Peptide binding specificity of major histocompatibility complex class I resolved into an array of apparently independent subspecificities: quantitation by peptide libraries and improved prediction of binding.

Considerable interest has focused on understanding how major histocompatibility complex (MHC) specificity is generated and characterizing the specificity of MHC molecules with the ultimate goal being to predict peptide binding. We have used a strategy where all possible peptides of a particular size are distributed into positional scanning combinatorial peptide libraries (PSCPL) to develop a highly efficient, universal and unbiased approach to address MHC specificity. The PSCPL approach appeared qualitatively and quantitatively superior to other currently used strategies. The average effect of any amino acid in each position was quantitated, allowing a detailed description of extended peptide binding motifs including primary and secondary anchor residues. It also identified disfavored residues which were found to be surprisingly important in shaping MHC class I specificity. Assuming that MHC class I specificity is the result of largely independently acting subsites, the binding of unknown peptides could be predicted. Conversely, this argues that MHC class I specificities consist of an array of subspecificities acting in a combinatorial mode.