Progressive assembly of multi-domain protein structures from cryo-EM density maps

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Cryo-electron microscopy (cryo-EM) enables large protein structure determination.
• Modeling inter-domain orientations in multi-domain proteins remains a significant challenge, limiting success rates.
• Accurate structural models are crucial for understanding protein function and drug development.

Purpose of the Study:

• To develop an automated method for assembling multi-domain protein structures from cryo-EM maps.
• To address the limitations in modeling inter-domain orientations for large and complex proteins.
• To provide a reliable pipeline for large-scale multi-domain protein structure modeling.

Main Methods:

• Developed Domain Enhanced Modeling using Cryo-Electron Microscopy (DEMO-EM), an automated approach.
• Employed a progressive structural refinement procedure combining rigid-body domain fitting.
• Integrated flexible assembly simulations with deep-neural-network inter-domain distance profiles.

Main Results:

• DEMO-EM achieved 97% accuracy in modeling correct inter-domain orientations on a benchmark set (TM-score >0.5).
• The method outperformed existing state-of-the-art approaches for medium- to low-resolution cryo-EM maps.
• Applied to SARS-CoV-2, DEMO-EM generated highly accurate models (avg. TM-score 0.97, 1.3 Å RMSD).

Conclusions:

• DEMO-EM provides an efficient and automated pipeline for reliable large-scale multi-domain protein structure modeling.
• The method significantly advances the ability to determine complex protein structures from cryo-EM data.
• This work facilitates structural studies of large proteins, crucial for biological and medical research.