Inhibitor binding to active and inactive CDK2: The crystal structure of CDK2-cyclin A/indirubin-5-sulphonate

Background: Cyclin-dependent kinase 2 (CDK2) is an important target for structure-based design of antitumor agents. Monomeric CDK2 is inactive. Activation requires rearrangements to key structural elements of the enzyme's active site, which accompany cyclin binding and phosphorylation. To assess the validity of using monomeric CDK2 as a model for the active kinase in structure-based drug design, we have solved the structure of the inhibitor indirubin-5-sulphonate (E226) complexed with phospho-CDK2-cyclin A and compared it with the structure of E226 bound to inactive, monomeric CDK2. Results: Activation of monomeric CDK2 leads to a rotation of its N-terminal domain relative to the C-terminal lobe. The accompanying change in position of E226 follows that of the N-terminal domain, and its interactions with residues forming part of the adenine binding pocket are conserved. The environment of the ATP-ribose site, not explored by E226, is significantly different in the binary complex compared to the monomeric complex due to movement of the glycine loop. Conformational changes also result in subtle differences in hydrogen bonding and electrostatic interactions between E226's sulphonate and CDK2's phosphate binding site. Affinities calculated by LUDI for the interaction of E226 with active or inactive CDK2 differ by a factor of approximately ten. Conclusions: The accuracy of monomeric CDK2 as an inhibitor design template is restricted to the adenine binding site. The general flexibility observed for the glycine loop and subtle changes to the phosphate binding site suggest a need to study interactions between inhibitors and active CDK2 in structure-based drug design programs.