Abstract
Cryo-correlative light and electron microscopy (cryo-CLEM) facilitates in situ imaging and structural analysis by combining the molecular specificity of fluorescence microscopy with the ultrastructural resolution of cryo-electron microscopy. By further combining single molecule localization with cryo-CLEM, molecular positions of individual emitters can be revealed in the context of the electron density map of a cell, providing unique insights to profound questions in cell biology and virology. However, cryogenic fluorescence light microscopy (cryo-FLM) suffers from severe and spatially heterogeneous optical aberrations that distort the point spread function, limiting the accuracy of molecular localizations as well as downstream cryo-transmission electron microscopy workflows. Here, we present a systematic and quantitative analysis of optical aberrations in a commercial cryo-FLM system, uncovering the sources of significant distortions such as system imperfections, refractive index mismatches, and sample-induced heterogeneities. These system and sample induced aberrations lead to localization errors up to 90 nm laterally and over 300 nm axially, challenging the feasibility of precise molecular positioning within the vitrified specimen. We demonstrate that these errors are partially mitigated by spatially matched or adaptive point spread function models pushing the error rate down to ten nanometers or less, offering practical guidance for aberration-aware cryo-FLM and cryo-CLEM strategies. Our findings highlight the necessity of accurate, in situ point spread function modeling to achieve nanometer-scale localization in cryo-FLM. The experimental pipeline developed in this work establishes a novel tool to assess optical performance in cryo-CLEM and cryogenic focused ion beam milling workflows as the field strives toward accurate and precise molecular localization.
Introduction: The "Optical Mess" of Cryo-Imaging
Cryo-correlative light and electron microscopy (cryo-CLEM) represents a powerful "best of both worlds" technique, merging the molecular specificity of fluorescence with the high-resolution ultrastructural context of electron microscopy. To achieve true molecular-level understanding, researchers must pinpoint fluorescent tags with nanometer precision to map them accurately onto electron density maps.
However, cryogenic fluorescence light microscopy (cryo-FLM) is plagued by severe optical aberrations. These distortions arise from a combination of hardware imperfections, the massive refractive index mismatch between air-immersion objectives and frozen samples, and physical warping of the sample grid during the plunge-freezing process. Until now, these aberrations were poorly quantified, leading to significant "localization fidelity" errors that compromised structural analysis.
Methodology for Quantifying Distortions
To isolate and quantify these distortions, researchers developed a comprehensive experimental and analytical pipeline:
- Sequential Imaging: The team designed a strategy to image the exact same 200 nm fluorescent beads first under cryogenic conditions (CryoT) and then at room temperature (RT). This allowed for the separation of inherent system errors from cryo-induced aberrations.
- Phase Retrieval and Zernike Decomposition: Using a Maximum Likelihood Estimation (MLE)-based algorithm, they reconstructed the 3D Point Spread Function (PSF) from noisy image stacks. These distorted wavefronts were then decomposed into Zernike polynomials to categorize errors into specific optical terms like astigmatism, coma, and spherical aberration.
- Simulation and Modeling: Using retrieved PSFs, they generated thousands of simulated images to test how different PSF models (theoretical vs. experimental) affected localization accuracy across varying photon counts.
Key Results: Restoring Precision
The study revealed that aberrations at cryogenic temperatures are significantly more severe and complex than at room temperature. While refractive index mismatch typically suggests spherical aberration as the primary culprit, the researchers found that astigmatism and coma were often more dominant and unpredictable, varying even between different squares on the same TEM grid.
Key findings include:
- Massive Localization Errors: Standard theoretical models resulted in localization biases of up to 90 nm laterally and over 300 nm axially—errors that could exceed the thickness of the entire sample.
- The Matched Model Solution: By using "spatially matched" PSF models (derived from a bead proximal to the target), the error rate was pushed down to approximately 6 nm laterally and 30 nm axially at typical photon counts. Under high-photon conditions, accuracy reached sub-nanometer lateral and sub-5 nm axial levels.
- Resolution Loss: Cryogenic conditions caused a 1.4 to 1.7-fold increase in the Full Width at Half Maximum (FWHM) of the PSF compared to room temperature, representing a significant loss in raw resolution.
Conclusion and Future Directions
This systematic characterization proves that while cryo-FLM suffers from severe spatial heterogeneity and aberrations, these errors are not insurmountable. The resilience of lateral precision to sample-induced errors is encouraging, though axial precision remains extremely sensitive.
Future work will focus on moving toward adaptive in situ PSF estimation, where the microscope "learns" the specific distortions of a biological sample in real-time. This study provides the foundational tools and guidance necessary for the field to achieve reliable, nanometer-scale molecular localization within vitrified specimens.
Summary updated