While the Watson-Crick (WC) base pairs (bps) are best-known and most abundant in nucleic acid structures (including RNA), the so-called reverse WC bp variants have received little attention. In the well-established Saenger scheme (see figure below), there are 28 possible bps for A, G, U(T), and C in their cononical (keto- and amino-) tautomeric forms and involving at least two H-bonds. The reverse A·T/U and G·C WC pairs are asymmetric, and are numbered XXI and XXII respectively (middle of right-hand side in the figure below).
In 3DNA, the WC bps are of type M–N and listed as A–T and G–C, consistent with the conventional notation. The reverse WC bps, on the other hand, are of type M+N and listed as A+T and G+C; the ‘+’ signifies the parallel z-axes of the two base reference frames, therefore their dot product is positive (see figure 2 in post Hoogsteen and reverse Hoogsteen base pairs).
As of this writing, a Google search of the phrase “reverse Watson Crick base pair” does not come up with anything informative — the top hit is the Jena Library page titled Nucleic Acid Nomenclature and Structure showing the same set of 28 possible bps only with explicit base chemical structures, as compiled by Tinoco Jr. et al. (1993).
However, once I look into this special type of bps, a quick search in PDB entry 1jj2, the Haloarcula marismortui large ribosomal subunit solved at 2.4 Å resolution, revealed nine reverse WC bps as shown below:
__U.U..0.205._ __A.A..0.437._ [U+A] __C.C..0.1186._ __G.G..0.1190._ [C+G] __C.C..0.1377._ __G.G..0.1683._ [C+G] __C.C..0.1856._ __G.G..0.1873._ [C+G] __A.A..0.2054._ __U.U..0.2648._ [A+U] __U.U..0.2109._ __A.A..0.2467._ [U+A] __A.A..0.2301._ __U.U..0.2306._ [A+U] __A.A..0.2321._ __U.U..0.2378._ [A+U] __C.C..0.2510._ __G.G..0.2564._ [C+G]
The following figure shows a representative reverse WC A+U bp (0.A437 with 0.U205, top), and a representative reverse WC G+C bp (0.G1683 with 0.C1377, bottom). For easy comparison, the two reverse WC bps are orientated in the reference frames of A and G, respectively.
In future releases of 3DNA, presumably starting from v2.2, we plan to provide a new component to classify bps according to the Saenger scheme, the Leontis/Westhof notation, and the geometric parameter-based strategy. Overall, the three bp classification methods are complementary in functionality, but with increased sophistication and applicability.