Nonbonded Force Field Parameters Derived from Atoms-in-Molecules Methods Reproduce Interactions in Proteins from...

Area of Science:

• Computational Chemistry
• Biophysics
• Molecular Modeling

Background:

• Noncovalent interactions are fundamental to biological processes like protein-protein interactions and DNA folding.
• Accurate and computationally inexpensive force fields are needed to simulate large biomolecules and their functions.
• Existing force fields require improvement to precisely capture the quantum nature of atomic interactions.

Purpose of the Study:

• To derive nonbonded protein force field parameters from partitioned electron density using the atoms-in-molecules (AIM) approach.
• To validate these AIM-derived parameters against first-principles calculations for key amino acid side-chain interactions.
• To introduce a constrained AIM method for optimizing protein force fields.

Main Methods:

• Utilized the atoms-in-molecules (AIM) approach to partition electron density and derive force field parameters.
• Validated electrostatic and van der Waals interaction energies against absolute localized molecular orbital-energy decomposition analysis (ALMO-EDA).
• Employed the minimal basis iterative stockholder (MBIS) scheme for atomic charges and C6 coefficients.

Main Results:

• AIM-derived atomic charges accurately reproduced electrostatic interactions (4-7 kJ/mol MAE).
• MBIS-derived C6 coefficients effectively predicted dispersion interactions (mean error -2 kJ/mol).
• A constrained AIM method was developed to optimize side-chain interactions while maintaining backbone compatibility.

Conclusions:

• Validated AIM methods provide cost-effective force fields for depicting noncovalent interactions in proteins.
• These methods achieve chemical precision, enabling better prediction of protein conformational changes and function.
• The approach enhances the accuracy of molecular dynamics simulations for biological systems.