Genetic Algorithm Approach for the Optimization of Protein Antifreeze Activity Using Molecular Simulations

Area of Science:

• Biochemistry
• Computational Biology
• Materials Science

Background:

• Antifreeze proteins (AFPs) are crucial for inhibiting ice crystal growth.
• Enhancing AFP activity is vital for applications in cryopreservation and agriculture.
• Predicting AFP activity computationally aids in designing improved variants.

Purpose of the Study:

• To develop a genetic algorithm for designing AFP mutants with enhanced antifreeze activity.
• To utilize a neural network model for predicting thermal hysteresis (ΔTc) of AFP mutants.
• To identify structural modifications that improve AFP ice-binding capabilities.

Main Methods:

• Implemented a genetic algorithm coupled with a neural network predictor for AFP mutant design.
• Applied the algorithm to three diverse AFPs: wfAFP, rQAE, and RiAFP.
• Validated predicted improvements through molecular simulations and analysis of thermal hysteresis.

Main Results:

• Discovered significantly improved AFP mutants for rQAE and RiAFP.
• Observed increases in thermal hysteresis of up to 0.572 K for rQAE and 1.33 K for RiAFP.
• Identified two key strategies for enhancing antifreeze activity: ordering surface water and increasing ice-binding surface area.

Conclusions:

• The genetic algorithm effectively designs AFP mutants with superior antifreeze properties.
• Structural modifications, including internal water channels and planar ice-binding surfaces, enhance AFP function.
• Specific nonpolar residues are critical for improving antifreeze activity at the ice-binding interface.