Exploring RNA Drug Binding Using Consistent Charge Models

Abstract

RNA has emerged as a prospective drug target for a wide range of diseases. However, existing drug discovery tools optimized for protein targets have been largely unsuccessful in producing novel small molecules that target RNA due to RNAs distinct chemical and structural characteristics. Prevalent screening methods assay catalytic activity and are therefore unsuitable for RNA, as few RNAs are catalytically active. Furthermore, RNA has flexible binding sites, precluding the causal relationship between strong binding and inhibition of activity. Computational methods, including molecular docking, can overcome some of these limitations; however, RNA can adopt radically different conformations upon binding small molecules which can be difficult to model computationally. An additional challenge is that current force fields remain underdeveloped in modeling polyanionic nucleic acids with complex electrostatic interactions. Accurately capturing these interactions is crucial to determining precise free energy calculations of binding. We have developed a force field extension model that allows the same charges and force field parameters to be used for both the receptor and ligand, significantly improving the accuracy of capturing interactions between receptor RNAs and potential drug-like small molecules. Here we present details on our force field extension model, predicting the hydration free energy of 503 organic molecules using free energy perturbation demonstrating that our model reproduces experimentally confirmed results. Furthermore, we present docking results of predicted binding affinities and ligand-bound poses for a set of 60 known RNA-ligand complexes from the Protein Databank and compare results with other docking programs. Finally, we predict the binding affinities of 65 small molecules to the HIV-1 TAR RNA and compare results with a high throughput screen.