MOLECULAR-DYNAMICS AND STRUCTURE-BASED DRUG DESIGN FOR PREDICTING NONNATURAL NONAPEPTIDE BINDING TO A CLASS-I MHC PROTEIN

Starting from the known three-dimensional structure of the class I maj or histocompatibility complex-encoded HLA-B2705 protein, three non-na tural nonapeptides were designed to fit optimally the HLA-B2705-bindi ng groove. The optimization was performed using structure-based drug d esign methods and the fact that all the possible interactions of the s econdary anchor residue (position 3) with its human leukocyte antigen- binding pocket (pocket D) in nature are not entirely utilized. 150 ps molecular-dynamics (MD) simulation in water was employed to study the stability of the bimolecular complexes with three non-natural peptides (P3 = homophenylalanine, beta-naphthylalanine, alpha-naphthylalanine) as well as with the two natural homologues (P3 = Gly, Leu). Various s tructural and dynamical properties (atomic fluctuations, solvent acces sible surface areas, peptide C alpha-atom positions) of the simulated bimolecular complexes were used to compare the three non-natural with the two natural ligands. Since the various molecular properties have b een shown previously to be related to the binding affinity of nonapept ide ligands to the major histocompatibility complex (MHC) HLA-B2705 p rotein, the MD data predict a rather higher stability of MHC-ligand co mplexes with the three non-natural peptides, suggesting that the unnat ural peptides studied show an enhanced binding affinity to the HLA-B2 705 protein. These results are in agreement with the experimental valu es of a semi-quantitative in vitro assembly assay, performed on the fi ve nonapeptides (P3 = Gly, Leu, homophenylalanine, beta-naphthylalanin e, alpha-naphthylalanine), which shows their ability to stabilize the native conformation of the HLA-B2705 heavy chain and also shows that the three non-natural ligands bind with higher affinity (0.5 mu M) to the MHC protein than the two natural homologues (40 mu M). Thus, this study demonstrates that structural information combined with rational design and molecular-dynamics simulations can illustrate and predict M HC binding and potential T-cell epitope properties as well as contribu te to the design of new non-peptidic MHC inhibitors that may be useful for the selective immunotherapy of autoimmune diseases to which HLA a lleles are directly associated.