Prediction of protein-protein interface sequence diversity using flexible backbone computational protein design

Area of Science:

• Computational biology
• Protein engineering
• Structural biology

Background:

• Identifying functional protein sequences is a key challenge in computational protein design.
• Protein conformational plasticity, or changes in shape due to mutations, often leads to design failures.
• Understanding and predicting these conformational changes is crucial for advancing protein design.

Purpose of the Study:

• To develop a computational method that predicts protein sequences tolerated at a protein-protein interface by incorporating conformational flexibility.
• To improve the accuracy of computational protein design by accounting for protein structural plasticity.

Main Methods:

• Developed a method using flexible backbone ensembles to model protein structural plasticity.
• Inspired the model on coupled side chain-backbone "backrub" motions observed in protein crystal structures.
• Applied the method to predict sequences for the human growth hormone and its receptor interface.

Main Results:

• Predictions from the flexible backbone ensemble method were enriched in functional proteins.
• The method's performance surpassed that of a phage display screen for the human growth hormone-receptor interaction.
• The backrub sampling approach captured significant localized conformational changes relevant to sequence plasticity.

Conclusions:

• Flexible backbone ensembles can effectively predict sequences tolerated at protein-protein interfaces.
• The "backrub" motion model captures essential aspects of protein conformational plasticity for improved design.
• This method has significant implications for generating sequence libraries for complex protein engineering tasks.