Methods for predicting three-dimensional structures for alpha helical membrane proteins and their use in design of selective ligands

Abstract

A method for practical prediction of the three-dimensional structure of α-helical membrane proteins (HMPs) is described. The method allows one to predict the binding site and structure for strongly bound ligands. The method combines a protocol of computational methods enabling a complete ensemble of packings to be sampled and systematically reducing this ensemble to progressively more accurate structures until at the end there remain a few that might be functionally relevant and likely to play a role in all binding and activation processes. This method is well suited to automatic operation making it practical to obtain, for example, the ensemble of important structures for all human GPCRs. With this ensemble of all active GPCR structures in the human body, an infimum method is presented to maximize efficacy toward the selected target while minimizing binding to all other GPCRs to eliminate toxicity arising from cross-reacting with other GPCRs (a most common source of drug failure). This infimum method is broadly applicable to any set of proteins where a ligand is desired to be able to modulate the function of one protein while not affecting the function of other proteins.