Atomic accuracy in predicting and designing noncanonical RNA structure

Abstract

We present fragment assembly of RNA with full-atom refinement (FARFAR), a Rosetta framework for predicting and designing noncanonical motifs that define RNA tertiary structure. In a test set of thirty-two 6-20-nucleotide motifs, FARFAR recapitulated 50% of the experimental structures at near-atomic accuracy. Sequence redesign calculations recovered native bases at 65% of residues engaged in noncanonical interactions, and we experimentally validated mutations predicted to stabilize a signal recognition particle domain.

Figure 1

Successes of de novo modeling of non-canonical RNA structure with Fragment Assembly of RNA with Full Atom Refinement (FARFAR). Two-dimensional annotations and three-dimensional representations are shown for (a) the E. coli signal recognition particle Domain IV RNA, (b) the bulged-G motif from the E. coli sarcin-ricin loop, (c) the E. coli loop E motif, (d) the kink-turn motif from the SAM-I riboswitch (T. tengcongensis), and (e) the hook-turn motif. (PDB codes are 1LNT, 1Q9A, 354D, 2GIS, and 1MHK respectively.) Each panel depicts the experimentally observed structure (left) and the best of five low-energy cluster centers (right). In (a), a conserved A-C interaction that was missed by automated annotation is shown in gray. (f) All-heavy-atom RMSD for the best of five final predictions (low-energy cluster centers) plotted against the number of residues in the modeled motif. Filled symbols denote atomic accuracy models (see text).

Figure 2

Computational and experimental tests validate sequence design and thermostabilization. (a) Sequence recovery over 15 high resolution side-chain-stripped RNA structures optimizing the Rosetta full-atom energy (black bars) was better than chance (25%, dashed line) and better than tests with the FARNA score function (gray bars). (b) Sequence preference predicted from 1000 redesigns (top) compared to an alignment of SRP Domain IV RNA sequences drawn from all three kingdoms of life , in sequence logo format . Two mutations (I and II) predicted by the Rosetta redesigns to stabilize folding are indicated. (c) Dimethyl sulfate (DMS) modification data probes the structure and thermodynamics of the SRP motif and variants. Sites of chemical modification were read out by reverse transcription of modified RNA with fluorescently labeled DNA primers, separated by multiplexed capillary electrophoresis. (d) Schematic of the construct's tertiary structure. Wedges mark residues that remained accessible to dimethyl sulfate in high Mg²⁺ folding conditions for the wild type RNA; the pattern for the mutant construct is indistinguishable except at the sites of mutation. (e) Folding isotherms by Mg²⁺ titration for four separate residues involved in the SRP motif's noncanonical structure (cf. symbols in c & d) overlay well and indicate that the Rosetta-predicted double mutant folds more stably than the wild type sequence. The left-most symbols represent conditions without Mg²⁺. Full electrophoretic profiles and single mutant fits are presented in Supplementary Fig. 6.