Issues of resolution and polymorphism in single-particle reconstruction

Area of Science:

• Structural biology
• Biophysics
• Electron microscopy

Background:

• Three-dimensional reconstruction from electron microscopic (EM) images is a powerful technique for analyzing macromolecular complexes.
• Assessing reconstruction quality lacks stereochemical constraints, unlike X-ray crystallography.
• Common quality assessments include comparing reconstruction projections with class averages and resolution statistics (e.g., Fourier shell correlations).

Purpose of the Study:

• To investigate the reliability of quality assessment metrics in single-particle EM.
• To demonstrate how artefactual reconstructions can arise and be erroneously validated.
• To highlight the need for critical evaluation of EM reconstruction quality.

Main Methods:

• Simulated noise-based image analysis to assess resolution statistics.
• Generation of independent reconstructions from different initial models.
• Analysis of real EM images of DnaB rings exhibiting polymorphism.

Main Results:

• Simulated noise can generate impressive resolution statistics, potentially overestimating the true resolution.
• Artefactual reconstructions can be produced that match class averages, even with heterogeneous images.
• Polymorphism in real EM data can lead to artefactual reconstructions that appear valid.

Conclusions:

• Resolution statistics alone are insufficient for validating EM reconstructions.
• Artefactual reconstructions can be generated and validated using standard metrics when dealing with heterogeneous or polymorphic data.
• Critical evaluation of single-particle EM reconstructions is essential to avoid structural misinterpretations.