De novo 3D models of SARS-CoV-2 RNA elements and small-molecule-binding RNAs to guide drug discovery

Ramya Rangan; Andrew M. Watkins; Wipapat Kladwang; Rhiju Das

We generated ensembles for SARS-CoV-2 RNA elements using Rosetta's FARFAR2 protocol, a collection of models we term the FARFAR2-SARS-CoV-2 dataset. Beginning with a sequence and secondary structure, FARFAR2 generates models through Monte Carlo substitutions of 3-residue fragments sampled from previously solved RNA structures, followed by refinement in a highresolution physics-based free energy function, which models hydrogen bonding, solvation effects, nucleobase stacking, torsional preferences and other physical forces known to impact macromolecules structure. 10 FARFAR2 3D modeling requires an input secondary structure. Here, most secondary structures for SARS-CoV-2 RNA elements were obtained from homologies to secondary structures previously probed in related betacoronaviruses, with the exception of stem loops 6-7 in the extended 5′ UTR, 4 which were generated from computational secondary structure modeling. We emphasize that these 3D model collections reflect a single secondary structure for each RNA element (or two secondary structures for FSE; see below). Future experimental work that elucidates additional secondary structures or tertiary contacts may prompt further 3D modeling of alternative states, but such wet lab experiments are currently not possible in our laboratory due to pandemic-related shutdowns.