Poster Presentation 2014 International Biophysics Congress

Computer-aided drug discovery: development of a method for G protein-coupled receptor binding pocket refinement (#317)

Thomas Coudrat 1 , John Simms 1 , Denise Wootten 1 , Arthur Christopoulos 1 , Patrick Sexton 1
  1. Monash Institute of Pharmaceutical Science, Parkville, VIC, Australia

G Protein-Coupled Receptors (GPCRs) are a superfamily of membrane proteins that mediate cellular responses to their environment upon binding of an effector to their extracellular- or transmembrane domain-binding pocket. With over 800 human GPCRs playing key roles in modulating tissue/cell physiology and homoeostasis, they represent a major target for pharmaceutical intervention.

Structure-Based Drug Discovery (SBDD) uses the complementarity of shape and electrochemical properties between a binding pocket and its active ligands. With Virtual Screening (VS), one can identify drug leads by evaluating large small molecule libraries for complementarity with the binding pocket. The success of a VS highly depends on the conformation of the binding pocket, which is why a key step of SBDD is to refine the binding pocket within protein structures obtained from X-ray diffraction and/or homology models.

Here we present a new computationally efficient Ligand Directed Modelling (LDM) method for GPCR binding pocket refinement. This method aims to establish the global energy minimum of a GPCR binding pocket in complex with a known active ligand by using protein sampling, docking and scoring of ligand/binding pocket complexes in recursive steps.

To benchmark the method, we have used family A GPCR structures that have been crystallised with known ligands and compared LDM refined ligand/binding pocket complexes with known X-ray structure complexes. We tested a range of different LDM refinement scenarios and compared the results using three metrics: binding pattern between the ligand and the binding pocket, binding pocket shape and binding pocket recovery of known actives in a small scale VS. This benchmark is a guideline for the application of the LDM method in future SBDD projects.