In silico discovery and in vitro validation of novel CDK5 and CDK9 inhibitors with glioblastoma therapeutic potential

Abstract

Glioblastoma (GBM) is the most common glioma in adults, with a poor survival rate of only ~15 months post-diagnosis. The aggressive and incurable nature of this tumour, together with the poor efficacy of current treatment methods, underlines the importance of the discovery of novel treatment strategies for GBM. Dysregulation of both Cyclin-Dependent Kinases (CDK) -5/-9 have been implicated in a number of cancers and neurological disorders, including GBM, and previous research has demonstrated that drug-like selective inhibitors may have therapeutic potential for GBM.
Two types of CDK5/9 inhibition were explored within this thesis: type-I and -II inhibition. Type-I inhibition is the most common mechanism of kinase inhibition where inhibitors bind at the ATP-binding site when the kinase is active. However, this approach may be limited because the highly conserved nature of kinase ATP-binding sites results in selectivity and off-target toxicity concerns. Type-II inhibitors bind to the inactive state of a kinase, occupying both the ATP-binding site and a nearby hydrophobic allosteric pocket. The allosteric pocket is much less conserved, allowing the potential of increased selectivity. Importantly, type-II inhibition has been much less studied for kinases (compared to type-I) and has not previously been directly investigated for CDK5/9.
In this work, the study of type-I inhibition for CDK5 and CDK9 revealed six low micromolar flavonoid inhibitors (compounds 1, 2, 3, 6, 7 and 9) of CDK5 (IC50 < 10 μM) and six low micromolar diverse inhibitors (compounds 31, 34, 37, 40, 42 and 44) of CDK9 (IC50 < 20 μM). These were identified using in silico docking studies and validated using in vitro binding assays using isolated purified forms of CDK5/9. For CDK5, 1 (fisetin) (IC50 = 0.9 μM) and 6 (apigenin) (IC50 = 1.2 μM), were found to be the most potent. In the case of CDK9, five unique scaffolds were identified, with compound 37 and 42 (IC50 = 7.6 μM and 6.2 μM) both with a 4H,8H,9H,10H-pyrano[2,3-h]chromene-4,8-dione scaffold found to be the most potent.
In the absence of a solved crystallographic structure of CDK5/9 with a type-II inhibitor, using Desmond metadynamics (MetaD) and molecular dynamics (MD) simulations, a DFG-out/ inactive conformation of CDK5 was created for in silico screening to identify potential type-II inhibitors. Selection of a Phase I set of compounds for experimental validation resulted in the identification of two hit compounds with IC50s of 13.9 μM (compound 49) and 9.8 μM (compound 50) for CDK5/p35 inhibition, both possessing a [pyrimidin-2-yl]amino-furo[3,2-b]-furyl-urea scaffold. SAR analysis of a Phase II set of analogues using the hit scaffold revealed ten compounds with CDK5/p35 IC50s < 56 μM. Four of the most potent type-II compounds (50, 73, 82 & 88) were screened against eleven other homologous kinases revealing selective inhibition for CDK5, CDK9, CDK2 and GSK-3α/β, all of which are potential targets for GBM treatment. IC50 determinations for these four compounds against the kinases revealed low nanomolar range for CDK9 and low micromolar range for CDK5, CDK2, and GSK-3α/β. The investigation of type-II inhibition for CDK5/9 has been discussed in depth for the first time, identifying [pyrimidin-2-yl]amino-furo[3,2-b]-furyl-urea/amide as a potent scaffold for CDK5/9 lead optimisation.
Cellular studies of the best identified kinase inhibitors (six CDK5 type-I, six CDK9 type-I and four CDK5/9 type-II) were conducted to determine cell viability following 24-, 48-, and 72- hours treatment against three GBM cell lines (U87-MG, T98G and, U251-MG), investigating the effects of the type-I/-II compounds at the cellular level. The compounds were generally found to decrease cell viability in a time- and concentration- dependent manner. In terms of the type-I inhibitors, compound 31 was the most potent with IC50 of < 13.5 μM across the GBM cell lines following 72 hours treatment. Other type-I inhibitors of note were compound 1(fisetin) and 37 with IC50s of 22-40 μM and 58-87 μM following 72 hours, respectively. In the case of the four aforementioned CDK5/9 type-II inhibitors, the IC50s against the three GBM cell lines and primary cultures were determined to be ~20-35 μM for 73, 82 and 88, and ~35-85 μM for 50, following 72 hours treatment. This, therefore, reveals the ability of type-II dual inhibition of CDK5/9 as well as other kinases
that are targets for GBM (CDK2, GSK-3α/β) to result in potent effects against GBM at the cellular level, making these compounds worthy of further investigation.
The research presented here has demonstrated the potential of natural product CDK5 flavonoid inhibitors such as fisetin as well as novel identified CDK9 type-I inhibitors against GBM. In particular, the CDK9 inhibitor scaffolds of 31 and 37, 3-phenyl-5-(2,3,4,9-tetrahydro-1H-pyrido[3,4-b]indol-1-yl)pyrimidine-2,4(1H,3H)-dione and 4H,8H,9H,10H-pyrano[2,3-h]chromene-4,8-dione, respectively, are proposed as novel lead compounds. Furthermore, the discovery of the type-II inhibitors with dual CDK5/9 inhibition promotes a promising anti-GBM potential. Select compounds did also show effects on the cell viability of the non-cancer derived glial SVG p12 cell line, which clearly needs further investigation, but the compounds were generally found to have promising calculated pharmacokinetic profiles. As a result of this study, therefore, compounds (both type-I and type-II inhibitors) for the development of the much needed new potential therapeutic options against GBM have been identified.