Fragment molecular orbital (FMO) calculations on DNA by a scaled third-order Møller-Plesset perturbation (MP2.5) scheme

Abstract

A series of fragment molecular orbital (FMO) calculations on DNA models was performed to assess the reliability of the scaled third-order Møller-Plesset perturbation method (MP2.5) (Pitoňák et al., Chem. Phys. Chem., 10 (2009) 282) for the incorporation of higher-order electron correlation. The Watson-Crick H-bonding (horizontal interaction) and the 1,2-stacking (vertical interaction) of four combinations of two base pairs (AT:AT, AT:TA, AT:GC, and CG:GC) were calculated at the levels of MP2, MP3 (nonscaled), MP2.5, and coupled cluster singles and doubles with perturbative triples (CCSD(T)). The comparison showed sufficient agreement between MP2.5 and the reference CCSD(T), and indicated that MP2.5 is an inexpensive but precise alternative to costly CCSD(T). The influence of basis sets including the model core potential (MCP) was also examined for the two base pairs. The Watson-Crick H-bonding interaction energies were fairly uninfluenced by these methods. In contrast, the values of the stacking interactions among bases quantitatively varied depending on the calculation options. As a cost-effective procedure, the FMO-MP2.5/MCP calculations were applied to a couple of realistic DNA models explicitly solvated by water: a 12-base-pair model and a 10-base-pair model cross-linked with cis-platin (Pt(NH3)2). The H-bonding and stacking interactions in them were discussed from a biochemical viewpoint.