In 1991, the first paper reporting a viral infection associated with mass mortalities in European sea bass (Dicentrarchus labrax L.) was published (Breuil et al. 1991). The viral infection, presently known as viral nervous necrosis (VNN), still represents a major threat for the sea bass industry. The lack of chemotherapeutics and vaccines that can be effectively used to control the disease and the significant additive genetic variation for resistance to VNN in sea bass (Doan et al. 2017; Palaiokostas et al. 2018; Faggion et al. 2021; Griot et al. 2021) make selective breeding directed to the enhancement of host resistance a promising and sustainable approach to prevent and control mortality derived from VNN outbreaks. Traditional selective breeding approaches rely on estimated breeding values (EBV) predicted using individual phenotypes of the breeding candidates or their relatives through estimated additive genetic relationships based on their pedigrees (Zenger et al. 2019). Routine individual phenotyping for complex traits such as disease resistance is either difficult or unfeasible due to high costs and time requirements, and the implementation of genomic selection procedures might be greatly beneficial. Recently, the integration of functional genomic information into genomic prediction models has been proposed as a strategy to improve genomic prediction accuracy: expressed and regulatory genomic regions are characterized and all the obtained resources are used to efficiently predict phenotypes or EBV (Clark et al. 2020). This strategy is expected to more efficiently detect causative variants and to enhance the prediction accuracy of the genetic merit of future breeding candidates across generations, when the reference population is likely to consist of gradually distant relatives of the animals to be predicted. In this study, we assessed whether the integration of functional data into genomic prediction models could improve the prediction accuracy of breeding values for VNN resistance in European sea bass.
Materials and methods
906 juvenile fish (body weight: 6 to 20 gr) produced in a full-factorial mating (25 sires × 25 dams) were subjected to a 29-days VNN challenge test. VNN resistance was recorded both as a binary trait and as time to death. The experimental fish (N = 906) were genotyped using a high-density SNP panel (Peñaloza et al. 2021; ~27,740 SNPs after quality control), while their parents (N = 50) were whole-genome sequenced and used to impute the offspring to whole-genome genotypes. A genome-wide association study was performed to identify SNP markers associated to the traits of interest. Loci that explain a fraction of the genetic variance of gene expression phenotypes were detected through an eQTL (expression quantitative trait loci) analysis, and ATAC-Seq analysis were performed to detect open chromatin regions corresponding to active regulatory or functional elements in the genome.
SNP genotypes were used as predictors of VNN resistance EBV and Bayesian models were fitted to the data. Genomic predictions were performed following different criteria: 1) pre-filtering genetic markers on the basis of their localization in open chromatin regions; 2) weighting genetic markers on the basis of their localization in regulatory regions; 3) weighting genetic markers on the basis of the p-value of eQTLs; 4) weighting genetic markers on the basis of chromatin status score; 5) combining all the aforementioned criteria in an index and using it to filter the SNPs.
Accuracies of the models were assessed in a cross-validation approach generating training and testing sets consisting of animals of varying genetic relatedness according to the genomic relationship matrix, to mimic genomic selection scenarios where the genomic prediction equation is generated by training models using data from a reference population that is either more closely or more distantly related to the animals to be predicted.
Genome-wide association analyses revealed a major QTL associated with VNN resistance phenotypes on linkage group 12. A total of 528 and 578 SNP markers were identified as significantly associated with VNN resistance phenotypes, both as a binary trait and time to death (false discovery rate, FDR, < 0.05). Using evidence from eQTLs and chromatin accessibility data, a putative causal variant was identified.
Results from the GWAS are consistent with those reported in the literature (Griot et al. 2021; Vela-Avitúa et al. 2022). Prediction accuracy of complex traits, such as disease resistance, can be increased by filtering genetic markers depending on whether the genetic variations are located in functional sequences and by developing prediction models that can take into account the biological priors. Integrating functional information into genomic prediction models have also the potential to deal with the decrease of predictive accuracy which occurs when the reference population consists of distant relatives of the animals to be predicted.
Research efforts on genome and functional annotation can be effectively made accessible and translated into application, promoting and facilitating the global implementation of genomic selection, either in the sea bass industry to enhance VNN resistance or in other aquaculture species for complex traits of economic importance.
Breuil, G.; Bonami, J.R.; Pepin, J.F.; Pichot, Y. (1991). Aquaculture 97:109-116. doi: 10.1016/0044-8486(91)90258-9.
Clark, E.L; Archibald, A.L.; Daetwyler, H.D.; Groenen, M.A.M.; Harrison, P.W.; et al. (2020). Genome Biol 21:285. doi: 10.1186/s13059-020-02197-8.
Doan, Q.K.; Vandeputte, M.; Chatain, B.; Haffray, P.; Vergnet, A.; et al. (2017). Aquaculture 478:1–8. doi: 10.1016/j.aquaculture.2017.05.011.
Faggion, S.; Bertotto, D.; Babbucci, M.; Dalla Rovere, G.; Franch, R.; et al. (2021). Genet Sel Evol 53:32. doi:10.1186/s12711-021-00625-2.
Griot, R.; Allal, F.; Phocas, F.; Brard-Fudulea, S.; Morvezen, R.; et al. (2021). Aquaculture 530: 735930. doi: 10.1016/j.aquaculture.2020.735930.
Ødegård, J.; Baranski, M.; Gjerde, B.; Gjedrem, T. (2011). Aquacult Res 42(s1):103-114. doi: 10.1111/j.1365-2109.2010.02669.x.
Peñaloza, C.; Manousaki, T.; Franch, R.; Tsakogiannis, A.; Sonesson, A.K.; et al. (2021). Genomics 113(4):2096-2107. doi: 10.1016/j.ygeno.2021.04.038.
Vela-Avitúa, S.; Thorland, I.; Bakopoulos, V.; Papanna, K.; Dimitroglou, A.; et al. (2022). Front Genet 13:804584. doi: 10.3389/fgene.2022.804584.
Zenger, K.R.; Khatkar, M.S.; Jones, D.B.; Khalilisamani, N.; Jerry, D.R.; et al. (2019). Front Genet 9:693. doi: 10.3389/fgene.2018.00693.
Acknowledgements. Funding: Horizon 2020 Grant n. 817923 (AQUA-FAANG). Animals used in the experiment: Valle Cà Zuliani Società Agricola srl (Italy).