Temperature-Dependent Twist of Double-Stranded RNA Probed by Magnetic Tweezer Experiments and Molecular Dynamics Simulations

Abstract

RNA plays critical roles in the transmission and regulation of genetic information and is increasingly used in biomedical and biotechnological applications. Functional RNAs contain extended double-stranded regions, and the structure of doublestranded RNA (dsRNA) has been revealed at high resolution. However, the dependence of the properties of the RNA double helix on environmental effects, notably temperature, is still poorly understood. Here, we use single-molecule magnetic tweezer measurements to determine the dependence of the dsRNA twist on temperature. We find that dsRNA unwinds with increasing temperature, even more than DNA, with ΔT wRNA = −14.4 ± 0.7°/(°C•kbp), compared to ΔT wDNA = −11.0 ± 1.2°/(°C•kbp). All-atom molecular dynamics (MD) simulations using a range of nucleic acid force fields, ion parameters, and water models correctly predict that dsRNA unwinds with rising temperature but significantly underestimate the magnitude of the effect. These MD data, together with additional MD simulations involving DNA and DNA−RNA hybrid duplexes, reveal a linear correlation between the twist temperature decrease and the helical rise, in line with DNA but at variance with RNA experimental data. We speculate that this discrepancy might be caused by some unknown bias in the RNA force fields tested or by as yet undiscovered transient alternative structures in the RNA duplex. Our results provide a baseline to model more complex RNA assemblies and to test and develop new parametrizations for RNA simulations. They may also inspire physical models of the temperature-dependent dsRNA structure.

Comprehensive Assessment of Force-Field Performance in Molecular Dynamics Simulations of DNA/RNA Hybrid Duplexes

“…However, many more groups are separately working on deriving and optimizing parameters for only either the RNA or DNA. ,− In those cases, the choice of the second nucleic acids ff is arbitrary and difficult to justify, especially when no specific recommendation is given by the ff developers and the hybrids are seldom among the systems used for benchmarking. In fact, there is currently no study systematically assessing the ff performance in MD simulations of hybrid duplexes available in the literature, although computational studies of specific hybrids have been previously carried out. − …”