Biophysical Society Thematic Meeting | Hamburg 2022

Biophysics at the Dawn of Exascale Computers

Poster Abstracts

3-POS Board 3 IMPROVED MONOVALENT CATION-PROTEIN INTERACTIONS IN THE CHARMM DRUDE POLARIZABLE FORCE-FIELD Kazi S Amin 1 ; Justin A Lemkul 3 ; Alexander D MacKerell Jr. 2 ; Dennis Salahub 4 ; Sergei Noskov 1 ; 1 University of Calgary, Biological Sciences, Calgary, AB, Canada 2 University of Maryland, School of Pharmacy, Baltimore, MD, USA 3 Virginia Tech, Department of Biochemistry, Blacksburg, VA, USA 4 University of Calgary, Department of Chemistry, Calgary, AB, Canada Explicit atomic polarizability is crucial to the accurate description of interactions between ions and biomolecules. The CHARMM Drude polarizable force-field models atomic polarizability by including charged particles with a small mass linked to non-hydrogen atoms through a harmonic bond. This complicates the parametrization process, and while partial charges and polarizabilities have been mostly parametrized, Lennard Jones (LJ) parameters for monoatomic ions still need to be revised. Here we present a new set of pair-specific LJ Rmin and Thole Coulomb screening parameters for each of Li + , Na + , and K + ions paired with several important biomolecular oxygen atom types, which include carbonyl, carboxylate, and phosphate oxygens. To obtain target data for parameter optimization, we computed single point quantum mechanical (QM) interaction energies of representative ion-molecule complexes, varying the ion position with respect to the molecule. We then used a novel scoring criterion to fit the Drude interaction energies to the target data. The scoring criterion prioritizes fitting the local shape of the minima of both QM and Drude energy surfaces, rather than the absolute energies, as is usually done. The parameter optimization was performed using a global search over a range of LJ Rmin and Thole screening factors. The parameters obtained for ion-carboxylate interactions agree very well with osmotic pressures in condensed phase MD simulations. Thus, we were able to select parameters which fitted both gas-phase QM data and condensed phase osmotic pressure data reasonably well. However, for other oxygen atom types belonging to uncharged functional groups, the same criterion was not as effective and we had to compare the energy surfaces more holistically to obtain reasonable parameters. This implies that careful criteria selection during force field parametrization is crucial, especially when exact reproduction of QM energy is impossible and when considering transferability of gas-phase derived parameters to condensed phase systems.

88