Combined Ligand and Structure Based Studies on Modulators of Akt Kinases

Abstract

Protein kinase B (PKB/Akt) belongs to the AGC superfamily of related serine/threonine kinases with three structurally homologous mammalian isoforms, Akt1 (PKBα), Akt2 (PKBβ), and Akt3 (PKBγ). Akt isoforms emerged as anti-cancer drug targets because of their constitutive hyperactivation in various oncological disorders. However, due to high intra-family similarity within ATP binding region, the development of isoform-selective modulators of Akt represent a challenging endeavor and thus, until now only a handful of compounds were selected for the clinical investigation. Yet none of them could reach the market for routine clinical use due to their off-target toxicity and poor pharmacokinetic properties. Recent reports on achieving isoform selectivity by designing inhibitors against less conserved pleckstrin homology (PH) domain offer the opportunity to reduce the major off-target toxic effects of Akt antagonists. Therefore, in this thesis, combined ligand- and structure-based in silico strategies have been utilized to probe the key structural features for the inhibition of the PH domain of Akt2 which is more commonly overexpressed in solid tumors. Toward this end, various predictive 2DQSAR (two-dimensional quantitative structure-activity relationship) and GridIndependent Molecular Descriptor (GRIND) and pharmacophore models using structurally diverse data set of 111 quinoline analogs have been developed. Our key findings demonstrate that the presence of three hydrogen bond donors (D1-D3) at a molecular distance of ~8.4, 21.6, 16.8 Å between D1–D2, D1–D3 and D2–D3, respectively is important for selective inhibition of PH domain of Akt2. In addition, our docking results indicated a crucial role of Lys30 for the optimal fit of quinoline-type inhibitors within the binding cavity of the Akt2 PH domain. Moreover, the structurebased pharmacophore model exhibited three hydrogen bond acceptors (A1-A3) at a distance of 4.05 Å (A1-A2), 11.58 Å (A2-A3) and 15.33 Å (A1-A3) that are complementary to the molecular distances identified by GRIND which further validate the reliability of our developed models. Additionally, identified hits through pharmacophoric-based virtual screening provided a new arsenal of potentially selective chemical scaffolds which have a broad structural diversity and less chemical similarity to any of the other known Akt2-PH domain inhibitors. Subsequently, selectivity profiling with the help of proteochemometric modeling revealed essential substructures such as Nmethylpent-3-en-2-amine for selective inhibition of Akt1, methylene amine, isoproenylterazol and 2H-tetrazole for Akt2, and formaldehyde hydrazine for the Akt3 selective inhibition. The present work also illustrates the substructure based similarity search of ChemBridge database to identify Akt2-selective hit compounds. In the present study, one of the selected carboxamide-type hit showed 1.2 and 2.1 fold selectivity against Akt2 as compared to Akt1 and Akt3, respectively.

Overall, the work described in this thesis could pave the way towards the identification of potential modulators of Akt2 against cell proliferation in cancer with high isoform-selectivity and limited side effects.