Background:  One of the most difficult problems facing the development of an effective HIV vaccine is the sequence diversity found in circulating strains. We hypothesize that any antigen sequence that embodies shared features of circulating viruses would be more broadly reactive than any single circulating virus or lab strain. We developed novel methods that capture protein fragments that are at high frequency in the population, and then compress them into short constructs, COT⁺. We tested such a construct for level of in vitro immunogenicity using the highly variable and cytotoxic T lymphocyte (CTL) -rich HIV Nef protein.

Methods:  We tested our COT⁺ construct against the “3-Best” natural variants and a consensus sequence of clade B HIV-1 sequences from the Los Alamos National Laboratory (LANL) HIV database. We synthesized a comprehensive set of 10-mer peptides corresponding to our construct, the 3-Best variants, and the HIV-1B consensus sequence from the LANL HIV database. The 3-Best combination was developed by searching the LANL HIV database for the 3 variants that cover the most diversity. T-cell reactivity ex vivo elicited by these test sets was assessed by interferon-gamma (IFN-g) ELISpot assays in a cohort of 27 HIV-1B-infected persons.

Results:  We found roughly the same number of responses using the 3-Best test set as there was from our synthetic construct. Specifically, of the 944 unique 10-mers derived from the 3 sequences, 561 peptides were found in our construct, 522 in the 3-Best, and 197 in the consensus. We detected 221 responses from our construct, 200 from 3-Best, and 95 from consensus.

Conclusions:  Despite the differences of the 3 constructs, all 3 sequences showed comparable immunogenicity. Detailed statistical analysis has identified cross reactivity as a likely explanation for the nearly equal coverage. Specifically, a peptide that is rare in the viral population, and thus unlikely to create a response directly, tends to react frequently when it differs from a common peptide by 1 or 2 amino acids. Specifically, cross reactivity is more likely when the amino acid changes are conservative and not at anchor positions. For example, peptide SRLAFHHVAR occurs in 42% of all wild strains of HIV-1 subtype B, while the peptide SRLAIHHMAR occurs in 1%; yet the same 9 individuals respond to both at approximately the same magnitude. Thus, the inclusion of multiple segments of well-chosen viral sequences can enhance cross reactivity and thereby immunogenicity of vaccine candidates.