Strong genome-wide HLA-associated selection evident in HIV-1 in a national contemporary US population: The ACTG 5142 study

Background: Knowledge of the qualitative differences between different epitope-specific and protein-specific T cell responses, especially between gag versus other genes, is needed for optimizing the design of T cell vaccine immunogens in current development. HLA-specific HIV sequence polymorphisms represent the evolutionary imprint of strong T cell responses within functional and structural constraints. HLA-HIV sequence associations have been demonstrated in selected HIV proteins but to date not in genomewide analysis where all proteins can be compared and in populations large enough for sufficient statistical power across most HLA genotypes and potential phylogenetic groupings. Objective: To determine all imprints of HLA-driven selection across the entire HIV genome and the extent to which these can serve to probe both functional constraint and functional effectiveness of HIV-specific T cell responses for all HLA-restricted CD8 T cell epitopes. Methods: High resolution (4 digit) HLA genotypes and bulk sequence of all expressed genes in pre-treatment plasma virus were characterized in 559 individuals recruited into the nation-wide AACTG 5142 study. Comparisons and pooling with previously collected data in a Western Australian cohort was examined (combined dataset n=631-781). Associations between HLA and viral sequence variations were determined using published methods and were analysed with respect to the general level of polymorphism in each protein and measures of natural control such as pre-treatment viral load. Results: The ACTG 5142 study population captures a high degree of global HLA diversity and therefore probes the adaptability of the virus in several potential ethnic/geographic settings. There were 673 unique HLA-sequence associations identified. The density of HLA associations across different proteins in decreasing order was nef>vpr>p17>vif>p2p7p1p6>rev>p24>gag/pol TF>vpu>integrase>protease>env>tat>RT. The hierarchy of selection density mirrored general polymorphism rate differences across the genome. Nef in particular showed evidence of the most intense HLA allele specific selection compared with gag and pol. Effects on viral load were epitope-specific. Conclusion: There are clear differences in the determinants of T cell selection between different proteins. These data can be used to inform epitope inclusion and exclusion criteria for polyvalent vaccine approaches