Toward Reproducing Sequence Trends in Phosphorus Chemical Shifts for Nucleic Acids by MD/DFT Calculations

This work addresses the question of the ability of the molecular dynamics-density functional theory (MD/DFT) approach to reproduce sequence trend in (31)P chemical shifts (δP) in the backbone of nucleic acids. δP for [d(CGCGAATTCGCG)]2, a canonical B-DNA, have been computed using density functional theory calculations on model compounds with geometries cut out of snapshots of classical molecular dynamics (MD) simulations. The values of (31)P chemical shifts for two distinct B-DNA subfamilies BI and BII, δP/BI and δP/BII, have been determined as averages over the BI and BII subparts of the MD trajectory. This has been done for various samplings of MD trajectory and for two sizes of both the model and the solvent embedding. For all of the combinations of trajectory sampling, model size, and embedding size, sequence dependence of δP/BI in the order of 0.4-0.5 ppm has been obtained. Weighted averages for individual (31)P nuclei in the studied DNA double-helix have been calculated from δP/BI and δP/BII using BI and BII percentages from free MD simulations as well as from approaches employing NMR structural restraints. A good qualitative agreement is found between experimental sequence trends in δP and theoretical δP employing short (24 ns) MD run and BI, BII percentages determined by Hartmann et al. or via MD with the inclusion of NMR structural restraints. Theoretical δP exhibit a systematic offset of ca. 11 ppm and overestimation of trends by a factor of ca. 1.7. When scaled accordingly, theoretical δP/BI and δP/BII can be used to determine the expected percentage of BII to match the experimental value of δP. As evidenced by the calculations on snapshots from Car-Parrinello molecular dynamics, the systematic offsets of the theoretical δP obtained by MD/DFT approach result primarily from the unrealistic bond lengths employed by classical MD. The findings made in this work provide structure-δP relationships for possible use as NMR restraints and suggest that NMR calculations on MD snapshots can be in the future employed for the validation of newly developed force fields.