Infectious environments select oysters resistant to POMS
In 2015, oyster broodstocks were collected in two distinct geographic areas (Brittany-Atlantic coast and Gulf of lion- Mediterranean coast) and two sampling sites (farming, high biomass and non-farming, low biomass) from each area (Fig. 1). These broodstocks were used to produce 12 biparental families (3 families from each origin and sampling sites). Three additional biparental families were produced from broodstocks originating from a mass selection program for higher survival for POMS [16] (Fig. 1). These 15 oyster families were subjected to four infectious challenges performed with two infectious environments (Atlantic and Mediterranean) and two experimental procedures (mesocosm and field infections) (Fig. 2 and Table 1).
High variability in percentages of mortality, ranging from 1% to 100%, was observed among families (Table 1). Family F15 showed the most susceptibility with a mortality rate higher than 97% for any infection trial. In contrast, Family F21 showed the highest resistance whatever the infection trial. Taking into account the 15 families and the 2 infectious environments, the percentage of mortalities observed in the field were not significantly different than those obtained in mesocosm conditions (Mann-Whitney test, p = 0.06). Overall, disease susceptibility was quite similar for the different families in the different infection trials (Table 1). Mortalities observed for the 15 families were 1.75 fold more important for the Atlantic infectious environment than for the Mediterranean infectious one (Mann Whitney, p = 0.02) (Table 1).
As expected, the 3 families (F21, F23 and F28) produced from broodstocks coming from the mass selection program [16] displayed low percentages of mortality (Fig. 3 and 4). Similarly, families obtained from wild oysters sampled in a farming area and putatively submitted to disease selection were also more resistant than those from non-farming areas ((Mann Whitney, p = 0.018), Fig. 4).
To confirm POMS disease in the different mesocosm experiments, we quantified OsHV-1 and total vibrio loads by qPCR (Additional file 1). We observed the colonization of oyster flesh by OsHV-1 and vibrios in both Atlantic and Mediterranean mesocosm experiments 72 hours post-exposure.
The 6 families, retained for the subsequent comparative transcriptomics, were the three best oyster families for POMS resistance (F21, F23 and F48 renamed RF21, RF23 and RF48, respectively) and the three worst (F11, F14 and F15, renamed SF11, SF14 and SF15, respectively) (Fig. 3).
Stress and immune functions are enriched in the basal transcriptome of resistant oysters
To identify putative transcriptomic determinants associated with POMS resistance, we compared the basal transcriptome profiles of the 6 selected families (RF21, RF23, RF48, SF11, SF14 and SF15), maintained in the same hatchery conditions and without disease challenge. We sequenced a total of 36 RNA-seq libraries (6 families, 2 independent experiments and 3 replicates for each experiment). Sequencing yielded between 30.1 and 39.3 million Illumina single reads per sample of which 70.1–74.9% mapped to the C. gigas V9 reference genome.
From these RNA-seq data, we compared the basal transcriptome of each resistant family to the three susceptible families using DEseq (DEseq p-value < 0.05). The differentially expressed genes (DEGs) common between each comparison were retained for further analysis. This strategy identified 3304, 2711 and 3259 DEGs modulated in the same way (up- or down-represented for the three comparisons) in RF21, RF23 and RF48, respectively (Fig. 5A). Among these DEGs, (i) 299 were differentially expressed by the three resistant families, (ii) 924 were differentially expressed by both RF21 and RF23, (iii) 261 were differentially expressed by both RF23 and RF48, and (iv) 308 were differentially expressed by both RF21 and RF48 (Fig. 5A and Additional file 2). The remaining 1773, 1227 and 2391 DEGs displayed a specific differential expression in RF21, RF23 and RF48, respectively (Fig. 5A and Additional file 3, 4 and 5). A previous study has evidenced that the resistance to POMS is associated to an early antiviral response that blocks OsHV-1 replication [17-19]. Indeed, 308 DEGs associated to antiviral defence are early induced in resistant oysters and among them, 61 are differentially expressed at basal level between resistant and susceptible families in the present study. These genes are highlighted in red in the additional files 2, 3, 4 and 5. A part of them are related to TLR-NF-κB, JAK-STAT and RLR-STING antiviral signalling pathways (indicated in red in Figure 6).To determine the enriched functions, we used a gene ontology (GO) enrichment analysis. As the mechanisms underlying resistance can be specific or shared by oyster families, we performed the enrichment analysis on DEGs for each resistant family separately and also on DEGs shared by at least two of the three resistant families. First, the analyses performed on DEGs of each resistant family separately showed a limited number of enriched functions (7, 6 and 5 for RF21, RF23 and RF48, respectively; Fig. 5B, 5C and 5D, respectively). Interestingly, four functional categories (“defense response to other organism”, “response to external stimulus”, “defense response” and “response to stress”) were enriched for the 3 resistant families. It is noteworthy that RF21 and RF23 shared two additional enriched categories (“receptor-mediated endocytosis” and “protein modification by small protein conjugation or removal”). Finally, a functional category related to the “actin polymerization and depolarization” showed enrichment in RF48 only, while a functional category related to “ubiquitin-dependent protein catabolic process” showed enrichment in RF21 only. Second, the GO enrichment analysis on DEGs shared by at least two of the three resistant families (1 792 DEGs, Additional file 2) revealed five enriched functional categories (Fig. 5E): “defense response to other organism”, “response to external stimulus”, “defense response”, “response to stress” and “protein modification by small protein conjugation or removal”. DEGs falling in these five enriched categories (374 genes) are shown in Additional file 2 (CGI indicated in yellow).
Resistant oysters differentially express common and specific immune genes
To further delineate the molecular mechanisms underlying POMS resistance shared by the different resistant families, we first analysed the 374 DEGs belonging to the above identified functions (ie. defense response, response to stress, defense response to other organism, response to external stimulus and protein modification by small protein conjugation or removal) and shared by at least two of the three resistant families (CGI indicated in yellow in Additional file 2). Among these genes, we found members of large multigene families known to be involved (i) in stress response like heat shock proteins (HSP) and glutathione S-transferases, (ii) in protein modifications like ubiquitin ligases and Tripartite Motif containing proteins (TRIM), (iii) in maintenance of DNA integrity and repair like Poly [ADP-ribose] polymerases (PARP), nucleases and helicases, (iv) in PAMP (Microbe Associated Molecular Pattern) recognition (PRR) like C1q domain containing proteins, lectins, scavenger receptors (SR), Fibrinogen domain containing proteins, hemagglutinins and (v) in antiviral defense like IFI44 proteins. We also identified a series of genes putatively involved in antiviral defense and signaling (TLR-NF-κB, JAK-STAT and RLR-STING pathways, Fig. 6A). A putative endosomal Toll-like receptor 13 was overexpressed in the 3 resistant families. A tRNA adenosine deaminases (ADAT) was over-represented in RF21 and RF23. A stimulator of interferon genes (STING) is under-represented in RF23 and RF48 (see Fig. 6A for CGI number for each DEG).
Finally, we analyzed DEGs for each resistant family belonging to the enriched categories described in Figure 5 (i.e. defense response, response to stress, defense response to other organism, response to external stimulus, protein modification by small protein conjugation or removal, receptor-mediated endocytosis, ubiquitin-dependent protein catabolic process via the multivesicular body sorting pathway and actin polymerization or depolymerisation; the corresponding CGIs are highlighted in yellow in Additional File 3, 4, and 5). This analysis highlighted specific processes associated with resistance in each resistant genotype. These genes represented 371, 251 and 315 DEGs in RF21, RF23 and RF48, respectively. In these specific sets of DEGs, we again found several genes belonging to the same large multigene families reported above (HSP, glutathione S-transferases, ubiquitin ligases, TRIM, PARP, nucleases, helicases, PRR and IFI44). In addition, several genes involved in antiviral and signaling pathways were also found differentially expressed in each resistant family specifically (TLR-NF-κB, JAK-STAT and RLR-STING pathways, Fig. 6B). Transcripts corresponding to a Toll-like receptor (TLR), 2 myeloid differentiation primary response 88 (MyD88), a TNF receptor-associated factor (TRAF), a deoxynucleoside triphosphate triphosphohydrolase (SAMHD1), and 2 stimulator of interferon genes (STING) were differentially represented in the RF21 family (see Fig. 6B for CGI number for each DEG). Transcripts corresponding to 2 TLRs, a MyD88, a TRAF, a 2’,5’- oligoadenylate synthase (2’,5’-OAS) and an interferon regulatory factor (IRF), were differentially represented in the RF23 family (see Fig. 6B for CGI number for each DEG). However, in these 2 families, the majority (9/13) of these DEGs was under-represented in comparison with susceptible oysters (Fig. 6B). In contrast, the majority (10/11) of DEGs in the RF48 family was over-represented in this resistant family. They corresponded to an interleukin-1 receptor-associated kinase (IRAK), a NF-κB p105 subunit, 2 Serine threonine- kinases TBK1, a signal transducer and transcription activator (STAT), 2 suppressors of cytokine signaling (SOCS), a STING and 2 IRFs (see Fig. 6B for CGI number for each DEG). Only a 2’,5’- oligoadenylate synthase (OAS) appeared under-represented in RF48.