A Coarse Grain RNA Model for Exploration of RNA Conformational Space

Abstract

Listen

Translate article

Several recent studies have suggested that RNA three-dimensional structure and dynamics are highly restricted to a small set of allowed conformations by topological constraints that are encoded at the secondary structure level. We have developed a coarse-grained model of RNA implemented within the CHARMM molecular dynamics package that allows us to further characterize the nature of RNA topological constraints. In this coarse grain model, each residue is represented using three pseudo-atoms for the phosphate, sugar, and base moieties respectively. Secondary structure is specified by modeling bonds between paired bases and parameterizing these regions to adopt A-form helical structure. All non-base paired residues are modeled without torsional potentials or attractive non-bonded forces, preserving only connectivity and repulsive steric terms. Thus, the energy landscape between different helical orientations is effectively flat, allowing efficient exploration of topologically allowed conformations. We benchmark our simulations using results from prior NMR and bioinformatics studies of two-way helix junctions. Moreover, simulations starting from a linear chain of the 76 residue tRNA-Phe molecule show that our model is able to sample the native conformation with minimal computational effort. We also show that the size of the conformational ensemble is reduced by over an order of magnitude when a limited set of three non-crystallographically determined tertiary contacts are used as restraints. In fact, the mean all phosphate RMSD over an ensemble of 100,000 structures has a value of 10 Å. We also present preliminary results of simulations done on RNAs with greater than 200 residues. These results suggest topological constraints alone, coupled with a few important tertiary contacts for larger RNAs, are enough to significantly constrain the available conformational ensemble and suggest a new approach to RNA structure prediction that is applicable to very large RNAs.