Receptor-guided 3D-QSAR approach for the discoveryof c-kit tyrosine kinase inhibitors

Studies of the the three-dimensional quantitativestructure–activity relationships for ninety-five c-kit tyrosinekinase inhibitors were performed. Based on a cocrystallizedcompound (1 T46), known inhibitors werealigned with c-kit by induced-fit docking, and multipletraining/test set splitting was performed to validate theselected pharmacophore model. The best pharmacophoremodel consisted of five features: one hydrogen-bond donorand four aromatic rings. Reliable statistics were obtained(R2=0.95, Rpred2=0.75), and the model was validated byusing it to select c-kit inhibitors from a database; 82.1% ofthe hits it retrieved were active. Accordingly, our model canbe reliably used to identify new c-kit inhibitors, and canprovide useful information when designing new inhibitors.