DESCRIPTION (provided by applicant): The extraordinary scale of HIV variation poses a major obstacle for AIDS vaccine development, and there is great need to optimize vaccines through considerations of diversity. An approach we have advocated for vaccine design is to use an artificially generated sequence, either consensus or reconstructed ancestral, that is "central" and more similar to regional circulating forms than they are to each other. Definition of such central sequences is, however, sample dependent and particularly problematic for key positions that are rapidly evolving as a consequence of immune selection. As an alternative, we propose here to select amino acids in highly variable positions for vaccine antigens based on a detailed evolutionary, immunological and population analysis that optimizes the number of cytotoxic T lymphocyte (CTL) epitopes that individuals in a vaccine target population could potentially recognize. We would start from a consensus sequence, and incorporate the most common amino acid in highly conserved positions. To select amino acids for inclusion in variable positions we will systematically integrate information from the Los Alamos HIV Immunology and Sequence Databases and incorporate evolutionary analysis and computational methods for prediction of epitope presentation and processing. We also will consider the human genetic variability in vaccine target populations of the immunologically relevant HLA class I genes, which are highly polymorphic. Integrating all of this information, we will attempt design a vaccine with improved immunogenicity and breadth of cross-reactivity for CTL responses in vaccine target populations for B, C, and CRF01 (E) subtype vaccines. The antigenic cross-reactive potential of peptides optimized for immunogenicity will be compared to consensus sequence peptides experimentally by Harvard collaborators, who have extensive collections of well-characterized samples from patients with known HLA types and autologous sequences that could be used as a baseline for comparison to new antigens generated based on our optimization strategies. The experiments will determine whether we were indeed able to enhance breadth and intensity of CTL immune response at the population level through rational design. At the end of the two-year period we propose to have the following major results of our analysis: (i) "epitope optimized" protein sequences for each protein in subtypes B,C and CRF01, (ii) a table and a summary explaining our reasoning and choice parameters for each protein sequence position, (iii) peptide cocktails of alternative amino acid choices for the most immunogenetic regions and most variable regions of the virus, and (iv) EliSpot assays to test for increased breadth of cross-reactivity. If our design proves advantageous, we would be in a strong position to apply for funding for further antigenic cross-reactivity and immunogenicity studies.