Immunological and virological parameters improve genotypic prediction of HIV-1 coreceptor usage

Poster number: 22

T Sing (1), T Lengauer (1), N Beerenwinkel (1), M Däumer (2), R Kaiser (2), P Cheung (3) and PR Harrigan (3)

  1. Max Planck Institute for Informatics, Saarbrücken, Germany
  2. Institute of Virology, University of Cologne, Cologne, Germany, 3BC Center for Excellence in HIV/AIDS, Vancouver, Canada

The new drug class of coreceptor antagonists tries to prevent viral entry into the cell by binding to one of the HIV-1 coreceptors. Inhibitors of CCR5 seem to be particularly promising, since people with a nonfunctional version of CCR5 (due to a 32-basepair deletion in both copies of the gene encoding for the receptor) are virtually immune to HIV-1 infection and not affected otherwise. Unfortunately, emergence of resistant variants has been observed in vitro, mediated by mutations in the viral envelope protein gp120, which interacts with the HIV-1 coreceptors during entry. The possible resistance mechanisms of gp120 in presence of a CCR5 antagonist are (a) increased affinity to the receptor, (b) altered binding mode, and (c) "switch" of the viral population to CXCR4 usage. This phenotypic switch, however, is associated with progression towards AIDS: CXCR4-using variants emerge during disease progression in 50% of the patients. Thus, even though the causal relationships are not known, treatment with coreceptor antagonists will have to be accompanied by careful monitoring of viral coreceptor usage. Several studies have investigated prediction of coreceptor usage based on the third hypervariable region of gp120, which could lead to a cheap and fast substitute of the laborious and expensive phenotypic assays. Recently, immunological and virological markers (CD4, CD4/CD8 ratio, viral load, CCR5-Delta32-heterozygosity) were shown to be associated with coreceptor usage in univariate analyses. Here we show, using data from over 1000 patients, that these markers also improve the performance of sequence-based models, by integrating them into our prediction system geno2pheno[coreceptor], available at http://www.geno2pheno.org. These findings add a novel perspective on sequence-based prediction of HIV-1 coreceptor usage, by showing the usefulness of integrating markers which can indicate the presence of undetected CXCR4-using strains.