RESPAC: Method to Determine Partial Charges in Coarse-Grained Protein Model and Its Application to DNA-Binding Proteins.

While coarse-grained (CG) molecular simulations for large biomolecular complexes have become popular, their electrostatic treatment is often rather simplistic. Here, for Cα-based CG models of globular proteins, we developed a method to obtain an optimal partial charge set and applied it to 17 proteins that bind to DNA. The method follows the restrained electrostatic potential (RESP) fitting method widely used for determination of atomic partial charges in all-atom (AA) molecular mechanics. The proposed method, called the RESPAC method, finds optimal partial charges on surface Cα CG beads so that these charges best approximate the electrostatic potential of the AA model under a restraint term. Comparison of the AA and CG electrostatic potentials showed that the RESPAC charges outperformed simplistic integer-valued charges. Then, the RESPAC method was applied to lac repressor binding to a nonspecific DNA sequence. We found that the CG simulations correlated well with AA molecular dynamics simulations. We also performed CG simulations of 16 other transcription factors. The differences in binding interfaces between nonspecific and specific DNAs were, on average, reduced by using the RESPAC charges. Yet, for several proteins, the nonspecific DNA binding interface was quite different from that of the specific binding interface, which is in accord with a previous report.