HLA Heterozygote Advantage against HIV-1 Is Driven by Quantitative and Qualitative Differences in HLA Allele-Specific Peptide Presentation

Abstract

Pathogen-mediated balancing selection is regarded as a key driver of host immunogenetic diversity. A hallmark for balancing selection in humans is the heterozygote advantage at genes of the human leukocyte antigen (HLA), resulting in improved HIV-1 control. However, the actual mechanism of the observed heterozygote advantage is still elusive. HLA heterozygotes may present a broader array of antigenic viral peptides to immune cells, possibly resulting in a more efficient cytotoxic T-cell response. Alternatively, heterozygosity may simply increase the chance to carry the most protective HLA alleles, as individual HLA alleles are known to differ substantially in their association with HIV-1 control. Here, we used data from 6,311 HIV-1-infected individuals to explore the relative contribution of quantitative and qualitative aspects of peptide presentation in HLA heterozygote advantage against HIV. Screening the entire HIV-1 proteome, we observed that heterozygous individuals exhibited a broader array of HIV-1 peptides presented by their HLA class I alleles. In addition, viral load was negatively correlated with the breadth of the HIV-1 peptide repertoire bound by an individual's HLA variants, particularly at HLA-B. This suggests that heterozygote advantage at HLA-B is at least in part mediated by quantitative peptide presentation. We also observed higher HIV-1 sequence diversity among HLA-B heterozygous individuals, suggesting stronger evolutionary pressure from HLA heterozygosity. However, HLA heterozygotes were also more likely to carry certain HLA alleles, including the highly protective HLA-B*57:01 variant, indicating that HLA heterozygote advantage ultimately results from a combination of quantitative and qualitative effects in antigen presentation.

Keywords: MHC evolution; antigen presentation; divergent allele advantage; human leukocyte antigen; major histocompatibility complex; pathogen-mediated balancing selection.

Fig. 1.

Viral load in HLA homozygotes and heterozygotes. Comparison of the spVL (log10 HIV-1 RNA copies per ml of plasma) between HLA homozygous (Hom) and heterozygous (Het) individuals for the three classical HLA class I loci. N indicates the number of individuals. Bonferroni-corrected P-value from Wilcoxon rank-sum test is shown.

Fig. 2.

Sequence divergence between individual’s HLA alleles and viral load. Correlation between spVL (log10 HIV-1 RNA copies per ml of plasma) and sequence divergence between individual’s HLA-A, HLA-B, and HLA-C alleles is shown (including homo- and heterozygotes; N = 6,311). The color indicates the density of individuals. Kendall’s estimate of correlation τ and Bonferroni-corrected P-value are shown.

Fig. 3.

HLA-bound peptides and viral load (spVL). Correlation between individual’s spVL (log10 HIV-1 RNA copies/ml of plasma) and the breadth of HIV-1 peptides predicted to be bound by the individual's HLA-A, HLA-B, and HLA-C alleles is shown (including homo- and heterozygotes; N = 6,311). The color indicates the density of individuals. Kendall’s estimate of correlation τ and Bonferroni-corrected P-value are shown.

Fig. 4.

Evolution of HIV sequence diversity in response to HLA-B diversity. (A) Rooted phylogenetic trees of autologous virus sequences (blue dots) from HLA-B homozygous (N = 4) and heterozygous (N = 36) individuals. We used HIV-2 (red dot) as outgroup to identify the root during tree construction. (B) Virus sequences exhibit higher sequence diversity (measured root-to-tip patristic distance, i.e., sum of branch lengths) among HLA-B heterozygous individuals (blue dots) compared with homozygotes (yellow dots). N indicates the number of individuals. (C) The number of predicted HLA-B bound peptides correlates positively with the diversity of HIV sequences among individuals (N = 40, correlation coefficient τ and P-value from Kendall rank correlation are shown).

Fig. 5.

Heterozygote advantage versus allele-specific effect for HLA-B*57:01. Variation in (A) set point viral load (spVL, log10 HIV-1 RNA copies/ml of plasma) and (B) the number of HIV-1 peptides bound by HLA-B in HLA heterozygous individuals not carrying HLA-B*57:01 (B57:01−/−), individuals carrying one copy of HLA-B*57:01 (B57:01+/−), and individuals homozygous for HLA-B*57:01 (B57:01+/+). N indicates the number of individuals. Bonferroni-corrected P-values from Wilcoxon rank sum test are shown.