Co-Evolutionary Fitness Landscapes for Sequence Design

Area of Science:

• Biochemistry
• Structural Biology
• Computational Biology

Background:

• Predicting protein sequence fitness is crucial for protein design.
• Statistical potentials derived from coevolutionary data offer insights into protein fitness landscapes.

Purpose of the Study:

• To explore the use of statistical potentials and Monte Carlo simulations for designing novel, high-fitness protein sequences.
• To validate the designed sequences by assessing their structural stability and ligand-binding capabilities.

Main Methods:

• Extracted statistical potentials from known protein sequences to define a coevolutionary fitness landscape.
• Employed Monte Carlo simulations to search this landscape for high-fitness sequences.
• Designed sequences for three distinct protein folds: GA (all-α), GB (α/β), and SH3 (all-β).
• Determined the structures of representative designed proteins and assessed their folding stability.

Main Results:

• Most designed protein sequences were found to fold stably into their target structures.
• Experimental structures were determined for representative GA, GB, and SH3 designed proteins.
• Several designed proteins exhibited native ligand binding, with some showing higher affinity than wild-type proteins.
• The study demonstrated the successful application of statistical fitness landscapes for protein design.

Conclusions:

• Searching statistical fitness landscapes is a highly effective method for identifying novel, stable protein sequences.
• The integration of statistical potentials and Monte Carlo simulations provides a powerful tool for de novo protein design.
• Designed proteins can achieve stable folds and functional ligand-binding properties, opening avenues for protein engineering.