Solvated interaction energy (SIE) for scoring protein-ligand binding affinities. 1. Exploring the parameter space

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Accurate prediction of protein-ligand binding free energies is crucial for drug discovery.
• Existing models often struggle to consistently account for solvation effects in binding calculations.
• The dielectric constant dependence of hydration differs significantly from binding free energies, necessitating specialized calibration.

Purpose of the Study:

• To develop and calibrate a novel binding free energy function incorporating force field and solvation terms.
• To systematically optimize model parameters for accurate prediction of absolute binding affinities.
• To assess the necessity of including bound water in calculations and evaluate the model's virtual screening capabilities.

Main Methods:

• Developed a binding free energy function combining force field and solvation terms.
• Self-consistently calibrated solvation and binding interaction parameters using 99 protein-ligand complexes.
• Systematically scanned five model parameters, including solute dielectric constant (Din) and van der Waals scaling factors.
• Investigated the impact of protein-ligand data preparation methods, including the presence of bound water.

Main Results:

• Achieved a mean unsigned error of 1.29 kcal/mol for predicted absolute binding affinities.
• Optimal performance was observed with high solute dielectric values (Din ≥ 20) and van der Waals scaling factors between 0.03 and 0.15.
• Retaining bound water in protein structures was found to be unnecessary; continuum solvation models suffice.
• Demonstrated good virtual screening enrichment for estrogen receptor and thymidine kinase targets.

Conclusions:

• The developed binding free energy function provides accurate predictions of protein-ligand binding affinities.
• The model's parameters suggest enthalpy-entropy compensation plays a role in ligand binding.
• Continuum solvation is sufficient for accurate binding free energy calculations, simplifying the process.
• The function shows potential for effective virtual screening in drug discovery pipelines.