Identification of GPCR Ligands through Docking-based Virtual Screening: Not only Crystal Structures but also Homology Models

COSTANZI, STEFANO; FERINO, GIULIO; VILAR, SANTIAGO; BERK, BARKIN; ENGEL, S.; TIKHONOVA, IRINA; GERSHENGORN, M.C.; HARDEN, T.K.; JACOBSON, K.A.; CAVASOTTO, CLAUDIO N.

Camerino

Simposio; 28th Camerino-Cyprus-Noordwijkerhout Symposium ?Trekking through Receptor Chemistry?; 2010

Three-dimensionalstructures of target proteins are remarkably useful tools for thecomputer-assisted discovery of novel lead compounds. Due to the paucity ofstructural information that characterizes the superfamily of G protein-coupledreceptors (GPCRs), the scientific community has amply resorted to theconstruction of homology models which have been applied to the identificationof lead compounds through virtual screenings.  Rhodopsinhas been for years the only available template. However, recent breakthroughsin GPCR crystallography have led to the solution of the structure of a fewadditional receptors. In light of these newly elucidated crystal structures, wehave been able to produce a substantial amount of data that demonstrate thataccurate models of GPCRs in complex with their ligands can indeed beconstructed through homology modeling followed by fully flexible moleculardocking. These results have been confirmed by our success in the first blindassessment of GPCR modeling and docking, organized in coordination with thesolution of the X-ray structure of the adenosine A2A receptor. Using the b2-adrenergic receptor as a case study, we have also investigatedthe applicability of crystal structures and homology models with differentlevels of accuracy to the identification of GPCR ligands. In particular, ourcontrolled docking-based virtual screening experiments resulted very effectivenot only when based on the experimental structure, but also when conducted atthe homology models, yielding to a neat prioritization of a pool of knownbinders dispersed within a large set of decoy compounds. Remarkably, theretrieval yields can be further improved by generating an ensemble ofalternative conformations that accounts for the flexibility of the receptor.These results are very much in line with our actual successes in theidentification of lead compounds for several GPCRs, including the thyrotropinreleasing hormone (TRH) receptor,the free fatty acid receptor 1 (FFA1,or GPR40) and the nucleotide P2Y1 receptor, on the basis of virtualscreenings conducted at their homology models.