3D-QSAR CoMFA on cyclin-dependent kinase inhibitors.

Several series of cyclin-dependent kinase inhibitors previously prepared in our laboratory were compared using 3D-QSAR (CDK1) and docking (CDK2) techniques. Evaluation of our own library of 93 purine derivatives served to establish the model which was validated by evaluation of an external library of 71 compounds. The best predictions were obtained with the CoMFA standard model (q(2) = 0.68, r(2) = 0.90) and with the CoMSIA combined steric, electrostatic, and lipophilic fields (q(2) = 0.74, r(2) = 0.90). The CDK1 3D-QSAR model was then superimposed to the ATP/CDK2 binding site, giving direct contour maps of the different fields. Although too few compounds were evaluated on CDK5 to derive a 3D-QSAR model, some interesting SARs have been deduced. Comparison of the results obtained from both methods helped with understanding the specific activity of some compounds and designing new specific CDK inhibitors.