Abstract

There is an urgent need for more effective treatment protocols for glioblastoma (GBM), the most common malignant brain tumour in adults. Kinase inhibitors have demonstrated significant potential, with cyclin-dependant kinase 5 (CDK5) recognised as a promising target. Here, we present an in silico led approach towards the discovery of CDK5 inhibitors with anti-GBM potential. Active/decoy benchmarking computations considered the available CDK5–ligand complex crystal structures, to design and optimize a virtual screening protocol that involved 3D pharmacophore pre-filtering of compounds and Glide docking calculations. Screening of the ZINC15 biogenic and Analyticon Discovery natural product databases led to the identification of six low micromolar CDK5/p35 flavonoid inhibitors, with fisetin (IC50 = 0.90 µM) and apigenin (IC50 = 1.19 µM) validated as the most potent. Additionally, three diverse chemical scaffolds with xanthone, aristolactam and β-carboline cores were identified as hit compounds for further optimization. Structure activity relationship (SAR) analysis of sixteen flavonoids tested in the CDK5/p35 in vitro binding assays unravelled structural features governing the observed potencies, while selectivity screening against five homologous kinases revealed cirsiliol as the most selective flavonoid for CDK5 (IC50 = 4.73 µM). Meanwhile, fisetin demonstrated the most potent effects on cell viability (IC50s ∼ 20-40 µM at 72 h) against three GBM cell lines (U87-MG, T98G and U251-MG), in a generally time- and concentration-dependent manner.