Direct Calculation of Protein Fitness Landscapes through Computational Protein Design

Area of Science:

• Protein biochemistry and structural biology
• Computational biology and bioinformatics
• Molecular evolution

Background:

• Protein function and structure are primarily determined by amino acid sequences.
• Naturally selected or experimentally derived sequences offer limited scope of protein variants.
• Understanding sequence variability requires exploring unexplored protein sequences.

Purpose of the Study:

• To demonstrate computational methods for large-scale characterization of alternative protein sequences.
• To define a protein fitness landscape by measuring stability and binding changes.
• To identify reasons for protein sequence robustness or variability.

Main Methods:

• Utilized dead-end elimination and A(∗) search algorithms.
• Analyzed low-energy single mutant variants and their structures for a G-protein heterotrimer.
• Measured changes in structural stability and binding interactions.

Main Results:

• Established consistency between computational algorithms and known biophysical/evolutionary trends.
• Recapitulated known protein side-chain interactions.
• Successfully predicted novel protein side-chain interactions.

Conclusions:

• Computational methods provide an efficient, large-scale mechanism to explore protein sequence space.
• This approach defines protein fitness landscapes and reveals sequence-interaction dynamics.
• The study successfully predicts novel interactions and enhances understanding of protein sequence variability.