Integrating Enhanced Molecular Sampling, RNA-Specific Scoring Functions, and SILCS Technology for Small-Molecule Targeting of RNA

Abstract

Increasing knowledge of the wide range of biological functions of ribonucleic acids (RNAs) has made them a critical target in drug development. Accordingly, modifying their function by targeting them with small, drug-like molecules for the development of therapeutic agents is attractive. However, the rational design of RNA-targeting small molecules has been hindered by the intrinsic structural complexity and dynamics of the RNAs. Here, we develop an in silico method to facilitate RNA-targeting small-molecule discovery that integrates enhanced sampling molecular dynamics (MD) simulations using the classical Drude polarizable force field with site identification by ligand competitive saturation (SILCS) technology. Enhanced sampling MD simulations initiated with apo RNA structures guided by reaction coordinates selected from order parameters identified through machine learning combined with survey data of ligand-bound structures of TAR RNA enable identification of druggable RNA conformations. Two custom RNA scoring functions are developed to (1) select RNA conformations with a high probability of containing druggable sites and (2) identify nucleotides comprising binding sites that are able to participate in interactions with small molecules. Subsequently, selected conformations are subjected to the SILCS method for the final selection of potential binding sites. Across three validation RNA systems, our method successfully identifies all experimentally known binding sites for four RNAs with both helical and noncanonical regions. In addition, novel potential binding sites were identified, supported by high-affinity SILCS FragMaps and docking analysis of FDA-approved ligands. However, only one of two binding sites in the THF riboswitch was identified, indicating limitations in the application of the method to RNAs with tertiary contacts. The developed method is anticipated to enable efficient virtual screening and lead optimization for RNA-targeted drug discovery.