Distance Dependent Hydrogen Bond Potentials for Nucleic Acid Base Pairs from ab Initio Quantum Mechanical Calculations (LMP2/cc-pVTZ)

Abstract

Hydrogen bonding between base pairs in nucleic acids is a key determinant of their structures. We have examined the distance dependence of the hydrogen bonding of AT-WC (Watson-Crick), GC-WC, and AT-H (Hoogsteen) base pairs using ab initio quantum mechanics, LMP2/cc-pVTZ(-f) energies at HF/cc-pVTZ(-f) optimized geometries. From these curves, we have extracted Morse potentials between the H atoms and the acceptor atoms that accurately reproduce the quantum mechanical energies for a range of geometries. Using these parameters, we have calculated the complexation energies of the remaining 26 possible pairwise combinations, and the agreement with previously reported ab initio calculations is excellent. We have also extracted off-diagonal Lennard-Jones 12-6 parameters to be used with the popular AMBER95 and CHARMM95 force fields that significantly improve their descriptions of the base-pairing energy and optimum geometry. Hydrogen bonds play a key role in maintaining structure and specificity of biological systems. In particular, the base pairing of nucleic acids (stemming from the specific formation of hydrogen bonds between Watson-Crick base pairs) is essential for the transfer of genetic information. We examine here the three of greatest biological relevance; the Watson-Crick base pairs, in which adenine hydrogen bonds to thymine (AT-WC) or guanine hydrogen bonds to cytosine (GC-WC), and the Hoogsteen adenine-thymine pair (AT-H). Given the importance of hydrogen bonding in biological systems, considerable theoretical attention has been focused on exploring the nature and strength of these interactions. Extensive calculations have been carried out on the nucleic acid base pairs using semiempirical or ab initio quantum mechanical (QM) methods. For a recent review covering the current state of the field and discussing the extensive progress from semiempirical to ab initio QM methods, see ref 1. It is now established that the determination of accurate hydrogen bond energies and geometries requires a large diffuse basis set and the inclusion of electron correlation. 2,3 Many of the earlier studies of the nucleic acid base pairs did not include electron correlation effects, or estimated correlation energies with empirical methods. Recently, a series of studies have been reported in which dispersion energies have been evaluated using second-order Moller-Plesset perturbation theory (MP2) 4-8 or density functional theory (DFT). 9