Amino-acid co-variation in HIV-1 Gag subtype C: HLA-mediated selection pressure and compensatory dynamics

Abstract

Background: Despite high potential for HIV-1 genetic variation, the emergence of some mutations is constrained by fitness costs, and may be associated with compensatory amino acid (AA) co-variation. To characterize the interplay between Cytotoxic T Lymphocyte (CTL)-mediated pressure and HIV-1 evolutionary pathways, we investigated AA co-variation in Gag sequences obtained from 449 South African individuals chronically infected with HIV-1 subtype C.

Methodology/principal findings: Individuals with CTL responses biased toward Gag presented lower viral loads than individuals with under-represented Gag-specific CTL responses. Using methods that account for founder effects and HLA linkage disequilibrium, we identified 35 AA sites under Human Leukocyte Antigen (HLA)-restricted CTL selection pressure and 534 AA-to-AA interactions. Analysis of two-dimensional distances between co-varying residues revealed local stabilization mechanisms since 40% of associations involved neighboring residues. Key features of our co-variation analysis included sites with a high number of co-varying partners, such as HLA-associated sites, which had on average 55% more connections than other co-varying sites.

Conclusions/significance: Clusters of co-varying AA around HLA-associated sites (especially at typically conserved sites) suggested that cooperative interactions act to preserve the local structural stability and protein function when CTL escape mutations occur. These results expose HLA-imprinted HIV-1 polymorphisms and their interlinked mutational paths in Gag that are likely due to opposite selective pressures from host CTL-mediated responses and viral fitness constraints.

Figure 1. Relationship between CTL targeting and viremia.

Protein-specific protective ratios were plotted as a function of the mean entropy of each HIV-1 protein. Protective ratios were calculated as the Log₁₀ of the viral load of all the individuals who did not mount a CTL response against a protein over the viral load of all the individuals who had one or more CTL response(s) directed against that protein.

Figure 2. Amino Acid associations in HIV-1 subtype C Gag.

Associations are depicted with a circular map: AA interactions among residues are represented with arcs, which are color-coded with a white to purple gradient – white corresponding to the strongest associations (i.e., lower q-values). HLA-restricted sites are identified by the HLA allele designations around the circle.

Figure 3. Distances between co-varying residues.

The larger graph shows the distribution of the 2D distances between all the pairs of co-varying AA in Gag. The inset represents the relationship between the 2D and 3D distances.

Figure 4. Hubness across HIV-1 Subtype C Gag.

The number of co-varying partners and the Shannon Entropy are represented for each site along the Gag protein. The blue (lower) part of the bars represent the number of AA-to-AA associations and the red (upper) part of the bars represent the number of HLA-to-AA associations at each site. The secondary axis refers to the Shannon Entropy at each site in Gag (continuous black line).

Figure 5. Sub-network associated with the HLA-B*81 epitope TPQDLNTML.

The tridimensional structure shows three p24 chains of HIV-1 subtype C (colored in light blue, golden yellow and pink) and the epitope TPQDLNTML is highlighted in bright blue on chain A. Side-chain atoms for the residues E177, Q182 and T186 are represented in sphere format.

Figure 6. Fitness competition assays between viruses mutated at residues in the sub-network associated with the HLA-B*81 epitope TPQDLNTML.

The relative fitness of viruses presenting a mutation at site 177, 186 or at both sites is compared to that of the wt COT virus. Fitness competition assays were performed against a wt COT virus; the proportion of viral RNA from the mutant and wt viruses was measured at day zero, three and five (see methods).