Identifying vulnerable sites of the HIV-1 capsid protein

One very challenging aspect of developing HIV vaccines and therapies is to overcome the high evolutionary rate and consequent sequence diversity of the virus. HIV is notorious for rapidly acquiring drug resistant and immune escape mutations, which allow the virus to survive and persist against antiviral drug suppression and host immune responses. I hypothesized that functionally and structurally conserved elements of the viral proteome with little-to-no tolerance to mutations would be good candidates for targets of vaccine-induced responses and antiretroviral drugs. To identify such regions in the HIV-1 capsid protein, I analyzed data from 5000 HIV capsid sequences from Genbank and the Los Alamos HIV sequence database to estimate sequence conservation and mutation frequencies. The structural location for each amino acid residue was determined based on the high-resolution X-ray crystal structure of the hexameric form of the capsid protein (the major morphological subunit of the mature HIV-1 capsid). In vitro pairwise growth competition assays were then carried out to determine the relative fitness cost of the most frequently observed mutations at capsid hexamerization interface sites and non-interface sites. Only a weak relationship between sequence conservation and the relative fitness was detected. On the other hand, the most frequently observed mutations at interface sites had larger fitness costs than the mutations at the non-interface sites, suggesting that interface sites are could be suitable targets for HIV vaccines and therapies. In addition to protein interface sites, I used homology protein modeling and two protein stability prediction methods to investigate links between changes in protein stability and the impact of mutations in the capsid protein on viral replication. I found that mutations predicted to induce large alterations in the dimerization of or the structural stability of the CA hexamer were far less likely to be found in the HIV sequence database than those not predicted to alter stability. Destabilizing mutations were also associated with deleterious phenotypes. Compared to mutation frequency, predicted protein stability was a better classifier of deleterious and non-deleterious mutations. However, a newly derived simple composite score, which takes into account both mutation frequency and the proteins stability score, performed better than both protein staibilty and mutation frequency alone. These results suggest that both sequence conservation and in silico structural stability should be used to identify potentially inactivating mutations. Utilizing optimized in vitro pairwise growth competition assays and in silico mutation and protein stability predictions, along with sequence conservation, I identified potentially important sites in the HIV-1 capsid protein that warrant further investigation as candidates for drug and CTL vaccine targets.