Multilevel virtual screening approach for discovery of thymidine phosphorylase inhibitors as potential anticancer agents

Abstract

Thymidine phosphorylase (TP) is an enzyme which catalyses the conversion of thymidine to thymine and 2-deoxy-α-d-ribose-1-phosphate. The 2-deoxy-α-d-ribose-1-phosphate is then dephosphorylated to the 2-deoxy-α-d-ribose, which can promote angiogenesis, block apoptotic pathway and activate kinase pathway. Therefore, inhibitors of TP are described as potential anti-neoplastic agents. In this study, we discovered new thymidine phosphorylase inhibitors using multilevel virtual screening. Initially, ZINC database was screened to find analogs of some known TP inhibitors. Obtained records were docked to the enzyme active site. Ligand selection was performed according to the GoldScore value, binding mode and required physicochemical properties. All in silico studies allowed for selection of 39 promising ligands from nearly 22 000 records obtained from ZINC database. Molecular modeling helped us to discuss their interactions within the TP active site. Selected 7 hits were then tested in vitro using spectrophotometric method, among them three inhibitors showed micromolar biological activity comparable to the reference ligand, 7-deazaxanthine (IC50 = 58 µM). Found hits VS-03, VS-04 and VS-05 had IC50 values of 98, 96 and 103 µM, respectively. The obtained results constitute a contribution into development of novel anticancer agents.