Computational protocol: HIV 1 adaptation to NK cell mediated immune pressure

Similar protocols

Protocol publication

[…] We initially focussed on the 4 viral variants (6 amino acid polymorphisms) studied in depth in []. All 4 variants are enriched in KIR2DL2+ individuals. One of the variant strains, Vpu(71M/74H) overlaps with Env(17W/20M); there are thus three independent variants: Gag(138L), Nef(9S) and Vpu(71M/74H)-Env(17W/20M). KIR2DL2 is an inhibitory receptor. It is postulated that the variant peptide causes stronger inhibitory signalling via KIR2DL2 than wild type [–]. This model requires that the variant peptide binds one or more of the KIR2DL2-ligating HLA class I molecules and that the amino acid polymorphism either enhances KIR signalling (i.e. the polymorphism affects KIR-peptide contact) and/or enhances peptide-HLA binding (i.e. the polymorphism affects ligand availability). We define individuals who carry both HLA class I and KIR alleles that meet these conditions as ‘selectors’, individuals where, according to our current understanding of NK cell activation, the variant virus could have a selective advantage. KIR2DL2 binds HLA C molecules which have an asparagine at position 80 (designated the C1 group of alleles), HLA-B*46:01 and B*73:01 (which have a HLA C-type motif at residues 77–83), and with weaker affinity, HLA C molecules with a lysine at position 80 (C2 group alleles) []. KIRs contact the C-terminal end (specifically positions PC-1 and PC-2) of the bound peptide [, , , –]. We calculated the proportion of selectors in the HIV-1-infected population in the USA (the population studied by Alter et al []), by using the HLA-gene frequency in different ethnic groups [], the frequency of these groups in the HIV-1-infected population [] and the binding affinity of 8- to 11mer variant peptides estimated using NetMHCpan v2.8 [] ().For Env(17/20) the carrier frequency of selecting HLA class I molecules (fH) was ≤77% but <40% for the other 4 polymorphisms considered (, column A rows 1–4). If we relax our definition of what KIR2DL2 can recognise and allow the variant amino acid on all peptide positions except position 2 and the C-terminal position, which are hidden in the HLA binding pockets, the frequency of selecting HLAs increases, reaching a median of 42% across all 6 amino acid polymorphisms (, column B, rows 1–4). KIR2DL2 mainly binds HLA-C1 group molecules and has a preference for 9mers []. As expected, if we only allow presentation by HLA-C1 group molecules or analyse binding of 9mers only, the frequency of selecting HLAs is significantly reduced (, columns C & D rows 1–4).Repeating the analysis with an alternative definition of epitope binding (based on the rank of a variant peptide relative to all other peptides of the HIV-1 proteome []; and ) and with alternative, independent epitope prediction software (Epipred; ) confirmed our finding that the frequency of selecting HLA class I molecules is low.Next, we extended the analysis to the other viral polymorphisms associated with inhibitory KIR genes identified in []. As for the variants initially focussed on, the frequency of selecting HLAs for the additional polymorphisms was low (median 14% if we require the polymorphism at position PC-1 or PC-2, median 26% if we allow the polymorphism at any non-anchor position, , rows 5–6).In summary, with the exception of Env (17/20), the proportion of the population where the variant has a selective advantage is low for all of the viral polymorphisms associated with the presence of inhibitory KIR genes in []. [...] Polymorphisms in neighbouring sequences can alter the processing of the peptides they flank []. So another possible explanation for the advantage conferred by a variant is that the mutation increases the production of nearby binding peptides and thus indirectly affects the level of HLA class I:peptides available for KIR binding. To test this possibility, we used NetMHCpan to predict the binding of peptides within 20 amino acids of the six polymorphisms (Env(17/20), Vpu(71/74), Gag(138) and Nef(9)) to HLA-C alleles, HLA-B*73:01 and HLA-B*46:01 and found 38 flanking binders. We then used NetCTLPan v1.1 to predict the ability of these peptides to be cleaved and transported with both the variant and the wildtype flanking region. We found that cleavage was only increased in the presence of the variant for one peptide that binds HLA-C*15:08. HLA-C*15:08 has not been reported in African American, white or Hispanic populations (which together constitute the vast majority of HIV-1-infected individuals in the US). We conclude that the impact of the polymorphism on flanking peptide processing is unlikely to contribute to the fraction of selectors in the population. […]

Pipeline specifications

Software tools NetMHCpan, Epitope Prediction, EpiPred, NetCTLpan
Application Immune system analysis
Diseases HIV Infections