Whole genome sequencing of extreme phenotypes identifies variants in CD101 and UBE2V1 associated with increased risk of sexually acquired HIV-1

Abstract

Host genetic variation modifying HIV-1 acquisition risk can inform development of HIV-1 prevention strategies. However, associations between rare or intermediate-frequency variants and HIV-1 acquisition are not well studied. We tested for the association between variation in genic regions and extreme HIV-1 acquisition phenotypes in 100 sub-Saharan Africans with whole genome sequencing data. Missense variants in immunoglobulin-like regions of CD101 and, among women, one missense/5' UTR variant in UBE2V1, were associated with increased HIV-1 acquisition risk (p = 1.9x10-4 and p = 3.7x10-3, respectively, for replication). Both of these genes are known to impact host inflammatory pathways. Effect sizes increased with exposure to HIV-1 after adjusting for the independent effect of increasing exposure on acquisition risk.

Trial registration: ClinicalTrials.gov NCT00194519; NCT00557245.

Fig 1. Genomic structure of CD101 and UBE2V1, by-variant association significance, Primary Replication Variants, and protein features.

Panel A: The Manhattan plot positioned below the chromosomal region map captures by-variant association significance for all single nucleotide variants (SNVs) present in CD101 for both Discovery and Replication stage samples (red dots indicate the–log₁₀(p) value for Discovery stage, and blue dots for Replication stage). rsIDs are shown below exon symbols only for the 5 Ig-like domain variants for CD101 and the single 3’-UTR variant for UBE2V1 that were associated with HIV-1 acquisition in the Replication analysis. The remaining Primary Replication Variants (PRVs) are indicated as “var”. The rsIDs for the CD101 PRVs are: var2 = rs142460852, var7 = rs12097758, var8 = rs12067543, var9 = rs34248572, var10 = rs150494742, var11 = novel, var12 = rs2296448, var13 = novel, and var14 = rs35163967. Panel B: The Discovery and Replication stage SNVs for UBE2V1. rsIDs are shown below the single 3’-UTR variant (rs6095771) that was associated with HIV-1 acquisition risk in the Replication analysis. The remaining Primary Replication Variants (PRVs) are indicated as “var”. The rsIDs for the UBE2V1 PRVs are: var1 = rs186621934, var2 = rs115164526, var3 = rs6095755, var4 = rs41283596, var5 = rs187204768, var6 = rs193169918, var7 = rs372425380, var9 = rs73278517, var10 = novel, var11 = rs185632114.

Fig 2. Time to seroconversion by Protected sex Index (PI) and tertiles of partner HIV-1 viral load distribution.

The protected sex index (PI) is the proportion of visits at which the female partner reported always having protected sexual intercourse with the HIV-1 infected partner (= 1 –proportion of interviews with reported unprotected sex by the female partner.) The figures show time to seroconversion stratified by tertiles of plasma viral load of the HIV-1 infected partners in log₁₀ copies/mL: Low = (0, 3.8], Medium = (3.8, 4.5], and High = (4.5,8]. Panels show this relationship for by PI strata: Panel A: PI = (0, 0.3], Panel B: PI = (0.3 to 0.6], Panel C: PI = (0.6 to 1.0), and Panel D: PI = 1.0. PI shows reasonable accuracy in predicting overall risk of seroconversion, indicating this variable can be used to eliminate individuals with probably little or no exposure from the study samples although a few more highly exposed individuals might also be dropped from the analysis.

Fig 3. Results of Primary Replication Variant group tests for CD101 and UBE2V1.

Panel A: CD101 Primary Replication Variant (PRV) groups—All three PRV groups present in the Replication stage are shown in the Kaplan-Meier plot. The CD101 Ig-like PRV group is significantly associated with a higher rate of seroconversion after adjustment for multiple testing (p = 1.9 x 10⁻⁴), while the 3’-UTR and cytoplasmic PRV tests were not significant. Panel B:UBE2V1 PRV groups—Four variants with False Discovery Rates (FDRs) < 0.05 in a by-variant analysis of Replication stage variants were grouped (red) for visual exploration of effect size. Panel C: CD101 PRVs by FDR grouping–CD101 variants grouped by observed FDR in Replication stage. Panel D: UBE2V1 UTR-5’ PRV, Females—The hazard ratio for presence of the minor allele for rs6095771 in females is shown, which is significant after adjustment for four tests (two variant groups and two sexes, HR = 6.4, nominal p = 1.2x10^-3, adjusted p = 4.8x10^-3).

Fig 4. Estimated hazard ratio by Protective Index (PI) cut-off defining the analysis sample.

Panel A: CD101 Ig-like missense variant group hazard ratio (HR) estimate increases with increased sexual exposure among the individuals from the augmented sample; Panel B: UBE2V1 5’-UTR Primary Replication Variant HR estimate increases with sexual exposure among the individuals from the augment sample. Two variants contribute to this analysis; see S7 Table. Numbers at the top of each plot indicate total sample size (N) included at that level of the plot, and the p-value (P) for the reported HR.