Understanding the role of base stacking in nucleic acids. MD and QM analysis of tandem GA base pairs in RNA duplexes

Abstract

Preceding NMR experiments show that the conformation of tandem GA base pairs, an important recurrent non-canonical building block in RNA duplexes, is context dependent. The GA base pairs adopt “sheared” N3(G)–N6(A), N2(G)–N7(A) geometry in the r(CGAG)₂ and r(iGGAiC)₂ contexts while switching to “imino” N1(G)–N1(A), O6(G)–N6(A) geometry in the r(GGAC)₂ and r(iCGAiG)₂ contexts (iC and iG stand for isocytosine and isoguanine, respectively). As base stacking is likely to be one of the key sources of the context dependence of the conformation of GA base pairs, we calculated base stacking energies in duplexes containing such base pairs, to see if this dependence can be predicted by stacking energy calculations. When investigating the context dependence of the GA geometry two different conformations of the same duplex were compared (imino vs. sheared). The geometries were generated via explicit solvent MD simulations of the respective RNA duplexes, while the subsequent QM energy calculations focused on base stacking interactions of the four internal base pairs. Geometrical relaxation of nucleobase atoms prior to the stacking energy computations has a non-negligible effect on the results. The stacking energies were derived at the DFT-D/6-311++G(3df,3pd) level. We show a rather good correspondence between the intrinsic gas-phase stacking energies and the NMR-determined GA geometries. The conformation with more favorable gas-phase stacking is in most cases the one observed in experiments. This correlation is not improved when including solvent effects via the COSMO method. On the other side, the stacking calculations do not predict the relative thermodynamic stability of duplex formation for different sequences.