The Potential, U, is Essential: The many conformations from a simulation produce a free energy surface. The surface is realistic, if the force field, FF, is. The database of RNA sequences is exploding, but knowledge of energetics, structures, and dynamics lags behind. All-atom computational methods, such as molecular dynamics, hold promise for closing this gap. New algorithms and faster computers have accelerated progress in improving the reliability and accuracy of predictions. Currently, the methods can facilitate refinement of experimentally determined NMR and x-ray structures, but are less reliable for predictions based only on sequence. Much remains to be discovered, however, about the many molecular interactions driving RNA folding and the best way to approximate them quantitatively. The large number of parameters required means that a wide variety of experimental results will be required to benchmark force fields and different approaches. As computational methods become more reliable and accessible, they will be used by an increasing number of biologists, much as x-ray crystallography has expanded. Thus, many fundamental physical principles underlying the computational methods are described. This review presents a summary of the current state of molecular dynamics as applied to RNA. It is designed to be helpful to students, postdoctoral fellows, and faculty who are considering or starting computational studies of RNA.

The design of compounds that can discriminate between closely related target proteins remains a central challenge in drug discovery. Specific therapeutics targeting the highly conserved myosin motor family are urgently needed as mutations in at least 6 of its members cause numerous diseases. Allosteric modulators, like the myosin-II inhibitor blebbistatin, are a promising means to achieve specificity. However, it remains unclear why blebbistatin inhibits myosin-II motors with different potencies given that it binds at a highly conserved pocket that is always closed in blebbistatin-free experimental structures. We hypothesized that the probability of pocket opening is an important determinant of the potency of compounds like blebbistatin. To test this hypothesis, we used Markov state models (MSMs) built from over 2 milliseconds of aggregate molecular dynamics simulations with explicit solvent. We find that blebbistatin’s binding pocket readily opens in simulations of blebbistatin-sensitive myosin isoforms. Comparing these conformational ensembles reveals that the probability of pocket opening correctly identifies which isoforms are most sensitive to blebbistatin inhibition and that docking against MSMs quantitatively predicts blebbistatin binding affinities (R2=0.82). In a blind prediction for an isoform (Myh7b) whose blebbistatin sensitivity was unknown, we find good agreement between predicted and measured IC50s (0.67 mM vs. 0.36 mM). Therefore, we expect this framework to be useful for the development of novel specific drugs across numerous protein targets.

To benchmark RNA force fields, we compared the folding stabilities of three 12-nucleotide hairpin stem loops estimated by simulation to stabilities determined by experiment. We used umbrella sampling and a reaction coordinate of end to end (5′ to 3′ hydroxyl oxygen) distance to estimate the free energy change of the transition from the native conformation to a fully extended conformation with no hydrogen bonds between non-neighboring bases. Each simulation was performed four times using the AMBER FF99+bsc0+χOL3 force field and each window, spaced at 1 Å intervals, was sampled for 1 μs, for a total of 552 μs of simulation. We compared differences in the simulated free energy changes to analogous differences in free energies from optical melting experiments using thermodynamic cycles where the free energy change between stretched and random coil sequences is assumed to be sequence independent. The differences between experimental and simulated ΔΔG° are on average 0.98±0.66 kcal/mol, which is chemically accurate and suggests analogous simulations could be used predictively. We also report a novel method to identify where replica free energies diverge along a reaction coordinate, thus indicating where additional sampling would most improve convergence. We conclude by discussing methods to more economically perform these simulations.

The ε4-allele variant of apolipoprotein E (ApoE4) is the strongest genetic risk factor for Alzheimer’s disease, although it only differs from its neutral counterpart ApoE3 by a single amino acid substitution. While ApoE4 influences the formation of plaques and neurofibrillary tangles, the structural determinants of pathogenicity remain undetermined due to limited structural information. Previous studies have led to conflicting models of the C-terminal region positioning with respect to the N-terminal domain across isoforms largely because the data are potentially confounded by the presence of heterogeneous oligomers. Here, we apply a combination of single-molecule spectroscopy and molecular dynamics simulations to construct an atomically detailed model of monomeric ApoE4 and probe the effect of lipid association. Importantly, our approach overcomes previous limitations by allowing us to work at picomolar concentrations where only the monomer is present. Our data reveal that ApoE4 is far more disordered and extended than previously thought and retains significant conformational heterogeneity after binding lipids. Comparing the proximity of the N- and C-terminal domains across the three major isoforms (ApoE4, ApoE3, and ApoE2) suggests that all maintain heterogeneous conformations in their monomeric form, with ApoE2 adopting a slightly more compact ensemble. Overall, these data provide a foundation for understanding how ApoE4 differs from nonpathogenic and protective variants of the protein.