Conformational Biases of Simple Helical Junctions Influence RNA Folding Stability and Specificity?”

In the hierarchical model of RNA folding, the flexibility of the junction regions is the dominant influence on the conformational ensemble of the helices, which are treated as as collection of rigid bodies. To account for this in our modeling, we created multiscale representations of the dPEG and sPEG constructs by coarse-graining the helices while retaining atomistic detail in the junction regions (Fig. 1). In our multiscale representations, each helix is replaced by a collection of “dummy” atoms and fiducial atoms that allow us to replace the coarse-grained helices with their atomistic representations. Each dummy atom (blue atoms aligned along the helical axis in Fig. 1) was assigned a Van der Waals radius of 4 Å. Parameters for the bond lengths, angles, and dihedrals for the PEG linker were obtained from experimental studies [1, 2]. We used the Gromacs software package to conduct long stochastic dynamics simulations of the sPEG and dPEG constructs [3]. The great reduction in the number of atoms in the simulated construct resulted in a substantial decrease in the computational cost of the simulation. For each simulation, the friction constant for the SD simulation was set at 0.01 ps. Bonds and angles where held constant using the SHAKE algorithm. We first began by equilibrating both constructs at 300 K for 1.6μs. After this initial step, another simulation of 1.6μs was conducted and ten evenlyspaced snapshots from this simulation were used to seed ten other simulations. From these ten simulations, snapshots of the constructs were taken every 16 ps of simulation time. The total number of recorded frames represents ∼45μs of simulation for each construct. Convergence of the simulation was checked using block averaging (data not shown).