Computer-aided Design of Novel CDK5 Inhibitors; Towards New Treatments of Glioblastoma

Abstract AIMS Design and statistically optimize an in silico screening protocol; apply the screening protocol to a compound database; in vitro validation of selected inhibitors (isolated CDK5 enzyme binding assay). METHOD Two large compound databases, ZINC15 and Analyticon Discovery, were used to screen for potential CDK5 ATP-binding site inhibitors. They were first filtered using a generated pharmacophore model and then virtually screened using Glide SP docking with a solved CDK5-p25 structure (PDB: 3O0G). The selected candidates were validated using enzyme binding assays to determine their inhibitory activity on CDK5, as well as against a panel of kinases. RESULTS Over 17,000 compounds were screened during molecular docking studies and of these, ~11,600 compounds returned which are predicted to bind to CDK5. The top 10% of docked compounds were analysed and 30 candidates were selected for single concentration screening at 50 µM concentrations. The 9 most potent compounds were selected for IC50 determinations and revealed 6 compounds with IC50 <10 µM. The selected novel compounds were also screened against a panel of kinases (CDK2, CDK5, CDK9, GSK-3β) with varying selective profiles observed. CONCLUSION Low micromolar potent compounds have been identified through in silico screening and validated in using in vitro binding assays. One flavonoid compound in particular, cirsiliol (IC50 = 5.90 µM), displayed best selectivity towards CDK5, a challenge in kinase inhibitor design. The identified compounds will now pass to cellular studies to determine their effect on Glioblastoma cell lines