Multiscale three-dimensional ultrastructural mapping of intestinal tissues and organoids

Problem
Understanding cell biology within native environments requires imaging subcellular organization in three dimensions. In the intestinal epithelium, cell-cell interfaces and subcellular architecture are critical for differentiation and tissue function, yet resolving these features volumetrically remains a significant challenge. Conventional light microscopy offers molecular specificity but suffers from limited resolution and optical scattering, while electron microscopy (EM) provides ultrastructural detail but is poorly suited for volumetric acquisition combined with specific protein labeling. Consequently, tissue-scale organization and subcellular ultrastructure are often analyzed separately using different modalities. Furthermore, while intestinal organoids are widely used as in vitro models, their ultrastructural validation has largely been restricted to sparse two-dimensional EM images, lacking the high-resolution volumetric imaging necessary to fully validate tissue mimics.

Methodology
The authors established an optimized Ultrastructure Expansion Microscopy (U-ExM) workflow tailored for mouse small intestine tissue sections and organoid models. Key methodological adaptations included:

• Fixation: Use of 4% paraformaldehyde (PFA) in PHEM buffer to better preserve cytoskeletal architecture and ultrastructure compared to standard PBS-based fixation.
• Gel Infiltration and Homogenization: Extended incubation times for gelation solutions to improve monomer penetration in dense tissues. For tissue sections, samples were denatured in an autoclave (126°C for 2 hours); for organoids, heat-block incubation (98°C for 3 hours) was used to achieve complete homogenization while preserving epithelial organization.
• Labeling Strategies: The workflow integrated both pre-expansion and post-expansion antibody labeling. Pre-expansion labeling targeted epitopes before structural modification, while post-expansion labeling (including tertiary staining) reduced linkage error and enhanced signal density. Total protein labeling was achieved using NHS-ester dyes.
• Segmentation: A custom "human-in-the-loop" segmentation workflow was developed, combining NHS-ester total protein labeling with membrane-localized CAAX-tdTomato and DAPI signals. This utilized tools such as SAM2, Cellpose, and seeded watershed algorithms to reconstruct individual cell bodies within entire crypt volumes.

Key Contributions and Results

• Multiscale Imaging of Native Tissue: The optimized workflow enabled the imaging of multiple continuous crypt-villus units in a single stitched overview. It preserved tissue architecture, allowing for the reconstruction of individual crypts and the visualization of the crypt niche, including Paneth cells and intestinal stem cells (ISCs). At subcellular resolution, the method resolved brush border organization (varying from dense at the villus tip to less organized near the crypt) and complex lateral membrane interdigitations between enterocytes.
• Organoid Validation: The workflow was successfully adapted to mouse small intestine organoids. Expanded organoids retained their overall architecture, revealing multiple crypt-like domains within a continuous epithelial structure. The method allowed for direct comparison of epithelial organization between native tissue and organoid systems, confirming that organoids recapitulate key architectural and ultrastructural features, including cell type-specific morphologies (enterocytes, goblet cells, Paneth cells) and apico-basal polarity.
• Molecular Annotation: The study demonstrated compatibility with targeted antibody labeling both before and after expansion. Specific markers such as GM130 (Golgi), Lysozyme (Paneth cells), and OLFM4 (ISCs) were successfully visualized within the ultrastructural context provided by total protein staining.
• Quantitative 3D Segmentation: The authors achieved nanoscale volumetric segmentation of entire crypt epithelia. This revealed distinct subsets of small, highly elongated ISC-like cells that were difficult to identify in 2D sections due to limited cross-sectional area. The 3D reconstruction highlighted extensive interfaces between elongated cells and Paneth cells, as well as "scutoid-like" geometries associated with apico-basal neighbor exchange.
• Disease Modeling: The workflow was applied to a Microvillus Inclusion Disease (MVID) organoid model (inducible Myo5b knockout). U-ExM successfully visualized disease-associated ultrastructural phenotypes, including reduced apical microvilli, accumulation of subapical vesicular structures, large (auto)lysosomes, and intracellular microvillus inclusions. These features, previously requiring EM for detection, were resolved volumetrically within the intact organoid context.

Significance
The paper establishes U-ExM as a key technology for studying three-dimensional cell biology within intestinal tissue and tissue mimics. By bridging the gap between light microscopy and electron microscopy, the workflow enables molecularly annotated, quantitative multiscale imaging of intact epithelial systems. The authors claim that this approach provides a platform for validating organoid models beyond transcriptional profiling, offering a direct volumetric comparison of native tissue and in vitro systems. Furthermore, the ability to visualize disease-associated ultrastructural defects in 3D without the complexity of EM suggests the method's utility for studying pathological states involving altered epithelial ultrastructure, such as Celiac Disease and Inflammatory Bowel Disease. The study positions U-ExM as an accessible intermediate space that combines specific antibody staining, volumetric context, and ultrastructural readout, facilitating new insights into the organizational complexity of specialized cell types during differentiation and disease.