This Title All WIREs
How to cite this WIREs title:
Impact Factor: 9.957

How to understand atomistic molecular dynamics simulations of RNA and protein–RNA complexes?

Full article on Wiley Online Library:   HTML PDF

Can't access this content? Tell your librarian.

We provide a critical assessment of explicit‐solvent atomistic molecular dynamics (MD) simulations of RNA and protein/RNA complexes, written primarily for non‐specialists with an emphasis to explain the limitations of MD. MD simulations can be likened to hypothetical single‐molecule experiments starting from single atomistic conformations and investigating genuine thermal sampling of the biomolecules. The main advantage of MD is the unlimited temporal and spatial resolution of positions of all atoms in the simulated systems. Fundamental limitations are the short physical time‐scale of simulations, which can be partially alleviated by enhanced‐sampling techniques, and the highly approximate atomistic force fields describing the simulated molecules. The applicability and present limitations of MD are demonstrated on studies of tetranucleotides, tetraloops, ribozymes, riboswitches and protein/RNA complexes. Wisely applied simulations respecting the approximations of the model can successfully complement structural and biochemical experiments. WIREs RNA 2017, 8:e1405. doi: 10.1002/wrna.1405 This article is categorized under: RNA Structure and Dynamics > RNA Structure, Dynamics, and Chemistry
Left: Time development of H‐bond heavy atom distances (listed in the box) in two simulations of the U1A protein/RNA complex, a typical system which is unable to simultaneously keep all the native H‐bonds. The protein/RNA interface is perturbed right after the start and although all the individual H‐bonds are capable to sample their native distances and the structure does not further degrade, there are no trajectory portions with all H‐bonds formed simultaneously. Right: application of the HBfix local potential (present work result) nicely stabilizes the protein/RNA interface. The individual H‐bonds are shown in three separate graphs; the colored points indicate the initial values and the black horizontal line marks 3.2 Å.
[ Normal View | Magnified View ]
Minimalistic workflow for executing simulations of protein/RNA complexes. Some systems may require additional actions to be taken. The simulations require a very careful preparation and interpretation, which need to be done case by case. A single incorrectly protonated histidine side‐chain may ruin all the effort. Blind reliance on popular simple analytical tools such as principal component analysis or cross‐correlation diagrams without full understanding of the trajectories may easily lead to false‐positive results, when simulation artifacts caused by the starting structures and force fields are overlooked or even misinterpreted as signs of biochemically relevant dynamics.
[ Normal View | Magnified View ]
Profile of the glycosidic χ torsion energy term in the force field. Ladder‐like transition is connected with χ transition from the anti region to the high‐anti region.
[ Normal View | Magnified View ]
Illustrative applications of the simulation techniques on medium‐sized RNA systems showing abilities and limits of these theoretical methods. (a) Modeling of the reactive form of the HDV ribozyme active site. Simulation techniques were not able to model the correct reactive conformation based on X‐ray structures of C75U mutant, which likely introduced too large structural rearrangement. The successful remodeling of the active site was finally achieved by Hammes‐Schiffer and coworkers starting from deoxy‐U‐1 inhibited X‐ray structure. (b) The enhanced sampling methods might investigate stability of various molecular contacts as illustrated on modeling of P1 stem (un)zipping and formation/breaking of the kissing‐loop interaction between P2 and P3 stems of adenine‐sensing add A‐riboswitch. The careful work by Bussi and co‐workers is instrumental to understand convergence limitation of the MD techniques, which are very often not properly considered in the literature.
[ Normal View | Magnified View ]
Local auxiliary (HBfix) potential supporting selected H‐bonding interactions that are underestimated in force fields. It biases the simulation only in a narrow range of heavy atom distances from rbeg to rend (e.g., from 3 to 4 Å) while there are zero bias forces elsewhere. The η is another tunable parameter adjusting the level of energy support for a given interaction.
[ Normal View | Magnified View ]
(a) The overlay of RNA and protein backbones of RNA Duplex‐Quadruplex Junction Complex with FMRP RGG peptide in the NMR (red), X‐ray (blue), and an average MD structure from the last nanoseconds of a 2 µs simulation (green). (b) The NMR structure of the mixed tetrad upon manual insertion of a K+ ion (left) and the final conformation after ~620 nanoseconds. (c) The simulation time development of H‐bond heavy atom distances of the four protein/RNA H‐bonds present in the NMR structure. The colored dots indicate the distance value in the initial structure. The black lines indicate 3.2 Å. The interactions were unstable in the initial simulations (left) but were eventually stabilized after insertion of the K+ ion into the mixed tetrad (right). (d) The simulation time development of eight H‐bond heavy atom distances of the interactions predicted by the MD and seen in the X‐ray structure which are not seen in the starting NMR structure (left). Note that for the last two interactions the heavy atom distance was under 3.2 Å in the NMR structure but their donor/hydrogen/acceptor atom angle was outside of the 120 –180° range typical for H‐bonds; thus, the simulation time development of these angles is shown (right), the red line indicates 120°.
[ Normal View | Magnified View ]
Native structure (a) and examples (b, c) of non‐native RNA structures stabilized by hydrogen bonds between phosphate non‐bridging oxygen atoms and ribose hydroxyl groups or nucleic acids bases (in the circles). The structure in C is an example of so‐called inverted stacking conformation. Snapshots from a simulation of r(CCCC) with χOL3 force field.
[ Normal View | Magnified View ]
Overlay of the original NMR ensemble (a) and the MD‐adapted ensemble (b) of structures of the SRSF1 protein/RNA complex. The protein backbone is in cyan, the RNA backbone is in orange, the phosphate groups are red and the RNA bases are blue. The amino acids constituting the protein/RNA interface are labeled and displayed in green. The MD‐adapted ensemble captures the flexibility observed in simulations, including a previously unknown G6/R142 interaction (c) which was first identified by MD simulations and subsequently confirmed by experiments.
[ Normal View | Magnified View ]

Browse by Topic

RNA Structure and Dynamics > RNA Structure, Dynamics, and Chemistry

Access to this WIREs title is by subscription only.

Recommend to Your
Librarian Now!

The latest WIREs articles in your inbox

Sign Up for Article Alerts