Abstract
Regulator of G protein Signaling-14 (RGS14), an intracellular inactivator of G protein-coupled receptor (GPCR) signaling, is considered an undruggable protein given its shallow and relatively featureless protein-protein interaction interface combined with a distal allosteric site prone to nonspecific inhibition by thiol-reactive compounds. Here, we identify and validate a tractable chemotype that selectively and non-covalently inhibits RGS14 GTPase-accelerating protein (GAP) activity. Combining structure-guided virtual screening, ligand docking across multiple receptor conformers, and enrichment validation, we progressed from a first-generation active, Z90276197, to over 40 second-generation analogs with improved potency. These inhibitors are predicted to engage the solvent-exposed "canyon" in the RGS14 RGS-box that interacts with the Gα switch I region. Binding pose predictions underscored the importance of non-polar interactions and shape complementarity over polar interactions in engaging this Gα-binding canyon and revealed an "ambidextrous" pattern of R1- and R2-group orientations. GAP inhibition was confirmed in fluorescence-based and gold-standard radioactive GTP hydrolysis assays. Two second-generation analogs, Z55660043 and Z55627844, inhibited RGS14 GAP activity in both assays and without measurable cytotoxicity. Deep learning-based scoring of predicted docking poses further supported observed affinity gains from R3-group additions. One analog demonstrated favorable in vivo pharmacokinetics and CNS penetration. Collectively, our findings establish tractable, non-covalent, small molecule inhibition of a G protein regulatory interface and illustrate how machine learning-enhanced docking can guide ligand optimization for shallow protein surfaces. This work opens the door to future development of RGS14 inhibitors as potential therapeutics for central nervous system and metabolic disorders.