We propose a self-consistent approach to analyze knowledge-based atom–atom potentials used to calculate protein–ligand binding energies. Ligands complexed to actual protein structures were first built using the SMoG growth procedure (DeWitte & Shakhnovich, 1996) with a chosen input potential. These model protein–ligand complexes were used to construct databases from which knowledge-based protein–ligand potentials were derived. We then tested several different modifications to such potentials and evaluated their performance on their ability to reconstruct the input potential using the statistical information available from a database composed of model complexes. Our data indicate that the most significant improvement resulted from properly accounting for the following key issues when estimating the reference state: (1) the presence of significant nonenergetic effects that influence the contact frequencies and (2) the presence of correlations in contact patterns due to chemical structure. The most successful procedure was applied to derive an atom–atom potential for real protein–ligand complexes. Despite the simplicity of the model (pairwise contact potential with a single interaction distance), the derived binding free energies showed a statistically significant correlation (∼0.65) with experimental binding scores for a diverse set of complexes.

B*2701 differs from all other HLA-B27 subtypes of known peptide specificity in that, among its natural peptide ligands, arginine is not the only allowed residue at peptide position 2. Indeed, B*2701 is unique in binding many peptides with Gln2 in vivo. However, the mutation (Asp74Tyr) responsible for altered selectivity is far away from the B pocket of the peptide binding site to which Gln/Arg2 binds. Here, we present a model that explains this effect. It is proposed that a new rotameric state of the conserved Lys70 is responsible for the unique B*2701 binding motif. This side chain should be either kept away from pocket B through its interaction with Asp74 in most HLA-B27 subtypes, or switched to this pocket if residue 74 is Tyr as in B*2701. Involvement of Lys70 in pocket B would thus allow binding of peptides with Gln2. Binding of Arg2-containing peptides to B*2701 is also possible because Lys70 could adopt another conformation, H-bonded to Asn97, which preserves the same binding mode of Arg2 as in B*2705. This model was experimentally validated by mutating Lys70 into Ala in B*2701. Edman sequencing of the B*2701(K70A) peptide pool showed only Arg2, characteristic of HLA-B27-bound peptides, and no evidence for Gln2. This supports the computational model and demonstrates that allowance of B*2701 for peptides with Gln2 is due to the long-range effect of the polymorphic residue 74 of HLA-B27, by inducing a conformational switch of the conserved Lys70.