Structural and Energetic Features of Base–Base Stacking Contacts in RNA

Abstract

Nucleobase π−π stacking is one of the crucial organizing interactions within three-dimensional (3D) RNA architectures. Characterizing the structural variability of these contacts in RNA crystal structures will help delineate their subtleties and their role in determining function. This analysis of different stacking geometries found in RNA X-ray crystal structures is the largest such survey to date; coupled with quantum-mechanical calculations on typical representatives of each possible stacking arrangement, we determined the distribution of stacking interaction energies. A total of 1,735,481 stacking contacts, spanning 359 of the 384 theoretically possible distinct stacking geometries, were identified. Our analysis reveals preferential occurrences of specific consecutive stacking arrangements in certain regions of RNA architectures. Quantum chemical calculations suggest that 88 of the 359 contacts possess intrinsically stable stacking geometries, whereas the remaining stacks require the RNA backbone or surrounding macromolecular environment to force their formation and maintain their stability. Our systematic analysis of π−π stacks in RNA highlights trends in the occurrence and localization of these noncovalent interactions and may help better understand the structural intricacies of functional RNA-based molecular architectures.

Structural Mapping of the Base Stacks Containing Post-transcriptionally Modified Bases in RNA

“…E AB is the energy of the base stack, and E A and E B are the single point energies of isolated bases A and B, respectively. Interaction energy calculations were carried out at the ωB97X-D/6-311+G­(2df,p) level of theory, in accordance with previous studies. ,,− Although a range of QM methods have been applied in the literature to study base stacking, , our comparisons on a selected set of stacked structures reveals that ωB97X-D/6-311+G­(2df,p) interaction energies are within 0.4 kcal mol –1 of the values calculated at the (higher) B3LYP-D3­(BJ)/def2-TZVP level of theory (Table S5). More importantly, the trend in interaction energies is not affected by the change in the DFT methodology, which points to the robustness of the QM data even at a lower, less computationally expensive, level of theory.…”

“…A third tie breaker was not necessary. For the QM calculations, the ribose was replaced by a hydrogen cap in accordance with previous studies. ,− Although the sugar-phosphate backbone plays an important role in stabilizing nucleic-acid structures, , we did not include the backbone in our calculations, mainly since the consecutive stacks involve the covalent phosphate linkage between interacting nucleotides (monomers). As a result, the calculated interaction energy will include contributions from both stacking interaction energy and the covalent bond energy.…”

“…Trial calculations involving replacement of the covalent linkage with capping groups to remove the effect of covalent bonding revealed significant steric clashes, which hinder the unambiguous estimation of stacking interaction energy. Further, we chose our models to remain consistent with our previous study on unmodified base stacks, to allow meaningful comparisons to understand the effect of base modifications on stacking interaction energies. Similar to earlier works, ,,− the geometry was optimized using density functional theory, employing the long-range corrected hybrid functional (ωB97X-D) with the 6-31G­(d,p) basis set.…”

“…Further, we chose our models to remain consistent with our previous study on unmodified base stacks, to allow meaningful comparisons to understand the effect of base modifications on stacking interaction energies. Similar to earlier works, ,,− the geometry was optimized using density functional theory, employing the long-range corrected hybrid functional (ωB97X-D) with the 6-31G­(d,p) basis set. The chosen level of theory is effective for dealing with weak non-covalent interactions. , These optimized monomer geometries were then used to replace the monomer coordinates present in each crystallographically identified representative stack, after superposition.…”

“…The chosen level of theory is effective for dealing with weak non-covalent interactions. , These optimized monomer geometries were then used to replace the monomer coordinates present in each crystallographically identified representative stack, after superposition. This procedure was adapted from prior studies on stacking between nucleobases and aromatic amino acids in RNA/protein complexes and between canonical bases in RNA structures . These alignments were carried out using the Open Babel software …”