Scientists hope to “sing along” with RNA molecules. Keeping in time with the internal dynamics of RNA molecules lets scientists appreciate structure-function relationships, which often illuminate the biological roles RNA fulfills. Scientists just have to do the computational equivalent of following the bouncing ball, an old-time cinema technique that visually indicates the rhythm of a song. (With this technique, the bouncing ball lands on the syllables of a lyric as they are to be sung.)
RNA molecules, however, pose a complication. They usually require scientists to keep track of multiple balls—even as many balls as there are atoms in a molecule, if the greatest possible fidelity is desired. The maximum-fidelity approach, called molecular dynamics, is costly and computationally intensive, taking days or weeks to complete a calculation.
An alternative approach, one in which fewer balls are followed, is called elastic network modeling (ENM). Instead of painstakingly simulating all the atom-atom interactions, ENM treats groups of atoms like beads, and it treats the connections between groups of beads like springs. By using a larger or a smaller number of beads to represent a particular molecule, ENM can provide a relatively coarse- or fine-grained simulation.
ENM has proven to be valuable and effective tool in the study of proteins, but it has been little used in the study of RNA molecules. Researchers are still trying to determine the optimal balance of computational simplicity and dynamical fidelity. That is, they are working out just how many beads should be followed.
A recent study from the International School for Advanced Studies (SISSA) in Trieste proposes that an RNA model can be fairly coarse and still faithfully represent an RNA molecule’s internal dynamics, showing how the different parts of the molecule, which are rigid or elastic, which are connected or independent.
In their work with simulated RNA molecules, researchers have, for the most part, tried using different bead-per-nucleotide ratios. For example, some researchers have tried using a single bead per nucleotide. Sometimes the bead is centered on a phosphorus atom, sometimes at the center of the ribose sugar in the backbone.
The SISSA study, led by Givanni Bussi and Cristian Micheletti, decided that a three-bead-per-nucleotide ratio worked best. The SISSA team presented its results July 17 in Nucleic Acids Research, in an article entitled, “Elastic network models for RNA: a comparative assessment with molecular dynamics and SHAPE experiments.”
“The fluctuations predicted by the alternative ENMs are stringently validated by comparison against extensive molecular dynamics simulations and SHAPE experiments,” wrote the authors. “We find that simulations and experimental data are systematically best reproduced by either an all-atom or a three-beads-per-nucleotide representation (sugar-base-phosphate), with the latter arguably providing the best balance of accuracy and computational complexity.”
By SHAPE, the authors meant selective 2′-hydroxyl acylation analyzed by primer extension experiments. SHAPE reactivity, the authors pointed out, is empirically known to correlate with base dynamics and sugar pucker flexibility at the nucleotide level and hence is, in principle, well suited for validating predictions of RNA internal dynamics.
According to the current study, three-bead-per-nucleotide RNA representations “provide a satisfactory proxy for the nucleotide-level flexibility as captured by experimental SHAPE data.”
And so the ENMs explored by the SISSA researchers not only correlated with computationally intensive molecular dynamics simulations, but chemical probing experimental data. “In spite of its simplicity, our model is able to predict the structural fluctuations of RNA molecules with the accuracy of more complicated models,” asserted Dr. Bussi.
Essentially, the SISSA model surpasses sing-song simplicity, achieving the structure-function equivalent of choir-like complexity, without producing the sort of cacophony only the most powerful computers can appreciate.