Exome Sequencing Reveals a Putative Role for HLA-C*03:02 in Control of HIV-1 in African Pediatric Populations

Abstract

Human leucocyte antigen (HLA) class I molecules present endogenously processed antigens to T-cells and have been linked to differences in HIV-1 disease progression. HLA allelotypes show considerable geographical and inter-individual variation, as does the rate of progression of HIV-1 disease, with long-term non-progression (LTNP) of disease having most evidence of an underlying genetic contribution. However, most genetic analyses of LTNP have occurred in adults of European ancestry, limiting the potential transferability of observed associations to diverse populations who carry the burden of disease. This is particularly true of HIV-1 infected children. Here, using exome sequencing (ES) to infer HLA allelotypes, we determine associations with HIV-1 LTNP in two diverse African pediatric populations. We performed a case-control association study of 394 LTNPs and 420 rapid progressors retrospectively identified from electronic medical records of pediatric HIV-1 populations in Uganda and Botswana. We utilized high-depth ES to perform high-resolution HLA allelotyping and assessed evidence of association between HLA class I alleles and LTNP. Sixteen HLA alleles and haplotypes had significantly different frequencies between Uganda and Botswana, with allelic differences being more prominent in HLA-A compared to HLA-B and C allelotypes. Three HLA allelotypes showed association with LTNP, including a novel association in HLA-C (HLA-B^∗57:03, aOR 3.21, Pc = 0.0259; B^∗58:01, aOR 1.89, Pc = 0.033; C^∗03:02, aOR 4.74, Pc = 0.033). Together, these alleles convey an estimated population attributable risk (PAR) of non-progression of 16.5%. We also observed novel haplotype associations with HLA-B^∗57:03-C^∗07:01 (aOR 5.40, Pc = 0.025) and HLA-B^∗58:01-C^∗03:02 (aOR 4.88, Pc = 0.011) with a PAR of 9.8%, as well as a previously unreported independent additive effect and heterozygote advantage of HLA-C^∗03:02 with B^∗58:01 (aOR 4.15, Pc = 0.005) that appears to limit disease progression, despite weak LD (r ² = 0.18) between these alleles. These associations remained irrespective of gender or country. In one of the largest studies of HIV in Africa, we find evidence of a protective effect of canonical HLA-B alleles and a novel HLA-C association that appears to augment existing HIV-1 control alleles in pediatric populations. Our findings outline the value of using multi-ethnic populations in genetic studies and offer a novel HIV-1 association of relevance to ongoing vaccine studies.

Keywords: AIDS; childhood HIV; genetics; genomics; long-term non-progression.

FIGURE 1

Circular bar plot of HLA class I distribution in Uganda and Botswana. Each bar represents a class I HLA allele. The heights represent the relative frequency of each allele per country and significantly different alleles are shown with asterisks.

FIGURE 2

Patterns of linkage disequilibrium (LD) between the most frequent class I HLA alleles. LD in our cohort is expressed as the R² values. LD plots were produced with (A) 56 alleles in the combined cohort, (B) 59 alleles in Uganda, and (C) 56 alleles in Botswana with a frequency ≥1%. The colors indicate increasing strength of LD from yellow to red. Alleles on the same HLA locus (no pairwise LD comparisons) are indicated in gray. Alleles associated with LTNP in the cohort are highlighted in red.

FIGURE 3

Distribution of the commonest HLA class I (A) A–B, (B) A–C and (C) B–C haplotypes in Uganda and Botswana. *-indicates alleles that are statistically different among both countries.

FIGURE 4

Circular bubble plot of the Class I HLA association with LTNP in the CafGEN cohort. Each bubble represents an HLA class I allele and the distance from the center is a measure of the negative log of the correct P (Pc) value. Protective alleles are represented with solid bubbles while the susceptible alleles are shown in blank bubbles. The size of the bubbles depicts the odds ratio from unity. The broken line represents the cut off of Pc value (<0.05).

FIGURE 5

Forest plot of HLA alleles associated with LTNP in Uganda and Botswana. Only alleles found to be statistically associated with HIV progression in either country are shown. Alleles highlighted in red and blue have protective and detrimental effects, respectively, in both populations. Log Odds Ratios (95% CI) are obtained from logistic regression adjusting the effect of gender on HIV disease progression in each country.

FIGURE 6

Effect of HLA-C*03:02 on other HLA class I alleles. Asterisk indicates allele combination that is statistically significant (Pc < 0.05) after correcting by a factor of 9 (Svejgaard and Ryder, 1994).