Refinement of DNA structures through near-edge X-ray absorption fine structure analysis: Applications on guanine and cytosine nucleobases, nucleosides, and nucleotides

In this work we highlight the potential of NEXAFS-near-edge X-ray absorption fine structure-analysis to perform refinements of hydrogen-bond structure in DNA. For this purpose we have carried out first-principle calculations of the N1s NEXAFS spectra of the guanine and cytosine nucleobases and their tautomers, nucleosides, and nucleotides in the gas phase, as well as for five crystal structures of guanine, cytosine, or guanosine. The spectra all clearly show imine (π1*) and amine (π2*) nitrogen absorption bands with a characteristic energy difference (Δ). Among all of the intramolecule covalent connections, the tautomerism of hydrogens makes the largest influence, around ±0.4-0.5 eV change of Δ, to the spectra due to a switch of single-double bonds. Deoxyribose and ribose sugars can cause at most 0.2 eV narrowing of Δ, while the phosphate groups have nearly negligible effects on the spectra. Two kinds of intermolecule interactions are analyzed, the hydrogen bonds and the stacking effect, by comparing "compressed" and "expanded" models or by comparing models including or excluding the nearest stacking molecules. The shortening of hydrogen-bond length by 0.2-0.3 Å can result in the reduction of Δ by 0.2-0.8 eV. This is because the hydrogen bonds make the electrons more delocalized, and the amine and imine nitrogens become less distinguishable. Moreover, the hydrogen bond has a different ability to influence the spectra of different crystals, with guanine crystals as the largest (change by 0.8 eV) and the guanosine crystal as the smallest (change by 0.2 eV). The stacking has negligible effects on the spectra in all studied systems. A comparison of guanosine to guanine crystals shows that the sugars in the crystal could create "blocks" in the π-and hydrogen bonds network of bases and thus makes the imine and amine nitrogens more distinguishable with a larger Δ. Our theoretical calculations offer a good match with experimental findings and explain earlier discrepancies in the NEXAFS analysis.