Magnesium Ion–Water Coordination and Exchange in Biomolecular Simulations

Abstract

Magnesium ions have an important role in the structure and folding mechanism of ribonucleic acid systems. To properly simulate these biophysical processes, the applied molecular models should reproduce, among other things, the kinetic properties of the ions in water solution. Here, we have studied the kinetics of the binding of magnesium ions with water molecules and nucleic acid systems using molecular dynamics simulation in detail. We have validated the parameters used in biomolecular force fields, such as AMBER and CHARMM, for Mg(2+) ions and also for the biologically relevant ions Na(+), K(+), and Ca(2+) together with three different water models (TIP3P, SPC/E, and TIP5P). The results show that Mg(2+) ions have a slower exchange rate than Na(+), K(+), and Ca(2+) in agreement with the experimental trend, but the simulated value underestimates the experimentally observed Mg(2+)-water exchange rate by several orders of magnitude, irrespective of the force field and water model. A new set of parameters for Mg(2+) was developed to reproduce the experimental kinetic data. This set also leads to better reproduction of structural data than existing models. We have applied the new parameter set to Mg(2+) binding with a monophosphate model system and with the purine riboswitch, add A-riboswitch. In line with the Mg(2+)-water results, the newly developed parameters show a better description of the structure and kinetics of the Mg(2+)-phosphate binding than all other models. The characterization of the ion binding to the riboswitch system shows that the new parameter set does not affect the global structure of the ribonucleic acid system or the number of ions involved in direct or indirect binding. A slight decrease in the number of water-bridged contacts between A-riboswitch and the Mg(2+) ion is observed. The results support the ability of the newly developed parameters to improve the kinetic description of the Mg(2+) and phosphate ions and their applicability in nucleic acid simulation.