Predicting and Experimentally Validating Hot-Spot Residues at Protein-Protein Interfaces

Area of Science:

• Chemical Biology
• Structural Biology
• Computational Biology

Background:

• Protein-protein interactions (PPIs) are fundamental to biological processes but challenging to study due to their dynamic nature and shallow interfaces.
• Understanding the thermodynamic stability of PPIs is crucial for advancing biological mechanism studies, understanding mutant phenotypes, and designing inhibitors.
• Computational alanine scanning (CAS) is a method for predicting hot-spot residues that drive protein association, but accuracy and throughput vary among methods.

Purpose of the Study:

• To comparatively analyze existing computational alanine scanning (CAS) methods for predicting protein-protein interaction (PPI) hot-spot residues.
• To introduce and validate a new method, BUDE Alanine Scanning, for enhanced accuracy in hot-spot prediction.
• To experimentally validate the predictive accuracy of improved CAS approaches across diverse PPI types.

Main Methods:

• Comparative analysis of multiple computational alanine scanning (CAS) methods.
• Introduction and application of the BUDE Alanine Scanning method to single structures and structural ensembles (NMR, molecular dynamics).
• Experimental validation of predicted hot-spot residues for three distinct protein-protein interactions: NOXA-B/MCL-1, SIMS/SUMO, and GKAP/SHANK-PDZ.

Main Results:

• Identification of effective CAS approaches for improving hot-spot residue prediction accuracy.
• Demonstration that BUDE Alanine Scanning accurately predicts hot-spots in various structural contexts.
• Experimental validation confirmed the accuracy of predicted hot-spots for diverse PPIs, including helix- and strand-mediated interactions.
• Successful application of the approach to predict hot-spots at a novel Affimer/BCL-xL interface.

Conclusions:

• Comparative analysis and the novel BUDE Alanine Scanning method significantly enhance the accuracy of predicting hot-spot residues in protein-protein interactions.
• Accurate hot-spot prediction is achievable across diverse PPI structures and types, facilitating further research in chemical biology and drug design.
• This validated approach provides a robust foundation for understanding PPI stability and engineering protein interactions.