Two physics-based models for pH-dependent calculations of protein solubility.

Area of Science:

• Computational biology
• Protein engineering
• Biophysics

Background:

• Protein solubility is a critical physicochemical property optimized during protein engineering.
• Accurate prediction of solubility is essential for successful protein design and formulation.

Purpose of the Study:

• To develop and validate novel computational methods for calculating pH-dependent protein solubility.
• To enable rapid ranking of protein mutant solubility and guide protein design strategies.

Main Methods:

• An empirical method incorporating electrostatic solvation energy (Generalized Born approximation), hydrophobic patches, protein charge, and stability changes.
• A force-field-based approach using CHARMm to calculate protein binding energy to crystal lattice components.

Main Results:

• The empirical method achieved over 80% prediction rate for mutations in globular proteins and antibodies, with high correlation (R=.83-.91) to experimental data.
• The force-field method accurately predicted pH-dependent solubility for Ribonuclease SA and its mutants without parameter adjustment.
• Both methods demonstrated utility in screening design candidates and optimizing formulation conditions.

Conclusions:

• The developed computational methods offer efficient and accurate tools for predicting protein solubility.
• These methods can significantly accelerate protein engineering workflows and improve the development of protein-based therapeutics and biologics.