RANKING POTENTIAL BINDING PEPTIDES TO MHC MOLECULES BY A COMPUTATIONAL THREADING APPROACH

In this paper, an approach developed to address the inverse protein fo lding problem is applied to prediction of potential binding peptides t o a specific major histocompatibility complex (MHC) molecule. Overlapp ing peptides, spanning the entire protein sequence, are threaded throu gh the backbone coordinates of a known peptide fold in the MHC groove, and their interaction energies are evaluated using statistical pairwi se contact potentials. With currently available tables for pairwise po tentials, promising results are obtained for MHC-peptide complexes whe re hydrophobic interactions predominate. By ranking the peptides in an ascending order according to their energy values, it is demonstrated that, in most cases, known antigenic peptides are highly ranked. Furth ermore, predicted hierarchies are consistent with experimental binding results. Currently, predictions of potential binding peptides to a sp ecific MHC molecule are based on the identification of allele-specific binding motifs. However, it has been demonstrated that these motifs a re neither sufficient nor strictly required to ensure binding. The com putational procedure presented here succeeds in determining the MHC bi nding potential of peptides along a protein amino acid sequence, witho ut relying on binding motifs. The proposed scheme may significantly re duce the number of peptides to be tested, identify good binders that d o not necessarily show the known allele-specific binding motifs, and i dentify the best candidates among those with the motifs. In general, w hen structural information about a protein-peptide complex is availabl e, the current application of the threading approach can be used to sc reen a large library of peptides for selection of the best binders to the target protein.