Binding affinities of 438 HLA proteins to complete proteomes of seven pandemic viruses and distributions of strongest and weakest HLA peptide binders in populations worldwide

Abstract

We report detailed peptide binding affinities between 438 HLA Class I and Class II proteins and complete proteomes of seven pandemic human viruses, including coronaviruses, influenza viruses and HIV-1. We contrast these affinities with HLA allele frequencies across hundreds of human populations worldwide. Statistical modelling shows that peptide binding affinities classified into four distinct categories depend on the HLA locus but that the type of virus is only a weak predictor, except in the case of HIV-1. Amongst the strong HLA binders (IC50 ≤ 50), we uncovered 16 alleles (the top ones being A*02:02, B*15:03 and DRB1*01:02) binding more than 1% of peptides derived from all viruses, 9 (top ones including HLA-A*68:01, B*15:25, C*03:02 and DRB1*07:01) binding all viruses except HIV-1, and 15 (top ones A*02:01 and C*14:02) only binding coronaviruses. The frequencies of strongest and weakest HLA peptide binders differ significantly among populations from different geographic regions, with Indigenous peoples of America showing both higher frequencies of strongest and lower frequencies of weakest binders. As many HLA proteins are strong binders of peptides from distinct viral families, we discuss this result in relation to possible signatures of natural selection on HLA promiscuous alleles due to undetermined past pathogenic infections. Although highly relevant for evolutionary genetics and the development of vaccine therapies, these results should not lead to forget that individual resistance and vulnerability to diseases go beyond the sole HLA allelic affinity and depend on multiple, complex and often unknown biological, environmental and other variables.