Comparing pairwise-additive and many-body generalized Born models for acid/base calculations and protein design

Area of Science:

• Computational chemistry
• Biophysics
• Protein engineering

Background:

• Generalized Born (GB) solvent models are widely used in computational chemistry for tasks like acid/base calculations and protein design.
• Standard GB models often simplify the protein/solvent boundary to reduce computational cost, which can neglect important many-body interactions.

Purpose of the Study:

• To develop and evaluate a novel method for GB solvent models that retains the many-body character of the protein/solvent boundary.
• To improve the accuracy of acid/base calculations and protein design by treating the dielectric boundary more precisely.

Main Methods:

• Utilized Monte Carlo simulations allowing side chains to explore rotamers, protonation states, and mutations.
• Implemented a numerically exact treatment of the fluctuating protein/solvent dielectric boundary within the GB framework.
• Integrated the method into the Proteus protein design software.

Main Results:

• The new method demonstrated a slight but consistent improvement in predicting acid/base constants for nine different proteins.
• Significant enhancements were observed in the computational design of three PDZ domains, indicating improved accuracy for protein engineering tasks.
• The approach successfully captured the many-body character of the GB model while maintaining residue-pairwise complexity.

Conclusions:

• The developed method offers a more accurate representation of the protein/solvent interface in GB calculations.
• This advancement reduces model uncertainty, facilitating further investigations into other limitations of computational protein design and modeling.
• The findings suggest a promising direction for enhancing the predictive power of computational tools in biochemistry and molecular biology.