LW base-pair classifications derived using DSSR

I recently come across the article FMN riboswitch aptamer symmetry facilitates conformational switching through mutually exclusive coaxial stacking configurations by Wilt et al. in the Journal of Structural Biology: X (JSBX). In the caption to Figure S1, “Secondary structure map of the FMN riboswitch”, the authors wrote:

Base-pairing is annotated using Leontis-Westhoff nomenclature (Leontis and Westhof, 2001), derived using 3DNA-DSSR (Lu and Olson, 2003), and the map was generated using VARNA (Darty et al., 2009).

It is a nice surprise to see that 3DNA-DSSR is cited this way. The LW scheme is based on the three edges of each base with potential for H-bonding interactions (Watson-Crick, Hoogsteen, and Sugar), and the two orientations (cis or trans) of the interacting bases with respect to the glycosidic bonds. The combinations of edges and orientations (3 × 2 × 2) “gives rise to 12 basic geometric types with at least two H bonds connecting the bases” (Leontis and Westhof, 2001). This geometry-based approach captures salient features of pairing interactions and strikes a balance between simplicity and expressiveness. The LW scheme is more widely applicable than the Saenger classification, and more intuitive to biologists. As a result, the LW classification has become a standard in RNA structural bioinformatics.

However, the RNA-centric LW classification has inherent limitations. For example, the Sugar edge explicitly includes the 2′-hydroxyl group, rendering it less applicable to DNA structures. Additionally, while the aromatic base can be taken as a rigid body with three fixed edges, the χ (chi) torsion angle characterizes the internal freedom between base and sugar (anti vs. syn). When χ is in the relatively rare (but not uncommon) syn conformation (especially abundant in G-quadruplexes), the Sugar edge, defined with reference to the common anti conformation, seems to no longer exist. The rich variety of RNA pairs extends beyond the 12 basic LW types. There are numerous pairs in RNA with only one H-bond or with bifurcated H-bonds, at boundary locations where the LW classification does not strictly apply. Lemieux and Major (2002) were the first to extend the LW classification. We noted the importance of the out-of-plane ‘backbone edge’ formed by an RNA-specific H-bond between O2′(G) and OP2 (Lu et al., 2010). Finally, the RNA 3D Hub website, hosted by the Leontis-Zirbel team, lists pairing interactions that do not fall into the 12 geometric types. For example, the page for 1msy contains pairing types ncSW, ntSH, and ntHH. Note that the terms nc (in ncSW) and nt (in ntSH/ntHH) do not have the normal meanings in literature; they stands for near cis and near trans respectively.

Summary of steps used to identify nucleic acid structural components using DSSR

As shown in the figure above, DSSR adopts a base-centric terminology for the three edges. In principle, M (Major groove) in the DSSR classification corresponds to the Hoogsteen/CH-edge (H) in the LW notation, and the DSSR m (minor groove) to the LW Sugar-edge (S) if χ is in the anti conformation. In practice, direct DSSR/LW correspondences M/H and m/S are assumed, regardless of anti/syn base conformation. Moreover, the cis/trans assignment is the same for both notations. Within the DSSR implementation, the LW and DSSR classifications are thus strictly parallel in terms of cis/trans orientation and interacting edges. The DSSR scheme has the extra ± for relative base orientations.

The LW classifications implemented in DSSR may differ from those listed in the RNA 3D Hub website or other resources. These discrepancies normally occur in boundary cases where the assignment of cis/trans and interaction edges can be ambiguous. For ‘authentic’ LW classification results, users should consult the original publication of Leontis and Westhof (2001) and use the RNAView (Yang et al., 2003) or FR3D (Sarver et al., 2008) tools instead of DSSR.



Commenting is closed for this article.



Thank you for printing this article from http://home.x3dna.org/. Please do not forget to visit back for more 3DNA-related information. — Xiang-Jun Lu