The structure of active opsin as a basis for identification of GPCR agonists by dynamic homology modelling and virtual screening assays

Most of the currently available G protein-coupled receptor (GPCR) crystal structures represent an inactive receptor state, which has been considered to be suitable only for the discovery of antagonists and inverse agonists in structure-based computational ligand screening. Using the β2-adrenergic receptor (B2AR) as a model system, we show that a dynamic homology model based on an “active” opsin structure without further incorporation of experimental data performs better than the crystal structure of the inactive B2AR in finding agonists over antagonists/inverse agonists. Such “active-like state” dynamic homology models can therefore be used to selectively identify GPCR agonists in in silico ligand libraries.

► We used an opsin structure as template for a homology model of an active GPCR. ► By MD simulations, we determined dynamic interaction fingerprints between receptor model and ligands. ► Agonist fingerprints selectively found agonists in a ligand test set. ► Our technique can be used to discriminate agonists and antagonists/inverse agonists in virtual ligand libraries.