The continuing discoveries of RNA chemical modifications and their critical roles in cellular and tumor machinery ? the `epitranscriptomics' revolution ? suggest novel routes to anticancer therapies based on disabling, manipulating, and repurposing chemical modifications at a molecular level. Unfortunately, our ability to computationally model chemically modified RNA lags behind the explosive rate of experimental discoveries; this absence of unbiased and automated tools to aid expert exploration is resulting in inconsistencies, gaps, and accumulating errors in our basic understanding of the epitranscriptome. To address this challenge, we have recently developed a framework for modeling chemically modified RNA nucleotides within the Rosetta package for high- resolution macromolecular structure prediction. Here, we outline how tools developed in this framework can resolve two fundamental bottlenecks in modeling chemically modified RNA. To address the unacceptably high frequency of stereochemical errors in experimental 3D structures of epitranscriptomic enzymatic complexes and RNA machines, we will expand and test ERRASER (Enumerative Real-space Refinement ASsisted by Electron density under Rosetta) to automatically rebuild these functionally critical regions in crystallographic and cryoelectron microscopy maps. To enable first automated calculations of secondary structure arrangements proposed to underlie epitranscriptomic gene regulation, we will expand and test the Reweighting of Energy- function Collection with Conformational Ensemble Sampling (RECCES) method for nearest-neighbor parameter prediction, focusing on the 10 most common modifications while enabling generalization to all other modifications, including ones that remain undiscovered. We will evaluate success through blind challenges, independent tests by expert biological and bioengineering collaborators, and by assessing adoption of our methods and software tools by the broader research community. In the same way that specially developed computational biology tools have helped establish a biomedically impactful understanding of proteins and their posttranslational chemical regulation, we propose that the technologies outlined here will strengthen our understanding of structure in posttranscriptionally modified RNAs, providing a firmer basis for their biomedical activation or disruption.
Non-standard RNAs with chemical modifications play fundamental roles in transmitting and regulating genetic information in all living systems. When aberrant, these molecules and their interactions induce inherited human disease and are associated with the proliferation of cancers. Potentially life-saving therapies that target or repurpose these chemical modifications are hindered by our imperfect ability to computationally model them. Our work aims to address this need for computational tools for chemically modified RNA in current biomedical research.
| Watkins, Andrew M; Geniesse, Caleb; Kladwang, Wipapat et al. (2018) Blind prediction of noncanonical RNA structure at atomic accuracy. Sci Adv 4:eaar5316 |
