Development of the ABEEMσπ Polarization Force Field for Base Pairs with Amino Acid Residue Complexes

DNA damage caused by oxidized bases can lead to aging and cancer in living beings. Luckily, a repair enzyme is able to repair the oxidized bases. The key step is to accurately recognize the oxidized bases, which mainly rely on complex hydrogen bond interactions. We have calibrated the charge parameters and torsional parameters of the ABEEMσπ polarization force field (ABEEMσπ PFF) to accurately describe the intermolecular and intramolecular interactions. Taking the experiment and quantum chemical method as the benchmark, a series of properties of base pair-amino acid residue systems, DNA and DNA-protein interaction systems were calculated and compared with those of other force fields. We have done a tremendous amount of tasks in testing, calibrations, and analyses. The ABEEMσπ PFF not only explicitly gives the position and the partial charge of lone-pair electrons but also introduces a function k_HB to fit special electrostatic interactions in hydrogen bond interaction regions. Therefore, it can accurately simulate the polarization effect and charge transfer of hydrogen bond interactions, especially for charged systems and sulfur-containing systems, such as the binding energy between amino acid and base pairs (24-28 kcal/mol), which is induced by charge transfer. The RMSD of ABEEMσπ PFF is 1.18 kcal/mol, whereas the RMSD of Amber OL15 is 8.21 kcal/mol. The relative positions of the amino acid residue have significantly changed, and the hydrogen bonds were broken when simulated by fixed charge force fields. In addition, owing to refitting the reasonable torsional parameters, the geometric structures optimized by ABEEMσπ PFF were well consistent with those of the M06-2X/6-311++G** method, but the simulations by fixed force fields have a large rotation of methyl and distortion of the plane of the base pair. After extensive MD simulation with four test DNAs and a DNA-protein system, we conclude that ABEEMσπ PFF shows better agreement when compared to experimental structures, which illustrates the reliability of our model and the transferability of the parameters.