Background:  Mutations that allow HIV-1 strains to escape MHC-restricted CTL killing arise during both acute and chronic infection, but in the absence of an immune response they can reduce fitness, and they can be selected against on transmission. Recently, from correlations of variability and MHC-restricted CTL epitopes in RT, it was suggested that human CTL responses have permanently altered the HIV genome. We have examined the persistence of this “immunological footprint” using a phylogenetic approach.

Methods:  From this perspective, within-individual adaptation is evolution at the “tips” of the tree. We used a suite of likelihood-based approaches to obtain maximum likelihood parameter estimates based on a codon model allowing separate synonymous, and 2 non-synonymous rates (internal and tip), independently at all sites. All analyses were run on an 18-processor Linux cluster.

Results:  Analyzing 92 clade C sequences from KwaZulu Natal and Zambia, we inferred amino acid substitutions at 32/99 codons in protease (PR) in terminal branches, but at only 16 codons internally (exact p paired data < 0.0001). Amino acid sites changing only at the tips have relatively few substitutions: the average d_N for these codons is about a third that of those with internal substitutions as well (0.87 vs 2.65, n = 16 for each class). In reverse transcriptase (RT) 81/312 codons had substitutions at terminal branches while only 26 codons had internal substitutions (exact p < 0.0001). The average terminal d_N for the former is only about 21% of the average d_N for codons with internal substitutions (0.574 vs 2.65).

Conclusions:  A high proportion of amino acid substitutions in HIV sequences are phylogenetically recent. This suggests that adaptations to individual hosts are often deleterious and do not persist at the population level, thus the “immunological footprint” may often be ephemeral. This approach provides a rapid, general, and efficient method for testing for differential selective impact on HIV sequences (e.g., in candidate vaccines) without requiring HLA typing of target populations.

Keywords: Positive selection; Immunological footprint; CTL