RNA backbone is rotameric

Abstract

Despite the importance of local structural detail to a mechanistic understanding of RNA catalysis and binding functions, RNA backbone conformation has been quite recalcitrant to analysis. There are too many variable torsion angles per residue, and their raw empirical distributions are poorly clustered. This study applies quality-filtering techniques (using resolution, crystallographic B factor, and all-atom steric clashes) to the backbone torsion angle distributions from an 8,636-residue RNA database. With noise levels greatly reduced, clear signal appears for the underlying angle preferences. Half-residue torsion angle distributions for alpha-beta-gamma and for delta-epsilon-zeta are plotted and contoured in 3D; each shows about a dozen distinct peaks, which can then be combined in pairs to define complete RNA backbone conformers. Traditional nucleic acid residues are defined from phosphate to phosphate, but here we use a base-to-base (or sugar-to-sugar) division into "suites" to parse the RNA backbone repeats, both because most backbone steric clashes are within suites and because the relationship of successive bases is both reliably determined and conformationally important. A suite conformer has seven variables, with sugar pucker specified at both ends. Potential suite conformers were omitted if not represented by at least a small cluster of convincing data points after application of quality filters. The final result is a small library of 42 RNA backbone conformers, which should provide valid conformations for nearly all RNA backbone encountered in experimental structures.

Fig. 1.

RNA backbone, with the six torsion angles α, β, γ, δ, ε, ζ labeled on the central bond of the four atoms defining each dihedral. The two alternative ways of parsing out a repeat are indicated: a traditional nucleotide residue goes from phosphate to phosphate (changing residue number between O5′ and P), whereas an RNA suite goes from sugar to sugar (or base to base).

Fig. 2.

A section with tertiary helix interactions inside the rr0033 23S rRNA at 2.4-Å resolution (chain 0 11–23, 520–57, 600–20), to show the contrasting accuracy with which bases vs. backbone can be determined for large RNAs. In a, the green and blue all-atom-contact dots show the nearly perfect van der Waals and H bond contacts of a well fit base stack, whereas in b, the red spikes mark impossible steric clashes of nonpolar atoms for residues with locally misfit backbone.

Fig. 3.

2D plots of the unfiltered data to 3-Å resolution for: α vs. β (a) and ε vs. ζ (b) torsion angles, to illustrate the severe signal-to-noise problem in the raw data. The large cluster in each plot is A form. The origin is at a corner rather than at the center to avoid splitting the data peaks, but angle values are labeled on the axes.

Fig. 4.

Plots of the heminucleotide angle triplets for data filtered by clashes and B > 60, with smoothed contours enclosing the top 7–10 peak clusters: (a) α–β–γ plot for adjacent sugar pucker C3′ endo. (b) α–β–γ plot for adjacent C2′ endo pucker. (c) δ–ε–ζ plot (filtered also for resolution >2.5 Å).

Fig. 5.

Identification and repair of an incorrect backbone conformation for suite 870–871 of the rr0033 23S rRNA. a shows the red spikes of bad clashes for the 5′ H atoms, in the original model with disallowed αβγδ values ptm3′; b shows good contacts for the refit model (magenta) in standard A form (mtp3′).