Résumé
The genetic determinism of viral traits is classically determined by comparing the genomes associated to distinct phenotypes (forward genetics) or by detecting phenotypic variations induced by genetic changes at particular loci (reverse genetics). These approaches might be limited for viral traits determined by several loci or genes which may additionally interact epistatically. Even in the presence of a major determinant, the iterative nature of these procedures can be very time-consuming if no candidate genomic region is identified. In eukaryotes, complex determinisms can be analyzed on recombinant progenies using population-based association studies based on quantitative trait loci (QTL) and fixation indexes (FST) measured at each locus. As QTL approaches need recombinant progenies, they have so far not been used for clonal organisms like viruses. However, these approaches might be extended to highly recombinogenic viruses or more generally to viruses for which genome-wide artificial recombination can generate infectious genomes. Here, we took advantage of the first virus library of genome-wide shuffled recombinant genomes (Proteus, Nimes) between Tomato yellow leaf curl virus (TYLCV-Mld) and Tomato leaf curl Mayotte virus (ToLCKMV). The bimodal distribution of virus accumulation in tomato plants for 47 recombinant genomes (Vuillaume et al. 2011) generated between Tomato yellow leaf curl virus (TYLCV-Mld) and Tomato leaf curl Mayotte virus (ToLCKMV) pointed towards a major genetic determinant of viral accumulation (Vuillaume et al. 2011) that we searched in a genome wide association study. Four QTL methods and the locus-by-locus FST approach were used to identify the regions of the begomovirus genome determining the within-host accumulation. Our results revealed that the viral accumulation in recombinant genomes was determined by loci at the 5' end of the Replication associated protein gene (rep) and the Intergenic Region upstream of the rep. Using classical virology approaches, this region was previously reported to be involved in viral accumulation. The consistency between our results and those previously generated with classical approaches suggests that the QTL and FST approaches could be valuable tools to identify loci involved in viral traits, which holds promises for the characterization of traits with a more complex genetic determinism when appropriate material and data are available. (Résumé d'auteur)