Improved CHARMM Additive Force Field Parameters to Accurately Model Tyrosine-Choline Cation-π Interactions

Abstract

Cation-π interactions between methylated ammonium groups and tyrosine amino acids have been shown to be important for epigenetic recognition motifs, choline-binding proteins, and for protein-lipid interactions. Accurately modeling cation-π interactions in biomolecular simulations remains a challenge due to the lack of explicit polarization or charge transfer effects. In this work, we investigate the nature of tyrosine-choline cation-π interactions by performing high-level Quantum Mechanical (QM) calculations and building Potential Energy Surfaces (PES). We benchmark QM levels of theory and find that SAPT2+/aug-cc-pVDZ level of theory performs well compared to large basis set CCSD(T). Further, we compared QM PES (using SAPT2+/aug-cc-pVDZ) to both additive CHARMM36 and Drude polarizable force field. With CHARMM36, the equilibrium distances are well captured while the interaction energies are underestimated for various approach angles of TMA with respect to phenol. While using the Drude polarizable force field, the interaction energies deviate less compared to target QM data. However, the obtained equilibrium geometries are slightly underestimated. The best agreement between force field and QM PES is obtained by modifying the Lennard-Jones potentials for selected atom pairs involved in phenol-TMA cation-π interactions. We performed MD simulations of a bilayer-bound bacterial phospholipase and calculate the occupancies of tyrosine-choline cation-π interactions. The cation-π occupancies obtained with the modified set correlate better with experimental data than those obtained with the CHARMM force field.