Hemocyte transcriptome of oysters with an inherent capacity to survive Vibrio infections
A nonlethal sampling method for hemolymph collection [23] was applied to obtain the hemocyte basal gene expression profile before a challenge by pathogenic bacteria. Hemocyte samples were collected from 180 tagged oysters that were further experimentally infected after an 8-day recovery. Vibrio strains, V. splendidus LGP32 (8 x 107 CFU/oyster) or V. aestuarianus LPi 02/41 (2 x 107 CFU/oyster), which were isolated during 2006 summer mortality episodes in France, were used for experimental infections. Survival rates were 59 % and 33 % (at day 4) for animals infected by V. splendidus LGP32 and V. aestuarianus LPi 02/41, respectively. Hemocyte mRNA samples collected prior to Vibrio infection were then categorized according to oyster survival. Thus, two DGE libraries were constructed from hemocytes of oysters able or not able to survive Vibrio infections. They were named SVir-8d [GEO: GSM667902] and NSVir-8d [GEO: GSM667901], respectively.
A total of 3,605,503 and 3,850,124 tag sequences (17-bp) were obtained from the whole sequencing of SVir-8d and NSVir-8d libraries, respectively, corresponding to 56,879 unique tags. A comparison of tag occurrences (incidence of each tag in libraries) revealed that 5,212 unique tags were differentially represented between libraries (>5-fold change) (Figure 1A). Tag-mapping was performed against the C. gigas EST database (29,745 unique sequences: 7,940 contigs and 21,805 singletons) containing 13,898 ESTs from hemocytes [18]. From the 1,049 EST-matched sequences, only 385 sequences showed homologies to known proteins. Thus, no striking relationship between biological functions and oyster capacity to survive infections was clearly evidenced. However, the SVir-8d library appears to be more enriched in genes related to genetic information processing and immune response compared to NSVir-8d, which was enriched in metabolism and cytoskeleton reorganization (Figure 1B).
Figure 1
Hemocyte transcriptome of oysters with distinct capacities to survive Vibrio infections. (A) General characteristics of DGE libraries generated from oysters able (SVir-8d) and not able (NSVir-8d) to survive infections by the pathogens Vibrio splendidus LGP32 and V. aestuarianus LPi 02/41. (B) Functional annotation of EST-matched sequences differentially represented between DGE libraries (>5-fold change).
Oyster hemocyte transcriptome-wide analysis reveals a high polymorphism in basal gene expression at individual level
To assess whether gene expression profiles were related to survival capacity phenotypes, we have explored the DGE data on individual oysters by high-throughput microfluidic qPCR approach [22]. Three new independent experimental infections were conducted to obtain individual hemocyte samples from oysters able (S) and not able (NS) to survive infection, as previously done for DGE library construction. We focused on V. splendidus which has been shown dominant in field mortalities, since 2009, comparatively to V. aestuarianus. Final survival rates of 50 % were obtained and, among the 180 hemolymph samples issued from the three independent infections, 90 samples, corresponding to 45 oysters for each phenotype, yielded high quality RNA suitable for further analyses.
A total of 280 hemocyte-expressed genes were chosen from the 1,049 sequences differentially expressed (>5-fold change) between SVir-8d and NSVir-8d DGE libraries. Both GenBank annotated (62 %) and non-annotated (38 %) sequences were considered. The technical efficiency of three independent 96.96 dynamic arrays was assessed by measuring the expression of the three reference genes in each qPCR chip (ANOVA test). Then, the expression profile of the selected genes was assessed to compare their mRNA abundance according to the oyster survival phenotypes (S versus NS) at the individual level. Surprisingly, the analysis of the large qPCR data set revealed an extraordinary interindividual variability in gene expressions for most of the transcripts measured. As illustrated in Figure 2A, striking inter-individual differences in basal gene expression were seen for some genes with up to 20-fold change, for example, for the follistatin. To our knowledge, such a high inter individual variability in basal gene expression has never been previously reported in oysters.
Figure 2
Basal gene expression polymorphism. Relative mRNA abundances of four hemocyte genes in eight individual oysters are shown as examples of (A) individual differences in basal gene expressions (follistatin and intersectin-1); (B) presence or absence (Ø) of basal expression for an interferon-induced 44-like protein (IFI44L) and a baculoviral inhibitor of apoptosis repeat-containing protein 3 homologue (BIRC3). Note that each individual displays a unique profile of gene expression for these four transcripts. Relative expression levels were calculated according to the 2-ΔΔCq method normalized with the ribosomal L40 protein (Cg-rpl40).
Additionally, the high-throughput qPCR analyses revealed numerous genes whose basal levels of expression could not be detected in some individuals (Figure 2B). Interestingly, those GenBank annotated sequences appeared to correspond to genes involved in immune responses, including an interferon-induced protein 44-like, a tumor necrosis factor ligand superfamily member 10 (TNFSF10), a member of oyster lysozymes (Cg _lysoz2), a baculoviral inhibitor of apoptosis repeat-containing protein 3 homologue (BIRC3) and members of the big defensin antimicrobial peptide family (Cg-BigDef) [24]. However, no significant relationships could be found between the expression detection of these genes and the S or NS phenotypes (ANOVA test).
A hemocyte 14-gene expression signature as a predictor of oyster survival capacity
We have investigated the correlation between the profiles of hemocyte basal gene expression and the oyster survival capacity by the analysis of the mRNA abundance of genes presenting exploitable quantitative values (5 ≤ Cq ≤ 25). Due to well known oyster DNA sequence polymorphism, rigorous melting curve analysis was required to ensure qPCR specificity and avoid misinterpretations of gene expression data [25]. Thus, we only considered the qPCR data of 233 genes, from the 280 initial ones, expressed in the 45 S and 45 NS individual oysters. In spite of the high gene expression polymorphism, statistical partial least-squares (PLS) regression [26, 27] analysis revealed that the basal expression levels of a core set of genes were associated to S or NS phenotypes (Figure 3A). To refine our analysis, we further considered only those genes that did not display high individual variability of basal expression among the oysters from a same phenotype. Thus, the combination of the mRNA abundance of 50 genes was identified that comprises a set of 19 genes highly transcribed in the S phenotype and a set of 31 genes highly transcribed in the NS phenotype (Additional File 1).
Figure 3
Partial least-squares (PLS) discrimination between hemocyte-expressed genes and oyster survival phenotypes (Survival and Non-Survival). (A) The graphic shows the X-loadings (w*) of the X variables (qPCR data) and the Y-loadings (c) of the Y variables (S and NS phenotypes), and thereby shows the relationship between X and Y variables. The variables X (black triangles) and Y (black squares) combine in the projections, and the variables X relate to the variables Y. (B) Relationship between hemocyte-expressed genes and oysters from S or NS phenotypes using PLS regression. The cross-validation led to two components represented as t[1] and t[2].
The expression profile of these 50 hemocyte-expressed genes (from two technical replicates) was further used to foresee oyster survival using the expression data from oysters from both survival phenotypes. The X-loadings (w*) corresponding to the data generated by qPCR and the Y-loadings (c) corresponding to the oyster S and NS phenotypes are presented in Figure 3B. The applied PLS model segregated most of the 90 individuals according to their survival capacity (R²Y = 0.831) (Figure 3B). To identify the hemocyte transcripts statistically related to each phenotype, we additionally applied appropriated fold change cut-offs (x > 1.7-fold change for overexpression and x
Table 1
List of the hemocyte-expressed genes comprising the
Crassostrea gigas
14-gene survival signature
mRNA abundance of the 14-gene markers is revealing an inherent genetic trait rather than a particular physiological status
We confirmed that the survival signature revealed an inherent genetic trait in oysters rather than a pre-stimulation or a physiological response. Indeed, high mRNA abundance of the genes of the signature could have resulted from an increase in gene transcription for some individuals before the experiment. For that, oysters (3 pools of 10 animals/conditions) were injected with a sublethal dose of V. splendidus LGP32 (5 x 107 CFU/animal) or sterile sea water (SSW). Non injected oysters were used as a control. As previously reported, a great variability in transcript rates was observed for both unchallenged and immune challenged (Vibrio or SSW) oysters. Two main types of expression profiles were obtained among the signature genes (Figure 4). Nine genes displayed no changes in transcript abundance at 24 h following either bacterial or SSW injection, compared to unchallenged animals as illustrated for multiple EGF-like domains 10 and cytochrome c (Figure 4). Besides, five genes, including the unknown BQ427036.1.a.cg2 and zinc finger HIT domain-containing protein 3 shown in Figure 4 as examples, displayed a decrease of transcript abundance in circulating hemocytes following both the bacterial and SSW injection, thus independently of an acute response to the Vibrio. No up-regulation was seen for any of the 14 hemocyte-expressed genes following the bacterial challenge or SSW injection. Thus, mRNA abundance of the 14 genes of the survival signature reveals, in oysters, basal gene expression independent from their physiological response.
Figure 4
Relative expression profile of transcripts from the 14-gene survival signature following oyster challenge. The two types of expression profiles obtained are illustrated taking 4 genes among the signature, as examples. Hemocyte gene expression was analyzed by qPCR in unchallenged oysters (white bars) and in oysters 24 h after Vibrio splendidus LGP32 (black bars) or sterile sea water (grey bars) injection. Results are presented as mean values from 10 individuals per conditions. Relative expression levels were calculated according to the 2-ΔΔCq method normalized with the ribosomal L40 protein (Cg-rpl40). Asterisks (*) indicate significant differences of gene expression between conditions (P
Polymorphism in hemocyte basal gene expression can be associated to variations in gene copy numbers in oyster genome
In an attempt to understand the genetic basis involved in the transcriptional regulation of the genes associated to oyster survival, we investigated whether variability in gene expression would reflect gene copy number variations (CNV) in oyster genome. A relative quantification approach was developed to verify a potential correlation between gene copy abundance and mRNA expression of the oyster survival signature. For that, both genomic DNA and total RNA were extracted from ten oysters showing striking differences in gene expression of seven genes from the 14-gene survival signature for which pairs of primers present same efficiency and specificity on gDNA and cDNA. Variations in gene copy number were shown for all analyzed genes. Higher variations in gene copies (from 4.5 to 120-fold changes) than in expression levels (from 2.2 to 16-fold changes) were seen for four genes (as example Figures 5A and 5D). Conversely, for the other three genes, the variations in expression were higher (from 4.5 to 250-fold changes) than in gene copy number (from 2 to 6-fold changes) (as example Figures 5B and 5 C). Finally, among those seven genes, only two presented positive correlation (p
Figure 5
Examples of correlations between relative abundance of gene copy and basal gene expression levels for four gene markers. Relative expression and gene copy abundance were estimated by qPCR at individual level (n = 10) for (A) wy0aba11yc21fm1.1.a.cg.2, (B) cdn20p0004b10.f.1.a.cg.2, (C) Multiple EGF-like domains 10, and (D) wy0aba18yb03fm1.1.a.cg.2. Relative expression levels were obtained according to the 2−ΔΔCq method using the Cg-rpl40 as constitutively expressed gene. Relative abundances of gene copy were obtained with the same method but using the single copy Cg-bpi gene as reference. Grey line represents linear fit, significant correlations are indicate by P