42 Computer folding of RNA tetraloops? Are we there yet?

Abstract

The tetraloops (Tls) are basic and unusually stable building blocks of RNA structure that often participate in a variety of biochemical processes – including nucleation in RNA, folding and formation of tertiary contacts. Moreover, they are known to play roles in transcription and translation as well as serving as recognition sites for RNA-binding proteins. Understanding the folding of small RNA hairpins is a critical first step in understanding the folding of larger RNA molecules. In present study, we investigate the folding and unfolding of two RNA Tls at the atomic level based on replica exchange molecular dynamics (MD) simulations. Using the most recent reparametrisation of the Amber family of RNA force fields (ff99bsc0OL3 (Banáš et al. 2010, Zgarbová et al. 2011)), we have, for the first time, folded tetraloop structures to within 2 Å all-atom RMSD of the native structure involving all signature interactions of native fold, starting from fully unfolded conformations. The most native-like structures reported by previous studies were only ∼4 Å from native; due to force field deficiencies, Tl structures degraded even in short MD simulations initiated from folded (Banáš et al. 2010). We described the folding pathway, which is very similar for both tetraloops, and it is comparable with experimental results. Further, we can use our results to address several questions related to the biological function of tetraloops. For example, an open question is whether the antibiotic restritocin induces the conformational change of the GNRA TL or binds to a transiently unstructured GNRA TL. We observed the misfolded state of UUCG Tl recently suggested by mutation experiments, which most likely acts as a kinetic trap during UUCG folding. We can also relate our results to ultrafast spectrosopy experiments that identified some features of misfolded states of GNRA TLs reminiscent of the misfolded-compact GAGA structure identified in this study. Nevertheless, our simulations show that despite REMD technique being powerful tool which significant increases the sampling, obtaining an equilibrium between the unfolded, misfolded and folded states is still very challenging.