STRUCTURE-BASED THERMODYNAMIC DESIGN OF PEPTIDE LIGANDS - APPLICATIONTO PEPTIDE INHIBITORS OF THE ASPARTIC PROTEASE ENDOTHIAPEPSIN

The prediction of binding affinities from structure is a necessary req uirement; in the development of structure-based molecular design strat egies. In this paper, a structural parameterization of the energetics previously developed in this laboratory has been incorporated into a m olecular design algorithm aimed at identifying peptide conformations t hat minimize the Gibbs energy. This approach has been employed in the design of mutants of the aspartic protease inhibitor pepstatin A. The simplest design strategy involves mutation and/or chain length modific ation of the wild-type peptide inhibitor, The structural parameterizat ion allows evaluation of the contribution of different amino acids to the Gibbs energy in the wild-type structure, and therefore the identif ication of potential targets for mutation in the original peptide. The structure of the wild-type complex is used as a template to generate families of conformational structures in which specific residues have been mutated, The most probable conformations of the mutated peptides are identified by systematically rotating around the side-chain and ba ckbone torsional angles and calculating the Gibbs potential function o f each conformation according to the structural parametrization, The a ccuracy of this approach has been tested by chemically synthesizing tw o different mutants of pepstatin A, In one mutant, the alanine at, pos ition five has been replaced by a phenylalanine, and in the second one a glutamate has been added at-the carboxy terminus of pepstatin A, Th e thermodynamics of association of pepstatin A and the two mutants hav e been measured experiment ally and tile results compared with the pre dictions, The difference between experimental and predicted Gibbs ener gies for pepstatin A and the two mutants is 0.23 +/- 0.06 kcal/mol. Th e excellent agreement between experimental and predicted values demons trates that this approach Gall be used in the optimization of peptide ligands. (C) 1998 Wiley-Liss, Inc.