Investigating the Folding Dynamics of RNA Pseudoknot Structural Motif via Massively Parallel Molecular Dynamics

Abstract

RNA pseudoknots compose a three-dimensional structural motif that is present in the catalytic cores of some ribozymes, and are also capable of stimulating ribosomal frameshifts. Their complex topology and non-canonical hairpin-loop composition make pseudoknots an ideal structural motif with which to study the RNA folding process. Here we report our analysis of nearly 20,000 independent all-atom molecular dynamics simulations of the ribosomal frame-shifting pseudoknot of Luteovirus and the tmRNA pseudoknot from Aquifex aeolicus, which share global topology but have only ∼50% sequence similarity. Using the Folding@Home distributed computing network and a novel Pathway Enumeration sampling method, a cumulative sampling time of over 115 μs was achieved for each of these pseudoknots. K-means clustering identified 27 conformational microstates for each pseudoknot, which reached conformational equilibrium after ∼6 ns of ensemble sampling. Multiple folding metrics were used to identify 9 macrostates participating in the folding process, including previously undescribed misfolded and intermediate states. In agreement with our previous studies of tetraloop hairpins and tRNA, the similar folding behavior of these pseudoknots suggests that native state topology is a predominant factor in the RNA folding mechanism.