Polarizable AMOEBA Model for Simulating Mg²⁺·Protein·Nucleotide Complexes.

Area of Science:

• Computational chemistry
• Biophysics
• Structural biology

Background:

• Molecular mechanics (MM) simulations offer detailed insights into enzyme mechanisms involving nucleotide cofactors and divalent cations.
• Modeling divalent cations, particularly Mg2+, in MM simulations remains a significant challenge, with existing polarizable force fields showing large errors in interaction energies and failing to reproduce experimental structures of Mg2+·Protein·ATP complexes.

Purpose of the Study:

• To systematically assess and critically revise the polarizable AMOEBA model for improved simulation of Mg2+·Protein·ATP complexes.
• To enhance the predictive performance of MM simulations for enzymes utilizing Mg2+ and nucleotide cofactors.

Main Methods:

• Revision of the AMOEBA protein model with high field corrections (AMOEBABIO18-HFC) to improve Mg2+-protein interactions.
• Inclusion of many-body Nonbonded-fix (NB-fix) corrections to further reduce interaction energy errors.
• Development of a new AMOEBA model for Adenosine Triphosphate (ATP) with revised polarization, van der Waals (vdW), and dihedral parameters.
• Benchmarking interaction energies against vdW-inclusive density functional theory (DFT) and coupled cluster (CCSD(T)) calculations.
• Performing molecular dynamics (MD) simulations of Mg2+·Kinase·ATP complexes using the improved models.

Main Results:

• The revised AMOEBABIO18-HFC model significantly reduced Mg2+-protein interaction energy errors (MAE from 17 to 10 kcal/mol).
• Incorporating many-body NB-fix corrections further decreased MAE to 6 kcal/mol (<2% error).
• The new ATP model accurately predicted experimental Mg2+-ATP binding free energy and provided insights into Mg2+ association.
• MD simulations with the improved models showed better agreement between simulated and experimentally determined structures (X-ray crystallography) of Mg2+·Kinase·ATP complexes.

Conclusions:

• The refined AMOEBA model, incorporating AMOEBABIO18-HFC and many-body NB-fix corrections, substantially improves the accuracy of simulating Mg2+ interactions in biological systems.
• The new ATP model enhances the simulation of Mg2+-ATP binding.
• These advancements enable more reliable MD simulations of Mg2+-dependent enzymes, leading to better structural predictions that align with experimental data.