 0.3 mg g 21 fresh weight in l2 e3779. PSII efficiency in fruit of M82 was similar to that of leaves (data not shown). In contrast, the fluorescence level of l2 e3779 fruit was lower than in M82 and the efficiency of PSII was significantly impaired (Fig. 3A). The determinate growth habit of the M82 variety enables quantitative analysis of fruit yield and overall biomass during harvesting. Plants of M82 and l2 3779 were grown in the field during April to July, 2011. Individual plants were analyzed at time of harvesting with the l2 3779 mutant exhibiting a 17% lower yield and biomass production than M82 (Fig. 3B).
The Onset of Fruit Ripening Is Delayed in lutescent Mutants
Together with the altered chlorophyll content in the fruits of the lutescent mutants ( Fig. 1, A and B), it was observed that fruits of each mutant ripened slightly, but consistently, later than AC fruit. This was confirmed through determining the duration from anthesis to the onset of ripening (breaker stage) that revealed an approximate delay of 6 d in the mutants compared with wild type (Fig. 4A). This delay in the onset of ripening was also observed in detached fruits harvested at the mature green stage of development and allowed to ripen off the vine. The onset of ethylene synthesis in detached fruits was delayed in both l1 and l2 and failed to reach the same levels as that observed in AC fruits (Fig. 4B). This slow-ripening phenotype appeared to be specific to a delay in the onset of the ripening process as once ripening was initiated the rate of progression appeared identical in wild-type and mutant plants with lycopene accumulation progressing at the same rate with both mutant and wild-type fruits becoming fully ripe at approximately 7 d after the onset of ripening (Supplemental Fig. S1). Furthermore, ethylene treatment of wild-type AC, l1, and l2 fruits at the mature green stage of development led to the onset of ripening within 2 d of treatment in each genotype, indicating that ethylene responsiveness is not significantly altered in mutant fruits (Supplemental Fig. S2). Similarly, the ethylene responsiveness of l1 and l2 dark-grown seedlings was normal (Supplemental Fig. S2). Together, these data suggest that mature green l1 and l2 fruit are fully competent to ripen yet are perturbed in a signal that initiates ethylene synthesis and therefore the onset of fruit ripening.
Positional Cloning of the l2 Locus
Classical linkage mapping indicated that the l2 mutant mapped in close proximity to the tangerine and hairs absent loci on linkage group VII that was subsequently identified as the long arm of chromosome 10 (Kerr, 1956). Using an F2 population of 60 individuals derived from an interspecific cross l2/l2 (tomato) 3 L2/L2 (Solanum galapagense), the l2 locus was mapped to a 7.5 centimorgan (cM) region of the long arm of chromosome 10 between the flanking RFLP markers T1682 and T802 (data not shown). Upon increasing the size of the F2 mapping population to 799 individuals by screening for recombinant individuals between T1682 and T0802 the position of the L2 locus was refined to a 0.88 cM interval between the RFLP markers T0615 and T0736 (Fig. 5A). T0615 and T0736 were used to screen three high M r tomato genomic DNA libraries made from DNA extracted from Solanum pennellii and S. galapagense Chen et al., 2007). Five individual clones were found to hybridize to T0615 and a single 140-kb clone, 31F20, was isolated following screening with T0736. The ends of these clones were isolated by direct DNA sequencing and specific primers were designed to amplify each end by PCR. These ends were converted to RFLP probes to enhance the resolution of the genetic map and to position the genomic clones relative to one another. A probe derived from the reverse end of 31F20 (31R) hybridized to clone 178I9, indicating that the L2 locus was contained on two overlapping clones at a maximum physical distance of approximately 200 kb. 31R was used as a probe to screen a tomato cosmid library and a single clone, LeCOS196E17, was identified. The ends of this clone were isolated as previously described and converted to RFLP markers. Genetic mapping indicated that the l2 locus lies within clone LeCOS196E17, 0.06 cM from 196R and 0.19 cM from 196F. LeCOS196E17 was sequenced revealing five predicted genes, corresponding to the tomato genome locus identifiers Solyc10g081450 through Solyc10g081490 (SL2.40Ch10:61831485-61868274). Sequence similarity searches with each predicted gene revealed that Solyc10g081450 encodes an a/b-fold hydrolase-related protein, Solyc10g081460 encodes a predicted amino acid transporter, Solyc10g081470 encodes a chloroplasttargeted zinc metalloprotease, Solyc10g081480 encodes a protein of unknown function, and Solyc10g081490 encodes a MYB transcription factor. Fragments of each gene were converted into RFLP markers and were mapped in the F2 population. Solyc10g081470 cosegregated with the mutant phenotype whereas Sol-yc10g081480 and Solyc10g081460 were 0.06 and 0.13 cM from the L2 locus, respectively, suggesting that Sol-yc10g081470 is the candidate gene for the l2 locus.
Sequencing of the predicted full-length cDNA of Solyc10g081470 amplified by reverse transcription (RT)-PCR from L2/L2 and l2/l2 genotypes revealed a single T → A substitution at base 1574 of the cDNA clone (exon 10 in the genomic clone) in the l2 mutant. The substitution results in the conversion of Leu 525 into a stop codon, truncating the predicted protein by 23 amino acids. Confirmation that the single base pair substitution in Solyc10g081470 is the underlying cause of the l2 mutant phenotype was achieved through complementation analysis. The full-length Solyc10g081470 cDNA was expressed under the control of the cauliflower mosaic virus 35S promoter in the l2 mutant background. Nineteen out of 22 independently transformed primary transformants recovered showed complementation of the l2 phenotype, determined by chlorophyll content of mature green fruit. Three lines were selected for subsequent analysis in the T1 generation (Fig. 5B). Typical phenotypes of AC and l2 mature green fruits are shown along with three independent T1 lines segregating for the NPTII selectable marker. Fruits from plants in the top section contained NPTII and showed complementation of the l2 mutant phenotype. Conversely the fruit on the bottom section came from plants that had segregated out the NPTII gene and had reverted to the l2 mutant phenotype.
L2 Encodes a Tomato Homolog of Arabidopsis EGY1 L2 (Solyc10g081470) encodes a predicted protein of 547 amino acids with a molecular mass of 58.4 kD. A BlastP search using the predicted L2 protein as the query sequence revealed homology to proteins from several higher plant species and other photosynthetic organisms, including bryophytes, algae, and cyanobacteria. Significantly, the Arabidopsis protein displaying the highest similarity to L2 (73% identical/ 82% similarity) was encoded by the EGY1 locus, a chloroplast-targeted member of the zinc-dependent M50 family of metalloproteases ( [URL]. sanger.ac.uk) that is required for normal plastid development and gravitropic response in Arabidopsis (Chen et al., 2005). The M50 family is diverse, with members from bacteria, archaea, protozoa, plants animals, and some fungi. The family is characterized by two conserved domains, HEXXH and NXXPXXXLDG that are present in L2 and related homologs (Fig. 6). These two domains are predicted to form the active site with Glu acting as the catalytic residue and the histidines and Asp residues coordinating the zinc ions. Mutant studies have demonstrated the importance of these domains for catalytic activity (Rawson et al., 1997;Rudner et al., 1999). A second feature of this Figure 5. Positional cloning of the L2 locus. A, Genetic and physical map of the l2 locus derived using an interspecific F2 population of 799 plants. Distances between adjacent RFLP markers are given in cM. BAC clones isolated using the RFLP markers T0615 and T0736 are shown as gray bars. BAC ends were isolated as described in the text and given F and R designations. A cosmid clone, LeCOS196E17, isolated using 31F20-R is represented as a striped bar. Vertical dashed lines are shown to indicate relative positions of genetic markers and genomic clones. LeCOS196E17 contains five predicted genes (Solyc10g081450-Solyc10g081490). The genetic distance between these genes is shown in cM. Solyc10g081470 cosegregates with the l2 phenotype and has a structure comprised of 10 exons separated by nine introns. The T → A substitution in exon 10 that converts Leu 525 to a stop codon in l2 is shown. B, Complementation of the l2 mutant phenotype. Mature green fruit from three independently transformed segregating T1 lines containing the transgene (top section) and without the transgene (bottom section). AC and l2 fruits are shown for phenotypic comparison.
family is that they are integral membrane proteins with the active site domains lying within, or in close proximity to, the membrane-spanning domains (Fig. 6). The Aramemnon Plant Membrane Database ( [URL]), which collates the membrane topology predictions of several algorithms, predicts that Arabidopsis EGY1 contains eight transmembrane-spanning domains (Fig. 6). A search utilizing the TMHMM server version 2.0 ( [URL]) revealed that the predicted L2 protein has six transmembrane-spanning domains with an additional two domains lying slightly below Figure 6. Amino acid alignment of L2-related proteins. Alignments based on the conserved region of L2 related proteins were generated using MUSCLE. Conserved amino acids are indicated by shaded squares. The highly conserved motifs characteristic of these proteins, GNLR, HEXXH, and NXXPXXLDG are underlined with solid black lines. The solid gray lines depict the predicted transmembrane helices for EGY1 as determined by the TmConsens algorithm ( [URL]. de/index.ep). Accession numbers of the sequences are as follows: L2 (JQ683149), Vitis vinifera GSVIVT01029948001 (XP_002269447), Populus trichocarpa POPTRDRAFT_773851 (XP_002319720), rice Os03g0792400 (EEE60079), Synechocystis sp. PCC 6803 sll0862 (NP_440944). The sequence of Arabidopsis EGY1 is based on annotation available at The Arabidopsis Information Resource ( [URL]) and the sequences of Cre19.g754700 and Ppls517_1V6.2 are based on annotations of the C. reinhardtii and Physcomitrella patens genomes available through the Phytozome database ( [URL]/). the threshold of probability. Similarly, the selection of EGY1/L2 homologs depicted in Figure 6 are all predicted to contain between six and eight transmembrane-spanning domains lying in relatively conserved positions (data not shown). Localization experiments utilizing an Arabidopsis EGY1-GFP fusion demonstrated that EGY1 is targeted to the chloroplast (Chen et al., 2005). Similarly, in silico prediction of L2 subcellular localization using the Predotar server ( [URL]. versailles.inra.fr/predotar/predotar.html) resulted in a probability estimate of 0.79 for plastid localization.
Identification of Additional Alleles at the l2 Locus
A second putative l2 allele was reported as a spontaneous mutation in a trial plot of the cultivar Fireball in 1960 and was designated Lutescent Fireball (Kerr et al., 1975;Supplemental Fig. S3). An allelism test confirmed that F1 progeny of a cross between Lutescent Fireball and the original l2 allele in the Long Red variety displayed a typical lutescent phenotype (Kerr et al., 1975). RT-PCR amplification of L2 from cv Fireball yielded the expected transcript of 1,641 bp that was identical to that obtained from cv AC. In contrast, amplification of the L2 cDNA from Lutescent Fireball revealed multiple products of a slightly altered size (Fig. 7A). Sequence analysis of these RT-PCR products revealed the existence of four distinct L2 transcripts in the Lutescent Fireball cultivar ranging in size between 1,572 and 1,846 bp, each with aberrant splicing occurring between intron 7 and exon 8. Three of the splice variants (SV2, SV3, and SV4) cause frame-shift mutations, leading to incorporation of several spurious amino acids followed by premature stop codons that truncate the predicted L2 protein between 79 and 85 amino acids (Fig. 7B). In contrast the SV1 splice variant causes an in-frame deletion of the entire eighth exon, leading to the removal of 23 amino acids between Q455 and L479 (Fig. 7C). Furthermore, the wild-type transcript was not detected in the Lutescent Fireball cultivar, suggesting that the L2 protein may be nonfunctional in this variety. To further explore the possibility that a splicing defect in L2 is the underlying cause of the Lutescent Fireball phenotype, a 750-bp genomic fragment spanning exons 7 and 8 from in both Fireball and Lutescent Fireball revealed a G → A nucleotide change at position 7000 of the L2 genomic clone in the Lutescent Fireball cultivar (Fig. 7B). This single nucleotide substitution disrupts the consensus splice acceptor site at the boundary between intron 7 and exon 8 (CAG → CAA). Together, these data suggest that the phenotype of Lutescent Fireball is caused by aberrant splicing, resulting in the failure to produce a mature L2 transcript. A third mutant resembling l2, e3779 that figures prominently in the characterization described here, was recovered from an EMS mutagenized population of the M82 cultivar (Menda et al., 2004;Supplemental Fig. S2). Sequence analysis of the L2 cDNA amplified from e3779 revealed the deletion of a G nucleotide at position 1575 that results in a frame-shift mutation that removes Gly 488, resulting in the incorporation of 15 spurious amino acids before terminating in a premature stop codon (Fig. 7, B and C).
The Role of L2 in Chloroplast Function
The nonallelic l1 and l2 mutants display pleiotropic phenotypes that are consistent with defects in chloroplast development (Figs. 1 and 2). These data are supported by the finding that L2 encodes an ortholog of Arabidopsis EGY1, mutations that disrupt plastid number and development, leading to impaired formation of the thylakoid membranes and reduced accumulation of chlorophyll a/b-binding proteins of the light-harvesting complexes I and II (Chen et al., 2005;Guo et al., 2008). However, a notable difference between l2 and egy1 mutant alleles is that while egy1 mutant alleles have reduced chlorophyll content and are pale throughout development (Chen et al., 2005), leaves of l2 mutants are slow to deetiolate (Fig. 2) but eventually accumulate close to wild-type levels of chlorophyll prior to undergoing premature chlorophyll loss later in development (Fig. 2C). Furthermore, altered patterns of chlorophyll accumulation are evident between leaf and reproductive tissues of lutescent mutants. For example, while l1 and l2 leaves accumulate chlorophyll, the pistils and fruits of the mutants display greatly impaired chlorophyll accumulation from early in development and therefore resemble the pale phenotype of egy1 mutant alleles (Fig. 1). The basis for these differences, particularly the seemingly less significant role that L2 plays in chlorophyll accumulation in tomato leaves, is currently not understood.
In addition to a role in mediating chlorophyll accumulation during development, evidence suggests that EGY1/L2 and related homologs may play a role in mediating chloroplast responses to environmental conditions. For example, both l1 and l2 mutants are hypersensitive to darkness and high-light stress, leading to enhanced chlorosis (Fig. 1, G and H) and EGY1 protein levels in Arabidopsis are influenced by dark/light cycles with increased abundance evident in light-grown tissues (Chen et al., 2005). Similarly, iron deficiency, a stress known to reduce photosynthetic electron transfer, resulted in increased accumulation of the EGY1/L2 homolog, C_770040, from Chlamydomonas reinhardtii (Naumann et al., 2007). Furthermore, publically available Arabidopsis gene expression data revealed that EGY1 is induced under both high-light treatments and iron deficiency together with other biotic and abiotic stresses (Zimmermann et al., 2004). Together these data suggest a role for EGY1/L2 in both the early steps of chloroplast development and the maintenance of chloroplast homeostasis in response to environmental perturbation. EGY1 and L2 belong to the M50 family of metalloproteases and each contains motifs known to be required for the activity of functionally characterized members of this family (Fig. 6). While considerable progress has been made on elucidating the role of chloroplast-localized proteases (Kato and Sakamoto, 2010), the exact biochemical function of EGY1/L2 and other plant M50 proteases, including the identity of their in vivo substrates remains unknown. Members of the M50 family of proteases are integral membrane proteins with their active sites residing within or adjacent to the membrane (Fig. 6). The M50 family belong to a category of proteases referred to as intramembrane cleaving proteases that are involved in the regulation of cellular responses to metabolic status and stress that participate in a general phenomenon known as regulated intramembrane proteolysis (RIP). For example, the founder member of the M50 family, the sterol-regulatory element binding protein (SREBP) site-2 protease regulates sterol biosynthesis and uptake in mammalian cells through release of the membrane-bound transcription factor SREBP in a pathway that involves trafficking of the sequestered SREBP from the endoplasmic reticulum to the Golgi under low-sterol conditions (Osborne and Espenshade, 2009). Similar two-step proteolysis occurs in response to stress in bacteria, allowing the release of membrane-bound transcription factors (Rudner et al., 1999;Alba et al., 2002;Kanehara et al., 2002). Far less is known about RIP in plants. Membrane-tethered transcription factors have been identified in plants and in a few instances proteolytic cleavage of these transcription factors has been demonstrated but none of these have been reported to involve transcription factors that are sequestered in the chloroplast (Seo et al., 2008;Kim et al., 2010;Liu and Howell, 2010). Therefore, it is possible that EGY1/ L2 proteases participate in chloroplast-localized phenomena such as processing or degradation of proteins at the thylakoid membrane during photosystem assembly or in response to environmental perturbation and disruption of photosynthetic function. Examples of proteolysis and processing in the chloroplast are already well characterized and are involved in the turnover and repair of photosystem proteins in response to photoinhibition and abiotic stress (Kato and Sakamoto, 2010;Sun et al., 2010).
lutescent Mutants and Their Impact on Fruit Development and Ripening
Both l1 and l2 mutants exhibited a delay in the onset of fruit ripening (Fig. 4). However, once ripening is initiated in l1 and l2 fruits the rate of ripening appears normal and mutant fruits accumulate wild-type levels and composition of carotenoids (Supplemental Fig. S1), suggesting that a fully functional chloroplast is not required for chromoplast development in tomato. The concept of separation of chloroplast and chromoplast development is also observed in stay-green mutants of tomato and pepper that undergo typical chromoplast development in the absence of chlorophyll breakdown and thylakoid membrane dissolution (Cheung et al., 1993;Roca et al., 2006).
As l1 and l2 mutants have disrupted chloroplast function, it is likely that the delayed fruit ripening phenotypes observed in mutant fruit arise as an indirect consequence of an altered chloroplast-derived signal that promotes the onset of ripening. The recent finding that disruption of Glu-1-semialdehyde aminotransferase activity in tomato fruits alters early seed development (Lytovchenko et al., 2011) provides a possible mechanism that could ultimately lead to a delay in the onset of ripening. It will be of interest to examine the rates of ripening in Glu-1-semialdehyde aminotransferase-silenced lines and early seed development in l1 and l2 mutants, particularly as fruit of lutescent mutants are also compromised in their photosynthetic capacity (Fig. 3A).
Alternatively, lower chloroplast numbers in developing fruits of l1 and l2 could be supportive of reduced levels of a factor that promotes ripening. The nature of the signals that stimulate ripening are not fully understood, although multiple factors including ethylene are known to play a role in ripening induction (Srivastava and Handa, 2005;Barry and Giovannoni, 2007;Zhang et al., 2009;Barry, 2010). The ability of ethylene to trigger rapid and normal ripening of l1 and l2 fruit at the mature green stage of development, coupled with the delay in the onset of ethylene synthesis suggests that l1 and l2 fruit possess a typical response to ethylene but may be perturbed in the ability to initiate the ripening-related increase in ethylene production ( Fig. 4; Supplemental Fig. S1). Recent characterization of the Orr DS mutant of tomato that encodes the M subunit of the chloroplastlocalized NADH dehydrogenase complex, also supports a role for a chloroplast-derived signal that promotes ripening (Nashilevitz et al., 2010). The Orr DS mutant exhibits a delay in the onset of fruit ripening and ripening-related ethylene biosynthesis. Although the reason for this delay is not fully understood, the Orr DS mutant has a 30% to 40% reduction in the levels of both Ser and Asp, two chloroplast-derived amino acids that are precursors of Met, which serves as the precursor of ethylene (Nashilevitz et al., 2010). It will be of interest to determine whether chloroplast metabolism in l1 and l2 fruit influences seed development and precursor pools for the synthesis of hormones and to define whether a link exists between these factors and the observed delay in the onset of fruit ripening.
CONCLUSION
Cloning of the L2 gene has identified a link between chloroplast development and as-yet-unidentified factors that influence the onset of fruit ripening. The mechanism of this interaction, together with the specific biochemical activity of EGY1/L2 and the potential role of these proteins in RIP remain to be identified. In this regard, we have shown that the l1 mutant, which resides at a distinct locus on chromosome 8 (Tanksley et al., 1992), possesses an identical phenotype to that of l2, suggesting the strong possibility that the underlying gene may be functionally related to L2. Therefore, identification of the gene underlying the l1 mutant phenotype may provide insight into the molecular function of EGY1/L2 in plants.
Plant Material and Treatments
Tomato (Solanum lycopersicum) seeds homozygous for the l1 (LA3717) and Evaluation of the deetiolation response of tomato seedlings was performed as follows. Surface-sterilized seeds were sown on 1% water agar supplemented with 0.53 Murashige and Skoog salts pH 5.6 and incubated in the dark for 7 d at 25°C. The germinated seedlings were then exposed to light for up to 24 h and the cotyledons excised and either immediately frozen in liquid N 2 or immersed in fixative for transmission electron microscopy (described below). Experiments to evaluate the triple-response phenotype in dark-grown tomato seedlings were performed as previously described (Barry et al., 2005) with the exception that seedlings were measured at 8 d after sowing. Ethylene treatments were performed by sealing fruits at the mature green stage of development in glass canning jars of a known volume and exposing to 10 mL L 21 of ethylene for 24 h. Following ethylene treatments fruits were held at room temperature for 72 h and ripening evaluated. For evaluation of response to high-light conditions plants were grown under a 16 h photoperiod at 27°C/20°C day/night temperature and a light intensity of 500 mmol m 22 s 21 . Plants were photographed 3 weeks after germination.
Ethylene Measurements
Mature green fruit were harvested and held at room temperature for 24 h. For ethylene determination, fruits were sealed each day for 2 h in glass jars and a 1-mL headspace sample was analyzed by gas chromatography using a Carle AGC series 100 gas chromatograph equipped with an activated alumina column and flame ionization detector, as previously described (Watkins et al., 2004).
Measurements of Photosynthetic Efficiency
The efficiency of PSII was estimated by measuring chlorophyll fluorescence with a computer-controlled pulse amplitude modulated fluorometer IMAG-MAX/L (Walz). Detached fruit were dark adapted for 20 min. A 2-cm-thick slice was cut and placed in the measuring chamber with the fruit surface facing up toward the measuring beam. A low (0.1 mmol photons m 22 s 21 ) modulated (1.6 kHz) measuring light was used to determine the minimal fluorescence level, followed by a 1 s of saturating white light pulse of 1,500 mmol photons m 22 s 21 to obtain the maximal fluorescence level. The fruit was then illuminated with continuous actinic red light (650 nm, 40-mmol photons m 22 s 21 ), until a steady-state fluorescence was reached (up to 10 min). A second saturating pulse was then applied to determine the maximal fluorescence level in the light-adapted state.
Pigment Analysis
Chlorophyll was extracted from tissues in acetone and quantified as previously described (Lichtenthaler, 1987). Fruit carotenoids were analyzed by HPLC as previously described (Ronen et al., 1999).
Confocal Laser-Scanning Microscopy
The chloroplasts of tomato fruit were imaged using an Olympus FluoView FV1000 laser-scanning confocal microscope. Chlorophyll autofluorescence was detected with a 650 to 750 nm band pass filter following excitation at 488 nm with an argon laser line. Cells were visualized using differential interference contrast transmitted light microscopy using a 203 UPlan SApo objective with a numerical aperture of 0.75. Images were merged using Olympus FluoView software.
Transmission Electron Microscopy
Cotyledons were fixed in a mixture of 2.5% glutaraldehyde and 2.5% paraformaldehyde in 0.1 M cacodylate buffer at 4°C for 24 h, postfixed in 1% osmium tetroxide, and dehydrated in a graded acetone series. Samples were infiltrated and embedded in Poly/Bed 812 resin (Polysciences Inc.). Thin sections (70-nm thickness) were obtained with a PTXL ultramicrotome (RMC, Boeckeler Instruments) on 200 mesh copper grids stained with uranyl acetate and lead citrate. Sections were imaged using a JEOL 100CX transmission electron microscope (JEOL) at a 100-kV accelerating voltage.
Genetic and Physical Mapping of the l2 Locus
An interspecific F2 mapping population was generated from the following cross tomato (l2/l2) (LA3581) 3 S. galapagense (L2/L2; LA0483). Genomic DNA isolation and molecular mapping using RFLP markers was performed as previously described (Barry et al., 2005). Details of tomato genetic maps and the chromosome 10 molecular markers can be accessed through the Solanaceous Genomics Network ( [URL]/). Restriction enzymes yielding RFLPs between tomato and S. galapagense for given DNA probes are as follows; T736: HaeIII, CT57: NdeI, T615: NlaIV, 95R: AvaII, 196E17F: XbaI, 196E17R: EcoRV, cLET10H23: RsaI, L2: ApoI, cTOF24I6: NsiI. A physical contig spanning the L2 locus was obtained via screening and characterization of ordered bacterial artificial chromosome (BAC) and cosmid libraries derived from tomato, Solanum pennellii, and S. galapagense Chen et al., 2007). Clones LA0483BAC69H22, LA0483BAC69J23, and LA0483BAC178I9 were isolated from an S. galapagense library. Clones LpenCOS31F20, Lpen-COS95L3, and LpenBAC137I4 were isolated from two S. pennellii libraries. The approximate size of each BAC clone was determined by pulsed-field gel electrophoresis. BAC and cosmid ends were isolated by direct DNA sequencing and converted to RFLP markers for further genetic and physical mapping. Subcloning of the tomato cosmid clone LeCOS196E17 was achieved by three separate restriction enzyme digests (BamHI, EcoRI, and HinDIII) followed by ligation into pBluescript SK2 (Stratagene), previously linearized with the same restriction enzymes. The resultant clones were sequenced with vector primers to yield a number of anchor points that were subsequently utilized to extend and complete the sequence of LeCOS196E17 using multiple rounds of primer extension.
DNA Constructs and Plant Transformation
The L2 cDNA insert was released from pBluescript SK2 by digestion with XbaI and SacI. The fragment was ligated downstream of the cauliflower mosaic virus 35S promoter in the binary vector pBI121, previously modified by removal of the UidA coding region by digestion with the same restriction enzymes. Transgenic tomato plants were generated through cotyledonderived explants via Agrobacterium tumefaciens mediated transformation (strain LBA4404), using previously described methods (Fillatti et al., 1987). Transgene integration was determined by a combination of gel-blot and PCR analysis.
Statistical Analysis
Statistical analyses were performed using SAS (SAS Institute, www.sas. com). The genotypic constituents were evaluated using least squares means.
Sequences reported in this manuscript have been deposited in GenBank under the following accession number: JQ683149.
Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Carotenoid composition in ripe fruits of the lutescent mutants.
Supplemental Figure S2. The lutescent mutants have a normal ethylene response.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2018-04-03T04:55:51.545Z", "created": "2012-05-22T00:00:00.000", "id": "29686048", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "http://www.plantphysiol.org/content/plantphysiol/159/3/1086.full.pdf", "pdf_hash": "9931a90265c1be51701e1da6c8d96f7e7c2fea84", "pdf_src": "Highwire", "provenance": "peS2o-0028.json.gz:6586", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "39b8460199ce73971fa38e6ba4984dc384802d0c", "year": 2012}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Mammary Gland Transcriptome and Proteome Modifications by Nutrient Restriction in Early Lactation Holstein Cows Challenged with Intra-Mammary Lipopolysaccharide
The objective is to study the effects of nutrient restrictions, which induce a metabolic imbalance on the inflammatory response of the mammary gland in early lactation cows. The aim is to decipher the molecular mechanisms involved, by comparing a control, with a restriction group, a transcriptome and proteome, after an intra-mammary lipopolysaccharide challenge. Multi-parous cows were either allowed ad libitum intake of a lactation diet (n = 8), or a ration containing low nutrient density (n = 8; 48% barley straw and dry matter basis) for four days starting at 24 ± 3 days in milk. Three days after the initiation of their treatments, one healthy rear mammary quarter of 12 lactating cows was challenged with 50 µg of lipopolysaccharide (LPS). Transcriptomic and proteomic analyses were performed on mammary biopsies obtained 24 h after the LPS challenge, using bovine 44K microarrays, and nano-LC-MS/MS, respectively. Restriction-induced deficits in energy, led to a marked negative energy balance (41 versus 97 ± 15% of Net Energy for Lactation (NEL) requirements) and metabolic imbalance. A microarray analyses identified 25 differentially expressed genes in response to restriction, suggesting that restriction had modified mammary metabolism, specifically β-oxidation process. Proteomic analyses identified 53 differentially expressed proteins, which suggests that the modification of protein synthesis from mRNA splicing to folding. Under-nutrition influenced mammary gland expression of the genes involved in metabolism, thereby increasing β-oxidation and altering protein synthesis, which may affect the response to inflammation.
Introduction
Milk is synthesized in mammary glands (MG) involving a large number of genes, the expression of which is modulated at a nutritional level [1] and by the health status [2]. Mastitis is the inflammatory response of the mammary gland to pathogens. This pathology is one of the most prevalent disease and has considerable economic impact due to decreased milk production, discarded milk, cost of body weight (BW), plasma metabolite, and insulin concentrations. Feeding the ration containing 48% of straw (restricted group: REST) induced an immediate depression of dry matter intake (DMI) and decreased the energy balance from 5.2 ± 8.9 to −67.2 ± 18.9 MJ/day one day before (corresponding to day 23), and on the last day of restriction (day 27), respectively. The plasma concentrations of glucose and insulin decreased, whereas non-esterified fatty acids (NEFA) and beta-hydroxy butyrate (BHBA) increased dramatically in REST (Supplementary Table S1). The 96 h of nutrient restriction decreased milk yield from 37.9 to 22.4 kg/day (p < 0.001) and milk protein yield from 1.12 to 0.62 kg/day (p < 0.005) in REST. Milk fat percentage increased during feed restriction in REST from 4 to 5.5% (p < 0.001) and returned to pre-restriction concentrations on the same day of re-feeding a regular diet. In CONT cows, milk fat content was greatest only during the 48 h following LPS injection (p < 0.05) compared with all other DIM. These variables were unchanged in the control (CONT) cows [25]. Within 2 to 6 h after injection with lipopolysaccharide, we noticed the edema of the challenged quarter of all cows, an increase in rectal temperature up to 39.5 • C (temperature increment was +2.1 ± 0.15 • C). The effect of inflammation also was confirmed by milk somatic cell count (SCC). The day before the LPS challenge, whole udder composite milk (from PM and AM milking) SCC was 78 000/mL and 92,000/mL (p = 0.51) in CONT, and in REST, respectively. The SCC response was greater in REST compared to CONT cows (6919 versus 1956 × 1000 per mL, respectively) in composite milk samples from the two milkings that followed LPS injection. SCC returned to pre-LPS counts, within less than 7 days post-LPS challenge and biopsy. Moreover, quarter milk IL-8, IL1-β, TNF-alpha, and CXCL3 at time zero (before LPS challenge) did not differ between CONT and REST, but their concentrations increased in response to LPS in both groups. This data shows that these indicators of mammary inflammation did not differ between CONT and REST before or after the challenge.
Effects of Under-nutrition on Expression of Genes Involved in Inflammation Response by RT-qPCR Analysis
RT-qPCR analysis was performed to quantify candidate genes (CCL5, LAP, RBP4, IL8, IL1, STAT3, CD36, and TAP) chosen on the basis of their implication in inflammation [26] Expression of these genes in mammary gland (MG) did not differ between REST and CONT (p ≥ 0.1) 24 h after the LPS challenge, except for the defensin Tracheal Antimicrobial Peptide (TAP) gene which tended to decrease in REST (p = 0.07). The expression of INSIG1, and CSN2 genes, which are involved in the biosynthesis of milk components and linked to MG metabolism did not differ between CONT and REST ( Figure 1). Figure 1. Effects of nutrient restriction and intra-mammary lipopolysaccharide (LPS) challenge on gene mRNA expression quantified by RT-qPCR and presented as ∆ CT . Comparison of the gene expression does not show a difference between control (CONT; n = 6) and restricted (REST; n = 6) Holstein cows (p ≥ 0.1). The expression of the TAP gene tended to differ (p = 0.07). UXT2, CLN3, and EIF3K were used as housekeeping genes.
Microarray Analysis
Mammary gene expression analyzed by a microarray assay allowed the identification of 33 differentially expressed genes, including 25 known genes (corrected p adj < 0.05), between CONT and REST, 24 h after the inflammatory challenge by LPS (Table 1). The expression increased for 19 and decreased for 6 genes in REST compared with CONT. All these DEG presented a fold change (FC) greater than 1.4, with two genes (PDK4 and SLC25A34) presenting an FC greater than 4. Gene ontology and function analyses revealed that most DEG are involved in metabolism, including the regulation of fatty acid (FA) oxidation, glucose, and protein metabolism, and in immune responses ( Figure 2). The results obtained, using Pathway Studio ® software, were consistent with those from Panther software. We focused on the most represented functions, in particular, those involved in metabolism and immune response. We identified DEG in FA and glucose metabolism (CPT1A, PDK4, PFKFB4), carnitine shuttle (SLC25A20, CPT1A, SLC25A34), regulation of cellular ketone metabolic process (PDK4), and the key genes in those processes. A number of genes involved autophagy (PFKFB4, DNNED) and immune function (PGLYRP3, KLF13, PLEKHA2, WC7, TRIB2, CXCR7, and MBP) processes also were altered.
Discussion
This study assesses the effects of undernutrition and the resulting metabolic imbalance on the mammary gland (MG) inflammatory response in early lactation cows using a nutrigenomic approach. The effects of dietary treatments are confirmed by decreased intake, milk yield, energy balance in underfed (REST) cows, and by changes in blood metabolite and insulin concentrations (an increase in plasma NEFA and BHBA and a decrease in insulin and glucose concentrations; Supplementary Table S1; [25]). The inflammatory response to intra-mammary lipopolysaccharide is confirmed by clinical parameters, such as milk SCC, rectal temperature, and other classical clinical symptoms [25]. The effects of nutrient restriction and metabolic imbalance on the inflammatory response, at the RNA and protein levels, were evaluated using transcriptomic and proteomic analyses. These analyses were performed using MG samples were obtained by biopsies performed 24 h after the LPS challenge. We performed a single biopsy to avoid the potential interference of repetitive biopsies on the inflammatory response. Also, the adjacent quarter may not constitute a good control for the LPS challenged quarter, as inflammation cytokines may exert local effects and influence adjacent quarters [19]. However, one limitation of this design is that the present experimental design does not allow a kinetic data to follow the establishment of inflammation.
Gene Expression Changes at mRNA Level
The study at the mRNA level was performed using two complementary approaches. The first one was a targeted approach to focus our attention on the inflammatory response, then a global analysis was performed using microarray study. We investigated the effects on candidate genes by RT-qPCR, the majority, which are involved in the immune response of interest for inflammation. The expression of candidate genes did not differ between REST and with the control (CONT) group ( Figure 1). This result suggests that the expression of genes considered important in the inflammatory response [11,[26][27][28] are not altered by the nutrient restriction in MG 24 h after LPS administration. This experimental design does not allow the ability to evaluate the modification of the expression of these genes during early inflammatory response. A tendency for decreased expression of the TAP gene in REST was observed. The product of TAP gene is a member of the family of small cationic peptides that have widespread antimicrobial activity; TAP is expressed by bovine mammary epithelial cells [29] and has a broad-spectrum activity against different strains of bacteria, including E. coli [30]. The upregulation of TAP gene expression in REST may constitute a protective mechanism against pathogens.
To complete the candidate gene analyses, a global gene expression approach, using a bovine microarray was used to assess the molecular mechanisms underlying metabolic and inflammatory MG responses, potentially affected by undernutrition and negative energy balance. Transcriptomic analysis revealed 33 differentially expressed genes in MG 24 h after LPS challenge in REST compared to CONT. The number of DEG detected in our study are small compared to the research assessing the effects of inflammation on MG transcriptome [18,19,31]. This study does not compare normal versus inflamed MG, rather it aims to evaluate the effects of undernutrition during the inflammatory response, and therefore both REST and CONT were challenged with LPS. The DEG were classified in six functional classes. In this discussion, we mainly focus our attention on genes that play a role in the metabolic processes (FA, glucose and protein metabolism) and on those involved in immune function.
DEG Involved in Metabolism
The classification of genes by bioinformatics analyses indicate that metabolic process (FA oxidation, glucose, and protein metabolisms) is the class most altered by undernutrition after an LPS challenge conditions. Among the genes presenting the highest fold change (between 2.7 and 6.8) are PDK4, CPT1, and SLC25A34, which are involved in glucose and FA metabolism. PDK4 plays a key role by inhibiting the pyruvate dehydrogenase complex. This inhibition prevents the formation of acetyl-coenzyme A from pyruvate [32], resulting in an expected decrease in glucose and an increase in fat utilization in response to prolonged undernutrition [33]. The upregulation of PDK4 was also found in leucocytes of underfed ewes, however, its expression was downregulated during an intra-mammary inflammatory challenge [34]. The large increase of PDK4 expression in REST is in agreement with decreased insulinemia and with the upregulation of ESRRA. The gene ESRRA upregulates PDK4 expression [35]. Both genes spare glucose and promote FA β-oxidation, therefore, their upregulation in REST would allow MG to shift the metabolic pathways from glycolysis to β-oxidation. The upregulation of CPT1 gene expression would also promote the β-oxidation [36,37], since it is a rate-limiting step of FA entry in mitochondria [38]. This is in line with the increased expression of the CPT1 gene observed in whole blood transcriptome of underfed dairy sheep [34]. The increase in β-oxidation is further supported by an upregulation of SLC25A20 and SLC25A34 in REST, which are two members of SLC25 mitochondrial carrier family. SLC25A20 transports carnitine and carnitine-FA complexes across the inner mitochondrial membrane. SLC25A34 is supposed to act in a similar way, but its exact function still is not known totally [39]. MG seems to spare glucose (downregulating glycolysis) and promote FAs as an energy source (upregulating β-oxidation; Figure 4) in order to adapt to underfeeding. The mammary expression of genes involved in lipid metabolism is also modified in comparison with NEB (induced by caloric restriction) and the positive energy balance of cows after the peak in lactation [40]. Interestingly, the downregulation of genes linked to fat metabolism (FA biosynthesis) is observed 24 h after E. coli infection in MG of lactating cows [18]. This suggests that inflammation downregulates the FA biosynthesis and may increase the use of preformed FA, derived from other sources, such as from the mobilization of adipose tissue. These results suggest that energy metabolism modifications, in response to inflammation, are more marked in REST than in CONT, probably due to the expected limited availability of nutrients to support an acute inflammation in REST.
DEG Involved in Immune Response
The objective of this study was to identify the effects of an aggravated NEB on MG transcriptomic responses to inflammation. The experiment was not designed to compare gene expression in normal versus LPS challenged MG, which have been previously reported [2,11,18,31]. Microarray analysis shows that, the immune response together with the metabolic processes, are the main biological processes modified by undernutrition after LPS challenge (Figure 2), with modifications of genes different from those selected for RT-qPCR analyses. However, immune response was not the main biological process to be affected by restriction during inflammation. Among the upregulated genes is PGLYRP3, which belongs to a family of innate immunity pattern recognition molecules that are activated by LPS and bactericidal and bacteriostatic properties and are activated by LPS [41]. When the invading bacteria survive, neutrophil infiltration is replaced with T and B lymphocytes and monocytes [42]. The upregulation in REST of KLF13, PLEKHA2, WC7, and MBP, involved in the immune response by activating T and B lymphocytes, therefore, is in line with the expect recruitment of leukocytes by MG ( Figure 5). Additionally, the upregulation of PLEKHA2 in REST, a gene involved in the cell adhesion process [43], suggests that there is an increased migration of B leucocytes. Together, the upregulation of these genes suggests a different nature or a higher response to LPS stimuli in REST compared with CONT. Nevertheless, the deregulation of TRIB2 and CXCR7 suggests that the immune process could be impaired. Indeed, both genes participate in the activation of immune cells and influence IL-8 production, a chemokine upregulated in response to infection [10], where the concentration increased in milk, within 4 h after LPS infusion [ (25], and within 16-24 h after experimental infection with different strains of E.coli or LPS infusion [44,45]. The upregulation of TRIB2 and the downregulation of CXCR7 genes, however, suggests a potential IL-8 production alteration in response to inflammation in REST. These results contrast with the expected inflammatory response and could be a sign of deficient immune function under exacerbated NEB. Taken together, differences in gene expression might suggest a modified resolution of inflammation in response to LPS, due to aggravated NEB. During the course of an experiment with LPS, the inflammatory response usually declines within 24 h [44]. The REST cows might have experienced difficulties in restoring the MG homeostasis by 24 h after LPS challenge, due to the metabolic changes inherent in nutrient deficiency. However, this conception needs more detailed investigation, with a kinetic analysis, to be confirmed.
Proteins Involved in Protein Synthesis
Among the 53 differentially expressed proteins (DEP) in REST compared to CONT, 43 were downregulated. Most of these are involved in RNA and protein metabolism, with roles that vary from RNA splicing to translation. The downregulation of proteins involved in the splicing process, such as HNRNPH1, HNRPC, HNRNPA3, PCBP2, YBX1, SNRPA1, and DHX9, suggests that splicing is impaired in the MG of REST cows. This could explain in part the reduced synthesis and secretion of milk protein in REST compared with CONT [25]. Moreover, altered splicing and translation mechanisms might have a profound influence on protein biochemical properties and, ultimately, alter immune response to pathogens. Among the four proteins belonging to the HRNPs family (HNRNPH1, HNRPC, HNRNPA3, PCBP2), the first three are RNA binding proteins associated with pre-mRNAs in the nucleus, influencing pre-mRNA processing as well as other aspects of mRNA metabolism and transport. The dysfunction of HRNPs is linked to different proliferative and degenerative diseases [46], but the role of these proteins in the inflammatory response is still not fully understood. Some members of this family are reported to ensure resolution of inflammation [47]. However, the role of HRNPs could be associated with a mammary remodeling due to the restriction. In our study, YBX1 and DHX9 are downregulated in REST and the downregulation of these two genes is associated with impaired inflammatory responses [48]. Additionally, the loss of PP2AC function causes severe immunological disorders in Treg cells [49]. Thus, the downregulation of all these proteins in REST ( Figure 6) suggest a modified inflammatory response in underfed early lactation cows. A number of proteins involved in translation are downregulated in REST ( Figure 6; RPS27A, RPS15, RPS2, RPL10, EIF3H, RPN2, RPN1, FARSB, and NACA). Four riboprotein family members (3 RPS and 1 RPL) are part of a ribosome. Interestingly, protein building ribosomes alone are shown to affect the other cell processes outside the ribosome like development, apoptosis, and aging during their altered expression levels [50]. Additionally, the decrease of RPN1 and RPN2, which catalyze co-translational N-glycosylation, suggests an impaired post-translational protein modification process in the REST group. This process may play an important role in the immune system by creating the glycans on an immune cell s surface that helps migration of the cell or by glycosylating the various immunoglobulins [51]. Elsewhere, it is reported that this process can be defective during glucose deficit, leading to a reduction of protein glycosylation and harmful accumulation of unfolded proteins [52]. The decrease in RPN1 and RPN2, observed in the current study, could potentially lead to the creation of misfolded proteins in MG of REST cows.
Additionally, protein folding and its control might be modified in REST, due to the downregulation of chaperone proteins such as PDIA3, PDIA4, and CCT4 ( Figure 6). PDIA3 and PDIA4 are part of a larger super-family of a disulfide isomerase family of endoplasmic reticulum proteins that catalyze protein folding [53]. PDIA3 contributes to the correct folding of glycoproteins [54]. The loss of PDI activity and the consequent accumulation of misfolded proteins are associated with chronic inflammation [54,55]. Moreover, PDIA3 is a structural component required for the stable assembly of the peptide-loading complex of the major histocompatibility complex class I pathway. Its activity seems to play a role in lymphocyte T and B function [56]. Added to its role in folding, PDIA4 promotes the immunoglobulin G intermolecular disulfide bonding and antibody assembly in vitro [57]. Because CCT4 assists in the folding of newly translated polypeptides, this function might have been altered in REST [58]. Overall, proteomic data strongly suggest that protein synthesis is impaired by undernutrition at different levels (translation, folding, and post-translation modifications). The modifications of protein metabolism might partially explain the lower milk protein yield from 1.12 to 0.62 kg/ observed during restriction.
DEP Involved in Immune Response
Undernutrition downregulated FARSB protein expression. The decrease of this protein is linked with impaired acute inflammation responses in mice [59], suggesting an impairment in immune system function. In contrast, there was an upregulation of proteins such as SERPINA3, SERPINA3-5, F1MLW8, and Q1RMN8. SERPINA is an acute-phase protein, whose concentration can rise during acute and chronic inflammation [60]. F1MLW8 and Q1RMN8 proteins are similar to immunoglobulin lambda and typical for B-cells, and are important for its maturation from pre-B cells to mature ones [61]. The increase of these four proteins in REST MG, suggests that the resolution inflammation process was delayed in REST, compared to CONT 24 h after LPS challenge, whereas it could be considered that, potentially, it has already resolved in CONT. This is in line with the reported peak in SCC at 12 h that declines 24 h after LPS challenge [44].
The decreased translation process and post-translational protein modification (folding and glycosylation), that is observed at the protein level, might alter protein synthesis and activate an unfolded protein response [52]. This role could also be suggested to affect the proteins involved in the immune response.
Ethics Statement, Treatments and Sampling
The cows were housed at the Herbivore Research Unit of INRA Research Center of Auvergne-Rhone-Alpes. Animal procedures were performed in compliance with Regional Animal Care Committee guidelines CEMEAA: Auvergne, French Ministry of Agriculture and European Union guidelines for animal research C2EA-02. All procedures were approved by the regional ethics committee on animal experimentation (APAFIS #2018062913565518). The animals were in their second to the fourth days of lactation, with a body condition score (BCS) of 2.0 to 2.2 (0 to 5 scale), a week before feed restricted diet. All animals were observed for uterine disease and did not present any signs of abnormality. Additionally, the health history of each animal was inspected and only those without any health problems, within the last 6 months before calving, were chosen. At 24 ± 3 days in milk, sixteen multiparous Holstein cows were allowed ad libitum intake of a lactation diet CTRL, n = 8, 7.1 MJ/kg DM NEL, 17.4% Crude Protein. Their diet constituted of corn (24.2% dry matter), corn silage (29%), grass silage (25.5%), soybean meal (16.9%), and complemented with vitamins and minerals (0.9%). The underfed (REST) group received a ration diluted with barley straw (48% DM) for 96 h (RES, n = 8; 5.16 MJ/kg DM NEL, 12.2% CP). Therefore, the ratio of forage to concentrate differed from 58.0/42.0 in control (CONT) group to 79.2/20.8 in REST group [25]. Dry matter intake, milk yield, energy balance, plasma insulin, glucose, non-esterified fatty acids (NEFA) and BHB (β-hydroxybutyrate) concentrations did not differ between CONT and REST immediately before underfeeding (21.8 kg/day, 39.0 kg/day, -5.6 MJ/day, 22 µIU/mL, 3.78, 0.415, and 0.66 mM, respectively, at day -1), but were significantly altered in REST at 72 h of underfeeding (Supplementary Table S1). Following 72 h of restriction or control diet, one healthy rear mammary quarter was injected with 50 µg of lipopolysaccharide E. coli 0111:B4; (LPS-EB Ultrapure, InvivoGen, San Diego, CA, USA) diluted in 10 mL of sterile saline (CDM Lavoisier, Paris, France), containing 0.5 mg/mL BSA cell culture grade, endotoxin free, A9576, (Sigma-Aldrich, St. Louis, MO, USA), using a sterile disposable syringe fitted with a sterile teat cannula. Mammary biopsies were performed 24 h after the LPS injection, as previously described [62], corresponding to 96 h of feed restriction or control diet. Tissue samples were immediately frozen in liquid nitrogen and stored at −80 • C prior to RNA and protein analyses.
Throughout the study, milk samples were collected at 4 consecutive milkings each week before the beginning of the restriction and just before the LPS challenge and analyzed for SCC. Only healthy cows were included in the study. Additionally cows were screened for mastitis, one week before and immediately before the LPS challenge, using the California Mastitis Test (Neodis, Rambouillet, France) for all quarters, and somatic cell counts of rear quarter milk samples (Galilait, Theix, 63122 Saint Genès-Champanelle, France), one week before and immediately prior to the LPS challenge. Only cows with SCC lower than 100,000 cells/mL, in a rear quarter, were included in the study. Indeed, cows were considered healthy if the quarter SCC was inferior to 100,000 cells/mL and were free of any other signs of health problems [25]. Additionally, foremilk samples were collected from the LPS challenged quarters, immediately before the morning milking that preceded the LPS injection (time 0), and at 4, 6, 10, and 24 h after LPS injection. These quarter milk samples were analyzed for IL-8, IL1-β, TNF-alpha, and CXCL3 using Elisa [25].
RNA Preparation and Analyses
RNA and protein extractions were performed from the same mammary biopsy samples n = 16 animals, (8 CONT and 8 REST). The total RNA was extracted from 50 mg of the mammary gland (MG) by using the mirVana miRNA Isolation Kit (Thermo Fisher Sciences, Waltham, MA USA). The concentration and purity of RNA were estimated by spectrophotometry NanodropTH, (ND-1000, NanoDrop Technologies LLC, Wilmington, DE, USA), and by using the Bioanalyzer 2100 (Agilent Technologies Inc., Santa Clara, CA, USA), respectively. Once these validation steps were completed, only 12 cows (6 RES and 6 CTR) were kept for gene expression analyses at mRNA level, which presented a good and uniform quality of the samples for a microarray experiment.
RT-PCR Analyses
Reverse transcription (RT) was performed on 2 µg of total RNA using the "High Capacity RNA to cDNA" kit and following the manufacturer's recommendations (Applied Biosystems, Villebon Sur Yvette, France) in a final volume of 20 µL. In parallel, negative controls were performed without the matrices. The primers are described in Table 3. The genes, UXT2, CLN3, and EIF3K were used as housekeeping genes [63]. Real-time quantitative PCR was performed on the StepOnePlus™ PCR System (Applied Biosystems, Villebon Sur Yvette, France), using 5 µL of 50 fold-diluted single-stranded cDNA and the TFPower SYBRGreen PCR Master Mix, according to the manufacturer's instructions (Applied Biosystems, Villebon Sur Yvette, France). Subsequent to an initial denaturing step (95 • C for 10 min), the PCR mixture was subjected to the following two-step cycle, which was repeated 40 times: Denaturing for 15 s at 95 • C and annealing and extension for 45 s at 60 or 62 • C. The results were expressed as a fold change of Ct values relative to the control using the ∆ Ct method [64]. The significance was determined using a t-test with p < 0.05 considered as significant.
Microarray Analyses
Microarray analyses were performed on twelve animals (6 RES and 6 CTRL) using 100 ng of total RNA from each MG sample. The total RNA was amplified, fluorescently labeled, and hybridized to the bovine 4 × 44K microarray (Agilent Technologies, Inc. Santa Clara, CA, USA), and all the procedures described below were performed according to the manufacturer instructions (Agilent Technologies, Inc. Santa Clara, CA, USA). Briefly, for each hybridization, the total RNA was linearly amplified and labeled with Cy3 using the one-color Low Input Quick Amp Labeling Kit. Then, 1650 ng of Cy3-labeled cRNA was hybridized on the microarrays using the Gene Expression Hyb Kit. Hybridization was performed for 17 h at 65 • C in a rotating hybridization oven at 10 rpm. Following hybridization, all microarrays were washed and scanned using the Agilent Microarray Scanner G2565A. The resulting TIFF-images (Tagged Image File Format) were processed using Feature Extraction software Version 11 to obtain normalized data. Normalized with 75th percentile shift, the data were analyzed using GeneSpring software. The moderated t-test with Westfall-Young familywise error rate (FWER) correction was applied [70]. The differences were considered significant at an adjusted p < 0.05. The data were accessible through the GEO series accession number GSE114975. The classification and functional analyses of differentially expressed genes were performed using PANTHER [71] and confirmed using Pathway Studio ® software (Elsevier, The Netherlands).
Protein Preparation and Analyses
The proteins were extracted by homogenizing 80 mg MG tissue (n = 16; 8 RES and 8 CTRL) in 2 mL lysis buffer (8.3 M urea, 2 M thiourea, 2% CHAPS, 1% DDT). Following homogenization, the samples were incubated for 5 min at room temperature and centrifuged at 10,000× g for 30 min at 8 • C. The protein concentrations were measured in supernatant with Quick Start Bradford protein assay (BioRad, Marnes-La-Coquette, France), aliquoted and then stored at −20 • C, until further preparation. Sample supernatants were mixed with 1 volume of Laemmli buffer and heated at 60 • C for 5 min. Separation, by SDS-PAGE (12% acrylamide), was performed using a Mini-Protean II electrophoresis unit (BioRad, Marnes-La-Coquette, France) and 100 µg protein loaded per lane. To concentrate the samples, the gels were run at 80 V until the dye front reached the bottom of the concentration gel. Gels were stained overnight in Coomassie brilliant blue G-250. Excised lanes were reduced and alkyled before de-staining in 25 mM ammonium bicarbonate with acetonitrile (50/50 v/v). Following dehydration with 100% acetonitrile, gel pieces were dried in a Speed Vacuum and the samples were preserved at −20 • C until LC MS/MS analysis.
LC MS/MS Analysis
The proteins were hydrolyzed overnight at 37 • C, using 800 ng (80 µL) of sequence grade-modified trypsin (Promega, France) per band. Subsequent to extraction by 64 µL of acetonitrile 100% and sonication, the peptides were concentrated in a Speed Vacuum and volume was adjusted to 30 µL with an aqueous solution (99.9% H2O, 0.1% TFA). Peptide mixture (2.5 µL) was injected into the nano HPLC Ultimate 3000, (Thermo Fisher Scientific, Courtaboeuf, France) after a preliminary step of desalting and concentration in the pre-column 300 µm × 5 mm, (ThermoFisher, Courtaboeuf, France) for 6 min, and a second step of separation in an analytical C18 column 75 µm, 25 cm, (Pepmap Thermo Fisher Scientific, Courtaboeuf, France) with a 10-40% gradient (A: 0.1% FA in water, B: 0.1% FA in acetonitrile) at 450 nL/min. The eluate was electrosprayed through the CaptiveSpray ion source into the mass spectrometer QTOF impact II (Bruker, Wissembourg, France) operated in CID Data Dependent mode. Each MS analysis was succeeded by as many MSMS analysis as possible within 3 s.
Protein Identification and Label-Free Quantitation
The raw files were loaded, at the end of each LC-MS/MS analysis, into the Progenesis QI software Non-linear Dynamics, v 4.1 (Newcastle upon Tyne, UK) and label-free quantitation was performed using a proprietary workflow alignment, peak picking, normalization, design set up, quantitation, and protein identification.
Regarding protein identification (Supplementary File S2), the Mascot V.2.5, internally licensed version (www.matrixscience.com) was used with uniprot-ref_Bos taurus database 19.840 sequences (07/2015). The following parameters were considered for the searches: Peptide mass tolerance was set to 10 ppm; fragment mass tolerance was set to 0.05 Da and a maximum of two missed cleavages was allowed. Variable modifications were methionine oxidation (M), carbamidomethylation (C) of cysteine and Deamidated (NQ). Protein identification was considered valid, if at least two peptides with a statistically significant Mascot score were assigned, with a false discovery rate (FDR) less than 1%.
Concerning label-free quantitation, all unique validated peptides of an identified protein were included, and the total cumulative abundance was calculated by summing up the abundances of all unique peptides allocated to the respective protein. Statistical analysis was performed, using the "between subject design," and the p-values were calculated by an analysis of variance, using the normalized abundances across all runs. Differential proteins were conserved for interpretation if the peptides' individual abundances showed a good correlation with protein abundance. All differential proteins were inspected manually with these correlation criteria. To extract the maximum biological information of differentially expressed proteins, PANTHER [71], Pathway Studio ® software (Elsevier, The Netherlands) and UniProt [72], were used.
Conclusion
Undernutrition affected multiple aspects of MG function, as demonstrated by modifications of milk secretion, and MG mRNA and protein expression. During this study, expression analyses were performed 24 h post-LPS challenge corresponding to the period of inflammation resolution. The effects of undernutrition on studied candidate genes, known as major genes relating to the innate immune responses, were weak. Therefore, the transcriptomic and proteomic analyses pointed out modifications of energy metabolism (fatty acid and glucose), and protein metabolism (synthesis and post-translational modification), respectively, but relatively few genes involved in immune response were affected. Our nutrigenomic analyses have suggested that undernutrition of early lactating cows modified the mammary gland metabolism. The holistic analyses of the systemic reaction in the mammary gland expands the knowledge of the effects of NEB and metabolic imbalance occurring in early lactation, during inflammation. These identified genes may be relevant for quantitative trait loci studies and genomic selection.
Supplementary Materials: Supplementary Materials can be found at [URL]/1422-0067/20/5/ 1156/s1. Table S1: Plasma insulin and metabolite concentration at the day of biopsy after dietary treatment and response to LPS challenge. Occurring at day 24 ± 3 of lactation, animals were assigned to a control (CONT, n = 8) or restricted (REST, n = 8) group. The REST animals received the ration diluted with barley straw (48% DM) for 4 days when cows from CONT were allowed to continue ad libitum intake of a lactation diet (7.1 MJ/kg DM NEL, 17.4% CP). Occurring on day 3, corresponding to the 27th day of lactation, the rear mammary quarter of animals from both groups was injected with 50 µg of LPS. Mammary biopsies were performed 24 h after LPS challenge. p < 0.01 for all variables. File S2: Protein analysis report.
Funding: Financial support for this research was provided by GISA meta-program of INRA (Ruminflame project).
Acknowledgments:
The authors thank the staff at Herbipole INRA (UE1414, Theix, France), Simon Collange for performing biopsies, Sebastien Bes, Arnaud Delavaud, Caroline Soulard, Martine Tourret, Didier Bany, Emile Tixier, Cyril Labonne, and Jacques Rouel for technical assistance, and for sample collection and laboratory analyses. The authors are grateful to the Galilait laboratory (Clermont-Ferrand, France) for the milk component, SCC and microbiological analyses. Special appreciation is extended to Gilles Foucras and Pascal Rainard for their helpful discussion regarding LPS and choice of candidate genes. Transcriptomic analyses were performed at the "PlateForme d'Exploration du Métabolisme" (PFEM, INRA, Theix, France).
Conflicts of Interest:
The authors declare no conflict of interest.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2019-03-11T17:25:24.133Z", "created": "2019-03-01T00:00:00.000", "id": "73513089", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1422-0067/20/5/1156/pdf", "pdf_hash": "eaf318eeaf20453438e8cf4662832f0854f4be4b", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:3553", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "eaf318eeaf20453438e8cf4662832f0854f4be4b", "year": 2019}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Fine mapping and candidate gene analysis of qTAC8, a major quantitative trait locus controlling tiller angle in rice (Oryza sativa L.)
Rice tiller angle is an important agronomic trait that contributes to crop production and plays a vital role in high yield breeding. In this study, a recombinant inbred line (RIL) population derived from the cross of a glabrous tropical japonica rice D50 and an indica rice HB277, was used to investigate quantitative trait loci (QTLs) controlling rice tiller angle. Two major QTLs, qTAC8 and qTAC9, were detected. While qTAC9 mapped with a previously identified gene (TAC1), using a BC2F2 population qTAC8 was mapped to a 16.5 cM region between markers RM7049 and RM23175. Position of qTAC8 was narrowed to a 92 kb DNA region by two genetic segregating populations. Finally, one opening reading frame (ORF) was regarded as a candidate gene according to genomic sequencing and qRT-PCR analysis. In addition, a set of four near isogenic lines (NILs) were created to investigate the genetic relationship between those two QTLs, and one line carrying qTAC8 and qTAC9 presented additive effect of tiller angle, suggesting that these QTLs are involved in different genetic pathways. Our results provide a foundation for the cloning of qTAC8 and genetic improvement of the rice plant architecture.
Introduction
Rice (Oryza sativa L.) is one of the most important food crops in China and the world. It is the most effective safeguard for food security and agricultural sustainable development through high-yielding rice breeding. Ideotype breeding strategy is an important approach to increase grain yield potential in rice breeding. Tiller angle, the angle between the main culm and its side tillers [1], is a decisive factor for building ideal plant architecture, whereby neither spreadout rice nor compact type rice is beneficial for grain production [2]. With a spread-out architecture, plants can decrease humidity and escape from some diseases, but they occupy too much space and increase shading and lodging, consequently decreasing photosynthetic efficiency and grain yield per unit area. On the other hand, compact plants prejudice in capturing light and prevention of plant diseases and insect pests, thus appropriate tiller angle is beneficial for improving rice production [3,4]. Although rice tiller angle has long attracted attention of a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 breeders due to the significant contribution to plant architecture and yield potential, the genetic mechanisms determining its characteristics are not fully understood.
Rice tiller angle has been recognized as a complex quantitative trait, which is not only controlled by genetic factors but also greatly influenced by environmental conditions, such as light intensity, climate, soil, planting density, watering and fertilizing [4]. In the past two decades, a number of QTLs for tiller angle have been identified on chromosomes 1,2,5,7,8,9,11,12 in various mapping populations of rice. Using an F 2 population helped to identify three major genes controlling tiller angle [5]. A major QTL (ta9) on chromosome 9 flanked by RZ228 to RG667 together with other four QTLs (QTa1, QTa2, QTa3, QTa8) were separated in a F 2:4 genetic segregating population generated from a cross between Lemont and Teqing [6]. In addition, a doubled haploid population generated from a cross between Zhaiyeqing 8 (loose plant architecture) and Jingxi 7 (compact plant architecture) was used and a total of three controlling tiller angle QTLs named qTA-9a, qTA-9b, qTA-12, that account for 22.7%, 11.9% and 20.9% of the variance, respectively, were detected [7]. Moreover, two major QTLs, qTA8-2 and qTA9-2 were determined in a recombinant inbred line population derived from Xieqingzao B/ Miyang 46, and no G×E interaction effect was detected for the additive effect of these two QTLs [1]. Five major QTLs including qTA-9, qTA-2, qTA-7a, qTA-7b and qTA-11 for tiller angle were present in a RIL population from a cross between Asominori and IR24. Then a major QTL, qTA-9, was singled out in a 15 cM region between RFLP markers C609 and C508 by using a CSSLs population [8]. To date, only two major QTLs for tiller angle have been cloned. One is TAC1, a major QTL for tiller angle was isolated by using a large F 2 population that derives from the cross between indica rice IR24, and an introgression line IL55. TAC1 harbors three introns in its coding region and a fourth 1.5 kb intron in the 3' untranslated region, and encodes a 259-amino-acid unknown protein. A mutation in the 3' splicing site of the fourth intron from 'AGGA' to 'GGGA' decreases its transcript levels, resulting in compact plant architecture. tac1 has also been extensively utilized in densely planted rice [4]. The second cloned QTL for tiller angle corresponds to PROG1 (PROSTRATE GROWTH 1), a semidominant gene encoding a newly identified Cys2-His2 zinc-finger transcription factor located on chromosome 7. PROG1 is predominantly expressed in the axillary meristems and the site of tiller bud formation, and disrupting the prog1 function and inactivate prog1 expression can lead to erect growth, increasing grain number and higher grain yield in cultivated rice [9,10]. Dong identified three major QTLs (TAC3, DWARF2 (D2) and TAC1) controlling tiller angles by genome-wide association studies. TAC3 encodes a conserved hypothetical protein of 152 amino acids that is preferentially expressed in the tiller base [11].
Previous reports indicate that the tiller angle is not only affected by QTLs but also could be controlled by single genes following the Mendel's genetics. Thus far, several genes controlling tiller angle have been cloned. LAZY1 (LA1), a novel grass-specific gene, is expressed during gravitropism sensitivity and plays negative role in polar auxin transport (PAT). Loss-of-function of LA1 enhances PAT causing alteration of endogenous IAA distribution in shoots, leading to reduced gravitropism and a tiller-spreading phenotype of rice plants [12,13]. LPA1 (Loose Plant Architecture 1), identified on chromosome 5, encodes a plant-specific INDETER-MINATE DOMAIN protein that regulates tiller angle by controlling the adaxial growth of tiller nodes [3]. PIN-FORMED1 and PIN-FORMED2 are auxin efflux transporters, and suppressing the expression of rice PIN-FORMED1 or enhancing the expression of rice PIN-FORMED2 can alter PAT and increase tiller angles [14,15].
To further elucidate the genetic control of tiller angle, we used a RIL population that derived from a cross between a japonica and an indica rice cultivars to pinpoint two major QTLs for this trait. qTAC8 and qTAC9 were detected, and qTAC8 was narrowed down to a 92 kb region where one candidate ORF was determined as the one encoding for qTAC8.
Plant materials
A recombinant inbred line (RIL) population of 190 lines was generated from the cross between D50 and HB277 as described previously by Shao [16]. Briefly, one line (RIL-77) that carried the homozygous target segments RM339-RM210 on chromosome 8 and RM201-RM7306 on chromosome 9 from HB277 (S1A Fig) was chosen from the RILs to backcross with D50. The resultant BC 2 F 1 was selfed to obtain a BC 2 F 2 population containing 178 plants for genotyping and phenotyping, and a progeny (BC 2 F 2:3 ) was used for phenotyping (S1B Fig). The RILs and BC 2 F 2 population were grown at the experimental site of Fuyang district of Hangzhou in 2012, and the BC 2 F 2:3 plants were grown at the experimental site of Lingshui (Hainan Province, China) and Fuyang district in 2013, respectively. Six plants per row were transplanted with a distance of 18 cm between plants within a row, and 18 cm between rows, and four rows were used to grow each line.
Two large segregating populations BC 2 F 4 and BC 2 F 5 carrying the heterozygous target segment RM339-RM210 on chromosome 8 and the D50 target segment RM201-RM7306 on chromosome 9, were picked for screening recombinant individuals in our target region. As a result, the homozygous recombinants were used for genotyping and phenotyping. One plant from BC 2 F 1 carrying the heterozygous target segment RM339-RM210 and the HB277 target segment RM201-RM7306 was selected to consecutive backcross with D50 and produce BC 4 F 1 and 120 BC 4 F 2 individuals. Finally four NILs designated NIL-qTAC8qtac9, NIL-qtac8qtac9, NIL-qTAC8qtac9 and NIL-qTAC8qTAC9 were developed. This set of four NILs was planted in the experimental site of Lingshui in 2014 following a randomized block design with three repeats. Each line was grown in a six-row plot with 6 plants in each row and spacing of 18 x 18 cm.
PCR and development of molecular markers
DNA was extracted following the protocols described by Murray and Rogers [17,18]; Each 10 μL PCR reaction system contained 1 μL 10X PCR buffer, 1 μL dNTP (2 mM), 1μL of primer (1 mM), 0.1 μL Taq DNA polymerase (5 U/μL) and 1 μL template DNA. Polymerase chain reaction (PCR) comprised an initial denaturation step (95˚C for 3 min), followed by 35 cycles of 95˚C for 30 s, 55˚C for 30 s and 72˚C for 45 s, and ending with an extension step of 72˚C for 10 min. PCR products were separated by electrophoresis and silver staining procedure. Simple sequence repeats (SSR) markers were selected covering the target region based on the published linkage map of rice ( [URL]). InDel (Insertion and Deletion) markers used for fine mapping of qTAC8 were designed based on the reference Nipponbare and 93-11 genomic sequences.
Sequencing and identification of candidate genes
The target gene in the candidate genomic region was predicted using the SIGnAL package ( [URL]/). The full-length genomic DNA sequence of the candidate gene was determined by dividing it into several overlapping segments. Sequencing primers were designed according to the sequence of cv. Nipponbare in the target region. The PCR products were sequenced directly. Primer sequences used in this study are listed in S1 Table. RNAiso Plus (Takara), following the manufacturer's instructions. The first cDNA strand was synthesized from 3 ug RNA using the First Strand cDNA synthesis Kit-Rever Tra Ace-α (ToYoBo). qRT-PCR analysis was performed on a Roche Light Cycler 480 device using various gene-specific primers. The rice Ubi gene (LOC_Os03g0234200) was chosen as reference gene. Reactions containing SYBR premix Ex TaqII (TaKaRa) were carried out in a final volume of 20 ul. The 2 -44CT method [19,20] was used to calculate relative levels of transcription. The PCR reaction implied an initial denaturation step (95˚C for 4 min), followed by 50 cycles of 95˚C for 15 s and 55˚C for 30s. Three technical replicates were analyzed for each cDNA sample.
Data analysis
Rice linkage maps were constructed using MAPMAKER/Exp Ver. 3.0, and genetic distances were converted into cM by using the Kosambi function. Composite interval mapping (CIM) analysis of QTL in the RILs and the BC 2 F 2 population was performed with QTL cartographer Ver. 2.5 (statgen.ncsu.edu/qtlcart/WQTLCart.htm). QTLs were called where their LOD value exceeded 2.5. Mean phenotypic values were compared using the Student's t test. Multiple comparison test and the correlation between genotypes and phenotypes were carried out by the SAS statistical software package.
Primary mapping of tiller angle
Rice tiller angle serves as an important trait in rice. In this work, a glabrous tropical japonica rice D50 cultivar which exhibits relative compact plant type and an indica rice HB277 cultivar displaying a relative loose architecture were used in this study (Fig 1A and 1B). Then a RIL population was used to detect QTLs for this trait. Measurement of the tiller angle showed a continuous and normal distribution whose variation range was 0.267-1.010 rad (S2 Fig). QTL mapping strategy was conducted and two major QTLs for tiller angle named qTAC8 on chromosome 8 and qTAC9 on chromosome 9 were identified in this RIL population. qTAC8 and qTAC9 were mapped within the region of RM339-RM210 and RM201-RM7306, respectively. qTAC8 could account for almost 33.4% of the variance, while qTAC9 could explain around 17.4% of the variance in tiller angle ( Table 1). The positive allele (increasing tiller angle) qTAC9 is the same gene as TAC1 In accord to reports in the literature, we found that qTAC9 was closely linked to TAC1 flanked by the SSR loci RM201 and RM1026 [4], and a SNP in the 3' splicing site of the fourth intron of TAC1 was detected after sequencing the TAC1 alleles in D50 and HB277 (S3A Fig). qRT-PCR analysis showed that the expression of TAC1 in HB277 is significant higher than D50 (S3B Fig). These results coincided with the report that a mutation in the 3' splicing site of the fourth intron from 'AGGA' to 'GGGA' can decrease the expression of TAC1 and lead to a compact plant. Hence, we can anticipate that qTAC9 and TAC1 may correspond to the same gene.
Characterization of qTAC8
In order to investigate qTAC8, we measured the tiller angle of NIL-qTAC8 and NIL-qtac8. The results showed that the tiller angle of NIL-qTAC8 is significant larger than that in NIL-qtac8 at the ripening stage, and the angle of tiller base in NIL-qTAC8 is significant larger than that in NIL-qtac8 (Fig 2A and 2B). We also found that the two NILs almost exhibited the same tiller angle at tillering stage, but NIL-qTAC8 showed a greater tiller angle than that in NIL-qtac8 since heading stage (Fig 2C). There were no significant differences in tiller numbers, plant height and spikelet fertility between NIL-qTAC8 and NIL-qtac8 (data not shown). Fine mapping of qTAC8 To further investigate the QTL qTAC8, a BC 2 F 2 population containing 178 individuals was established (S1B Fig). The tiller angle of the BC 2 F 2 population showed a normal distribution.
According to the progeny test, the BC 2 F 2 individuals could be classified into three subgroups of homozygotes for D50 (DD), for HB277 (HH) and heterozygotes at the targeted region. According to the tiller angle performance in the progeny test, paired t test was used to compare the difference between subgroups. Significant differences occurred between DD and the other two subgroups, HH and DH, suggesting that the DD genotype presented a larger tiller angle (0.462 rad) than HH (0.294 rad), while the tiller angle of the heterozygotes was an intermediate value ( S4 Fig). These results showed that qTAC8 is a QTL in D50 that presents a partial dominance and a positive additive effect. Seven more SSR markers were used to genotype the BC 2 F 2 population and to construct a local linkage group covering 26.2 cM. The phenotype of two populations BC 2 F 2 and BC 2 F 2:3 were investigated in this study and qTAC8 was mapped within a 16.5 cM interval flanked by RM23097-RM23201. qTAC8 could account for 26.3% of the phenotype variance with additive effect of 0.07 rad in BC 2 F 2 population, while in the BC 2 F 2:3 , the QTL could explain 47.1% and 51.8% of the variation in Hainan and Hangzhou, respectively ( Table 2). In order to test and verify this, four informative homozygous recombinants were identified with four markers within RM23097-RM23201 and grouped into four genotypes according to the positions of recombinant breakpoints and allelic composition. Multiple comparisons between the tiller angle and recombinant individual genotypes, using the two non-recombinant lines as controls (h9 and h5), reflected that qTAC8 was narrowed down to a 1.4 cM interval flanked by RM7049 and RM23175 (Fig 3A and 3B).
To further fine mapping qTAC8, a segregating population with 2,000 individuals derived from BC 2 F 4 lines that carried a heterozygosis segments at the qTAC8 region, were used to identify the recombinants between RM7049 and RM23175 ( Fig 4A). Next, the identified recombinants were analyzed by seven more markers located between RM7049 and RM23175. Multiple comparisons were conducted and qTAC8 was placed in a 199 kb region between In12 and RM2767. To position qTAC8 more precisely, a large segregating population of 6,000 individuals was introduced from BC 2 F 5 , and a total of 40 recombinants were identified with the help of two markers In12 and RM2767, and four polymorphism markers within this region were developed to genotype these individuals. Multiple comparisons were also conducted between the genotypes of these homozygous recombinants and the phenotypes of their progeny. Finally, qTAC8 was placed spanning on BAC P0431A03, in a 92 kb region flanked by In2 and In36 (Fig 4B).
Analysis of candidate genes for qTAC8
Based on the genome annotated database ( [URL]/), the critical 92 kb region contains seven predicted ORFs. They encode a TCP family transcription factor, an ATPdependent Clp protease adaptor protein, a transposon protein, two retrotransposon proteins, a zinc knuckle family protein and a basic helix-loop-helix protein (Fig 4C and Table 3). First, the transposon protein and retrotransposon proteins were excluded as candidates for qTAC8, leaving ORF1, ORF2, ORF6 and ORF7. Genomic sequencing of these four candidate genes in D50 and HB277 revealed that nucleotide diversity occurred except for ORF2. No products of ORF6 genomic DNA were identified in two parents (data not shown), which might be explained by a deletion of this gene during rice evolution, thus ORF6 was also excluded as candidate gene for qTAC8 (data not shown). Quantitative real-time PCR was used to characterize the transcripts of the three candidate genes left in the tiller base at heading stage of NIL-qTAC8 and NIL-qtac8. No significant difference in expression was observed for ORF1 and ORF2, while RNA expression level of ORF7 in NIL-qTAC8 was significant higher than in NIL-tac8 (Fig 5A). This result indicates that ORF7 is the possible candidate for qTAC8. Next we also found that the expression level of ORF7 was also increased at heading and ripening stages, but not at tillering stage, which coincides with the phenotypes of one NIL pair NIL-qTAC8 and NIL-qtac8 at three developing stages (Fig 2C and Fig 5B). All together, these data suggest that ORF7 is the candidate gene for qTAC8. Its cDNA stretched 1038 bp, comprises two exons, and encodes a 345 amino-acid protein containing a putative bHLH conserved domain. Although we found that seven single nucleotide polymorphisms (SNPs) occurred in ORF7 between the NIL pair, they didn't cause any variation at the amino acid level. Further genomic sequencing results indicated that several nucleotides difference in the promoter and 3' UTR region of ORF7 might be responsible for the alteration of gene expression (S5 Fig). Expression analysis other tiller angle-related genes Tiller angle is known to be controlled by TAC1, LPA1, LAZY1 and PROG1. To investigate the expression pattern of those genes in the NILs used in this study, qRT-PCR analysis was Fine mapping and candidate gene analysis of qTAC8, controlling tiller angle in rice conducted. We found that LPA1, LAZY1 and PROG1 were all affected by the positive function of the ORF7 allele, whereas there was no difference in expression of TAC1 between the NIL pair (Fig 6).
Genetic relationship between qTAC8 and qTAC9
To further study the genetic relationship between qTAC8 and qTAC9, a set of four NILs including NIL-qTAC8qTAC9, NIL-qtac8qTAC9, NIL-qtac8qTAC9 and NIL-qtac8qtac9 was produced to evaluate tiller angle (Fig 7A-7E). The results showed that the tiller angle of NIL- Fine mapping and candidate gene analysis of qTAC8, controlling tiller angle in rice qTAC8qTAC9 was significantly larger than in the other three NILs, with NIL-qtac8qTAC9 and NIL-qtac8/qtac9 presenting the smallest. The tiller angle of NIL-qTAC8qTAC9 displayed an additive effect, indicating that qTAC8 and qTAC9 may be participating in different genetic pathways (Fig 7F).
Discussion
Plant ideotype has been recognized as an advanced breeding concept and is regarded to be highly associated with high grain yield in rice breeding [21]. Rice traits for plant ideotype include plant height, stem strength, leaf morphology, panicle morphology, tiller angle among other critical traits. Ideal Plant Architecture 1 (IPA1) was a major gene affecting rice productivity. Introduction of the IPA1 semi-dominant gene into the japonica rice Xiushui 11 cultivar resulted in increased seed yield [22]. Tiller angle also plays a central role in rice production formation, and appropriate tiller angle is beneficial for improving rice production [3,4]. Thus [URL]007 Fine mapping and candidate gene analysis of qTAC8, controlling tiller angle in rice exploration of new genes controlling tiller angle would facilitate strategies to manipulate plant ideotype and increasing rice yield.
In this study, we identified two major QTLs controlling tiller angle on chromosomes 8 and 9 which were named qTAC8 and qTAC9, respectively. Using an F 7 RIL population derived from a cross between D50 and HB277 (Table 1), qTAC9 was located between the SSR loci RM201 and RM7306 and the positive allele (increasing tiller angle) at qTAC9 originated from HB277 with a loose plant architecture (Table 1). Previous studies revealed that the target region near qTAC9 was a hot site controlling rice tiller angle on the long arm of chromosome 9 [1,4,6,7], and one QTL named TAC1 was isolated within this interval and cloned using a large F 2 population derived from a cross between indica rice IR24 and an introgressiong line IL55 [4]. Sequencing analysis indicated that a SNP ('AGGA' to 'GGGA') occurred between D50 and HB277, and qRT-PCR analysis revealed that the RNA level of TAC1 was significant reduced in D50 compared to HB277, suggesting that qTAC9 was the same allele of TAC1, a gene that can form a plant with spread-out architecture and is wildly used within indica rice varieties (S3A and S3B Fig). A partially dominant gene qTAC8 originated from D50 with a compact plant architecture and positive allele of the other QTL identified for tiller angle nearby. qTA8-2 which controls rice tiller angle was firstly mapped between R1394 and RZ66 [1], which locates on the same region with qTAC8, suggesting that qTA8-2 may be an allele of qTAC8. However primary QTL mapping, fine mapping, candidate gene prediction and qRT-PCR analysis indicated that ORF7, which encodes a basic helix-loop-helix protein, might be the gene underlying this QTL (Fig 4 and Table 3). Previous studies indicate that the semidominant gene PROSTRATE GROWTH 1 which affects plant architecture, also encodes a basic helix-loop-helix transcriptional factor, suggesting that qTAC8 might present a similar function as PROG1 [9,10].
Analysis of genetic interactions among genes for tiller angle is required to better understand pathways controlling rice tiller angle formation. qTAC8 (the positive allele from D50) and qTAC9 (positive allele from HB277) are two semi-dominant genes for tiller angle, and a double mutant NIL-qTAC8qTAC9 presented additive effect for that trait, suggesting that qTAC8 and qTAC9 are involved in different pathways. This was consistent with the similar expression of TAC1 observed in NIL-qTAC8 and NIL-qtac8 (Fig 6 and Fig 7). Interesting, we found that LAP1, LAZY1 and PROG1 were all down-regulated in NIL-qtac8 at ripening stage, implying that those genes might function in the same pathway with qTAC8 (Fig 6). In addition, qTAC8 encodes a predicted transcription factor, thus, whether qTAC8 acts directly or indirectly on LAP1, LAZY1 and PROG1 remains to be investigated.
Gene expression in plants can be basically of two types, constitutive or with specific expression pattern. Although some genes can be expressed during the whole plant life, it may function during specific growth stages and contribute to a certain phenotype. The genetic control of tiller angle is very complex and phenotypes can change during different developmental stages. The japonica rice ZH11 is a typical rice cultivar which displays large tiller angle at the seedling and tillering stages but usually presents a more compact architecture after the heading stage. The rice mutants lazy1 and prog1 present a large tiller angle during all growing periods, while lpa1 presents a large tiller angle at the seedling stage [3]. In this study, NIL-qTAC8 and NIL-qtac8 exhibit no differences in tiller angle at the tillering stage, but they show increasing TAC8 expression and TAC8 transcript levels are significant different during heading and ripening stages. It would be interesting to uncover the molecular mechanisms of how tiller angle is regulated during different rice development stages.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2018-04-03T02:29:19.135Z", "created": "2017-05-25T00:00:00.000", "id": "20482941", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0178177&type=printable", "pdf_hash": "d90db04c71bae1a85f1436b871573b41ba2e9aa7", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:4115", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "d90db04c71bae1a85f1436b871573b41ba2e9aa7", "year": 2017}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Wild and Domestic Differences in Plant Development and Responses to Water Deficit in Cicer
There is growing interest in widening the genetic diversity of domestic crops using wild relatives to break linkage drag and/or introduce new adaptive traits, particularly in narrow crops such as chickpea. To this end, it is important to understand wild and domestic adaptive differences to develop greater insight into how wild traits can be exploited for crop improvement. Here, we study wild and domestic Cicer development and water-use over the lifecycle, measuring responses to reproductive water deficit, a key Mediterranean selection pressure, using mini-lysimeters (33 L round pots) in common gardens under contrasting water regimes. Wild and domestic Cicer were consistently separated by later phenology, greater water extraction and lower water use efficiency (WUE) and harvest index in the former, and much greater yield-responsiveness in the latter. Throughout the lifecycle, there was greater vegetative investment in wild, and greater reproductive investment in domestic Cicer, reflected in root and harvest indices, rates of leaf area, and pod growth. Domestic WUE was consistently greater than wild, suggesting differences in water-use regulation and partitioning. Large wild-domestic differences revealed in this study are indicative of evolution under contrasting selection pressures. Cicer domestication has selected for early phenology, greater early vigor, and reproductive efficiency, attributes well-suited to a time-delimited production system, where the crop is protected from grazing, disease, and competition, circumstances that do not pertain in the wild. Wild Cicer attributes are more competitive: higher peak rates of leaf area growth, greater ad libitum water-use, and extraction under terminal drought associated with greater vegetative dry matter allocation, leading to a lower reproductive capacity and efficiency than in domestic chickpea. These traits strengthen competitive capacity throughout the growing season and are likely to facilitate recovery from grazing, two significant selection pressures faced by wild, rather than domesticated Cicer. While increased water extraction may be useful for improving chickpea drought tolerance, this trait must be evaluated independently of the other associated wild traits. To this end, the wild-domestic populations have been developed.
INTRODUCTION
Given genetic bottlenecks in most crop species, there is widespread interest in exploiting crop wild relatives (CWRs) that typically harbor much greater diversity (Tanksley and McCouch, 1997;Dempewolf et al., 2014;Gepts, 2014). Advocates of base broadening suggest that the greater genetic diversity of CWR will improve complex traits like yield by breaking linkage drag through the introgression of novel, previously unexploited alleles (Tanksley and McCouch, 1997), and/or be reflected in greater adaptive diversity (Dempewolf et al., 2017;Coyne et al., 2020). Indeed, ideas such as exploiting the climatic resilience of wild populations have gained considerable traction recently (McCouch et al., 2013;Dempewolf et al., 2014). Supporting this, there is an ever-increasing understanding of plant domestication in terms of history, geography, trait selection over time, and the genomic implications of these (Purugganan and Fuller, 2009;Meyer et al., 2012;Meyer and Purugganan, 2013;Abbo et al., 2014;Gepts, 2014;Larson et al., 2014). However, this work largely focuses on changes in the domesticate, rather than the attributes of the wild progenitor. As a result, when we want to turn to the wild relatives for crop improvement, we lack a framework to guide us. Genomic approaches maximizing genetic diversity to break up linkage drag advocated by Tanksley and McCouch (1997) are being deployed in breeding and pre-breeding programs in diverse ways from nested association mapping (von Wettberg et al., 2018) to MAGIC populations through to genomic selection (Bazakos et al., 2017). While these offer a methodology for increasing the genetic diversity of our breeding programs, they do not guide us in the identification of new adaptive traits and strategies that are missing in the cultigen, and which may or may not be present in the CWR (see discussion in Dempewolf et al., 2017). Thus, the big picture questions such as "where among wild populations do we find the adaptive traits with which to improve our crops and why?" remain important priorities (Thormann et al., 2014).
Clearly domestication has changed the adaptive capacity of crop compared to wild progenitor. In addition to the selection of domestication syndrome traits (lower seed dispersal and dormancy, larger seed size and more erect habit, and modified phenology), humans have altered many aspects of the cropping environment compared to that in which the progenitors evolved (Milla et al., 2015). Domesticated crops grow in managed (mostly) fields under reduced competition relative to the wild state (Gepts, 2014), their lifecycle is regulated (Milla et al., 2015), and they are not typically found as weedy escapes in the habitats, where their wild relatives occur. Ipso facto wild and domestic adaptive capacities do differ. But this does not mean that wild relatives will automatically contain all the adaptive traits lacking in the domesticate. Both wild and domestic plants are shaped by evolutionary responses to selection pressure which should be taken into account when evaluating their adaptive potential (Milla et al., 2015). There is a disconnect here between the agricultural and ecological approaches to this problem. While the former often samples widely (see examples in Singh et al., 1990Singh et al., , 1995Singh et al., , 1998, when screening for a shopping list of required traits ignores questions such as how representative is our collection and what selection pressures were imposed by the environment of origin, no greater insights into the biology of the species are made. While ecological approaches do address these issues, their experimental material may lack the depth to provide certainty. Recently, an argument has been advanced that under domestication crops evolved into resource-acquisitive, fast growing plants (competitors sensu Grime, 2006) as a result of their cultivation in well-managed, resource rich environments relative to those of their wild progenitors (Milla et al., 2015). By extension, wild progenitors should be relatively slower growing plants with traits tending toward the stress tolerator spectrum sensu Grime (2006). This is an interesting idea which has been tested quite widely across species, but with little depth (typically one wild and domesticated accession per species), and only in the early vegetative phase (Milla et al., 2014;Matesanz and Milla, 2018), where the strong selection for early vigor in domesticated crops (Evans and Dunstone, 1970;Berger et al., 2017) might be expected to influence the results.
These concepts are particularly tractable in wild and domestic Cicer. The genetic narrow base of chickpea as an adaptive constraint has been long recognized . Early work at the International Center for Agricultural Research in the Dry Areas (ICARDA) demonstrated that the wild Cicer had a wider, potentially useful range of responses to pests, diseases, and stresses than domestic chickpea, particularly for ascochyta blight, leaf miner, bruchids, cyst nematode, and vegetative cold (Singh et al., 1998). However, at that time the world collection of wild, Cicer was far too narrow to adequately characterize the adaptive potential of any species, including those that can readily cross with domestic chickpea, with only 18 independent accessions of Cicer reticulatum, the wild progenitor, and even less for its close relative, Cicer echinospermum . This has changed recently with extensive new collection across the habitat range of both these species in Southeastern Anatolia (von Wettberg et al., 2018) that is driving renewed interest in trait discovery in these CWRs (Reen et al., 2019;Newman et al., 2020).
In this paper, we investigate wild and domestic Cicer responses to reproductive water deficit (terminal drought), one of the principal selection pressures exerted by the Mediterranean climate, using mini-lysimeters in common gardens under contrasting reproductive water regimes. This approach allows us to compare wild and domestic responses to contrasting resource (water) availability as well as the underlying water-use patterns. This is followed up by a plant above-and below-ground development and water-use study to describe wild-domestic differences in greater depth across the lifecycle. We were interested to discover to what extent the wild and domestic Cicer would segregate along stress tolerator-competitor continuum of Grime (2006). Secondly, we were interested to explain a consistent field observation made during the collection that the annual wild Cicer species tend to have a longer lifecycle, reproducing and maturing considerably later than most of their sympatric annual plant competitors such as Lens, Pisum, and many Vicia species. To validate the evolution of "acquisitiveness" in domestication hypothesis, the following expectations should be met: Frontiers in Genetics | www.frontiersin.org 1. Lower growth rates, water-use, and above-and below-ground productivity in wild compared to domestic. 2. Improved stress tolerance in wild compared to domestic. 3. Lower response to resource (water) availability in wild compared to domestic. Table 1 for accession details) were randomized complete block designs (RCBDs, n = 3) randomized within water regime [terminal drought (TD) and well-watered (WW)]. Because of space limitations, the two water regime treatments were allocated individually in two contiguous areas along the rainout shelter bay separated by 2 m (see Figure 1 for layout). This was done so that the rainout roof would shelter only the TD treatment when rain was detected. Rainout shelter closure was an automatic sensor driven process whereby the roof covers the TD treatment during rainfall and then withdraws. The 2018-2019 experiments were smaller split plot designs (n = 4) with water regime as main plots, accessions as sub-plots, and all located in the same parcel of the rainout shelter bay (see Figure 1 for layout). Experimental plots were 33 L round pots (430 × 340 mm) containing ca. 38 kg of Gingin loam. Five evenly spaced plants were planted in each pot (one in the middle and one in each quadrant).
Germplasm
The 2016 and 2017 trials evaluated a wide range of wild germplasm (C. echinospermum and C. reticulatum) collected from a range of sites against domestic chickpea check varieties ( Table 1)
Water Regimes
Reproductive water deficits were set-up by with-holding water in the terminal drought (TD) treatment when pod set was underway, defined by the first appearance of enlarged, but unfilled pods. To this end, phenological observations (dates of first flowering, podding, and pod enlargement) were recorded Frontiers in Genetics | www.frontiersin.org three times a week. The WW treatment was irrigated three times weekly until the end of the experiment. Typically, the TD dry-down phase took 16-23 days to complete, at which point the WW treatment was also stopped. During the vegetative phase, all plants were largely rain-fed, and only manually watered occasionally when required. Vegetative phases across years were consistently wet and cool ( Table 2).
In the 2016, trial all plots were planted on 9th June. Large phenological differences among accessions meant that 10 separate dry-down groups were required to initiate the water deficit treatment from the onset of pod filling (Figure 2A; Table 2). The combined dry-down period in the 2016 trial was characterized by gradually rising temperatures punctuated by temperature spikes at approximately 10-day intervals such that Frontiers in Genetics | www.frontiersin.org 5 December 2020 | Volume 11 | Article 607819 later groups experienced more terminal drought stress than earlier groups (Figure 2A; Table 2).
To ensure consistent terminal drought stress later trials used a combination of vernalization (4 weeks at 4°C) and staggered sowing dates to make the onset date of water deficit as uniform as possible. To this end, the 2017 and 2018 trials were sown over six staggered occasions according to phenology from 7th June to 12th July and 18th June to 9th July, respectively. Despite our attempts to synchronize reproductive phases, in the 2017 trial, the onset of water deficit was still staggered across four groups starting from 2nd to 30th October. Over this period, there was a linear temperature increase punctuated by high temperature spikes driving sequentially increasing terminal drought stress in later dry-down groups ( Figure 2B; Table 2). The 2018 staggered sowing was more effective with the onset of water deficit staggered across three groups at weekly intervals from 15th to 29th October, with smaller temperature differences driven by a weaker linear temperature rise ( Figure 2C; Table 2). Nevertheless, isolated hot days (maxima > 30.0°C) in early-mid November 2018 added stress to the end of the dry-down period in groups 2 and particularly 3 ( Figure 2C).
In the 2019, trial sowing dates were widened even more to ensure reproductive synchrony. Fourteen staggered sowing dates over 37 days were used in each of three staggered starting dates (3rd June, 10th June, and 17th June) to grow three times as many pots as required for the experiment. In early October, plants with similar pod development were selected for all treatments and moved into the final split plot configuration such that the 2019 trial contained only a single, synchronous dry-down group exposed to a similar temperature range as previous early groups ( Figure 2D; Table 2). Unlike previous water deficit treatments, the 2019 TD treatment was irrigated on day 2 (receiving 50% of the water used calculated individually for each pot) to extend the dry down period because of a temperature spike ( Figure 2D). Thereafter, no water was added to the TD treatment, as in the 2017-2018 experiments.
Observations
Phenology was measured as described previously. Plant maturity (defined as 95% of pods ripe) was recorded in 2016 and 2018.
Plants were bagged prior to pod maturity to prevent seed loss during shattering. Subsequently, plants were counted and harvested, total biomass, seed weight, and number recorded. Seed size and harvest index were calculated from this data. A TD stress index was calculated based on the percentage of the well-watered value (Bouslama and Schapaugh Jr., 1984).
Reproductive water-use was measured by weighing pots at 2-day intervals after the onset of the terminal drought. Final pot weights were used to calculated plant available water (PAW) at each weighing. This data were used to calculate reproductive WUE, PAW% and transpiration per unit time, expressed in days, thermal time or cumulative evapotranspiration. In 2018 and 2019, water-use was measured in both water regimes.
Wild and Domestic Developmental Differences Over the Lifecycle (Exp 5)
To better understand behavior of wild and domestic Cicer under contrasting water supply, it was necessary to develop an understanding of developmental differences between the groups. To this end, we designed a glasshouse pot trial measuring above and below-ground dry matter partitioning and their effects on water extraction and WUE. To describe the effects of domestication independently of phenology, we studied plant development throughout the lifecycle in balanced early and late phenology subsets of wild and domesticated Cicer ( Table 3). These were grown as single plants in 120 × 16 cm split pots filled to 100 cm with Gingin loam using an RCBD (n = 3), with extra replication (n = 3 per harvest) allowing for destructive harvests at 35, 70, and 105 days and a final harvest at physiological maturity (155-170 days). The experiment was sown on June 20 2017 to coincide with the normal Mediterranean winter growing season after 4 weeks of vernalization at 4°C.
At sowing, the soil was filled to field capacity and water-use monitored at approximately 2-day intervals throughout the growing season. This was done by applying a known amount of water and calculating daily water-use by subtracting the overflow emerging from a tube at the base of each pot. To minimize evaporation, the soil surface layer was covered by plastic beads to a depth of 5 cm. Cumulative water-use curves were generated by fitting logistic functions to the daily summed water-use (see Statistical Analysis). Water inputs were matched to water use to avoid over-filling. Water-use data were used to calculate WUE of above ground biomass at each destructive harvest and reproductive WUE using pod weights at final harvest. Reproductive WUE was calculated on cumulative water-use between the start of podding and final harvest (as in the rainout shelter experiments), biomass WUE was based on water-use since the start of the experiment and each destructive harvest. Phenology observations were made on all pots throughout the growing season as described in the rainout shelter experiments.
Water inputs were stopped for those pots slated for destructive harvest 1 week before each harvest date to reveal water extraction Frontiers in Genetics | www.frontiersin.org profiles along the soil column. The above ground biomass was removed, processed into vegetative and reproductive tissue, and leaf area measured (only until day 105 because leaves were shed by physiological maturity). The split pots were carefully divided in two without disturbing the soil column, which was then separated into 20 cm segments (0-20, 20-40, 60-80, and 80-100 cm). Soil in each segment was sub-sampled, fresh and dry weights recorded (after 48 h oven drying at 60°C) to calculate the relative water content [(fresh-dry weight)/ dry weight]. Roots were carefully washed out of each soil segment, sieved, and dried. The data from each segment were used to explore relationships with depth (see Statistical Analysis). Root weights from each segment were also summed to provide a total root biomass, which was used to calculate root index (percent of total biomass attributable to roots), shoot to root ratio, vegetative, and total biomass (sum of above and below ground biomass). Below ground data were only available for the first three harvest dates because by physiological maturity the roots had started to decay.
Statistical Analysis
Genstat (V20) was used for all statistical analyses. Nested ANOVA and regression models were used to partition variance between species, between collection sites within species and finally between accessions within collection sites within species. Wild and domestic differences were analyzed using orthogonal contrasts. The same nested approach was used in the lifecycle development study, except that accessions were nested within phenology categories within species, rather than collection sites within species. Replications were fitted within water regime in the RCBDs used in 2016-2017, where the water regime treatments were established on separate contiguous areas of the rainout shelter. In the 2018-2019, split plot designs water regime was treated as the main plot, accessions as the subplot. The lifecycle development study was analyzed as a standard RCBD using reps as blocks.
Linear and non-linear {exponential, Y = a + br X ; logistic, Y = a + c/[1 + e −b(X − m) ]} regression was used to model changes in PAWC and biomass over time, and root growth and water extraction over depth in the development study. These regression models were chosen to best fit the trend in the data, producing residual plots with normally, and independently distributed errors. Changes in PAWC and leaf area over time, and root growth over time and depth were exponential. Water-use, vegetative matter, and total biomass followed logistic patterns over time, while WUE over the growing season was wellmodeled by quadratic regression.
Residual plots were used throughout to identify outliers and confirm that errors were normally and independently distributed.
Correlation-based principal components analysis (PCA) was used to integrate the results using two-way accession by water regime means and curve parameters generated by the analyses described above.
Cicer Species Responses to Terminal
Water Deficit: Productivity, Phenology, and Water-Use Nested ANOVA demonstrated consistent patterns among traits across years. In those phenological observations taken prior to the onset of water deficit treatments, the largest differences occurred between species, followed by collection sites within species, and finally accessions within collection sites (p < 0.001 for all). Observations made after the imposition of water deficit tended to follow a similar pattern, albeit with significant interactions with water regime. Domestic chickpea was characterized by a consistently earlier phenology than wild C. reticulatum and C. echinospermum (Table 4). Moving from a common (2016) to a staggered sowing date (2017 onward) to try to synchronize the reproductive phase shortened the vegetative phase, particularly in the 2019 trial, based on the widest combination of sowing dates ( Table 4).
The reduction of the vegetative phase had ramifications on biomass production. Total biomass decreased as the vegetative phase was shrunk from 2016 to 2019 (Table 4; Figure 3). Despite these scale changes, water deficit consistently reduced above ground biomass production in all trials, albeit with some differences between species. In 2016 and 2019, wild Cicer was more responsive to the WW treatment than domestic chickpea (p < 0.001), in 2017, there was no difference (p diff = 0.390), while in 2018, C. echinospermum was uncharacteristically unresponsive (Figure 3). These interactions notwithstanding, domestic chickpea tended to accumulate more above ground biomass than wild Cicer on most occasions (Figure 3, TD: 2016TD: -2018WW: 2017WW: -2018. Under terminal drought, wild and domestic Cicer had consistently similar reproductive investment (Figure 3: harvest index and seed weight). Indeed, only the 2019, TD treatment plants exhibited a significant harvest index difference in favor of domestic chickpea (p < 0.05), while seed weights were similar. However, there were clear domestic/wild differences in the reproductive response to the additional irrigation provided by the WW treatment. In 3 of the 4 trial years (2016-2018) harvest index and seed weight increased more in domesticated than in wild Cicer (Figure 3; p diff = 0.013-<0.001) in response to WW treatment irrigation.
The reduction of vegetative phase length and total biomass production over years had implications on plant reproductive water-use, measured gravimetrically during the dry-down cycle. This is clearly indicated by reduced water extraction from 2016 to 2019 in all species (Figure 4; Table 5). Nevertheless, there were remarkably consistent, contrasting water-use patterns between domestic and wild Cicer across all years (Figure 4). Thus, domestic chickpea tended to extract less water in the dry down cycle than either wild species, as indicated by significant y intercept differences (Figure 4; p < 0.001) in all years, captured by the exponential curve parameter B (see Table 5 for raw water uptake values). Surprisingly, the exponential rate of water-use (parameter R) was higher in chickpea than in wild Cicer from 2017 to 2018 (p < 0.05). These curve parameter differences are clearly evident across years in Figure 3, with chickpea varieties forming a tight cluster at the lower end of the water extraction range (except for ICCV 93929 as a high extracting outlier in 2016 only). Conversely, the two wild species were characterized by a wider range of water-use curves among accessions, corresponding to higher mean values than in domestic chickpea (Figure 4).
In 2018 and 2019, water-use was also measured in the WW treatment. ANOVA was dominated by very large wild and domestic differences (p diff = 0.005-<0.001), without water regime interaction (p diff = 0.2110.298). This is important because it indicates that while the wild Cicer species extracted more water under terminal drought, they also consistently used more water under the WW treatment ( Table 5).
Integrating the Results With Multivariate Analysis
Principal components analysis integrated the observations made in the TD treatment, capturing 55-66% of variance in two components in the ordinations performed for the 2016-2019 trials (Figure 5). Wild and domestic Cicer were consistently separated by phenology, water extraction, and WUE ( Figure 5; Table 5; all years), and to a lesser extent rate of use (2017-2019). Productivity traits (seed weight, biomass, etc.,) were consistently closely associated with harvest index and negatively correlated with water extraction and WUE. Thus, high yielding plants had high harvest index and WUE but extracted less water during the reproductive Frontiers in Genetics | www.frontiersin.org 8 December 2020 | Volume 11 | Article 607819 phase dry-down than low yielding plants. In 2 of the 3 years where TD stress indices were calculated (2016-2017) plants with high TD productivity achieved a high proportion of their WW productivity, that is to say they were less responsive to the more benign WW treatment than poorly yielding plants [TD stress indices were not calculated in 2019 because domestic chickpea performed so poorly in the WW treatment ( Figure 3D) rendering the wild comparison meaningless]. Principal components analysis confirmed the specific distinctions described earlier that wild Cicer tends to have later phenology, lower productivity, and greater water extraction coupled with lower maximum use rates than domestic chickpea. However, PCA also demonstrated considerable within species variation in these traits, particularly in 2016-2017, when a much larger wild cohort was investigated. In both these years, some C. reticulatum and C. echinospermum accessions had similar, or greater productivity, WUE and harvest index than the most productive domestic chickpea (Figures 5A,B). The ordinations for the WW treatment were remarkably similar (data not presented). Wild and domestic Cicer were separated by phenology, productivity, harvest index, and WUE (measured in 2018 only). As in the TD treatment, seed yield was positively correlated with harvest index and WUE, and negatively correlated with phenology. With very few exceptions, domestic chickpea was more productive than wild Cicer in the WW treatment, associated with greater reproductive investment and WUE.
Wild and Domestic Developmental Differences
To independently test the effects of phenology and "wildness" on above and below-ground dry matter vegetative and reproductive partitioning, and their effects on water extraction and WUE, we studied plant development throughout the lifecycle in balanced early and late phenology subsets of wild and domestic Cicer grown in the glasshouse. This experimental approach set vegetative phase limits within and between species. Figure 6 shows consistent phenology category differences between species for flowering and podding, not for end of flowering and particularly maturity. Thus, our phenology categories reliably determined the length of the vegetative phase and the onset of reproduction (e.g., short and early vs. longer and later) with no, or very minor species differences (Figure 6). However, domestic chickpea matured approximately 10 days later than wild Cicer (Figure 6; p < 0.001), with important flow-on effects on the length of the reproductive phase. While the reproductive phase was consistently 9-22 days longer in the early compared to late phenology groups, the domestic reproductive phase was 9-20 and 12-15 days longer than wild in the early and late categories, respectively (Figure 6).
These species and phenology category differences played out in plant development. Polynomial contrasts in nested ANOVA showed strong linear and quadratic interactions between species and phenology categories within species for most measured traits (p < 0.001). Because of developmental lag phases and ceiling values in most of these plant structures, non-linear logistic and occasionally exponential regression was generally more appropriate (Figure 7), confirming the significant interactions indicate by ANOVA.
The early harvests (35 and 70 days) were dominated by wild vs. domestic Cicer differences. Domestic chickpea had significantly greater early vigor than wild Cicer (p < 0.001), Frontiers in Genetics | www.frontiersin.org producing far greater leaf area, root, and shoot mass at 35 and 70 days, accounting for almost all significant differences (Figure 7). The only exception was at 35 days where the late kabuli variety Almaz had much greater biomass than the early desi ICCV 93929, a difference that disappeared by day 70. Thus, the early-mid vegetative phase growth rates were far higher in domestic chickpea than wild Cicer, particularly for leaf area (Figure 7). These wild-domestic early vigor differences were reflected in root development and water extraction down the soil profile. Root mass and water extraction decreased curvi-linearly with depth in all species (Figure 8). At 35 days, domestic chickpea (particularly the kabuli cultivar, Almaz) had far greater rates of root weight decline and water extraction over depth than wild Cicer (p < 0.001), driven by massive differences in the 0-20 cm soil layer, disappearing by 40-60 cm (p diff = 0.334), the limit of root exploration ( Figure 8A). Interestingly, water extraction of the two chickpea cultivars was similar, despite their differences in surface root production. Domestic water extraction was greater than wild in the upper soil layer (p < 0.001) but not at greater depths ( Figure 8B; p diff = 0.150). At 70 days, wild and domestic differences still dominated, albeit these were becoming smaller in terms of root distribution ( Figure 8C). Root decline rates over depth were similar across all accessions, regardless of species or phenology category, except for Almaz (p diff < 0.001) with its high surface layer root mass. Nevertheless, there were significant intercept differences between the desi cultivar ICCV 93929 and most of the wild Cicer (p diff = 0.127-0.001), reflecting root mass differences at most depths ( Figure 8C). These wild-domestic differences had a large impact on water extraction down the soil profile ( Figure 8D). Domestic chickpea depleted water from the top three soil layers, leaving progressively more water in the remaining two soil layers, captured by a similar upward trending quadratic curve in both the early desi and late kabuli varieties. Conversely, the wild Cicer water extraction curve was much more linear over depth, leaving significantly more residual water in all layers (p < 0.001).
Day 70 marked the end of the lag phase when all species showed rapid growth rates for above and below ground biomass, and new species differences and phenology category by species interactions were emerged. Thus, the early specific differences in root growth rates disappeared during the rapid growth phase, while phenology category by species interactions emerged, indicated by higher growth rates in the late compared to the early Cicer arietinum (p diff = 0.015) and C. echinospermum (p diff = 0.027), but not C. reticulatum (Figure 7, p diff = 0.733). This pattern was clearly evident in the root distribution over depth, with steeper declines in late compared to early C. arietinum (p diff = 0.065) and C. echinospermum (p diff = 0.002), but not C. reticulatum (Figure 8E, p diff = 0.259). Accordingly, there were no specific differences in water extraction by day 70, with a common, relatively flat curvi-linear response ( Figure 8F, p diff = 0.467-0.708), leaving only minor variety within phenology category differences. Above-ground vegetative biomass growth curves were clearly logistic across the range of harvest dates (Figure 7). As with root growth, there were no consistent specific differences.
In terms of leaf area growth species differences trumped phenology interactions, but now the wild Cicer had a higher rate of leaf area production than domestic chickpea (P diff = 0.048), such that by 105 days there were no specific differences in leaf area (Figure 7, P diff = 0.192).
Phenology and "wildness" were both important in the relative above-and below-ground dry matter partitioning, indicated by strong species by phenology category interaction (p < 0.001). While later types of all species invested more heavily in roots, as indicated by higher root indices and lower shoot:root ratios (Figure 9), the contrast was much stronger in wild than domesticated Cicer, particularly C. reticulatum. Thus, late wild Cicer have a much greater root index than late domestic chickpea, while there were no specific differences among the early group. Interestingly, this pattern was evident already at 35 days after sowing, well before the start of flowering, suggesting that these differences are not explained by differences in the vegetative phase length.
Pod growth rates also showed strong species by phenology category interaction (p < 0.001). Chickpea rates were considerably higher than wild, while late types of all species tended to fill pods at higher rates than early types, reaching similar final pod weights at maturity despite a later podding onset (Figure 7). Pod growth rate differences between early and late types were larger in C. reticulatum than in C. arietinum and C. echinospermum, accounting for the significant interaction. Ultimately at maturity, reproductive investment (harvest index) was far greater in domestic than wild Cicer (p < 0.001), and greater in early compared to late C. arietinum and C. echinospermum, but not C. reticulatum. The combination of rapid pod growth rates and high reproductive investment was responsible for higher rates of aerial biomass production in domestic vs. wild Cicer over the growing season (p < 0.001), with no differences between phenology categories within species (Figure 7). The final areal biomass values at maturity largely FIGURE 6 | Life cycle phenology in early (E) and late (L) flowering domestic (C. arie, C. arietinum) and wild Cicer (C. echi, C. echinospermum; C. ret, C. reticulatum) selected to study the role of "wildness" and phenology on plant development and water use under ad libitum water over the growing season. LSD bars (least significant difference) are presented for each trait individually.
Frontiers in Genetics | www.frontiersin.org reflected these rate differences: domestic larger than wild (p < 0.001), late C. echinospermum larger than early, whereas the opposite was the case for C. reticulatum (P diff = 0.05).
Plant water-use throughout the growing season followed a logistic curve largely, but not totally mirroring biomass production ( Figure 10A). Daily water-use rates remained flat for the first 70 days with minor wild vs. domestic differences (12 vs. 13 ml/ day, P diff = 0.014), turning sharply at 90 days, and peaking at 111 days. Moreover, the species/phenology characteristics described earlier also played out in the water-use curves: later C. arietinum and C. echinospermum used more water [parameter C (curve maximum value), p < 0.001] at higher rates (logistic growth rate k, p < 0.001) than early types, while the opposite was the case for C. reticulatum (Figure 10A; p < 0.001).
Nevertheless, the wild and domestic curves were remarkably similar, given differences in their aerial biomass. This is reflected in large wild vs. domestic differences in WUE across the life cycle ( Figure 10B). While all species became more efficient in their water-use for aerial biomass production over time, the rate of increase was much larger in domestic vs. wild Cicer (Figure 10B; p < 0.001). Domestic WUE was larger than wild (p < 0.001) at every point sampled throughout the lifecycle. Moreover, there were important differences in the shape of the response. In domestic chickpea (and early C. reticulatum), the rise in WUE over time was curvi-linear, peaking approximately 2/3 of the way through the life-cycle, whereas the response in the remaining wild Cicer groups was much linear, peaking at maturity ( Figure 10B). Nevertheless, Color coded markers, fitted curves and daily LSD bars (least significant difference) are presented for leaf area (-), root weight (-), vegetative matter (-), and total aerial biomass (-). Point values represent accession means from destructive harvests at 35, 70, 105, and 155-170 days after sowing. Logistic and exponential curves fitted for species/phenology categories captured 92.6-96.3% of variance, accounting for all differences between accessions. The area between the vegetative matter and aerial biomass curves represents pod weight. Root weights are presented as negative values to facilitate visual root-shoot comparisons over time.
Frontiers in Genetics | www.frontiersin.org at maturity the domestic WUE remained higher than wild ( Figure 9B; p < 0.001), accounting for all phenology/species category or variety differences (p diff = 0.901 and 0.561).
DISCUSSION
Our work shows that domestication has ramifications throughout the entire lifecycle in this Cicer example, but that domestication as a promotor of acquisitiveness is not particularly helpful to distinguish wild and domesticate. While domestic chickpea was indeed much more responsive to resource-rich conditions, and wild and domestic differences minimized under terminal drought (similar to Matesanz and Milla, 2018), these differences were not driven by greater resource acquisition in domestic chickpea. On the contrary, wild Cicer was able to extract more water under water deficit, but also used more water when it was freely available. Nor were there consistent wild-domestic growth rate differences across all plant organs over time, as would be predicted if wild and domestic occupied different ends of the stress tolerator-competitor continuum (Grime, 2006). While early vegetative growth and water extraction was much more Frontiers in Genetics | www.frontiersin.org rapid in domestic compared to wild Cicer, this is likely to be a function of greater early vigor, presumably a by-product of selecting large seed sizes. This is a common phenomenon in both domesticated cereals (Evans and Dunstone, 1970) and grain legumes (Berger et al., 2017), and underlined in this example by the early wild and domestic growth differences, and between desi and the much larger seeded kabuli type within domesticated chickpea. The fact that these growth rate differences disappeared over the growing season highlights the need to study development over the entire lifecycle. Indeed, in the late vegetative phase, as growth rates became exponential, rates of leaf area expansion were considerably greater in wild compared to domestic Cicer. Clearly, the slow-wild/fast-domestic dichotomy is not supported in this Cicer example. Unpacking the stress toleratorcompetitor continuum (Grime, 2006) in an agricultural context indicates why this may be. The triangle of Grime (2006) suggests that acquisitive traits such as rapid growth rates above and below ground are selected for in resource-rich environments, where there is strong competition for these resources (the "use it or lose it" scenario, see examples in Grime, 1988Grime, , 2006. Conversely, this strategy is risky and maladaptive in resource-poor environments, where periods of stress have to be tolerated. These environments select for slow, resource efficient growth, and stress tolerating physiology (Grime, 2006) that is strongly expressed in extremophiles such as desert cacti and succulents (Ehleringer and Mooney, 1983), and is less relevant to annual plants and agriculture (Berger et al., 2016). The lifecycles of annual plants and most agricultural crops balance stress escape (the ruderal strategy, third apex in the triangle of Grime, 2006) against acquisitiveness. Although wellmanaged fertile fields are likely to represent a more resourcerich environment for crops than the natural systems in which their wild progenitors evolved, this is unlikely to have selected for greater acquisitiveness (aka competitive capacity) because of the disparate selection pressures imposed by the two systems. Whereas wild plants are selected as individuals, crops are grown and selected as populations rather than as single plants. This has selected for crops that are poor competitors as individual plants, lifting the productivity of the community as a whole, rather than maximizing the fitness of the individual (Donald, 1963(Donald, , 1981Reynolds et al., 1994). This is exactly what we see in our results, where the wild Cicer consistently extract and use more water (which would otherwise be lost to their competitors) than their domesticated counterparts. In fact, far from being parsimonious and efficient stress tolerators, the wild Cicer seems to be profligate competitors compared to domestic chickpea. Indeed, the poor competitor aspect of chickpea water-use is being exploited in the development of cultivars for short season, stored soil moisture systems in the semi-arid tropics. ICRISAT is promoting the use of cultivars that use less water in the vegetative phase, leaving more residual soil water for seed filling (Zaman-Allah et al., 2011), a scenario that is difficult to envisage occurring in the natural world of inter-plant competition.
Instead of greater acquisitiveness Cicer domestication appears to have selected for greater reproductive efficiency, whether measured in terms of dry matter allocation or WUE. This is particularly apparent under high input conditions, where domestic chickpea is far more responsive than wild in terms of pod growth rates and harvest index. Conversely, these studies provide evidence of greater vegetative investment by the wild Cicer in terms of later phenology, higher rates of leaf area production, and greater relative investment in roots, particularly in late types. This combination facilitates greater water extraction under terminal drought, but also allows greater water-use when it is freely available. Here, there are interesting parallels with the Old World lupins, where there are similar wild-domestic differences in reproductive and vegetative investment, and flow-on effects on water use (Berger and Ludwig, 2014;Berger et al., 2020). The cereals may be more conservative in terms of relative dry matter partitioning (Wacker et al., 2002;see Wang et al., 2017 for counter-indication), but do show a similar trend in tillering along the domestication series. Thus, post anthesis tillering decreases in domestic tetra and hexaploid wheat compared to the diploid wild progenitors (Evans and Dunstone, 1970). Interestingly, similar phenology-reproductive investment trade-offs were seen in wild emmer × durum RIL populations, highlighted by the remarkable similarity in the ordinations presented by Peleg et al. (2009) and those in this manuscript. These patterns align well with competitor-ruderal continuum of Grime (2006). By selecting for earlier phenology to fit crops into a time delimited production system, domesticated A B FIGURE 9 | Relative above and below-ground investment (A, Root index; B, shoot to root ratio) in early and late flowering domestic (C. arie, C. arietinum) and wild Cicer (C. echi, C. echinospermum; C. ret, C. reticulatum). Error bars represent LSD values.
crops took on ruderal attributes such as increased harvest index, and seedling establishment was improved by greater early vigor, associated with large grain size (Berger et al., 2017). This is a productive and efficient, but risky reproductive strategy that works in agriculture where the crop is protected from grazing, disease, and competition, circumstances that do not pertain in the wild. Life is less certain for wild Cicer, and flexibility appears to be more important than reproductive efficiency over the long term. Our field observations support this idea: we have seen wild accessions re-growing from the base after Ascochyta has killed the aerial plant parts, or after grazing and also after very late rains when the plant had seemingly matured, leaving only shattered pods, and dry straw. Apart from differences in vegetative and reproductive dry matter allocation, domestic water-use was also more efficient than wild. In cereals, this has been attributed to greater harvest index (Wang et al., 2017). While our later season results confirm this idea, the fact that this trend was already apparent in the vegetative phase when calculated over aerial (Figure 10) or total biomass (data not presented) indicates that these WUE differences are not entirely attributable to vegetative vs. reproductive partitioning. Given previous reports of wild-domestic differences in stomatal conductance and photosynthesis (Evans and Dunstone, 1970;Matesanz and Milla, 2018), this underlines the possibility of differential water-use regulation in the genus Cicer that warrant further investigation.
CONCLUSION
This study has demonstrated large wild-domestic differences in vigor, vegetative and reproductive investment, water extraction, and WUE in the genus Cicer indicative of evolution under contrasting selection pressures. Dry matter allocation in wild Cicer is more vegetative than in domestic, which appears to be responsible for greater water extraction under terminal drought, and also greater water-use when it is freely available, but leads to a lower reproductive capacity and efficiency. While increased water extraction may be useful for chickpea improvement in water limiting environments, the wild trait combination should be disassembled as much as possible to evaluate its potential independently. To this end, wild × domestic populations have already been developed. It will be fascinating to see to what extent it is possible to recombine wild and domestic trait assemblages, whether water extraction capacity can be evaluated without simultaneously introducing low harvest index, or whether this returns a similar wild-domestic cline as observed in wild emmer x durum populations (Peleg et al., 2009).
DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
AUTHOR CONTRIBUTIONS
JB conceived the research, helped to run the experiments, analyzed the data, and wrote the manuscript. RP, CL, SP, FB, and KW implemented the experimental program, discussed the results, and provided feedback on the manuscript. All authors contributed to the article and approved the submitted version.
FUNDING GRDC CSP00185 and CSIRO have co-invested in funding this research.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2020-12-04T14:08:53.317Z", "created": "2020-12-04T00:00:00.000", "id": "227257849", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.frontiersin.org/articles/10.3389/fgene.2020.607819/pdf", "pdf_hash": "29eacebd1a71ba939a285de1e37d7bd358c48095", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:2337", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "29eacebd1a71ba939a285de1e37d7bd358c48095", "year": 2020}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
This below document has 2 sentences that start with ' Also, it',
3 sentences that start with ' It was found that',
2 sentences that start with ' In fact',
2 sentences that start with ' It was shown that',
2 sentences that end with 'as shown in Fig',
2 sentences that end with 'and beef products',
2 paragraphs that start with 'Conjugated linoleic acid (CLA)',
2 paragraphs that start with 'Oxidative stability of'. It has approximately 4439 words, 159 sentences, and 36 paragraph(s).
===
Physico-chemical modifications of conjugated linoleic acid for ruminal protection and oxidative stability
Conjugated linoleic acid (CLA) is a mixture of positional and geometric isomers of octadecadienoic acid [linoleic acid (LA), 18:2n-6]. Although ruminant milk and meat products represent the largest natural source of CLA and therefore, their concentration in ruminant lipids are of interest to human health, chemical or physical modifications of CLA should be needed as a means to enhance oxidative stability, to improve post-ruminal bioavailability, and to increase the clinical application. In fact, CLA are rapidly decomposed to form furan fatty acids when its are oxidized in air, and the effectiveness of dietary supplements of CLA may be related to the extent that their metabolisms by rumen bacteria are avoided. For these reasons, many scientists have examined the effect of manufacturing and protection on the stability of CLA in ruminants and food products. In this review, physico-chemical modifications of CLA for ruminal protection such as calcium salt (Ca), formaldehyde protection (FP), lipid encapsulation (LE), and amide linkage (AL), and for oxidative stability such as green tea catechin (GTC), cyclodextrin (CD), arginine (Arg), amylase, and PEGylation are proposed.
Background
Conjugated linoleic acid (CLA) are a collective term for a group of positional (C8,C10; C9,C11; C10,C12; and C11,C13) and geometric (cis,cis; cis,trans; trans,cis; and trans,trans) isomers of octadecadienoic acid (linoleic acid, LA) with a conjugated double-bond system [1]. Also, the major and most important formation of CLA are the endogenous desaturation of vaccenic acid (VA) due to the action of stearoyl-CoA desaturase (SCD; also so-called delta-9 desaturase) [1,2]. The CLA are effective in protecting tissues from carcinogenesis [2], reducing the develop-ment of atherosclerosis [3], stimulating the immune system [4], and inducing enzyme change in mouse liver [5,6]. These effects appear to be mediated by two isomers of CLA, and the two biologically active isomers are the cis9, trans11 and trans10, cis12 [7,8]. Also, it is well known that the CLA content in ruminant-derived food are certainly more affected by the animal diet and production system than by food-manufacturing factors [9]. Several factors influence the CLA content of food products, such as temperature, protein, quality, choice of starter cultures, and period of aging [10]. Variations of CLA content in foods are also affected by the animal's diet, animal age and breed, and seasonal factors [10][11][12]. Unfortunately, physical or chemical modification of CLA should be needed when its are to be used in food systems as fortifiers or additives. In fact, a number of methods have been used to prepare "rumen-protected" feed supplements, and their efficacy can be described by the extent of protection from rumen bacteria as well as post-ruminal bioavailability [13]. Also, CLA are extremely unstable in air and cis, cis-CLA isomers are most susceptible to oxidative degradation while the four trans, trans-CLA isomers are most stable in air [14], indicating that the CLA must be protected from oxidation. This review is attempted to propose various physico-chemical modification methods of CLA for ruminal protection and oxidative stability, and the potential of clinical applications of CLA is also explained.
1. Origin of CLA
Polyunsaturated fatty acids (PUFA) other than CLA, with a conjugated double bond system and with more than two double bonds, occurred naturally in nature in various seed oils but their biological activity had to date not been extensively investigated when compared with data available for CLA [15][16][17][18]. The first step in the biohydrogenation of dietary LA resulted in the formation of the cis9, trans11 isomer, due to isomerization and transposition of the delta-12 double bond [16]. This was the most abundant natural isomer present in ruminant tissue fats (over 90% of total CLA) and had been termed rumenic acid (RA) [16,19]. Further hydrogenation of RA resulted in the production of trans11-18:1 VA which was the major transmonounsaturated fatty acid present in the fats of ruminant food products (milk, yoghurt, cheese, butter, and meats) [19,20]. This contrasts with commercial preparations of CLA where proportions of the two main isomers were usually equal, although the chemical method for synthesis will allow a variety of ratios for the two isomers in the final mixture [16]. Meat from ruminant animals, particularly the fat associated with meat, was also an important source of CLA, contributing in the region of 25 to 30% of the total food intake in Western populations [16,19]. Isomers of CLA could also be synthesized in the laboratory from C18:2 or from sources high in C18:2, such as sunflower, safflower, soybean, or corn oils by a reaction involving alkaline water isomerization [19][20][21][22] and isomerization in propylene glycol [20,21,23]. The predominant isomer in milk and other dairy products was the cis9, trans11 with minor but significant proportions of trans10, cis12 [16,19,20].
Structure of CLA
CLA were a series of positional and geometric isomers of LA where one or both of the double bonds are either in the cis or the trans configuration and transposed to differ-ent positions along the acyl chain with the bonds separated by a simple carbon-carbon linkage rather than by the normal methylene group [16,17]. A number of cis-cis, cis-trans, trans-cis, and trans-trans isomers with the double bonds at various locations along the acyl chain, from carbon-6 to carbon-15, had been identified by various chemical reductive, chromatographic, and spectroscopic techniques [16,18,19]. A total of natural CLA isomers had been found in milk, dairy products, beef, human milk, and human adipose tissue using silver ion-high performance liquid chromatography and gas chromatographyelectron ionization mass spectrometry [23][24][25][26].
Physiological and biological effects of CLA on health and disease
There have been few studies that have examined the effects of CLA or its isomers in humans. Recently, a wealth of literature available mainly from cell line and animal studies indicates that CLA and individual isomers (especially, C9,T11 and T10,C12) may have numerous health benefits. One of the first studies in healthy adult women examined the effects of 3 g/day intake of CLA for 64 days on fat-free mass, fat mass and percentage fat mass compared to sunflower oil (SFO) placebo, and the results showed that there were no differences in body composition or any of the parameters examined [27]. However, two studies from Norway in healthy exercising humans (CLA, 1.8 g/day) and in overweight and obese humans (CLA, 1.7, 3.4, 5.1, and 6.8 g/day) for 12 weeks showed that CLA can decrease fat mass without significantly affecting body weight [28,29]. Results of the first study in athletes were encouraging considering that a much lower dose of CLA (1.8 g/day) produced significant results compared to the second study where the authors concluded that a dose of 3.4 g/day of CLA was enough to cause reduction of body fat. Another study in 2001 evaluated the effects of intake of 4.2 g CLA/day for 4 weeks on changes in adipose tissue and cardiovascular risk factors in middle-aged obese men with signs of metabolic syndrome [30]. In an interesting study from Netherlands, 13-week intervention with 1.8 or 3.6 g/day CLA in overweight humans supplemented prior to this with a very-low-calorie diet (which induced weight loss) increased resting metabolic rate and lean mass without affecting body weight regain [31]. Other studies had previously showed a net decrease in body fat greater than net decrease in body weight, suggesting that lean mass may have increased in the subjects [28,29]. Volunteers in both these studies were either on intensive training programs or exercised for 90 min, three times per week. These studies seem to suggest that exercise could enhance the fat-lowering effects of CLA and also help improve lean mass in humans. However, there is very limited literature on human studies with individual and different ratios of isomers, which makes it difficult to clearly establish the protective role of the physiological and biologically effect of CLA isomers in improving human health. In fact, recent studies in animals and humans suggest that CLA may not have adverse effects with long-term intervention and may actually be beneficial in reducing fat mass and atherogenic lipids [32]. Also, very few clinical studies have focused on the effects of CLA on bone health and cancer, and there are also differences in the way CLA was supplemented. In addition, effect of CLA and its isomers on inflammatory mediators has not been the subject of extensive research in humans and need to be pursued urgently. Overall, more well-controlled studies are needed before CLA or enriched isomers can be recommended to humans with confidence to improve health and quality of life. Also, evidence for efficacy in humans is being steadily strengthened by the results from clinical trials as well as animal toxicology tests.
Ruminal protection of CLA
Since ruminant milk and meat products represented the largest natural source of CLA and therefore, their concentration in ruminant lipids was of interest to human health although only trans10, cis12 was primarily mentioned which was used in the studies to obtain milk fat depression [33]. A technology to reduce milk fat output in a controlled manner had a potential application as a management tool [34]. However, the effectiveness of dietary supplements of CLA may be related to the extent that their metabolism by rumen bacteria was avoided [13,34]. Hence, a number of processes had been used to manufacture "rumen-protected" feed supplements, and their efficacies were described by the extent of protection from rumen bacteria as well as post-ruminal bioavailability [13,35]. Several methods had been used to reduce the extent to which lipid supplements were metabolized by rumen bacteria. These included the formation of calcium salts [15,[36][37][38], amide linkages [13,36,37], [39][40][41], formaldehyde treatment [15], and lipid encapsulation [13,30,42,43]. The form of fatty acid required for the production of these supplements varies, and became an important consideration due to the variation in manufacturing processes and costs associated with the production of different lipid forms.
1. Calcium salt
Calcium salts (Ca) of fatty acids had been commercially used as a dietary lipid supplement for dairy cows, and they had been experimentally used to provide rumen-protected supplements of CLA [15,36]. Giesy et al. reported that milk fat percentage was reduced in a dose-dependent manner by feeding Ca-CLA [36]. The predicted equation followed data well and showed the expected decrease in milk fat percentage as dose increased, indicating that Ca-CLA provided with the opportunity to regulate milk fat synthesis with only a minor dietary addition (Fig. 1A).
(A) Relationship between dose of conjugated linoleic acid (CLA) as calcium salts and milk fat percentage in Holstein cows on the fifth days of CLA feeding A B C These effects were caused by the trans10, cis12 CLA. The extent of the effect was certainly higher by using the form of a Ca because it protected the fatty acid from microbial degradation. Moreover, supplementing dairy cows with rumen-protected forms of CLA such as Ca-CLA substantially reduced the yield and content of milk fat without altering other production responses [15]. Similar effects on milk fatty acid pattern had been reported in previous studies where cows had received Ca-CLA [37], and studies involving abomasal infusion of trans10, cis12 CLA, where effects on de novo synthesized fatty acids became more pronounced as the dose of trans10, cis12 CLA increased [38].
2. Formaldehyde protection
Formaldehyde protection (FP) enabled the use of either FFA or esterified fatty acids [15]. Figure 1B and Table 1 showed that the CLA supplements resulted in a progressive reduction in milk fat content through the first few days of the treatment period and the proportion of preformed fatty acids in milk fat was increased with CLA treatment, whereas short-and medium-chain fatty acids tended to decrease, indicating that both protection forms such as Ca and FP were effective methods for the formulation of CLA supplements to induce milk fat depression (MFD) in lactating dairy cows and the transfer of trans10, cis12 CLA into milk fat for both Ca-CLA and FP-CLA supplements was much lower than that previously reported when CLA were supplied post-ruminally. As expected, the CLA treatments resulted in increased concentrations of trans10, cis12 CLA in milk fat, with the increase for FP-CLA treatment being greater than the Ca-CLA treatment.
3. Lipid encapsulation
Recent work showed that the FFA and methyl ester forms of trans10, cis12 CLA were equally effective in reducing milk fat synthesis when supplied by abomasal infusion [39]. Also, Perfield et al. observed a gradual reduction in milk fat yield over the first few days of treatment and a return to control values when the supplement was terminated [13], and this was similar to changes observed when trans10, cis12 CLA were abomasally infused [42] or provided intravenously [43]. It was found that the lipid encapsulation (LE) of CLA supplement was manufactured by binding methyl esters of CLA to a silica matrix, and then coating this complex with hydrogenated soybean oil, which contained fatty acids in the triglyceride form.
Amide linkage
Although Ca and FP of fatty acids had been commercially used as a dietary lipid supplement for dairy cows, and they had also been experimentally used to provide rumen-protected supplements of CLA [36,37], only few studies had been performed to suggest the rumen-protected supplements of CLA on amide protection (AP) methods. A simple AP supplement required that the starting material be FFA, whereas other amide-protected supplements had been manufactured from oils, and esters or other forms could be used for FP or LE [40]. Perfield et al. examined the use of AP and LE as methods to supply CLA [13]. It was found that the AP supplement provided CLA as FFAs, whereas the LE supplement provided CLA as methyl esters, suggesting that the AP-CLAs supplements resulted in decreased secretion of milk fatty acids of all chain lengths, but the reduction was relatively greater for milk fatty acids containing ≤ 16 carbons (Fig. 1C and Table 2). These results indicated that the AP-CLA supplements were able to reduce milk fat in a controlled manner with no adverse effects. Recent work had also reported that the FFA and methyl ester (ME) forms of trans10, cis12 CLA were equally effective in reducing milk fat synthesis when supplied by abomasal infusion [39] and CLA-ME could be protected from ruminal metabolism and inclusion of RP-ME-CLA supplement in the diet reduced milk fat content by 35-40% and significantly increased the concentration of CLA isomers in milk [41]. Overall, supplementation of an AP-CLA or a LE-CLA product resulted in similar reductions in milk fat with no effect on feed intake or milk yield. Although reduction in milk fat yield had achieved nadir by the sixth day of supplementation, transfer of trans10, cis12 CLA into milk fat was similar for both rumen-protected CLA products, and the AP-CLA and LE-CLA supplements were able to reduce milk fat in a controlled manner with no adverse effects [13], as this was a short-term study with very limited animal numbers, further research with these supplements would be needed to verify and extend results.
Oxidative stability of CLA
Some of the studies had shown that CLA acted as an antioxidant although some other studies had been reported that CLA might be pro-oxidant. Anti-oxidant activity of CLA was observed in mammary gland tissues from rats fed with CLA, when evaluated by the thiobarbituric acid reactive substances (TBARS) values [44]. However, CLA did not reduce TBARS values in pork patties mixed with CLA [44,45] and in in vitro study; CLA were oxidized as rapidly as LA [12,46]. In fact, CLA were extremely unstable in air and cis, cis-CLA isomers were most susceptible to oxidative degradation while the four trans, trans-CLA isomers were most stable in air under same conditions [14]. Moreover, CLA were oxidized faster than LA, suggesting that a conjugated double bond was more vulnerable to auto-oxidation than a non-conjugated double bond. This was in agreement with those of previous observations [14,47,48], indicating that CLA must be protected from oxidation when it was to be used in food systems as fortifiers or additives.
1. Green tea catechins
Dietary supplements containing green tea extracts were expected to be comprised of polyphenolic compounds called catechins, which were commonly obtained from green tea. The catechins were believed to act as anti-oxidants and free radical scavengers having chemo-preventative behavior as well as protection against coronary heart disease and attenuation of high blood pressure [40]. The catechins common to all green teas were (-)-epigallocatechin (EGC), (+)-catechin (C), (-)-epigallocatechin-3-gallate (EGCG), (-)-epicatechin (EC), and (-)-epicatechin gallate (ECG) [49]. Also, it was known that green tea catechins were strong anti-oxidants [50]. Seo et al. reported that CLA was more stable than LA in the aqueous system when 2,2'-azobis(2-amidinopropan) dihydrochloride was used as a free radical initiator [51]. Also, jasmine green tea catechins (GTC) (200 ppm) were effective as an anti-oxidant in protecting CLA from oxidation and the inhibition of 200 ppm GTC on CLA oxidation was even stronger than that of 200 ppm butylated hydroxytoluene (BHT) under the same conditions [40]. The oxygen consumption test showed that the oxygen uptake by the CLA samples was considerably more unstable than LA, whereas addition of 200 ppm GTC significantly decreased the oxygen uptake by CLA as compared with the control CLA samples (Fig. 2A). It is noteworthy that 200 ppm GTC is more effective than 200 ppm BHT in protecting CLA from oxidation. It is known that canola oil contains α-tocopherol and it is also possible that GTC and α-tocopherol have a synergistic effect on the oxidation of CLA when added in canola oil [52]. Also, feeding canola seed to lactating dairy cows resulted in milk fat with higher proportions of healthful fatty acids without affecting milk yield or composition of milk [53].
Oxidative stability of CLA encapsulated in α-, β-, and γcyclodextrins (designated as CLA/CDs microencapsules) was studied by measuring the headspace-oxygen depletion in airtight serum bottles and by measuring the peroxide values (POV) [57]. It was found that CLA/α-CD microencapsules at a 1:4 mole ratio completely protected CLA from oxidation, when oxidized at 35°C and a 1:6 mole ratio of CLA/β-CD was required to give a protective effect similar to that exhibited by CLA/γ-CD microencapsules at a 1:4 mole ratio (Fig. 2B). The protective efficiency of CDs for CLA oxidation may be, in part, attributed to the hydrophobicity of the inner cavity of CDs, which facilitated insertion of the conjugated diene portion into the CD cavity, and the cavity of the CDs, which was possibly large enough to incorporate some oxygen molecules and create a miniature reaction chamber to facilitate the reaction between CLA and oxygen, suggesting that physical interference by CDs was not a negligible factor for the oxidative stability of CLA. Further study will be required to explore the structural features of CLA/CDs microencapsules. It was of significance to note that β-CD was an appropriate material with which to microencapsulate CLA for industrial use, because of its adequate protective effects for CLA oxidation and because it was much lower in cost than other CDs.
3. Arginine/CLA complex
Arginine (Arg), a water-soluble amino acid, was known to be a substrate of nitric oxide synthase and regulate vascular function and blood pressure homeostasis, and thus prevent cardiovascular disease [58]. Also, Arg had some protective roles against oxygen radical attack possibly due to its direct chemical interaction with oxygen radicals [59]. Kim et al. reported that hydroperoxide production from CLA increased about 12-fold at 10 hr after heat treatment at 100°C, whereas Arg/CLA complex did not exhibit significant hydroperoxide production [60]. The Arg/CLA complex showed synergistic anti-oxidant activity in a 2,2'azinobis(3-ethylbenzothiazoline)-6-sulfonic acid (ABTS) radical scavenging assay (Fig. 3). It was found that Arg/ CLA complex at 20 mM scavenged 89% of ABTS radicals in 3 h, whereas CLA alone quenched only 48% under the same conditions, suggesting that a hydrophilic Arg/CLA complex exhibited enhanced oxidative stability and antioxidant activity, which may expand the scope of CLA applications in various food industries. Recent research had been aimed to elucidate the physiological background of CLA-inducing reduction in adipose tissue mass [61,62]. However, some concerns had been aroused on the potential implication of the CLA in insulin resistance and fatty liver under certain conditions [63][64][65]. Because Arg infusion was known to have a preventive role in the insulin resistance by decreasing the total plasma homocysteine concentration [66] and antioxidant capacity [66,67], the formation of Arg/CLA complex could alleviate the potential side effects, if any, resulted from the high dose of CLA. Thus, the presence of Arg in the form of complex with CLA may expand the scope of the application of CLA as a health-promoting agent.
4. Amylase/CLA complex
Most of the interest in amylase/lipid complexes focused on their technological importance in the starch food system, since they modified the texture and structural stability of starch based-product (e. g., reduction in stickiness, improved freeze-thaw stability, and retardation of retro- A B gradation) [68,69]. Other researchers studied amylase/ lipid complexes in view of their contribution to the bioavailability of starch, in terms of its enzymatic digestion [70]. It was shown that the V-form could be produced from mono-and di-glycerides, and saturated fatty acids, as well as un-saturated fatty acids [70,71]. The protection against oxidation afforded to CLA by its inclusion in an amylase complex demonstrated the potential of the complexes, especially those created in water/dimethyl sulfoxide (DMSO) solution, to efficiently protect CLA from oxidation [72]. It was shown that the peaks of the diffractograms obtained from complexes made in KOH/HCl solution were narrower than those obtained from complexes created by water/DMSO solution, indicating that these complexes were composed of large crystals. Also, the complexes created in water/DMSO solution at 90°C revealed globural structures of heterogeneous nature with an average z-range of 71.6 ± 59 nm and diameter of 152 ± 39 nm. Fig. 4 shows oxidation of CLA on each water/ DMSO and KOH/HCl solution, indicating that the protective effect of complexes against CLA oxidation was higher for complexes created in water/DMSO solution than that for KOH/HCl solution, suggesting that complexes created in water/DMSO solution exhibited better protective ability from oxidation. Hence, these results indicate that the amylase/lipid complex system could serve as a vehicle for delivery of PUFA to the intestine and potential use of amylase/lipid complexes could be supplementation of various stable foods with PUFA.
PEGylation
Poly(ethylene glycol) (PEG) was a nontoxic, water-soluble polymer widely used for stabilizing colloids in foods and paints and in formulating pharmaceuticals and cosmetics [73]. Generally, covalent attachment of activated PEG to proteins altered protein properties, such as increased solubility and stability in organic solvents, increased thermal stability, and reduced immunogenicity and antigenicity, in ways that extend their potential uses [74][75][76]. Also, PEGylation provided a higher stability owing to the formation of core-shell type nanoparticels (NPs) when compared to the non-modified drug [77]. In fact, PEGylated drugs such as core-shell type polymeric NPs increased stability in air, light, pH, and temperature [74,76]. Support for these results, as shown in Fig. 5A, the concentration of intact all-trans retinoic acid (atRA) in the methanol solution rapidly decreased during incubation at room temperature under light exposure whereas the rate of PEGylated atRA (PRA) degradation was very slow [78]. Also, we previously reported that PEGylated CLA (PCLA) had increased bioavailability than CLA due to the biocompatible and hydrophilic properties of PEG, and peroxisome proliferator activated receptor gamma 2-induced adipogenesis was reduced by PCLA [79]. In addition, we further demonstrated that a time-dependent effect on lipolysis and p-extracellular signal-related kinases (ERK) expression was observed for PCLA-treated, but not for CLA-treated cultures [80], suggesting that the induction by PCLA of mitogen-activated protein kinase kinase (MEK)/ERK mitogen-activated protein kinase (MAPK) activation was linked to secretion of adipo-cytokines, interleukin-6 (IL-6), and interleukin-8 (IL-8), in timedependent manners. Our findings provide support for a role for PCLA as a pro-drug in the regulation of metabolism in adipose tissue. Moreover, we showed that the level of headspace oxygen decreased by PCLA was lower than that decreased by CLA, indicating that the oxidation of CLA was significantly protected by PEGylation, as shown in Fig. 5B (unpublished data). Hence, observations with PEGylation regarding their biological and CLA-protected effects are encouraging and their use in a variety of biological activity, oxidative stability and post-ruminal bioavailability must not be ignored.
Summary
CLA have numerous potential health benefits, and the fats from milk and meats of ruminants are the richest natural dietary sources of CLA. Manipulating the diets of dairy and beef cattle and altering management practices on the farm could enhance the CLA contents of milk and dairy products and beef products. Also, the CLA contents of milk, dairy products, meat, and meat products vary widely, and the CLA intake by humans have the potential to increase to a level that have been shown to reduce the incidence of cancer in animal models through the consumption of CLA-enriched dairy and beef products. However, CLA are rapidly decomposed to form furan fatty acids when its are oxidized in air, and the effectiveness of dietary supplements of CLA may be related to the extent that their metabolism by rumen bacteria is avoided. To overcome with these phenomena, many investigations have examined the effect of manufacturing and protection on stability of CLA in ruminants and food products such as rumen-protected methods including Ca, Fa, AP, and LE, and oxidative stability methods such as GTC, CD, Arg, and amylase. All of these modifications, however, there are short-term study with very limited animal numbers; further research with these supplements will be needed to verify and extend results. Also, before CLA supplementation are recommended for human beings, controlled research studies using single isomers of CLA need to be completed to determine its efficacy and safety. Moreover, it is still questionable whether high levels of trans10, cis12 in milk are indeed desired-especially taking into consideration that trans10, cis12 is not really a naturally occurring fatty acid although CLA, specifically trans10, cis12, might be detrimental to human health. However, suffice it to say that observations with CLA regarding their possible health effects are encouraging and their use in a variety of functional foods are a distinct possibility that should be interested. Headspace oxygen in the sample bottles was analyzed by injecting a 100 μL headspace air sample into a HP 5890 GC (Avondale, PA), equipped with a stainless steel molecular sieve column (13×, 80:100; Alltech, Deerfield, IL) and a thermal conductivity detector. High purity helium (99.99%) was used as the carrier gas. The flow rate was 40 mL/min. The GC oven temperature was maintained at 40°C. The injector port and detector temperatures were maintained at 120 and 150°C, respectively. Each sample was analyzed in triplicate. Oxygen contents were quantified by an HP 3396A integrator (unpublished data).
Authors' contributions
MHS designed and wrote this manuscript. LHG, CCS, CYJ, and CCS conceived of the study, and participated in its design and co-ordination. All authors read and approved the final manuscript.
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2014-10-01T00:00:00.000Z", "created": "2008-06-01T00:00:00.000", "id": "16466685", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://nutritionandmetabolism.biomedcentral.com/track/pdf/10.1186/1743-7075-5-16", "pdf_hash": "1c78d9bc815a2be0ddf5e10ae933b7228d7b824d", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:4050", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "1c78d9bc815a2be0ddf5e10ae933b7228d7b824d", "year": 2008}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Proteomic analysis of shoot tissue during photoperiod induced growth cessation in V. riparia Michx. grapevines
Background Growth cessation, cold acclimation and dormancy induction in grapevines and other woody perennial plants native to temperate continental climates is frequently triggered by short photoperiods. The early induction of these processes by photoperiod promotes winter survival of grapevines in cold temperate zones. Examining the molecular processes, in particular the proteomic changes in the shoot, will provide greater insight into the signaling cascade that initiates growth cessation and dormancy induction. To begin understanding transduction of the photoperiod signal, Vitis riparia Michx. grapevines that had grown for 35 days in long photoperiod (long day, LD, 15 h) were subjected to either a continued LD or a short photoperiod (short day, SD, 13 h) treatment. Shoot tips (4-node shoot terminals) were collected from each treatment at 7 and 28 days of LD and SD for proteomic analysis via two-dimensional (2D) gel electrophoresis. Results Protein profiles were characterized in V. riparia shoot tips during active growth or SD induced growth cessation to examine physiological alterations in response to differential photoperiod treatments. A total of 1054 protein spots were present on the 2D gels. Among the 1054 proteins, 216 showed differential abundance between LD and SD (≥ two-fold ratio, p-value ≤ 0.05). After 7 days, 39 protein spots were more abundant in LD and 30 were more abundant in SD. After 28 days, 93 protein spots were more abundant in LD and 54 were more abundant in SD. MS/MS spectrometry was performed to determine the functions of the differentially abundant proteins. Conclusions The proteomics analysis uncovered a portion of the signal transduction involved in V. riparia grapevine growth cessation and dormancy induction. Different enzymes of the Calvin-Benson cycle and glutamate synthetase isoforms were more abundant either in LD or SD treatments. In LD tissues the significantly differentially more abundant proteins included flavonoid biosynthesis and polyphenol enzymes, cinnamyl alcohol dehydrogenase, and TCP-1 complexes. In the SD tissue photorespiratory proteins were more abundant than in the LD. The significantly differentially more abundant proteins in SD were involved in ascorbate biosynthesis, photosystem II and photosystem I subunits, light harvesting complexes, and carboxylation enzymes.
Background
Viticulture and enology have a rich history beginning over 7,000 years ago. With the growth of civilization grapevines became a prominent fruit crop and are now the most widely grown and economically important in the world. Even though the majority of wine production takes place in Mediterranean or oceanic climate areas, vineyards of continental regions contribute greatly to the diversity of viticulture. Grapevines grown in these temperate climates must be adapted to cold, dry winters in order to survive. Vitis riparia, the only grape species native to the upper Midwest region of the United States, is particularly adapted to colder climates [1,2].
Like many perennial plants, grapevines survive subzero winter temperatures by ceasing growth and entering dormancy. In many temperate woody species, the transition from active growth to dormancy is promoted by decreasing daylength [3]. Photoperiodic response is a stable annual cue that provides plants with a reliable timing mechanism to signal winter's onset [4]. Daylength sensing takes place in the leaves and a signal is believed to be transported to the shoot apex [5]. In tree species with photoperiodically induced dormancy, such as birch (Betula), the perception of decreasing daylengths results in cessation of growth, development of a terminal bud, and progression to a dormant and more freezing-tolerant state [6,7]. The decreased daylength also triggers other adaptive responses including nitrogen storage, stem growth cessation, and leaf senescence [8].
In contrast to tree species such as poplar (Populus) and birch, V. riparia does not set a terminal bud in response to decreasing daylength in the autumn. Upon reaching a critical daylength specific to a given V. riparia ecotype, shoot growth ceases and shoot tip abscission and latent bud dormancy are induced [9][10][11]. Shoot tip abscission coincides with bud dormancy induction in grapevines and occurs prior to leaf senescence. Full shoot tip abscission in V. riparia takes place after 28 days of short photoperiod (SD) [11]. The shoot tip begins to yellow from the 2 nd node to the apex and eventually dries up and falls from the plant. Autumn senescence, or programmed cell death, stimulates many changes in gene expression which are accompanied by a remobilization of nutrients, carbohydrate accumulation, and shedding of plant parts [12]. This study examined protein abundance during the transition from active growth to initiation of shoot tip abscission to begin unraveling SD programmed induction of growth cessation and shoot tip senescence in grapevines. Quantitative and qualitative differences in protein abundance were identified by employing a phenol-based extraction and 2D gel analysis [13,14].
Photoperiod regulation of shoot growth
Measurements of V. riparia primary shoot length and node number were initiated at day zero of the differential photoperiod treatments and were repeated every seven days (0, 7, 14, 21, and 28 days). Figure 1 shows that shoot length and node number were similar for the first 7 days of LD and SD. At day 14, the shoot length and node number were statistically different (p-value ≤ 0.05 and ≤ 0.001 respectively) between LD and SD treatments. By 28 days growth had ceased in the SD vines and there was a decrease in node number as tip abscission occurred. The LD treated V. riparia grapevines continued to grow and had a greater total shoot length and node number.
Differential photoperiod influence on protein abundance 2D PAGE analysis was used to examine the response to photoperiod change and the physiological alterations as the shoot tip yellows and abscission is initiated. Proteins were extracted from four node shoots harvested after 7 Figure 1 LD vs. SD physiological data. Primary shoot length and node number were determined for LD (circle) and SD (square) treatments at various time points. Solid lines indicate primary shoot length; dashed lines represent node number. and 28 days of differential photoperiod treatment (six replicates for each time point and photoperiod). There was no significant difference in the amount of recovered proteins that was observed between photoperiod treatments from the same harvest time point. Total protein recovery averaged 5.4 ± 1.4 mg per gram of tissue extracted.
In total, 1054 spots were detected among the four treatments sampled (two photoperiods × two time points). An average of 785 spots per gel with an intensity value greater than 0.01% of the total average spot intensity was observed. Faint spots were included in the gel analysis to maximize the number of proteins identified and to increase the potential of indentifying signaling-related proteins that are typically low in abundance. The inclusion of faint spots increased spot number per gel but also resulted in a comparatively high average coefficient of variation (CV) (7SD: 0.76; 7LD: 0.82; 28SD: 0.70; 28LD: 0.72). However, these CV values were within a range consistent with values previously reported for other plant proteomic analyses (0.26-0.31) [15] (0.47-0.75) [16] and (0.24) [17].
No significant difference in the number of spots with an intensity greater than 0.01% was observed at 7 days of differential photoperiod treatment. At 28 days, the LD treatment presented a significantly (p value = 0.0002) greater number of spots than the SD treatment (814 versus 742 spots). A few major proteins may have contributed to these differences as in 28LD where the top 10 most intense proteins accounted for 10.2% of the total spot intensity while in 28SD the top 10 most intense proteins accounted for 13.7% of the total spot intensity.
Proteome differences were analyzed at 7 days and 28 days of differential photoperiod treatment. At 7 days 69 spots displayed differential abundance (ANOVA, p-value ≤0.05) and ≥ two-fold ratio. Of these 69 spots, 39 were more abundant in LD ( Figure 2) and 30 were more abundant in SD (Figure 3). At 28 days, 147 spots displayed differential abundance (ANOVA, p-value ≤0.05) and ≥ two-fold ratio. Of these 147 spots, 93 were more abundant in LD ( Figure 4) and 54 were more abundant in SD ( Figure 5).
Identification of differentially abundant proteins between photoperiod treatments
Protein spots that displayed differential abundance (ANOVA, p-value ≤0.05) and ≥ two-fold ratio between Figure 2 2D PAGE analysis of V. riparia after 7 days of LD treatment. Proteins that exhibited a significant change (≥ two-fold ratio, p-value ≤ 0.05) between LD and SD are indicated by circles and standard spot numbers on a representative replicate gel. See Table 1 for a detailed listing of proteins. the two photoperiods were excised and analyzed by MALDI TOF/TOF. At 7 days, 68 of the 69 differentially abundant protein spots were positively identified (Table 1 and 2). At 28 days, 137 of the 147 differentially abundant protein spots were positively identified (Table 3 and 4). The identity of the majority of the protein spots was determined using the putative proteins from the homozygote Pinot Noir (PN40024) genome sequence; only 4 were identified from different Vitis data sources (tentative contig, EST or the heterozygote genome). Supplementary spot data are available in Additional File 1; including IDs of corresponding predicted proteins from genome sequencing data, data for other proteins with positive IDs, and abundance of each spot on each replicate gel.
Discussion
Decreasing daylength is the environmental signal utilized by many perennial plant systems to initiate growth cessation and to prepare for adverse environmental conditions associated with winter in temperate zones. In this study, V. riparia vines showed no difference in the rate of shoot growth in LD and SD during the first seven days of differential photoperiod treatments; Figure 3 2D PAGE analysis of V.riparia after 7 days of SD treatment. Proteins that exhibited a significant change (≥ two-fold ratio, p-value ≤ 0.05) between LD and SD are indicated by circles and standard spot numbers on a representative replicate gel. See Table 2 for a detailed listing of proteins.
thereafter, growth ceased in the SD treatment and the shoot apices senesced upon prolonged SD exposure. This data is in accordance with previous studies that found shoot length and node number were greater under long days [9,11,[19][20][21][22].
Several proteins identified in this V. riparia study are in common with proteins identified in shoot or leaf proteome profiles of several V. vinifera and V. rotundifolia cultivars [23,24]. However, those studies indicated that genotype was the most significant factor determining differences in protein abundance [23]. Therefore, this study presents only the differentially abundant proteins in response to LD growth maintenance and SD induced growth cessation in V. riparia. In contrast to photoperiod studies in peach bark (Prunus persica) [25], which showed a small number (66) of proteins differentially abundant in response to SD, V. riparia had 216 proteins (≥ two-fold ratio, p-value ≤ 0.05) that showed differential abundance in response to SD. There were very few differentially abundant proteins in common between peach bark (a storage tissue) and grape shoot tip (predominately photosynthetic tissue) in response to photoperiod treatment. A comparison of the proteomes of the V. riparia shoot tissue exposed to LD and SD indicated a greater number of proteins in LD than in SD. Since an individual spot intensity is relative to the total intensity, this difference could be related to a higher abundance of a few major proteins in SD treatments, thus reducing the share of lower abundant proteins. In addition to differences in the number of abundant proteins, a comparison of the proteomes identified several molecular parameters that could play significant roles in plant adaptation to decreasing photoperiod.
In barley shoot apices it was noted that the rate of carbohydrate production was considerably slower in 8 h than in 16 h photoperiods [26]. Similarly, in this study, enzymes involved in the reduction phases of the Calvin-Benson cycle are more abundant in LD shoot tips while enzymes involved in the carboxylation and regeneration phase are more abundant in SD shoot tips. In contrast, the greater recovery potential of ribulose-1, 5-bisphosphate exhibited in SD treatments may be related to an overall decrease in available carbon in comparison to the LD treatments.
Under LD conditions it appears that the carbon surplus promotes tissue growth by increasing the pyruvate pool. Roeske and Chollet [27] found that pyruvate accumulation was light dependent. The LD treated tissue had a greater abundance of sucrose synthase (SSP7708) enzymes. Similarly, an increase of sucrose synthase activity was observed in LD in soybean leaves [28], and a higher abundance of sucrose was observed in LD in tobacco leaves [29]. In Arabidopsis, enzymes in the glycolysis pathway showed a decrease in activity in conjunction with decreasing photoperiods, while activity of photosynthesis and starch synthesis remained high [30].
A greater abundance of enzymes leading to the accumulation of starch has been observed in SD shoot tips. Analysis identified these storage enzymes as fructose bisphosphate aldolase (SSP7317), a second glyceraldehyde-3-phosphate dehydrogenase (SSP6412), and glucose-1-phosphate adenylyltransferase (SSP4401). Table 4 for a detailed listing of proteins.
Previous reports illustrated that plants grown in shorter photoperiods or lower light intensities usually synthesize proportionally more starch [29,31,32]. The present study reveals a clear contrast in carbon utilization through its enzymatic steps. While more carbon is probably accumulated and used for the plant growth in LD, under SD plants appear to store the carbon as starch.
Amino acid metabolism
Most minor amino acid abundance in plants has shown poor correlation with short term photoperiod changes [29,33]. These insignificant associations suggested that the variation in minor amino acids cannot be traced to short-term changes in primary carbon and nitrogen assimilation [33]. However, glutamate, glutamine, glycine, asparagine, alanine, threonine, and serine present daily variation in abundance in tobacco [29]. These authors also reported that all amino acids assayed were more abundant in LD than SD unless they could not be detected. Glutamate acts at the center of nitrogen flow by incorporating ammonia into the plant [34]. Glutamine synthetases are especially important in the transport of nitrogen in aerial parts of the plant, and play different roles according to their cellular localization. Two glutamine synthetases have been detected as differentially abundant. One, presumably cytosolic (SSP4315), was more abundant in 7LD (Table 1), and the second, likely chloroplastic (SSP5407), was more abundant in 28SD. A third glutamine synthetase (GSVIVP00030210001) has been identified on two proximal spots (SSP3311; SSP3313); SSP3311 was more abundant in 7LD and SSP3313 was more abundant in 7SD. The differentiation between these glutamine synthetase spots is not likely caused by phosphorylation because SSP3313 has a slightly higher molecular weight (Mw) and the impact of phosphorylation on Mw is not generally noticeable in 2D gels. Over abundance of the cytosolic isoform in 7LD could be related to a greater nitrogen uptake in LD [35]. The major role of the chloroplastic isoform of glutamine synthetase in leaves is thought to be re-assimilation of the NH 3 generated in photorespiration [36]. Glutamine synthetases are known to interact with 14-3-3 proteins [37]. Seven 14-3-3 proteins have been identified as differentially abundant in the present study, but only three correlate strictly with glutamine synthetase abundance. One 14-3-3 protein in LD (SSP0224) ( Table Table 2 Proteins whose abundance was significantly more abundant in SD than LD at 7 days 1) and two in SD (SSP0227; SSP0109) ( Table 2) could also be involved in glutamine synthetase regulation during photoperiod. In addition to chloroplastic glutamine synthetase, other enzymes involved in photorespiration [38] have been seen as more abundant in SD shoot tips.
Decarboxylating glycine dehydrogenase (SSP8713) and two phosphoglycolate phosphatases (SSP2202; SSP3111) ( Table 4) were differentially abundant in the 28 day treatments. Increased photorespiration in plants has been observed in the dark [39,40]. Photorespiration commonly produces reactive oxygen species (ROS), such [41], which can be toxic to plants at certain concentrations [42]. SD plants are known to cope better with H 2 O 2 toxicity than LD plants [43]. Overabundance of enzymes in SD tissue related to ascorbate metabolism, which is involved in the detoxification of reactive oxygen species [41], also supports the hypothesis that the grapevine leaves have a higher level of peroxides under SD treatments. Monodehydroascorbate reductase (NADH) (SSP7406), dehydroascorbate reductase (SSP5106), and L-galactose 1-phosphate phosphatase (SSP2209) ( Table 4), enzymes related to ascorbate biosynthesis, were all found in greater abundance in SD shoot tips. ROS such as H 2 O 2 often elicit various physiological processes as signal molecules. H 2 O 2 is produced during photosynthesis and photorespiration, and interacts with thiol-containing proteins. H 2 O 2 directly activates numerous signaling pathways and transcription factors that regulate gene expression. Most research discusses the role of hydrogen peroxide in photorespiration and stress signaling, but it was not until recently that H 2 O 2 was linked with cell growth and other cellular processes [41,44]. Hydroxyl radicals may have an active role in cell wall loosening [45]. Fry and colleagues suggest that ascorbate, H 2 O 2 , and copper ions (Cu +2 ) could interact to form OH radicals that actively loosen cell walls [46][47][48]. Additional enzymes involved in the metabolism of amino acids have been identified as more abundant in LD shoot tips (Table 1 and 3), possibly linked to a greater requirement of metabolites during growth. Aspartate semialdehyde dehydrogenase (SSP4320) forms an early branch point in the metabolic pathway forming lysine, methionine, leucine, and isoleucine from aspartate [49]. Enzymes involved downstream in the amino acids biosynthetic pathways have also been identified, including two ketol-acid reductoisomerase spots (SSP6614 and SSP6517, both matching GSVIVP00018719001) and dihydroxy-acid dehydratase (SSP7613), which are involved in the biosynthesis of isoleucine and valine. Furthermore, the 5-metetrahydropteroyltriglu-homocys S-Me-transferase (SSP8731;, SSP8726; SSP9702; SSP8718; SSP8723; SSP8706, all matching GSVIVP00003836001) and S-adenosylmethionine synthetase (SSP5408 matching GSVIVP00019707001; SSP6425; SSP5415 matching GSVIVP00028192001) are involved in methionine metabolism. Additionally, a cysteine synthase (SSP6307) has also been identified as more abundant in LD shoot tips.
Secondary metabolism
Phenylpropanoid biosynthetic pathways provide anthocyanins for pigmentation, which are important compounds for protection against UV photo-damage in plants [50]. Effects of light treatment on phenylpropanoids have been widely studied in grape berries because of their important organoleptic properties. UV is known to increase phenolic composition in grape berries [51], and photoperiod has been identified as directly affecting the flavonoid composition. Flavonoid compounds decreased in SD versus LD in Xanthium, including anthocyanidin (quercitin), caffeoyl quinic acid, and bulk phenols [52]. In this study, three enzymes involved in the flavonoid biosynthesis were more abundant in LD shoot tips (Table 1 and 3): chalcone synthase (SSP8417), chalcone isomerase (SSP2120), and leucoanthocyanidin dioxgenase (SSP6413). Polyphenols, which also play an important role in protection against oxidation, and anthocyanidin reductase (SSP7313; SSP7325) were more abundant in LD shoot tips. Cinnamyl alcohol dehydrogenase (SSP6205), an enzyme that catalyzes the final step for production of lignin monomers, was also more abundant in LD shoot tips. Both cinnamyl alcohol dehydrogenase and lignin content have been shown to be enhanced by light in Pinus radiata callus cultures [53] and Arabidopsis roots [54].
Energy metabolism
Surprisingly, a large number of proteins involved in photosystem II (PSII) (SSP2206; SSP1121; SSP1116; SSP3010), light harvest complex (LHC) subunit (SSP2101; SSP1008; SSP0008; SSP0006; SSP1002), and one involved in photosystem I (SSP1006) were more abundant in SD shoot tips (Table 2 and 4). These observations were unexpected since photoassimilate incorporation related proteins are more abundant in LD shoot tips (see previous carbon fixation section). However, several explanations are possible for these observations. Light stress-related oxidative damage causes protein degradation in PSII [55] and it could potentially be more dramatic under LD, leading to fewer PSII proteins. It is also noted that the leaves in the SD shoot tip are older than those in LD, since shoot growth ceases in the SD treatment and the LD shoot continues to grow and initiate new leaves. Thus the SD leaves may simply contain a greater number of photosystem complexes. Fewer photosystems does not necessarily reflect a decreased efficiency of the photosynthetic system, but rather an indication of leaf maturity and the fact that the photoassimilates are exported from the older mature leaves to the shoot tips. Mor and Halevy [56] and Lepistö et al. [57] observed a similar pattern in LHC proteins in rose (Rosa) shoots and Arabidopsis leaves respectively and showed that the photochemical efficiency of PSII was not affected by day length.
Protein fate
The chaperonin TCP-1 is involved in cytoskeleton organization and keeps cytoskeletal proteins folded. Six of the eight subunits of the chaperonin TCP-1 complex were more abundant in 28LD shoot tips (SSP6609; SSP5617; SSP8609; SSP6606; SSP4517; SSP8605), (Table 3). Actin and tubulin monomers both interact with TCP-1 in order to reach their native states. Brackley and Grantham [58] and Himmelspach et al. [59] observed that abundance of TCP-1 subunits is age dependent but not growth dependent. This suggests that the greater abundance of TCP-1 subunits in the LD shoot tips was more related to the fact that the tissues are younger in the actively growing LD shoot tips than in the SD shoot tips. Consistently, tubulin proteins (SSP2406; SSP2516; SSP2404; SSP1516) ( Table 1 and 3) were more abundant in LD shoot tips. Also, seven proteasome subunits were more abundant in LD (SSP2221; SSP4520; SSP8103; SSP7108; SSP2008; SSP6014; SSP7520). Proteasome plays an important role in plant life cycle processes; among them, cell division, growth and light signaling, which would all be higher in the actively growing LD shoot tip [60].
Conclusions
Previous woody plant studies on photoperiod influence of protein abundance have focused predominately on specific bark storage and dehydrin proteins [61,62]. Studies of photoperiod induced changes in proteins during the induction of poplar bud dormancy showed many changes in protein profiles; however they did not identify the metabolic pathways involved in response to SD [63]. The proteome of V. riparia shoot tip tissue changes dramatically upon exposure to shorter photoperiod, although effects were more visible at 28 days than at 7 days. V. riparia grapevines seem to shift the direction of carbon flux from metabolites for shoot growth in LD to starch accumulation when shoot growth ceases in SD. Both cytoskeletal proteins and protein fate enzymes were more abundant in LD shoot tips, suggesting turnover and production related to cell development. In addition, under LD there was a greater abundance of phenylpropanoids which may contribute to increased cell wall synthesis as a result of active growth. In contrast, photosystem proteins were more abundant in SD shoot tips which may be a factor of difference in leaf age as growth ceases in the SD treatment while the LD shoots continue to grow and produce new leaves. Abundance of photorespiratory enzymes was higher in SD shoot tips suggesting that reactive oxygen species were more abundant. This suggestion is also supported by an abundance of ascorbate metabolite enzymes which are involved processes for detoxifying reactive oxygen species.
Plant material and growth conditions
Potted, spur-pruned two to six-year-old V. riparia grapevines were removed from cold storage on 3/26/ 2007 and 3/24/2008. The plants were repotted and grown under a long photoperiod (LD, 15 h) with 25/ 20°C + 3°C day/night temperatures and 600 to 1400 μmol m-2s-1 photosynthetic photon flux (PPF) in a climate-controlled, un-shaded glass greenhouse (En Tech Control Systems Inc., Montrose, MN) in Brookings, SD, USA (44.3°N). After 30 days, when grapevines reached 12-15 nodes, pots were randomized into two replicated treatment groups. Five days after randomization, differential photoperiod treatments began with one treatment group continuing in LD and the other receiving a short photoperiod (SD, 13 h). The SD was imposed using an automated, white covered black-out system (Van Rijn Enterprises LTD, Grassie, Ontario, Canada). At 7 and 28 days of differential photoperiod treatment, fournode shoot tips were collected between 8:30 and 10:30 AM, immediately frozen in liquid nitrogen, and placed at -80°C for future protein extraction. Three replications (5 vines/replication) were harvested in two consecutive years (2007 and 2008) resulting in a total of six replications analyzed by 2-D gel electrophoresis.
Growth measurement
Shoot growth was measured weekly in both photoperiod treatments. Primary shoot length (in centimeters) and node number were recorded on a random sample of 11 LD and 11 SD V. riparia grapevines.
Protein extraction
Protein extractions were performed in sets of four random samples. To reduce the effect of systematic variation in the extraction, only one randomly selected replicate of each condition was extracted at a time. These precautions reduced the occurrence of false positives but may have increased the variability between replicates. One gram of shoot tissue was ground to a fine powder in liquid nitrogen with a mortar and pestle. Extraction was adapted from the phenol-extraction protocol as described by Vincent et al. [14]. Ten mL of Hurkman extraction buffer [64] was added to each sample (0.7 M sucrose, 0.5 M Tris-HCl pH = 7.5, 50 mM EDTA, 0.1 mM potassium chloride, 2% 2-mercaptoethanol, 2 mM PMSF, 1 antiprotease tablet (Roche Diagnostics, Indianapolis, IN, USA)), homogenized for 30 sec, and incubated for 10 min at 4°C. After incubation an equal volume of 1 M Tris-saturated phenol (pH = 7.5) was added to each sample. The mixture was homogenized and incubated at 4°C for 30 min. The phases were separated by centrifugation (30 min, -4°C, 3,650 × g). The upper phenol phase was collected and re-extracted with an equal volume of Hurkman extraction buffer. The sample was vortexed, incubated at 4°C for 30 min, and centrifuged (30 min, -4°C, 3,650 × g). The upper phenol phase was collected, and five volumes of 0.1 M ammonium acetate in cold methanol (MeOH) were added to precipitate proteins. The samples were incubated overnight at -20°C and then centrifuged (30 min, -4°C, 3,650 × g). The pellet was washed twice with 5 mL of cold 0.1 M ammonium acetate/MeOH, twice with 10 mL of cold acetone, and once with 1.5 mL of cold acetone. The pellet was then vacuum-dried for 2 min and resolubilized in 1.5 ml of Rehydration Buffer (7 M urea, 2 M thiourea, 4% CHAPS, 20 mM DTT, 1% IPG buffer pH 4-7). Each sample was vortexed, allowed to stand at 4°C for 2 h to resolubilize proteins, and subsequently stored at -80°C.
Protein assays
Protein concentrations were determined using an EZQ™ Protein Quantitation Kit (Invitrogen, Carlsbad, CA, USA), with ovalbumin as a standard according to manufacturer's instructions. Protein concentrations ranged from 1.6 to 5.9 mg/ml.
2D and gel staining
The 2D SDS-PAGE protocol was adapted from O'Farrell [65]. Iso-electric focusing (IEF) was carried out using immobilized pH gradient (IPG) strips (24 cm, pH 4-7, Immobiline™ DryStrip, GE Healthcare, Piscataway, NJ, USA). Samples were thawed on ice and centrifuged (15 min, 4°C, 10,000 × g) prior to loading on IPG strips. A loading volume of 450 μL of protein extract, corresponding to a protein amount of 1.0 mg, was added to each strip. Three mL of mineral oil was added to each well before IEF. Protein IEF was performed at 20°C using a Protean® IEF Cell (Bio-Rad, Hercules, CA, USA) as follows: active rehydration at 50 volts (V) for 12 h, 200 V for 30 min with a linear increase in voltage, 500 V for 30 min with a linear increase in voltage, 1000 V for 1 h with a linear increase in voltage, and 10,000 V with a rapid increase in voltage until a total of 95,000 Volt-hours (Vh) was reached. Strips were then stored at -20°C until further use. Once thawed, the strips were washed for 20 min in an Equilibration Buffer (6 M urea, 30% v/v glycerol, 2 M Tris-HCl pH 8.8, 2% w/v SDS) containing 1% w/v DTT, followed by washing for 20 min with an Equilibration Buffer containing 2.5% w/v iodoacetamide. SDS-PAGE was performed using noncommercial 12% polyacrylamide gels (25 cm × 20 cm × 1 mm) and run at 40 V for 2 h and 120 V for 13 h in a Bio-Rad Protean® II XL 2D Multi-Cell. A coomassie brilliant blue (CBB) G-250 procedure was used to stain the 2D gels. The gels were washed twice in 50% ethanol (EtOH)/2% phosphoric acid/de-ionized water (diH 2 O) v/ v/v for 1 h, transferred to 2% phosphoric acid for 60 min, then washed in 17% ethanol/2% phosphoric acid/ 15% ammonium sulfate v/v/w for 1 h, and finally agitated for 3 days in 17% EtOH/15% ammonium sulfate/ 2% phosphoric acid/0.02% CBB G-250/diH2O v/w/v/w/ v. The 2D gels were imaged using a ScanMaker 9800XL with TMA scanner (Microtek, Hsinchu, Taiwan).
Protein analysis and statistical analysis
Gels from 7LD, 7SD, 28LD and 28SD treatments were compared using PDQuest™ Gel Analysis SW (Bio-Rad) with six replicates (three from each year of harvested plant material). Spots were matched within gels. If no spot was detected, a background value was used for the corresponding area in order to limit the rate of false positives. Average CV was calculated for each experiment. Differences in spot abundance were statistically evaluated using the ANOVA method with GeneANOVA software [66]. The number of detected spots showing differences with a p-value of ≤0.05 was determined. The spots were conserved only if their normalized intensity was higher than 0.01% of the total spot intensity. Differentially abundant spots were manually curated with respect to spot quality (e.g., sharpness, resolution) and the quality of spot matching to reduce false positives. Only spots with ≥ two-fold ratio between photoperiod conditions were conserved.
Protein identification
Spot excision was performed manually, and then trypsin digested according to Rosenfeld et al. [67] using the Investigator™ ProPrep™ (Genomic Solutions, Ann Arbor, MI, USA). The tryptic fragments were analyzed using an ABI 4700 Proteomics Analyzer (Applied Biosystems, Foster City, CA, USA) MALDI TOF/TOF™ mass spectrometer (MS). A 0.5 mL aliquot of a matrix solution containing 10 mg/mL alpha-cyano-4-hydroxycinnamic acid (Sigma-Aldrich, Inc., St. Louis, MO, USA) and 10 mM ammonium phosphate (Sigma-Aldrich) in 70% acetonitrile was co-spotted with 0.5 mL of sample. Data were acquired in reflector mode from a mass range of 700 to 4,000 Da, and 2,500 laser shots were averaged for each mass spectrum. Each sample was internally calibrated if both the 842.51 and 2211.10 ions from trypsin autolysis were present. When both ions were not found the instrument used the default calibration. The twenty most intense ions from the MS analysis, not present on the exclusion list, were subjected to MS/MS analysis. The mass range was 70 to precursor ion with a precursor window of 21-3 Da and an average of 5,000 laser shots for each spectrum. The resulting file was then searched by using automated MASCOT software [URL]:// www.matrixscience.com/ through the IDQuest (Bio-Rad) interface to search the putative proteins obtained from the grapevine PN40024 homozygote genome [68], the Pinot Noir heterozygote genome [69], and the tentative contigs from the DFCI gene index ver. 5.0 [URL]/; ver. 18_9_2006, 23,871 sequences). Peptide tolerance was 20 ppm; 1 missed cleavage was allowed; MS/MS tolerance was 0.8 Da.
Additional material
Additional file 1: additional data for differentially expressed spots. SSP, standard spot number; SD/LD, normalized spot volume in the SD divided by the normalized spot volume in the LD, from 6 different plants; Pval, p-value; Average SD, average intensity value in SD; Average LD, average intensity value in LD; Exp Mr, experimental molecular mass; Exp pI, experimental pI; Th Mr, theoretical molecular mass; Th pI, theoretical pI; Pep, number of peptides mass and in ( ) the number of MS/MS ions matching the query; M score, MOWSE score; % Cov, percentage of coverage; Function, description of protein identity. 8×, protein ID in the gravevine genome with a 8× coverage; 12×, protein ID in the grapevine genome with a 12× coverage; Gels nomenclature: first character, 7 or 28 for the date; second character, S or L for SD or LD; third character, 1, 2, 3 for the replicate number; fourth character, 7 or 8 for the harvested year (2007 or 2008) Additional file 2: Cytoscape session containing the VitisNet molecular networks with proteins presenting outstanding evolution between LD and SD. The session contains a subset of 15 VitisNet molecular networksshowing differential LD and SD protein expression. Tutorial for using VitisNet in Cytoscape can be obtained at [URL].
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2017-06-24T16:06:46.376Z", "created": "2010-08-12T00:00:00.000", "id": "1660350", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://proteomesci.biomedcentral.com/track/pdf/10.1186/1477-5956-8-44", "pdf_hash": "d28b335860fe636891fb38743de7672b17fa4ded", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:4915", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "d28b335860fe636891fb38743de7672b17fa4ded", "year": 2010}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Mechanism of resistance to mesotrione in an Amaranthus tuberculatus population from Nebraska, USA
Amaranthus tuberculatus is a troublesome weed in corn and soybean production systems in Midwestern USA, due in part to its ability to evolve multiple resistance to key herbicides including 4-hydroxyphenylpyruvate dioxygenase (HPPD). Here we have investigated the mechanism of resistance to mesotrione, an important chemical for managing broadleaf weeds in corn, in a multiple herbicide resistant population (NEB) from Nebraska. NEB showed a 2.4-fold and 45-fold resistance increase to mesotrione compared to a standard sensitive population (SEN) in pre-emergence and post-emergence dose-response pot tests, respectively. Sequencing of the whole HPPD gene from 12 each of sensitive and resistant plants did not detect any target-site mutations that could be associated with post-emergence resistance to mesotrione in NEB. Resistance was not due to HPPD gene duplication or over-expression before or after herbicide treatment, as revealed by qPCR. Additionally, no difference in mesotrione uptake was detected between NEB and SEN. In contrast, higher levels of mesotrione metabolism via 4-hydroxylation of the dione ring were observed in NEB compared to the sensitive population. Overall, the NEB population was characterised by lower levels of parent mesotrione exported to other parts of the plant, either as a consequence of metabolism in the treated leaves and/or impaired translocation of the herbicide. This study demonstrates another case of non-target-site based resistance to an important class of herbicides in an A. tuberculatus population. The knowledge generated here will help design strategies for managing multiple herbicide resistance in this problematic weed species.
Introduction
Hydroxyphenylpyruvate dioxygenase (HPPD, EC1. 13.11.27) is a ubiquitous non-hemeoxygenase involved in the catabolism of the amino acid tyrosine [1,2]. Additionally, it is a key enzyme in the synthesis of homogentisate, a precursor of plastoquinone and tocopherol in plants [3]. PLOS population from Nebraska. We have also conducted detailed glasshouse and lab-based studies to determine the mechanism of resistance to mesotrione in this population.
Plant material
The suspected resistant A. tuberculatus population (NEB) was sampled from a seed corn field in Nebraska (near Tarnov), USA in 2010. Seeds were collected from plants surviving a postemergence field application rate of mesotrione (105 g ai ha -1 ). The owner gave permission to collect A. tuberculatus seeds from his land. The field received a soil application of s-metolachlor + atrazine before crop and weed emergence followed by a full labelled rate of tembotrione prior to the mesotrione rescue treatment. A standard sensitive biotype (SEN) was sourced from Herbiseed (Twyford, UK) and used for comparison in all glasshouse and laboratory-based studies.
Confirmation of resistance to mesotrione
Seed treatment. To ensure maximum germination, the seeds from the SEN and NEB populations were sterilised in a 50:50 mix of sodium hypochlorite (Sigma) and water for 10 minutes. Then after, they were rinsed thoroughly with distilled water and placed on 0.4% plant agar (Duchefa) in 10 cm 2 Petri-dishes (Fisher Scientific) and sealed with Parafilm (Sigma). The Petri-dishes were stored at 4˚C for 21 days following which the seeds were dried on tissue paper overnight before sowing.
Pre-and post-emergence mesotrione dose response tests. The seeds of NEB and SEN were sown directly in sandy loam soil to a density that would provide around 15 plants per 10 cm-diameter pots. The pots were watered and maintained in a glasshouse providing a 16 H photoperiod of 180 μmol m -2 s -1 with temperatures of 24˚C day and 18˚C night and 65% relative humidity. For post-emergence application, the plants were treated with herbicides when they were 7 cm tall. For pre-emergence application, the seeds were sown as above and covered with soil prior to herbicide treatment. For both pre and post-herbicide treatments, mesotrione (Callisto 1 480 SC, Syngenta) was applied at 0.8, 1.6, 3.3, 6.7, 13.1, 26.3, 52.5, 105, 210, 420 and 840 g ai ha -1 using a track sprayer fitted with a Teejet nozzle delivering 200 L ha -1 . All herbicide treatments included ammonium sulphate (Sigma) at 2.5% w/v and Agridex 1% v/v (Helena Chemical). Three replicate pots were used per herbicide treatment and population. Following herbicide treatment, the pots were arranged in a randomised block design and maintained in the same glasshouse conditions as before for 21 days, at which time they were assessed for percentage visual biomass reduction compared to an untreated control.
Cross-resistance to other corn selective HPPD-inhibiting herbicides
To determine cross or multiple resistance to other HPPD herbicides, NEB and SEN plants were produced as described above and sprayed post-emergence with tembotrione (Laudis 1 , Bayer CropScience) at 50, 100, 200 and 400 g ai ha -1 and topramezone (Armezon TM , BASF) at 5, 10, 20 and 40 g ai ha -1 , and pre-emergence with isoxaflutole (Balance 1 Pro, Bayer CropScience) at 50, 100, 200 and 400 g ai ha -1 . All herbicide treatments included Agridex at 1% v/v. Three replicate pots per herbicide treatment and population were included. Following herbicide treatment, the pots were arranged in a randomised block design and maintained in the same glasshouse conditions as before for 21 days, at which time they were assessed for percentage visual damage compared to an untreated control.
Mechanism of resistance to mesotrione
Both potential target-site and non-target-site resistance mechanisms were investigated as part of this study. Target-site resistance studies consisted of (1) HPPD gene sequencing to identify potential mutations associated with resistance, (2) testing for HPPD gene duplication, (3) testing for constitutive and inducible HPPD gene over-expression following mesotrione treatment. Non-target-site resistance studies encompassed measuring relative uptake, translocation and metabolism of mesotrione between the sensitive and resistant populations.
HPPD gene sequencing. Twelve individual plants from the NEB population that survived the post-emergence application of mesotrione at 105 g ai ha -1 from the dose response test described above and twelve untreated plants of the SEN population were used in HPPD gene analysis. A RT-PCR approach was adopted to sequence the HPPD gene from the resistant NEB and sensitive SEN plants. For each of the 24 plants, 1 cm 2 of leaf tissue was placed in a 24-well plate, frozen and then ground in a SPEX SamplePrep Geno/Grinder for 1 min at 1000 rpm following which 1.5 ml Tri-Reagent was added, mixed, incubated for 5 minutes and transferred to a 2 ml tube. The samples were centrifuged for 5 minutes at 10,000xg. Subsequently, RNA was isolated from the supernatant using the Zymo Research Direct-Zol RNA mini-prep kit.
HPPD gene duplication and constitutive or inducible gene over-expression. Seeds of NEB and SEN were sown in seed trays (modular soil) to a high density and covered with vermiculite. The seed trays were watered and maintained in a glasshouse providing a 16 h photo period of 180 μmol m -2 s -1 with temperatures of 24˚C day and 18˚C night and 65% relative humidity. One week later, 24 seedlings each from NEB and SEN were individually transplanted to 7 cm pots (to 50:50 soil) and maintained in the same glasshouse conditions as above. Each seedling was labelled individually for later identification. When the seedlings were 7 cm tall, 0.5 cm 2 leaf tissue was sampled from each of NEB and SEN plants, placed in a Costar™ 96-Well Assay Block (Fisher Scientific) and frozen at -80˚C for subsequent DNA analysis. A second lot of 0.5 cm 2 leaf tissues from each of NEB and SEN plants were collected in 2 ml micro-centrifuge tubes and immediately frozen at -80˚C for RNA extraction.
The 48 plants (24 each of NEB and SEN) were then treated with mesotrione (Callisto 1 480 SC) at 105 g ai ha -1 using a tracksprayer fitted with a Teejet nozzle delivering 200 L ha -1 . The herbicide treatment included AMSat 2.5% w/v and Agridex 1% v/v. Forty-eight hours later, 0.5 cm 2 leaf tissue was sampled from the 48 individual plants for future RNA analysis as described above. The plants were subsequently maintained in the same glasshouse conditions as before for 21 days at which time they were assessed for survivorship and visual biomass reduction compared to an untreated control. Twelve most sensitive SEN plants and 12 of the most resistant NEB individuals were chosen for HPPD gene copy number and overexpression experiments.
For gene duplication studies, DNA was extracted from leaf tissues from 12 each of the selected SEN and NEB plants collected prior to herbicide treatment using a Qiagen DNeasy Plant mini kit (Qiagen) as per the manufacturer's instructions. DNA concentrations were determined using a NanoDrop ND-2000 spectrophotometer (Thermo Scientific) and then normalised to 20 ng/μl using autoclaved RO water.
For constitutive and inducible gene over-expression studies, RNA was extracted from the 12 SEN and 12 NEB selected plants using leaf tissues collected before and after mesotrione treatments respectively. RNA from individual plants was isolated using RNAzol 1 RT (Sigma) as per the manufacturer's instructions and treated with DNAse to remove DNA contamination. RNA concentrations were determined using a NanoDrop ND-2000 spectrophotometer (Thermo Scientific) and 550 ng total RNA was employed for each sample for cDNA synthesis using SuperScript1 III Reverse Transcriptase First-Strand Synthesis System (ThermoFisher) with Oligo(dT) primers according to the manufacturer's recommendations. A negative RT control was included for each RNA sample to ensure no amplification occurred through DNA contamination.
Gene copy number and expression of HPPD were determined relative to two control genes: acetolactate synthase (ALS) and carbamoyl phosphate synthetase (CPS). DNA from the SEN population was diluted in water 1:1, 1:2, 1:4, 1:8, 1:16 and 1:32 ratios and used to determine efficiency for each primer pair. The same dilutions were employed for determining the cDNA primer efficiency with SEN cDNA. The primer pairs were HPPD-forward 5'-TGGATCAT GCTGTAGGGAATGTCCC-3' with HPPD-reverse 5'-CATTCATTGGAAACAACACCATTTC ATC-3' and ALS-forward CGCTGCTCAAGGCTACGCTCG with ALS-reverse GCGGGACTGA GTCAAGAAGTGCATC. The CPS primers used were as described in Ma et al. [30]. Each pair of primers was prepared in a Power SYBR1 green PCR master mix (Life Technologies Ltd) and used to amplify 1 μl of DNA and cDNA from each sample. A SEN-DNA bulk was made by mixing 10 μl DNA from the 12 individual SEN samples. Similarly SEN-cDNA before treatment and SEN-cDNA after treatment bulks were generated from corresponding mixtures of individual cDNAs. These three separate bulks were used as the reference controls in all qPCR runs.
The design for the qPCR experiment was as follows: (i) DNA and RNA samples from the 12 individual plants of each two populations were allocated to six 96-well qPCR plates in all; (ii) two biological replicates of each population were assayed per qPCR plate; (iii) each plant was tested in two technical replicates; and (iv) technical replicates from each plant sample were assigned to 18 wells on each plate (six wells for DNA, 6 for RNA before treatment and 6 for RNA following treatment). In these groups of six wells, two each were for HPPD, ALS and CPS gene analysis. The remaining wells were taken up by bulked control and negative control samples.
qPCR was conducted on a StepOnePlus™ Real-Time PCR System (Life Technologies Ltd.) with a program set to 95˚C for 10 min followed by 40 cycles of 95˚C for 15s followed by 60˚C for 1 min, with a final step used to carry out the melt-curve analysis of 95˚C for 15s then 60˚C for 1 min followed by 0.3˚C incremental increases every 15s until 95˚C. 14 C mesotrione uptake and translocation. NEB and SEN plants were grown in individual 7 cm pots at 24/18˚C day/night temperature, 65% relative humidity and irrigated as required. At the 4-leaf stage plants were treated with a [phenyl-U -14 C]-mesotrione (0.6 MBq with specific activity 4.266 MBq/mg) solution supplemented with Agridex at an inclusion rate of 1% v/ v. Unlabelled mesotrione was added to the radioactive solution to provide a treatment rate equivalent to 105 g ai ha -1 in a spray volume of 200 l ha -1 . The mesotrione treatment was delivered in 20 x 0.2μl microdroplets (4μl total) applied in a 1 cm band across the middle of the adaxial surface of selected leaves to give 5,000 dps (1.2μg) per plant. The droplets were applied using a 10 μl Hamilton syringe with a 50 X multi-stepper mechanism. Four replicate plants were treated for each population and time points. The plants were sampled at zero time (5 minutes after the droplets had dried) for recovery comparisons and then at 6, 24, 48 and 72 hours after treatment. Individual plants were sectioned into treated area, meristem, rest of foliage, and stem and roots. Foliar surface residues were recovered by washing the treated leaf with 2 ml of acetonitrile 80:20 water containing 0.1% v/v Tween 1 20 (Sigma). Radioactivity in the leaf rinsate was quantified by liquid scintillation counting (LSC) using a Perkin Elmer Tricarb 2900TR. The different plant sections were then accurately quantified by sample oxidation using a Harvey OX 500 Biological Oxidiser with attached Zinnser robot (R. J. Harvey Instruments). The samples were subsequently quantified by LSC. Percentage uptake was determined by the total amount of radioactivity detected in the plants x 100/total radioactivity applied (washes at T0). Relative herbicide translocation was determined as: (sum of radioactivity from meristem + rest of foliage + stem and roots) x 100/total amount recovered from the plant (including the treated area).
Unlabelled mesotrione metabolism. Individual NEB and SEN plants were grown in 7 cm pots in the aforementioned glasshouse conditions. A treatment solution containing 525 μg/ml of active ingredient was prepared in deionised water using mesotrione (Callisto 1 480 SC) and Agridex at 1% v/v. At the 4-leaf stage, plants were treated with 20 x 0.2μl microdroplets of the solution (using a 10 μl Hamilton syringe with 50 x multi stepper mechanism) on the adaxial surface of the newest fully expanded leaf to provide 2.1 μg mesotrione per sample. Four replicate plants were used per time point and population. The plant tissues were sampled at 5 minutes (after the droplets had dried) for recovery comparisons and then at 6, 24, 48 and 72 hours after treatment. Foliar surface residues were recovered by washing the treated leaf with 2 ml of acetonitrile 80:20 water. A 1 ml aliquot was sampled from the washes for quantification by Liquid Chromatography-Mass Spectrometry (LC-MS).
The plant tissues were dissected into two parts, namely, treated leaf and rest of the plant. The treated leaves were placed in 2 ml MP Bio Fast prep tubes (containing garnet lysing matrix A and a ¼ ceramic sphere) with 1 ml acetonitrile 80:20 water and then frozen overnight. The rest of the plant was placed in 15 ml Fisher Brand centrifuge tubes with 3 ml of acetonitrile 80:20 water and frozen overnight. Foliar samples were removed from the freezer and allowed to thaw before being macerated. The treated leaf extracts were then centrifuged at 10,000 rpm for 15 minutes (using a Thermo Scientific PICO 17 centrifuge) and the rest of plants extracts were centrifuged at 3000 rpm for 10 minutes on a Heraeus 3L kit. A second 1 ml extraction was used for the treated leaf and the supernatants combined. Aliquots of 1 ml were removed from each supernatant and placed in 1.5 ml Chromacol crimp top LCMS vials for quantification of parent mesotrione and known metabolites: 4-hydroxymesotrione and AMBA [2-amino-4-(methylsulfonyl) benzoic acid] using an Acquity BEH C18 column attached to an Acquity Binary solvent manager and sample organiser with Thermo Scientific TSQ Vantage mass spectrometer. The samples were analysed using a reverse phase Acquity UPLC BEH C18 Column (130 Å, 1.7 μm, 2.1 mm X 50 mm). The mobile phase comprised two solutions: eluent A was 0.2% formic acid and eluent B was acetonitrile. The elution profile was as follows: step 1, A:B (95:5, v/v) isocratic for 0.5 min; step 2, A:B (95:5, v/v) to A:B (5:95, v/v) linear gradient for 4 min; step 3, A:B (5:95, v/v) isocratic for 0.4 min; step 4, A:B (5:95, v/v) to A:B (95:5, v/v) linear gradient for 0.1 min; and step 5, A:B (95:5, v/v) isocratic hold for 1 min to re-equilibrate the column. Parent mesotrione and known metabolites were quantified against matrix matched standard calibration curves.
Uptake was defined as the sum (mesotrione + detectable metabolites (4-OH mesotrione + AMBA)) x 100/total applied (washes at T0) in the whole plant. The relative amounts of mesotrione and its metabolites detected in the two plant sections were expressed as a percentage of total mesotrione absorbed.
Statistical analysis
For the pre-and post-emergence dose-response tests on mesotrione, GR 50 estimates were obtained for each population by fitting a least squares logistic regression model of the form: where P denotes the visual percentage damage, x denotes log 10 (Rate), and μ and β denote the logGR 50 and slope parameters respectively [32]. Resistance indices for NEB versus the SEN population were estimated as the ratio of the respective GR 50 estimates and are quoted with 95% confidence limits. A statistically significant (P = 0.05) difference between the populations is concluded when the confidence interval for the resistance index does not include the value 1. This is equivalent to carrying out a t-test between the means of the logGR 50 estimates of the populations in question. Analyses were carried out using SAS version 9.4.
For the qPCR data from the HPPD gene duplication and gene expression tests, separate statistical analyses were carried out on the different DNA and RNA measurements. Prior to analysis, the C T values for each biotype and gene were averaged across the two technical replicates in each plate. When primer efficiencies are equal, testing for differences in HPPD gene copy number and HPPD gene expression between populations is equivalent to comparing populations in terms of the difference between the average C T value for the HPPD gene and that of the ALS and CPS genes in turn. Consequently, the data were analysed using the analysis of variance model: where y ijk denotes the difference between the average C T value for the HPPD gene and that of the ALS or CPS genes, for plant k of population j in plate i, μ denotes the overall true mean, γ i denotes the effect of plate i, β j denotes the effect of population j and ε k(ij) denotes the random error associated with plant k of population j in plate i. The comparison between the populations is then equivalent to carrying out a t-test using the pooled plant-to-plant variation within plates and populations as the source of 'error' variation. The statistical significance of the population comparisons are summarised by p-values, a value of 0.05 or less indicating a statistically significant result.
The uptake, translocation and metabolism measurements were analysed by factorial analysis of variance using the model: where y ijk denotes the transformed measured response for population j at time k in replicate i, μ is the overall true mean response, γ j is the true effect of population j, τ k is the true effect for time k, (γτ) jk denotes the population-by-time interaction and ε ijk is the error associated with each individual response. Since the data analysed were percentages, an arcsine transformation was applied prior to analysis in order to satisfy the assumption of variance homogeneity required for the validity of the pooled error estimate. Where the population-by-time interaction was not statistically significant, populations were compared averaged across time points. Otherwise, comparisons were made separately at each time point.
Mesotrione resistance confirmation test
The sensitive A. tuberculatus population SEN was fully controlled at 26 g ai ha -1 (0.25X field rate) mesotrione, demonstrating high efficacy of the herbicide when tested post-emergence under glasshouse conditions (Fig 1A). On the other hand, at this rate less than 10% biomass reduction was recorded for the resistant population NEB. At the commonly used post-emergence field rate of 105 g ai ha -1 , mesotrione achieved only 37% control of NEB, thereby confirming field resistance to the HPPD herbicide in this population. Some clear but highly stunted survivors were still visible at 420 g ai ha -1 and full weed control was only attained at 840 g ai ha -1 mesotrione. The GR 50 value for NEB was 162.1 (138.8-189.3) compared to 3.6 (3.1-4.1) for SEN amounting to a resistance index (RI) of 45.5 (37.1-55.8). When tested preemergence, lower levels of control were also observed for NEB as compared to SEN (Fig 1B). However, GR 50 values between NEB and SEN were in the same order of magnitude and were estimated at 31.0 (25.1-38.2) and 12.8 (10.8-15.1) respectively. Consequently, the calculated resistance index was much lower at 2.4 (1.9-3.2), with full weed control of the NEB population achieved at around half the labelled field rate (105 g ai ha -1 ) of mesotrione applied preemergence.
Cross or multiple resistance to other HPPD herbicides
The cross or multiple resistance profile was determined for one each of a triketone, isoxazole and pyrazolone herbicide commonly used to control Amaranthus species pre-emergence (isoxaflutole) and post-emergence (tembotrione and topramezone) in corn agro-systems in the USA. All three herbicides were efficacious, achieving full control of the SEN population at half the recommended field rates and above (Fig 2A, 2B and 2C). Unsatisfactory control was observed for NEB with tembotrione and topramezone applied post-emergence, both providing only 20% biomass reduction at the rates that killed the sensitive population. In all cases clear survivors were identified at the recommended field rates for these two latter herbicides. Similarly, isoxaflutole applied pre-emergence provided lower levels of control on NEB compared to SEN at the discriminating rate of 50 g ai ha -1 . The shift in the herbicide response was less pronounced compared to the two other herbicides tested post-emergence, with 100% control of NEB attained within the range of recommended field rates for this isoxazole herbicide.
Mechanism of resistance to mesotrione HPPD gene sequencing. Using primer pairs located on the 5' and 3' UTR of the HPPD gene, RT-PCR generated a 1350 bp fragment for all the 24 plants analysed. The 1305 translated region showed on average 99% and 80% identity to published HPPD sequences from A. tuberculatus (GenBank: JX259255) and Beta vulgaris (GenBank: XM_010690603) samples respectively, thereby supporting the identity of the target gene amplified in this study. Sequence comparison between the 12 SEN and 12 NEB individuals identified 52 nucleotide changes at Table 1. Low levels of variation in relative gene copy numbers were observed between plants within the SEN and NEB populations. The ratios in HPPD copy numbers for NEB vs SEN relative to the CPS and ALS genes were 1.04 and 1.12, respectively. Corresponding p-values were above the threshold for significance at the 5% significance level, indicating that HPPD gene duplication is not linked to resistance to mesotrione in NEB. In contrast to gene copy numbers, the HPPD gene expression levels varied appreciably from plant to plant within the SEN and NEB populations (Fig 3). This was true when expressed relative to both the CPS and ALS reference genes and for RNA samples extracted from leaf tissues before and after mesotrione treatment. Importantly, comparable HPPD expression levels were computed for NEB and SEN both before and after mesotrione treatment. This indicates that resistance to mesotrione in NEB is not due to higher levels of the target-site gene that is expressed constitutively or inducibly following mesotrione application at 48 HAT. 14 C uptake and translocation. On average, around 25% of mesotrione applied was absorbed six hours after treatment. The level of mesotrione uptake increased steadily to reach 60% at the end of the time course experiment (Fig 4A). There was no evidence of a difference in mesotrione uptake between NEB and SEN based on the factorial analysis across time (pvalue = 0.609). To determine whether differential translocation could account for resistance in NEB, the samples were sectioned into treated leaf, meristem, rest of foliage and root and stem. Radioactivity was predominantly recovered in the treated leaf for both the SEN and NEB populations throughout the experiment (Table 2). Nonetheless, relatively more radioactivity was retrieved outside the treated area for the SEN compared to the NEB population (p = 0.014) (Fig 4B). The significant difference in radioactivity recovery was mostly accounted for by the different amounts detected in the meristematic tissues (p = 0.0003). For instance, at 48 hours after treatment, 15.4% radioactivity was detected in the meristem of the SEN population and only 4.4% in NEB. Significantly higher levels of radioactivity were also observed in the rest of the foliage for SEN in comparison to NEB (p = 0.014) although the difference was not as pronounced as for the plant meristems.
Unlabelled mesotrione metabolism. Under our HPLC conditions, mesotrione and its polar metabolites, 4-hydroxymesotrione and AMBA, could be clearly resolved with retention times of 3.5 min, 2.9 min and 2.3 min respectively. Typical HPLC profiles for one each of NEB and SEN plants are provided in Fig 5. The AMBA peak is not visible on the profile given the low levels of this compound detected among all the plant samples analysed. Factorial analysis of the levels of mesotrione and 4-hydroxymesotrione shows clear evidence that the differences between NEB and SEN are time-dependent (p << 0.05) ( Table 3 and Fig 6A). Individual ttests carried out at the separate time points show no convincing evidence of any population differences at the 6h and 24h assessment times (P >> 0.05) but clear evidence of differences at the 48h and 72h assessments (P << 0.05). In this respect, the amount of parent mesotrione remaining in the treated area was as high as 38.1% of total applied for SEN and only 7.5% for NEB at 72 hours. Consequently, relatively higher levels of 4-hydroxymesotrione were detected in the NEB (80.8%) vs the SEN (36.3%) population in the treated leaves. A similar scenario unfolded when mesotrione and its metabolites were quantified in the rest of the plants with significantly higher levels of the parent compound detected for SEN in comparison with NEB at the 48h and 72h but not earlier time points (Table 3 and Fig 6B). For example, 12.5% of applied mesotrione was recovered in the rest of the plant for the SEN population and only 1% for NEB at the 72 hour assessment time.
Discussion
Evolution of resistance to post-emergence application of mesotrione Field resistance to mesotrione applied post-emergence was confirmed in an A. tuberculatus population from Nebraska via whole plant pot assays conducted under controlled glasshouse conditions. The estimated RI (45.5) was relatively high compared to those determined for two other A. tuberculatus populations from Illinois (10-35 fold resistance increase depending on the sensitive population used) and Iowa (RI = 8) [26,28]. The difference in the resistance indices in the three populations could be due to diverse sets of resistance genes and expression levels involved or different proportions of sensitive and recalcitrant seeds collected from field survivors.
Analysis of field treatment histories reveals that the Nebraska and Iowa sites were in seed corn/soybean rotation in alternate years whilst the Illinois population was in continuous seed corn production for seven years prior to resistance being confirmed in the different A. tuberculatus samples [26,28]. In any case, resistance in the three populations evolved relatively quickly, very likely because HPPD herbicides were highly relied upon for controlling emerged and possibly large (> 10 cm) A. tuberculatus plants in the seed corn production years. Inbred seed corn plants are generally not as competitive and are often more damaged by herbicide applications compared to hybrid field corn varieties [33]. Consequently, relatively large Amaranthus spp. plants are allowed to proliferate and accumulate 'creeping' resistance genes which, on their own, would not be sufficient to permit the individuals to survive an HPPD herbicide treatment but when accumulated in a few subsequent generations would lead to resistance to HPPD and other herbicides [34]. The multi-genic and complex nature of resistance to mesotrione is suggested from classical genetics studies in the HPPD recalcitrant A. tuberculatus population from Illinois [35] whilst the mode of inheritance and potential number of resistant genes remain to be determined in the Iowa and Nebraska populations.
Overall, resistance to HPPD herbicides in Midwestern USA has evolved slower and is not as problematic as with other single-site herbicide modes of action such as ALS, photosystem II and EPSPS inhibitors [25]. After more than 15 years of intensive HPPD herbicide use, resistance is fully established in only five A. tuberculatus and three A. palmeri populations since the first reported case in McLean County, Illinois, in 2009 [26]. More recently, a random survey of 187 samples from Missouri has identified three additional A. tuberculatus populations that survived a single discriminative rate of mesotrione in a glasshouse experiment [36]. Additional dose response tests using larger number of individuals per herbicide rate are required to confirm the resistance status of these latter three populations. The few instances of resistance to HPPD-inhibiting herbicides in corn agro-systems in the USA may be explained by the fact that they are typically used in two, three and even four-way mixtures with compounds belonging to other modes of action. One of the preferred mixing partner is the PSII-inhibitor, atrazine, which acts synergistically with HPPD herbicides [22]. Synergism between these two herbicide modes of action has been demonstrated in PSII-sensitive, PSII-resistant and PSII/ HPPD resistant populations, thereby contributing to the long-term sustainability of HPPD herbicides for controlling Amaranthus spp. [23,24,27]. Of concern, however, is the fact that the few A. tuberculatus and A. palmeri populations that are resistant to HPPD herbicides are also recalcitrant to two, three and even four other herbicide modes of action, consistent with the ability of these two dioecious and highly prolific species to accrue resistance to different classes of herbicides [25,37,38]. All the populations that were not controlled with HPPD compounds were also resistant to PSII and ALS herbicides, as these products have been widely used in corn/soybean production in Midwestern USA for over 25 years. In addition to HPPD, PSII and ALS inhibitors, the latest A. tuberculatus population identified in Champaign County, Illinois, is also recalcitrant to protoporphyrinogen oxidase inhibitors as well as synthetic auxin herbicides [39]. This scenario seriously limits the number of effective chemical options for managing such multiple resistant weed populations.
Mechanism of resistance to mesotrione applied post-emergence
Detailed mechanism studies have showed that resistance to mesotrione in the Nebraska population is not due to a target-site mutation or HPPD gene duplication, in agreement with previously published data in other A. tuberculatus and A. palmeri populations [30,31]. The absence of a target-site resistance mutation in all the Amaranthus spp. populations investigated to date contrasts with what was observed with other single-site herbicide modes of action following similar use intensity. Target-site insensitivity due to a subtle amino acid change in Amaranthus spp. was documented for ALS, PSII, PPO and EPSPS inhibitors only after a few years of widespread usage [40][41][42][43][44]. The lack of evolved HPPD target-site resistance mutations may be explained by the fact that HPPD herbicides are competitive inhibitors and the target enzyme may not tolerate many amino acid changes without compromising catalytic activity [16,45]. Additionally, mutagenesis studies in view of engineering HPPD tolerance in dicotyledonous crops have shown that individual mutations are not sufficient to confer resistance to HPPD herbicides as is the case for some ALS and PSII resistance mutations [46,47]. Indeed, target gene mutations and over-expression had to be introduced in an already tolerant monocotyledonous HPPD to endow sufficient levels of resistance to HPPD herbicides in soybean [10,48,49]. Another important contributing factor is that HPPD inhibitors are almost always used in mixtures with other overlapping herbicide modes of action, thereby limiting the risk of selecting for an HPPD target-site resistance mutation in Amaranthus spp. Resistance in the Nebraska populations was neither due to constitutive nor mesotrioneinducible over-expression of the HPPD gene. This differs with data generated on two A. palmeri populations from Kansas and Nebraska whereby mesotrione resistance appeared to be associated, in part, with higher numbers of HPPD transcripts compared to three sensitive populations [31]. The observed increase in target gene expression levels varied from 5-12 depending on the sensitive population being considered [31]. Contrary to gene copy number, the level of HPPD expression as measured by RT-qPCR was quite variable between plants within the SEN and NEB populations. For instance, up to 10-fold difference in HPPD gene expression was detected between resistant NEB plants whilst this figure was as high as 30-fold between two extreme SEN individuals. Unless a sufficiently large number of biological replicates are evaluated, HPPD gene expression data should be treated with caution when drawing conclusions about the potential contribution to resistance to HPPD inhibiting herbicides.
Resistance to mesotrione in NEB is due to enhanced detoxification of the parent compound into 4-hydroxymesotrione, thus mirroring the selectivity basis of the HPPD herbicide in naturally tolerant corn [2,16]. The cytochrome p-450 mediated hydroxylation occurs so rapidly in the crop that translocation outside the treated area is limited. In contrast, slow metabolism in sensitive weeds allows ample translocation of mesotrione to other parts of the plant by both acropetal and basipetal movement. Cytochrome p-450 mediated metabolism via hydroxylation of dione ring at the 4' position was also found to account for resistance to mesotrione in the A. tuberculatus population from Mclean County, Illinois [30]. Convergent resistance by the same mechanism in these two distant populations attests for the growing evidence that resistance in highly heterogeneous and prolific weed species such as A. tuberculatus and A. palmeri occurs primarily by spontaneous evolution from standing genetic variation in the field rather than by migration from an initial location [50]. Whilst increased detoxification of mesotrione at the treated area was clearly established in NEB, it remains to be determined whether resistance could also be due to impaired translocation of the parent compound to other plant parts as well. Indeed, lower levels of radioactivity were detected outside the treated area and in particular in the meristematic tissues for NEB compared to SEN. The difference in radioactivity levels outside the treated area could be due to reduced mesotrione translocation, as is unambiguously demonstrated for other herbicides such as glyphosate and paraquat in various resistant grass or broadleaf weeds and more recently as documented for 2,4-D in wild radish [51][52][53][54]. Conclusion of impaired transport in the latter studies was facilitated by the fact that differential metabolism of glyphosate, paraquat and 2,4-D was not a contributing factor in resistance. Resolution of the potential contribution of reduced translocation of mesotrione in NEB could be achieved by analysing the fate of other experimental HPPD inhibitors sharing similar physicochemical properties to mesotrione but blocked and metabolically robust on the aryl-dione ring [55].
Cross or multiple resistance to foliar-applied HPPD herbicides
The Nebraska A. tuberculatus population was multiply resistant to tembotrione and topramezone applied post-emergence. The cross-resistance between foliar-applied HPPD inhibiting herbicides is in line with what was observed for some other A. tuberculatus and A. palmeri populations [26,27]. When applied post-emergence at their commercial field rates, mesotrione, tembotrione and topramezone provided unsatisfactory control (27%, 31% and 58% respectively) for the HPPD resistant A. tuberculatus population from Illinois (MCR) whilst the two reference populations used for comparison were completely killed [26]. Similarly, a 4-23-fold resistance increase was estimated for mesotrione, tembotrione and topramezone for the A. palmeri population from Nebraska [27]. Since both mesotrione and tembotrione were used for dicotyledonous weed control in the Nebraska field, they may have co-selected for resistance to HPPD herbicides in NEB. Tembotrione and mesotrione belong to the same triketone HPPD herbicide subgroup and as such share a similar chemical structure and liability with regard to metabolism, especially on the aryl-dione moiety [2,11]. We therefore hypothesize that resistance to tembotrione in NEB also occurs by increased detoxification by 4-hydroxylation on the dione ring, very similar to what was established for mesotrione. This idea is supported by a genetic study on a cross between two corn varieties that are sensitive and resistant to HPPD herbicides [56]. Analysis of the progeny identified a single major resistance locus and, importantly, a close linkage between tolerance to mesotrione and tembotrione in corn.
It is noteworthy that NEB was never pressured with topramezone in the field, yet significant levels of resistance were observed for this HPPD inhibitor as well. It therefore appears that the gene(s) selected by mesotrione and/or tembotrione has conferred resistance to topramezone in the Nebraska population. Topramezone belongs to the pyrazolone HPPD herbicide subclass and the selectivity basis in corn is primarily through N-demethylation at the pyrazole ring [14]. It remains to be determined whether NEB impersonates corn and degrades topramezone by N-demethylation or via ring or alkyl hydroxylation at a liable position, as is the case for mesotrione.
Control of NEB with HPPD herbicides applied pre-emergence
A modest resistance index of 2.4 was computed between the Nebraska and the standard sensitive population when mesotrione was applied pre-emergence. NEB was fully controlled at half the recommended field rate of soil-applied mesotrione whilst plant survivors were recorded at up to 4X the commonly use rate of the herbicide applied post-emergence. Similarly, NEB was killed within the range of recommended rates of isoxaflutole applied pre-emergence. Therefore, it appears that the metabolic resistance mechanism to HPPD herbicides identified in emerged A. tuberculatus plants is not significantly expressed at the seed germination stage. An improvement in weed control was also observed when mesotrione was applied post-emergence on smaller and more vulnerable individuals compared to plants at later growth stages for the A. tuberculatus from Illinois [57]. The same was true for several other herbicides that are effective on Amaranthus spp. For instance, atrazine was more effective on a GST-based metabolic resistant A. tuberculatus population from Illinois when applied pre-emergence as opposed to post-emergence [58]. Up to a 10-fold gain in PPO herbicide efficacy was also reported for a target-site (210 codon deletion) resistant A. tuberculatus population treated at pre-emergence as opposed to 7 cm tall plants [59]. Therefore, targeting populations early in the season when the plants are small, or even more so at the seed germination stage, appears to be a good strategy for maintaining the efficacy of HPPD and other herbicides that are active on Amaranthus spp..
Conclusion and future research
We have confirmed high levels of resistance to mesotrione, tembotrione and topramezone applied post-emergence in an A. tuberculatus population from Nebraska, USA. Mesotrione and isoxaflutole applied pre-emergence are still effective on the NEB population, suggesting that the gene(s) endowing resistance to HPPD herbicides in emerged A. tuberculatus plants is not appreciably expressed at the seed germination stage in NEB. Resistance due to enhanced metabolic breakdown of mesotrione to 4-hydroxymesotrione has been clearly established, similar to what is documented in corn and the MCR population [30]. It remains to be determined whether other non-target-site resistance mechanisms apply, in particular reduced cellular transport or whole-plant translocation, by exploring the relative movement of metabolically blocked experimental triketones. Given the dissimilar structures and corn selectivity basis between mesotrione/tembotrione and topramezone, further research will investigate the physiological mechanism by which NEB is resistant to the pyrazolone herbicide topramezone.
Supporting information S1 File. Supporting information file for whole plant dose response, qPCR, uptake, translocation and metabolism tests. (XLSX)
==
Domain: Biology Medicine Agricultural And Food Sciences
|
[{"added": "2018-04-03T04:21:53.070Z", "created": "2017-06-29T00:00:00.000", "id": "23408218", "metadata": {"extfieldsofstudy": ["Biology", "Medicine"], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0180095&type=printable", "pdf_hash": "606cdf5c5c0263cca99a324297c41c50fabcde9e", "pdf_src": "PubMedCentral", "provenance": "peS2o-0028.json.gz:2496", "s2fieldsofstudy": ["Agricultural And Food Sciences"], "sha1": "606cdf5c5c0263cca99a324297c41c50fabcde9e", "year": 2017}, "source": "pes2o/s2orc", "version": "v3-fos-license"}]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.