Evaluation of the Performance of Non-Covalent Molecular Mechanical Models for Zn2+ and Mg2+ Metal Ions

Computer simulations using molecular dynamics (MD) on classical molecular mechanical (MM) interatomic potentials can provide valuable information and quantitative predictions about these systems. In MD calculations binding free energies of ions to host molecules can be studied if correct ion solvation free energies in aqueous solution are obtained. However, no MM models exist that parametrize various Zn2+ and transition metals consistently and consider both structural and thermodynamic data simultaneously. The first part of our work focused on MD free energy perturbation (FEP) simulations to derive MM interaction parameters for Zn2+ and Mg2+ ions in aqueous solution. To obtain these parameters the absolute solvation free energies were calibrated against the experimentally determined solvation free energies, using the TIP3P water model, which is integrated in the Q-package from Aqvist. In addition to the traditional single charge (SC) model a distributed charge model (DC) was developed and tested to study the impact of charged ghost atoms on the solvation free energy. Furthermore, the structural properties of the system were taken into consideration to obtain a parameter set that reproduces both the experimentally observed solvation free energies and the structure of the rst solvation shell. The results for Zn2+ in aqueous solution showed that solvation free energies and metal-oxygen radial distribution functions in aqueous solution are not coinciding with the experimentally observed data simultanously. The developed distributed charge (DC) model increased the solvation free energies substantially for various charge distributions compared to the traditional single charge (SC) model. However, the radial distribution functions were altered as well when performing molecular dynamics simulations with the DC model. As far as the simulation of the solvent is concerned, the continuum Born model overestimates the solvation free energy for realistic Zn2+ ionic radii whereas the explicit model underestimates solvation energies for realistic Zn2+-O distances. During the second part of this work a Metal Center Parameter Analyzer (MCPA) code was developed in python to automate gas-phase simulations for the ZnF2(H2O)3 complex for a given range of force eld parameter values. In particular, a random number generator has been integrated in our MCPA code. The results that were found with this MCPA code revealed that there are particular force field parameters sets which produce Zn-O and Zn-F mean distances that coincide with the distances obtained from ab initio calculations for some DC models. Our extensive numerical data will guide the design of the next-generation of the DC models of metal ions