TL;DR
This paper introduces a novel correlative microscopy pipeline that combines cryogenic super-resolution (SR) fluorescence microscopy with focused ion beam scanning electron microscopy (FIB-SEM) to image entire frozen cells in 3D. This approach overcomes previous limitations in sample preparation and imaging modalities, enabling unprecedented visualization of protein localization within cellular ultrastructure and revealing new biological insights.
What Problem Does This Paper Address?
Understanding the intricate organization of cells at the nanoscale requires visualizing both the precise location of specific proteins and the overall ultrastructure of the cell. Correlative light and electron microscopy (CLEM) aims to achieve this, but traditional methods often face trade-offs: chemical fixation can introduce artifacts, while preserving fluorescence and ultrastructure simultaneously is challenging. Furthermore, achieving high resolution across an entire cell's volume has been a significant hurdle. This paper addresses the need for a method that preserves native cellular states, allows for independent optimization of both fluorescence and electron microscopy, and can image large volumes with high resolution.
Methods
The core of this work is a new pipeline for correlative cryogenic super-resolution (cryo-SR) and focused ion beam milling block-face electron microscopy (FIB-SEM).
Key methodological advancements include:
- Cryo-fixation via High-Pressure Freezing (HPF): This technique rapidly freezes cells in their native state, preserving ultrastructure without chemical fixation artifacts.
- Cryogenic SR Microscopy: SR imaging (both Structured Illumination Microscopy - SIM, and Single Molecule Localization Microscopy - SMLM) is performed at cryogenic temperatures (down to 8K) on vitrified samples. This significantly improves fluorophore photostability, allowing for longer imaging times and higher photon counts crucial for precise localization. The use of a liquid helium cryostat is highlighted as enabling exploration of photophysics at very low temperatures.
- Decoupled Sample Preparation: The pipeline strategically places the cryo-SR step between cryofixation and freeze substitution/staining for FIB-SEM. This allows for independent optimization of sample preparation for each modality, avoiding the compromises often required in traditional CLEM.
- FIB-SEM for Whole-Cell 3D Imaging: FIB-SEM is used to mill away thin layers of resin-embedded cells and image the exposed surface, generating high-resolution (4-8 nm isotropic voxels) 3D reconstructions of entire cells.
- Advanced Registration Algorithms: Deformable registration algorithms (e.g., BigWarp) are employed to accurately align the high-resolution SR and FIB-SEM datasets, accounting for sample deformations introduced during processing.
Datasets
The paper does not explicitly mention training datasets for machine learning models. The "datasets" in this context refer to the biological samples imaged:
- U2OS cells: Used for initial characterization of fluorophores and for demonstrating correlative imaging of ER and mitochondria.
- COS-7 cells: Used to demonstrate whole-cell correlative imaging of ER and mitochondria, and to reveal intranuclear vesicles.
- HeLa cells: Used to study peroxisome morphology and interactions with other organelles.
- Mouse Cerebellar Granule Neurons (CGNs): Used to investigate cell-to-cell adhesions and nuclear chromatin organization during differentiation.
The paper also refers to extensive supplementary data, including numerous figures (Figs. S1-S49), tables (Tables S1-S2), and movies (Movies S1-S7), which collectively represent the imaging data generated.
Key Insights and Ideas
- Cryogenic SR Enhances Photostability: Imaging at very low temperatures dramatically improves the photostability of fluorophores, enabling more accurate and higher-resolution SMLM. This is a critical enabler for high-quality cryo-SR.
- Decoupling Sample Prep is Key: The strategic placement of cryo-SR before FIB-SEM preparation allows for optimal preservation of both fluorescence signals and ultrastructure, a major breakthrough for CLEM.
- Whole-Cell 3D ultrastructure reveals unexpected protein localization: By imaging entire cells, the method uncovers protein distributions and ultrastructural features that would be missed in smaller, sectioned samples. This leads to the discovery of novel protein-ultrastructure relationships.
- Morphological diversity of organelles: The technique reveals significant heterogeneity in the shape and organization of organelles like peroxisomes and endosomes, which may be linked to their function.
- Chromatin organization changes during differentiation: The study provides detailed 3D maps of chromatin organization in developing neurons, revealing distinct subdomains and dynamic changes associated with differentiation.
Key Results
- High Correlation Accuracy: The developed pipeline achieves high accuracy in correlating SR and FIB-SEM data, with median correlation errors as low as 27 nm in high-landmark-density regions.
- Revealed Novel Protein-Ultrastructure Relationships: The method uncovered unexpected localization patterns, such as intranuclear vesicles positive for an ER marker, intricate web-like structures of adhesion proteins at cell-cell junctions, and heterogeneity in chromatin organization.
- Detailed Organelle Morphology: The study provided detailed 3D reconstructions of peroxisomes, revealing diverse morphologies (spherical, plate-like, cup-shaped, hollow spheres) that correlate with their volume.
- Chromatin Dynamics in Neuronal Differentiation: The research mapped changes in chromatin organization (heterochromatin and euchromatin) during cerebellar granule neuron differentiation, identifying distinct subdomains and their correlation with specific histone modifications.
- Comprehensive Cellular Views: The pipeline enables the imaging of entire cells (up to ~5000 µm³), providing a global context for subcellular structures and protein localization.
Limitations and Future Work
- Limitations:
- Time and Data Volume: 3D cryo-SMLM and FIB-SEM are time-intensive processes, requiring days per cell for sample preparation and imaging. The resulting datasets are massive (tens of terabytes), posing challenges for analysis.
- Resolution Trade-offs: While high resolution is achieved, there are still differences in resolution between SR and FIB-SEM, requiring sophisticated registration.
- Artifacts: Although minimized, some artifacts from freeze substitution and resin embedding can still occur, requiring careful interpretation.
- Future Work:
- Live-Cell Imaging Integration: Incorporating live-cell imaging before freezing would allow correlation of dynamic processes with static ultrastructure, enabling studies of cellular responses to perturbations.
- Broader Specimen Types: Extending the pipeline to more physiologically relevant specimens like small organisms or organoids is feasible.
- Improved SR Resolution: Designing dual-window cryostats and advanced detection schemes could further improve axial resolution of SR imaging.
- Scalable Segmentation: Developing robust automated identification and segmentation algorithms is crucial for analyzing the massive datasets generated by FIB-SEM.
- Complementary Techniques: The authors suggest that FIB-SEM and cryo-EM tomography are complementary and developing a pipeline to combine both would be valuable.
Summary updated