Determination of in silico rules for predicting small molecule binding behavior to nucleic acids in vitro.

Abstract

The vast knowledge of nucleic acids is evolving and it is now known that DNA can adopt highly complex, heterogeneous structures. Among the most intriguing are the G-quadruplex structures, which are thought to playa pivotal role in cancer pathogenesis.Efforts to find new small molecules for these and other physiologically relevant nucleic acid structures have generally been limited to isolation from natural sources or rationale synthesis of promising lead compounds.However, with the rapid growth in computational power that is increasingly becoming available, virtual screening and computational approaches are quickly becoming a reality in academia and industry as an efficient and economical way to discover new lead compounds. These computational efforts have historically almost entirely focused on proteins as targets and have neglected DNA. We present research here showing that not only can software be utilized for targeting DNA, but that selectivity metrics can be developed to predict the binding mechanism of a small molecule to a DNA target. The software Surflex and Autodock were chosen for evaluation and were demonstrated to be able to accurately reproduce the known crystal structures of several small molecules that bind by the most common nucleic acid interacting mechanisms of groove binding and intercalation. These software IV were further used to rationalize known affinity and selectivity data of a 67 compound library of compounds for a library of nucleic acid structures including duplex, triplex and quadruplexes. Based upon the known binding behavior of these compounds, in silica metrics were developed to classify compounds as either groove binders or intercalators.These rules were subsequently used to identify new triplex and quadruplex binding small molecules by structure and ligand-based virtual screening approaches using a virtual library consisting of millions of commercially available small molecules. The binding behavior of the newly discovered triplex and quadruplex binding compounds was empirically validated using a number of spectroscopic, fluorescent and thermodynamic equilibrium techniques. In total, this research predicted the binding behavior of these test compounds in silica and subsequently validated these findings in vitro. This research presents a novel approach to discover lead compounds that target multiple nucleic acid morphologies.