Problems of robustness in Poisson-Boltzmann binding free energies

Area of Science:

• Computational chemistry
• Biophysics
• Molecular modeling

Background:

• Poisson-Boltzmann (PB) models are widely used for solvation free energy (ΔG) calculations.
• These models have shown limitations in accurately predicting protein-protein binding free energies (ΔΔG).

Purpose of the Study:

• To investigate the reliability of electrostatic components (ΔΔGel) of binding free energies predicted by PB models.
• To understand why PB models struggle with ΔΔG predictions despite success with ΔG.

Main Methods:

• Analysis of electrostatic contributions (ΔGel and ΔΔGel) to free energies for protein-protein complexes.
• Comparison of ranking abilities using ΔGel and ΔΔGel across different force fields and surface definitions.

Main Results:

• Ranking protein complexes by electrostatic binding energy (ΔΔGel) was significantly more challenging than by electrostatic solvation energy (ΔGel).
• ΔΔGel values were orders of magnitude smaller than ΔGel, leading to poor correlation in estimates.
• Despite high correlation for ΔGel estimates, ΔΔGel estimates showed a lack of correlation across different computational setups.

Conclusions:

• The small magnitude of electrostatic binding energy components limits the predictive power of current PB models for protein-protein interactions.
• Current PB models require refinement to accurately capture the subtle electrostatic contributions crucial for binding free energy predictions.