Evaluating the London Dispersion Coefficients of Protein Force Fields Using the Exchange-Hole Dipole Moment Model

London dispersion is one of the fundamental intermolecular interactions involved in protein folding and dynamics. The popular CHARMM36, Amber ff14sb, and OPLS-

AA force fields represent these interactions through the C6 /r 6 term of the Lennard-Jones potential. The C6 parameters are assigned empirically, so these parameters are

not necessarily a realistic representation of the true dispersion interactions. In this work, dispersion coefficients of all three force fields were compared to corresponding

values from quantum-chemical calculations using the exchange-hole dipole moment (XDM) model. The force field values were found to be roughly 50% larger than the XDM values for protein backbone and side-chain models. The CHARMM36 and Amber OL15 force fields for nucleic acids were also found to exhibit this trend. To explore how these elevated dispersion coefficients affect predicted properties, the hydration energies of the side-chain models were calculated using the staged REMD-TI method of Deng and Roux for the CHARMM36, Amber ff14sb, and OPLS-AA force fields. Despite having large C 6 dispersion coefficients, these force fields predict side-chain hydration energies that are in generally good agreement with the experimental values, including for hydrocarbon residues where the dispersion component is the dominant attractive solute–solvent interaction. This suggests that these force fields predict the correct total strength of dispersion interactions, despite C6 coefficients that are considerably larger than XDM predicts. An analytical expression for the water–methane dispersion energy using XDM dispersion coefficients shows that that higher-order dispersion terms(i.e., C 8 and C 10 ) account for roughly 37.5% of the hydration energy of methane. This suggests that the C 6 dispersion coefficients used in contemporary force fields are

elevated to account for the neglected higher-order terms. Force fields that include higher-order dispersion interactions could resolve this issue.