TL;DR
This paper introduces new methods within the RELION software package to automatically detect and correct for three common optical imperfections in cryo-electron microscopy (cryo-EM) data: symmetrical and antisymmetrical aberrations, and anisotropic magnification. By correcting these "hidden" distortions after initial data collection, the researchers demonstrate significant improvements in the resolution of reconstructed 3D molecular structures, making it easier to visualize fine details.
What Problem Does This Paper Address?
Cryo-EM is a powerful technique for determining the 3D structure of biological molecules at near-atomic resolution. However, the quality of the final 3D reconstruction is fundamentally limited by the optical system of the electron microscope. Imperfections in the microscope's optics, such as aberrations (distortions in how the electron beam is focused) and anisotropic magnification (uneven scaling of the image in different directions), can introduce subtle but significant errors. These errors, often referred to as "hidden" distortions, can degrade the resolution of the reconstructed map, making it difficult to discern fine structural details. Traditionally, correcting these issues required specialized test samples and meticulous microscope alignment, which is not always feasible or practical. This paper addresses the need for methods that can detect and correct these optical artifacts directly from the cryo-EM data of biological samples themselves, after the data has been collected.
Methods
The paper presents three distinct, yet complementary, methods implemented in RELION-3.1 to estimate and correct for:
- Antisymmetrical Aberrations: These affect the phase shift of the electron beam differently for different spatial frequencies, breaking the inherent symmetry of the contrast transfer function (CTF). The method models these aberrations using antisymmetrical Zernike polynomials and estimates their coefficients by minimizing the squared difference between observed and predicted particle images. This is achieved in a two-step process: first, calculating per-pixel optimal phase shifts and weights, and then using a downhill simplex method to find the best Zernike coefficients.
- Symmetrical Aberrations: These affect the amplitude of the CTF, causing deviations from the expected elliptical rings in the power spectrum. The approach here is to model these deformations using symmetrical Zernike polynomials. Similar to antisymmetrical aberrations, the method minimizes a sum of squared differences, but it's formulated as a linear problem after a trigonometric substitution, allowing for efficient optimization.
- Anisotropic Magnification: This refers to uneven scaling of the image in different directions. The method estimates this by comparing predicted particle images with observed ones. It models the distortion as a 2x2 matrix and uses a first-order Taylor expansion of the 3D reference map to linearize the problem, allowing for efficient estimation of the magnification matrix. This approach can also account for skew components, which are not typically considered in cryo-EM.
Crucially, these methods are designed to work a posteriori, meaning they can be applied to existing cryo-EM datasets without needing to recollect data or use special samples. The methods leverage a preliminary 3D reconstruction and the viewing parameters of individual particles.
Datasets
The researchers utilized several publicly available cryo-EM datasets for their experiments:
- Rabbit muscle aldolase (EMPIAR-10181)
- Thermoplasma acidophilum 20S proteasome (EMPIAR-10185)
- A 20S proteasome dataset collected with a Volta phase plate (EMPIAR-10078)
- Mouse heavy-chain apoferritin (EMPIAR-10216)
- Human apoferritin (EMPIAR-10200) (used for simulated anisotropic magnification experiment)
These datasets represent a range of biological molecules and were collected on different microscopes, allowing for a comprehensive evaluation of the methods. The use of these publicly available datasets also serves as a benchmark for future comparisons.
Key Insights and Ideas
- "Hidden" Distortions Matter: The paper strongly emphasizes that optical aberrations and magnification anisotropy are not just minor nuisances but can significantly limit achievable resolution in cryo-EM, especially at higher resolutions.
- A Posteriori Correction is Powerful: The ability to correct these distortions after data collection, using only a preliminary 3D reconstruction, democratizes high-resolution cryo-EM by reducing the reliance on perfect microscope alignment and specialized samples.
- Zernike Polynomials as a Universal Language: The consistent use of Zernike polynomials as a basis for modeling aberrations provides a standardized and interpretable way to describe these complex optical effects.
- Interplay of Aberrations and Refinement: The study shows that correcting aberrations can enable more effective downstream processing steps like Bayesian polishing and defocus refinement, leading to cumulative resolution gains.
- Anisotropic Magnification is More Than Just Scaling: The method's ability to detect and model skew components in magnification is a novel contribution, potentially capturing more complex image distortions.
Key Results
- Resolution Improvement: For datasets where these aberrations were present, correction led to significant resolution improvements. For example, the aldolase dataset saw resolution increase from 2.5 Å to 2.1 Å, and the proteasome from 3.1 Å to 2.3 Å after aberration correction.
- Measurable Aberrations: The study consistently detected antisymmetrical (trefoil/threefold astigmatism) and symmetrical (fourfold astigmatism) aberrations in multiple datasets, demonstrating their prevalence.
- Effective Anisotropic Magnification Correction: The methods successfully estimated and corrected for anisotropic magnification, even when it was relatively weak (e.g., 0.54% anisotropy in the apoferritin dataset).
- Implementation in RELION-3.1: The developed methods are now integrated into the widely used RELION software, making them accessible to the cryo-EM community.
- Quantitative Accuracy: The anisotropic magnification estimation could achieve precision to three significant digits, even with a relatively small number of particles.
Limitations and Future Work
- "Easy" Targets: The paper notes that the most significant resolution improvements were observed for datasets that already achieved resolutions beyond 3.0-3.5 Å before correction. It's less clear how impactful these methods would be for lower-resolution datasets.
- Phase Plate Artifacts: The methods are not designed to correct for all types of artifacts, particularly those arising from residual charges on Volta phase plates, which can create complex interference patterns not captured by the Zernike polynomial models.
- Source of Aberrations: While the methods correct for aberrations, they don't address their root cause. The paper suggests that understanding and mitigating these sources at the microscope level would be ideal, but until then, these correction methods are crucial.
- Combining Datasets: The paper mentions the potential of these methods for combining datasets with different magnifications, but this aspect requires further illustration.
- Low-Energy Electrons: The authors hypothesize that these methods will become even more relevant for lower-energy microscopes (e.g., 100 keV), which are being developed for cost-effectiveness and imaging thin specimens, as aberrations are more pronounced at lower electron energies.
Summary updated