Adding the AMBER 14SB Force Field to the Stochastic Titration CpHMD Method

Incorporating pH into molecular dynamics simulations
is vital for
accurately capturing the fully coupled conformational, energetic,
and protonation landscape of many systems. The constant-pH molecular
dynamics (CpHMD) methodologies represent state-of-the-art approaches
to achieve this, with stochastic titration CpHMD (st-CpHMD) currently
being one of the most well-developed and validated methods. St-CpHMD
is already compatible with both the GROMOS 54A7 and CHARMM 36m force
fields, and we extend it here to support the AMBER 14SB force field
available in the GROMACS software package. We introduce and validate
a minor modification to the official atomic partial charges of ff14SB
(to achieve neutralization of the main chain) to render them compatible
with st-CpHMD, and we benchmark the final implementation using lysozyme
and Staphylococcal nuclease proteins. Although the root-mean-square
error (RMSE) values of the predictions for pKa versus experimental data align closely with those obtained
using the other supported force fields, we also identified several
challenging cases where the method requires further improvement. AMBER
14SB simulations showed a lower computational cost compared to CHARMM
36m, despite being slightly higher than the GROMOS 54A7 simulations.
Our findings also indicate that to further enhance computational speed,
future efforts should concentrate on accelerating the PB/MC step.
With this extension, we have developed the first CpHMD method implementation
compatible with the three most widely used protein force fields, enabling,
for the first time, a direct performance comparison among them.