Searching of new natural DNA methyltransferase inhibitors using computational approach

Abstract

The searching of DNA methyltransferases inhibitors (DNMTs), therapeutic targets in cancer, is currently a scientific priority. However, only two DNMTs inhibitors have been approved as anticancer drugs, although with some restrictions. Natural products (NPs) are a promising source of drugs due to their wide molecular diversity and low toxicity. In order to find NPs with novel chemotypes for inhibitors of DNMTs, in this study an in silico searching was performed using a combination of QSAR and protein-ligand docking. A set of 47 compounds with known activity against these enzymes was used to construct a discriminating model of active/inactive molecules (93% accuracy), based on physicochemical descriptors generated by the DRAGON program. This classifier was used to identify potential DNMTs inhibitors on 800 NPs from the NatProd Collection (www.msdiscovery.com/natprod.html). 447 selected compounds were docked on two human DNMTs structures (DNMT1:3SWR and DNMT3A:2QRV) using AutoDock Vina and Surflex-Dock programs, with subsequent prioritization according to their score values, contact patterns at 4 Å and molecular diversity from clustering (k-means). Six NPs with novel chemotypes were identified as virtual hits against DNMTs, including 9,10-dihydro 12-hydroxygambogic, phloridzin, 2',4'-dihydroxychalcone 4'-glucoside, daunorubicin, pyrromycin and centaurein. The methodology proposed in this study is an innovative computational tool for identifying DNMT inhibitors, useful in the design of more potent and selective anticancer drugs.