Methods for three-dimensional reconstruction of heterogeneous assemblies.

Area of Science:

• Structural biology
• Biophysics
• Biochemistry

Background:

• Electron microscopy (EM), particularly single-particle cryo-EM, is crucial for determining 3D structures of biological complexes.
• Biological samples in solution often exhibit conformational heterogeneity, limiting the resolution and accuracy of 3D reconstructions.
• This heterogeneity can manifest as different molecular conformations or variations in assembly and ligand binding.

Purpose of the Study:

• To review image processing methods for detecting and sorting structural heterogeneity in cryo-EM data.
• To present and exemplify a multivariate statistical analysis approach for handling mixed structural states.
• To demonstrate the ability to obtain high-quality reconstructions from heterogeneous datasets.

Main Methods:

• Review of image processing techniques designed to address structural heterogeneity in cryo-EM.
• Application of multivariate statistical analysis (MSA) to discriminate image variations.
• Separation of heterogeneous images into homogeneous subsets for individual 3D reconstruction.

Main Results:

• MSA effectively discriminates variations arising from particle orientation, conformation, ligand binding, and noise.
• The method successfully sorts structural heterogeneity in both model and real cryo-EM data.
• High-quality, reliable reconstructions of multiple structures were obtained from mixed datasets.

Conclusions:

• Image processing methods, especially MSA, are vital for overcoming structural heterogeneity challenges in single-particle cryo-EM.
• These techniques enable the accurate determination of multiple 3D structures from heterogeneous biological samples.
• This advancement improves the resolution and reliability of structural biology studies using cryo-EM.