Computational tools help improve protein stability but with a solubility tradeoff.

Area of Science:

• Protein Engineering
• Computational Biology
• Biotechnology

Background:

• Accurate prediction of protein stability changes from amino acid substitutions is crucial for protein engineering.
• Stabilizing mutations are valuable for industrial and therapeutic applications, but random substitutions have a low success rate.
• Existing computational tools for predicting mutation effects on stability face challenges in accuracy and consistency.

Purpose of the Study:

• To develop a more accurate computational tool for predicting protein stability changes upon mutation.
• To experimentally validate the predictions of the developed tool and assess its utility in protein design.
• To investigate the relationship between protein stabilization strategies and protein solubility.

Main Methods:

• Combined 11 freely available computational tools into a meta-predictor.
• Validated the meta-predictor against approximately 600 experimental mutations.
• Used the meta-predictor to recommend mutations for a designed protein (ThreeFoil) and experimentally characterized the resulting mutants.

Main Results:

• The meta-predictor demonstrated improved performance compared to individual tools.
• Four recommended mutations increased the thermodynamic stability of ThreeFoil by >2 kcal/mol.
• While stability increased, most mutations decreased protein solubility, a common failure point in protein design.
• Surface mutations that increase stability often increase hydrophobicity, a trend favored by current prediction tools.

Conclusions:

• A meta-predictor combining multiple tools enhances the reliability of predicting protein stability changes.
• Computational tools can successfully increase protein stability, but current methods may compromise solubility.
• Improvements in the underlying potentials/force fields of prediction tools are needed to balance stability gains with solubility maintenance.