Enhanced Protein Complex Prediction via Rosetta, AlphaFold, and Nondifferential Covalent Labeling Mass Spectrometry

Area of Science:

• Biochemistry and Structural Biology
• Computational Biology and Bioinformatics
• Mass Spectrometry Techniques

Background:

• Covalent labeling (CL) mass spectrometry is a key technique for protein structure elucidation.
• Integrating CL with computational prediction aids in modeling protein complexes.
• Differential CL data (monomer vs. complex) is ideal but often difficult to obtain.

Purpose of the Study:

• To develop a framework for utilizing nondifferential CL data in protein complex prediction.
• To improve the accuracy of computational protein complex models.
• To leverage bound-state CL measurements when monomeric labeling is challenging.

Main Methods:

• Developed a computational framework for protein complex prediction using nondifferential CL data.
• Introduced a hydroxyl radical protein footprinting (HRPF)-derived scoring term.
• Applied the scoring term to penalize docked models based on agreement with experimental CL data.

Main Results:

• The HRPF-derived scoring term significantly improved the accuracy of protein complex models.
• Best-scoring models showed an average root-mean-square deviation (RMSD) improvement of 4.6 Å with HRPF data.
• Inclusion of CL data led to four top-scoring complexes with RMSDs below 5Å, a significant improvement.

Conclusions:

• Nondifferential CL data can be effectively utilized to enhance protein complex prediction.
• The developed framework improves model accuracy, particularly when only bound-state CL data is available.
• This approach expands the utility of CL mass spectrometry in structural biology.