RESUMO
G protein-coupled receptors (GPCRs) are valuable therapeutic targets for many diseases. A central question of GPCR drug discovery is to understand what determines the agonism or antagonism of ligands that bind them. Ligands exert their action via the interactions in the ligand binding pocket. We hypothesized that there is a common set of receptor interactions made by ligands of diverse structures that mediate their action and that among a large dataset of different ligands, the functionally important interactions will be over-represented. We computationally docked ~2700 known ß2AR ligands to multiple ß2AR structures, generating ca 75 000 docking poses and predicted all atomic interactions between the receptor and the ligand. We used machine learning (ML) techniques to identify specific interactions that correlate with the agonist or antagonist activity of these ligands. We demonstrate with the application of ML methods that it is possible to identify the key interactions associated with agonism or antagonism of ligands. The most representative interactions for agonist ligands involve K972.68×67 , F194ECL2 , S2035.42×43 , S2045.43×44 , S2075.46×641 , H2966.58×58 , and K3057.32×31 . Meanwhile, the antagonist ligands made interactions with W2866.48×48 and Y3167.43×42 , both residues considered to be important in GPCR activation. The interpretation of ML analysis in human understandable form allowed us to construct an exquisitely detailed structure-activity relationship that identifies small changes to the ligands that invert their pharmacological activity and thus helps to guide the drug discovery process. This approach can be readily applied to any drug target.