Multiscale natural moves refine macromolecules using single-particle electron microscopy projection images

Area of Science:

• Structural biology
• Computational biophysics
• Biochemistry

Background:

• Electron microscopy (EM) is crucial for determining macromolecular structures.
• Reconstructing 3D density maps from EM projections can be challenging due to sample heterogeneity or preferred orientations.
• Refining molecular conformations directly from 2D images offers a potential solution to these limitations.

Purpose of the Study:

• To develop and validate a novel method for refining molecular conformations directly against 2D class averages from cryo-electron microscopy.
• To bypass the need for 3D density map reconstruction, especially for conformationally flexible or difficult-to-crystallize samples.

Main Methods:

• Utilized Natural Move Monte Carlo simulations to model macromolecules as segmented chains with flexible loops.
• Optimized molecular conformation and projection orientation simultaneously.
• Applied the method to both artificial lysozyme data and experimental data from Methonococcus maripaludis chaperonin.

Main Results:

• Successfully refined a closed-state model of Methonococcus maripaludis chaperonin to an open-state conformation using a single 2D class average.
• Demonstrated iterative refinement against heterogeneous projection images, capturing mixed open and closed states.
• Validated the method's efficacy on artificial data.

Conclusions:

• The presented method provides a direct and effective approach for refining molecular conformations from 2D electron microscopy data.
• It is particularly well-suited for macromolecules exhibiting significant conformational flexibility.
• The associated algorithm is available in the Methodologies for Optimization and Sampling In Computational Studies (MOSICS) program.