1.2.1.

Anchoring links sample components to the hydrogel monomers

Currently, most ExM protocols rely on AcX as an anchor,¹⁵ which enables crosslinking of amine moieties in proteins and other molecules to acrylamide monomers in the gel. Anchoring with AcX is compatible with conventional fluorescent labeling techniques, including immunohistochemistry. Importantly, this method is compatible with not only genetically expressed fluorescent proteins and commercially available antibodies, but also nucleic acids incubated with an additional anchoring agent such as Label-IT, which covalently attaches amines to nucleic acids.²³ We note, however, that some fluorescence loss can occur in the digestion and polymerization steps and the extent to which signal is lost depends on the fluorophore chemistry. In particular, cyanine-based dyes (e.g., Alexa 647, cy2, cy3, and cy5) are associated with significant signal loss in proExM and thus are not recommended.¹⁵ N-acryloxysuccinimide (AX) has recently proven to be an anchoring agent similar to AcX and is available at a much-reduced cost.²⁰

1.2.2.

Gelation polymerizes and solidifies the hydrogel in which the sample is embedded

Gelation results in a 3D matrix of crosslinked polymers to which molecules of interest in the sample are chemically anchored. The gelation process occurs in a chamber that holds the sample and the solutions that form the gel. The chamber must be large enough to cover the sample in its entirety and establish the shape of the gel. The chamber is then sealed with a coverslip and incubated to allow polymerization.²² The size and shape of the chamber are additional critical features that determine the user’s ability to manipulate the gel and its structural integrity during processing. After the monomer solution diffuses throughout the sample in a 4°C environment, a change to 37°C triggers the polymerization of the sodium acrylate and acrylamide monomers.²⁴ The crosslinking polymers form a dense matrix, and the anchoring agent forms covalent bonds between amines in the sample and acrylamide in the gel.

Most existing protocols for ExM depend on bis-acrylamide, which cross-links acrylamide monomers in the gelation step. $N,N$-dimethylacrylamide (DMAA) is a self-polymerizing agent, which forms more structurally robust hydrogels that are able to withstand $10\times$ or even $20\times$ expansion.^20,25 However, DMAA is even more sensitive to atmospheric oxygen, which reacts with intermediate radicals and interferes with radical-dependent gel cross-linking. Nitrogen gas can be used to flush out ambient oxygen from a working environment for uniform gelation.²⁰

1.2.3.

Digestion homogenizes the sample through proteolysis to minimize the distortion of the sample during later expansion

Digestion is typically performed using proteinase-K.¹⁵ For experiments in which labeling is performed after the expansion step, this step can be replaced with autoclaving the samples in detergents such as SDS and Triton, or via the use of a milder protease, to preserve epitopes while denaturing protein aggregates to allow gel expansion.²¹

1.2.4.

The sample-embedded hydrogel absorbs water and expands

Water is attracted to the polarized ends of the hydrogel molecules and forms dipoles around the negative ionic charges in the gel mesh, causing the polymer chain molecules to extend and the gel to swell isometrically.²⁶ Samples typically expand approximately fourfold (although see additional variants described below).

1.3.1.

High expansion factors and resolution

One approach to obtain expansion factors higher than approximately fourfold, and thus higher spatial resolution, is the iterative application of the expansion process. Iterative expansion (iExM) involves two consecutive expansion rounds. In iExM (Fig. 2), an expansion factor of $\sim 20$-fold is obtained, yielding a resolution of 20 to 25 nm with standard confocal microscopy imaging.¹⁷

Fig. 2

Variations of Expansion Microscopy. ExM: In the original protocol, the monomer gel solution includes eight components—sodium acrylate; acrylamide; MBAA, NN′-methylenebisacrylamide; PBS, phosphate-buffered saline; APS, ammonium persulfate; TEMED, $N,N,N^{\prime },N^{\prime }$-tetramethylethylenediamine; 4HT, 4-hydroxy-tempo. Sodium acrylate and acrylamide form the polymer as they are cross-linked by MBAA. 4HT is a polymerization inhibitor that prevents premature gelation before the monomer solution diffuses throughout the sample tissue; TEMED serves as an accelerator; APS serves as the initiator and thus is added last to the solution.^7,27 ProExM: Acryloyl-X is added as an anchoring agent that binds proteins to the polymer.¹⁵ iExM: An iterative approach to ExM, where the initial polymer is broken down, and a second swellable polymer mesh is formed in the space newly opened by the first expansion. This was implemented using custom-made DNA oligonucleotides conjugated to secondary antibodies and polymer anchoring moieties,¹⁸ such that only the specifically labeled structures of interest are attached to the polymer. ExR: AcX was applied in iExM for broad anchoring of proteins and antibodies to the polymer before post-expansion staining.¹⁹ $10\times$ expansion: A higher expansion factor was obtained by replacing acrylamide with $N,N$-dimethylacrylamide acid (DMAA) as the main monomer together with sodium acrylate.¹⁷ 20ExM: A single-step twentyfold expansion protocol without iterative steps, using optimized hydrogel integrity and oxygen-controlled gelation.²⁰ TREx: By decreasing the crosslinker and monomer concentrations, the hydrogel maintains structural integrity and increases the expansion factor tenfold in a single gelation step.¹⁶ U-ExM: For expanding cultured cells and analyzing proteins at higher resolution. Anchors proteins to the polymer by applying acrylamide and formaldehyde rather than using AcX.²⁸ iU-ExM: Enables high-resolution visualization of cellular structures by combining ultrastructure expansion with iterative expansion, achieving near-SMLM resolution for detailed molecular imaging in tissues and organelles.²⁹ TissUExM: To penetrate thicker tissue, this protocol adds triton to the monomer solution and increases the concentration of acrylamide. In addition, a collagenase VII digestion step is introduced after gelation, followed by a denaturation step and post-expansion staining.^30,31 Whole-body ExM: to homogenize and expand an entire zebrafish with multiple tissue types such as bones, cartilage, muscles, and soft tissues, this protocol extended the proteinase-K digestion time, added collagenase I, II, IV digestion, and decalcification with ethylenediaminetetraacetic acid (EDTA).³² ExFISH: Preserves the integrity and spatial positions of RNA molecules in cells and tissue slices. The RNA is stained by fluorescence in situ hybridization (RNA-FISH) amplified with hybridization chain reactions (HCRs) that enable the resolving of single RNA molecules. Multiple rounds of FISH increase the number of target RNA sequences.²³ EASI-FISH: Replaces Label-IT in the ExFISH protocol with a smaller molecule linker, melphalan, to retain and anchor nucleic acids to the ExM polymer. This linker is cost-effective, improves signal-to-noise ratio, and optimizes reagent penetration.³³ ExLLSM: Addresses microscopy working distance that limits sample thickness with a light sheet microscopy pipeline.^34,35 ExCel: Adapted ExM to the nematode C. elegans by adding steps for cuticle permeabilization.^36,37 Magnify: Achieves higher expansion factors and enhanced imaging resolution by adding $N,N$-dimethylacrylamide (DMAA) to the monomer solution alongside acrylamide and sodium acrylate. This protocol combines the high expansion factor of $10\times$ expansion with protein retention strategies from ProExM, allowing for detailed imaging of proteins in large or thick samples.²⁵ ExSeq: Enables spatially resolved RNA sequencing by anchoring RNA molecules to the gel using reagents such as LabelX and maintaining a neutral pH during expansion to preserve RNA integrity. The protocol involves re-embedding the expanded gel in a non-expandable gel to facilitate sequencing procedures.³⁸ ExIGS: Facilitates in-gel sequencing of DNA or RNA molecules with higher expansion factors by using a modified polyacrylate gel incorporating DMAA and Tn5 transposase. By adding rolling circle amplification (RCA) and sequencing-by-synthesis (SBS) buffers with increased gel porosity, it supports the enzymatic diffusion necessary for sequencing. The protocol includes re-embedding the expanded gel in a non-expandable gel for sequencing procedures and is derived from Magnify and ExSeq.³⁹

In an alternative approach, $10\times$ expansion and TREx (Fig. 2), 10-fold expansion is achieved in a single step via optimization of the gel recipe.^16,17 The most recent improvement of this permits $20\times$ single-shot expansion,²⁰ by developing a super-absorbent hydrogel recipe, tightly controlling oxygen levels during gelation and optimizing the gelation times according to the biological sample type (Fig. 2).

1.3.2.

Compatibility with light microscopes: tradeoffs and considerations

When working with large, expanded samples, the working distance between the imaged plane and the microscope objective becomes an important consideration. Working distance depends on the sample, expansion factor, and the microscope. Water-dipping objectives and upright microscopes offer the best working distances and are thus advantageous over inverted microscopes and oil or water immersion objectives. However, inverted microscopes are in more common use, and the latter objectives are associated with higher numerical apertures. Superficial target molecules within the sample may be accessible with objectives that have working distances smaller than the gel thickness. Consequently, the orientation of the sample within the gel and the orientation of the gel are important when imaging. Moreover, it is important to fit the chamber height as precisely as possible to the sample depth to minimize empty space.²²

1.3.3.

Labeling target proteins and nucleic acids

To pinpoint spatial relationships among molecular structures, proteins and nucleic acids are tagged with fluorescent molecules or other markers that have high specificity to the molecules under study. ExM is typically paired with protein and nucleic acid labeling techniques, such as immunohistochemistry and in situ hybridization, to tag molecules of interest. By anchoring the molecules of interest to the gel using AcX for proteins, or label-IT together with AcX for nucleic acids, and then expanding the sample, ExM enhances the detection of labeled proteins and nucleic acids, making it easier to study complex and densely packed structures within cells and tissues.^15,20,23,40

1.3.4.

Pre- versus post-expansion labeling

Molecules of interest can be labeled either before or after expansion treatment depending on the scientific question and preferred imaging tool. Pre-expansion labeling is most common in established protocols and procedures; fluorophores are often better able to withstand the digestion step necessary to soften and homogenize samples for expansion, whereas native fluorescent proteins may be proteolyzed and unfit for labeling post-expansion. However, post-expansion labeling is ideal for densely packed targets that benefit from “decrowding,” an increase in target accessibility that has helped visualize nanostructures such as amyloid plaques of Alzheimer’s disease model mice samples.¹⁹ In specific cases, as the expansion process physically separates proteins, it can render previously inaccessible targets accessible for antibody labeling, as native inter-protein distances can be shorter than the size of antibodies.^19,20 Moreover, it can allow for more precise localization of molecules of interest by reducing the linkage error, an error in spatial visualization measured by the distance between the molecule of interest and the fluorescent reporter due to the size of antibodies used as reporters;⁴¹ the relative size of the antibody with respect to the structure of interest becomes significantly smaller when the antibody is applied to the expanded sample. In addition, post-expansion labeling avoids fluorescence signal loss during homogenization and polymerization, which can be significant.⁴² Although post-expansion labeling allows for flexibility in experimental design and has greater signal amplification, it requires a gentler digestion step to retain target isotopes and greater reagent quantities and thus costs more than pre-expansion labeling due to the increased size of the sample. To mitigate the signal loss during the expansion step, combining pre- and post-expansion labeling can further amplify the labeling signal.^20,36

1.3.5.

Limitations of ExM

Several technical challenges remain in ExM, which future advancements aim to address. One such challenge is consistency in protocol application and obtained expansion factors between samples of the same type and to a greater extent in different types of samples. Moreover, ExM protocols, particularly ones associated with high expansion factors, can involve complex steps and require adaptation for and validation in new types of samples. Nevertheless, recent developments in automated sample preparation and streamlined protocols have shown promise in reducing benchwork time and improving workflow efficiency.

Additional challenges arise from the size of the expanded samples, particularly when expansion factors are 10× or higher. This is a challenge for the imaging itself, as some microscope imaging chambers have limited sizes, high NA objectives have limited working distances and thus are ill-suited for imaging these samples—giving rise to a resolution limit (albeit this limit is far more than compensated by the expansion factor), and large samples require the use of large working distance objectives that poorly fit, e.g., inverted microscopes. Moreover, sample volumes can increase by three orders of magnitude, which can make imaging times very long even on fast microscopes. Light-sheet and upright spinning disk confocal microscopes are thus best suited for imaging expanded samples. Large sample volumes also dilute the fluorescent signal, requiring signal amplification or the use of post-expansion staining, which requires very large amounts of antibodies (see pre- versus post-expansion labeling). Finally, as ExM achieves higher resolutions and sample volumes increase, the size of resulting datasets grows substantially, increasing storage and computational demands and requiring the development of appropriate software solutions for data analysis.³⁴

2.4.1.

Spatial transcriptomics with ExM

Expansion fluorescence in situ hybridization (ExFISH) (Fig. 2) was the first ExM variant that introduced the anchoring of nucleic acids, such as RNA, to the gel.²³ Expansion-assisted iterative fluorescence in situ hybridization (EASI-FISH) (Fig. 2) improves spatial precision of in situ hybridization in thick tissue specimens via protocol optimization and the use of a distinct anchoring molecule.³³ EASI-FISH was used to spatially determine transcripts from many genes simultaneously across a $300\text{-}\mu \mathrm{m}$-thick slice and identified previously unknown morphological diversity in deep regions such as the lateral hypothalamic area of mice. Using EASI-FISH in combination with genetic data from single-cell RNA-Sequencing (scRNA-Seq), researchers characterized marker genes in the lateral hypothalamic area and classified nine spatio-molecular regions.³³

2.4.2.

Revealing hidden protein structures with ExM

In expansion revealing (ExR) (Fig. 2), proteins are “decrowded” by two iterative expansion steps such that the increased distance between proteins allows post-expansion applied antibodies, larger than the space between these tightly packed proteins, to access their targets. Decrowding allows the labeling of targets, which cannot be visualized well even using super-resolution techniques, as the targets are challenging to label precisely until they are expanded. Using ExR, researchers revealed the nanoscale coordination of presynaptic calcium channels with postsynaptic proteins and the periodic nanostructures of amyloid beta plaques in mouse models of Alzheimer’s disease.¹⁹ This was only possible due to ExR’s capacity to decrowd proteins spatially while imaging with resolution on par with super-resolution techniques.

2.5.1.

In Drosophila, ExM enables the comparison of neural circuit connectivity characteristics between populations

ExM was optimized for application to larval Drosophila brains. The presynaptic protein organization observed was consistent with that established with other super-resolution microscopy methods. In addition, ExM allowed more precise quantification of insertion of somatosensory neural dendrites into epithelial cells compared with standard confocal microscopy.⁷⁵

When the expansion was followed by lattice light-sheet microscopy (ExLLSM³⁴) (Fig. 2), it allowed for high-resolution imaging of the entire Drosophila brain, with the ability to distinguish features that are $\sim 60\mathrm{nm}$ apart in the $x$ and $y$ dimensions and 90 nm apart in the z dimension. Furthermore, ExLLSM image acquisition was $\sim 700$ times faster than other super-resolution fluorescent microscopy techniques and 1200 times faster than EM.³⁴ The technique makes it possible to zoom in on individual synapses and zoom out to see broader patterns in synaptic connectivity, facilitating comprehensive and comparative anatomical investigations.⁷⁶

Using ExLLSM, researchers imaged all dopaminergic neurons across the Drosophila brain, visualizing their morphologies in all major brain regions and tracing specific clusters to determine cell types.³⁴ They also quantified presynaptic active zones across the brain, providing insights into the local density of synapses and dopaminergic neuron-associated active zones.³⁴ ExLLSM has been used to trace axonal branches of olfactory projection neurons and study their bouton arrangements at the calyx and lateral horn across multiple Drosophila specimens.³⁴

Lillvis et al.³⁵ added the development of a high-throughput ExLLSM open-source image analysis pipeline to enhance the accessibility of studying neuronal circuits. Importantly, in this manner, they were able to produce a proof of concept for one of the greatest promises of the application of ExM to neural circuits: due to the high speed and applicability to volumetric samples, ExM can be applied for the comparison of neural circuit connectivity across animal populations. This is a major advance compared with EM connectome generation that provides higher resolution data but for a single animal. Thus, they demonstrated the quantification of synapses in male compared with female fly populations. In the future, this approach can be applied to reveal gender, development, and experience-dependent differences in neural circuits and synapse structure. They also revealed synaptic structural underpinnings of behavior variability in a population. Such questions can only be addressed with rapid, volumetric imaging beyond the diffraction limit.

2.6.1.

Adapting expansion microscopy for C. elegans

Super-resolution microscopy methods such as STORM, PALM, SR-SIM, or STED have been used to study intact and dissected C. elegans,^95–100 but the imaging depth of these techniques is insufficient to map the entire depth of the animal. Furthermore, the tough cuticle of C. elegans limits antibody penetration and immunohistochemistry important for STORM or STED. The technical difficulty of immunostaining in C. elegans likely contributed to the pioneering use of GFP as a reporter in the worm. However, in recent years, the first ExM protocol for C. elegans has introduced a cuticle permeabilization technique for immunostaining and expansion³⁶ (Fig. 2). Immunostaining allows robust labeling of structures of interest for high-resolution imaging of a large region of interest with reduced constraints from photobleaching of endogenously produced fluorophores.^36,37 In this initial proof-of-concept paper, the authors showed that ExM can resolve adjacent synaptic puncta, more than doubling the number of puncta counted pre-expansion.³⁶

Several technical challenges remain for a comprehensive optical connectome.¹⁰¹ The first challenge is in the complete and continuous segmentation of each neuron, where the identity of each neuron can be maintained across the reach of its neurites. A novel probe has recently been developed expressly for the purpose of dense and continuous membrane labeling in ExM; this azide probe binds to fixed tissue membranes and later is fluorescently labeled via click-chemistry.¹⁰² The next challenge is in labeling pre- and postsynaptic specializations while retaining contextual information about neuronal identities. Given the mere 20 nm space between opposed membranes of synapses¹⁰³ and the enormous synaptic density, conventional light microscopy cannot parse fluorophore-labeled synapses with the resolution and throughput required to build a connectome. However, with up to $20\times$ physical magnification of samples, ExM could disambiguate individual GFP molecules in the cytosol of neurons;³⁶ such resolution has the potential to unpack dense neuronal architecture for optical connectomics.

In the way of protocol optimization and proof of concept, Yu et al.^36,37 demonstrated two relevant neuronal features that could be resolved by ExM. One is the disambiguation of individual neurites within a bundle and through defasciculation, as tracking neurites and maintaining neuronal identity is crucial when collecting connectome data. The other feature is the localization of chemical synapses and gap junctions by immunostaining and fluorescence in situ. Although synaptic partners can be deduced from membrane proximity, assigning electric and chemical connections, the directions of synaptic connections, and even neurotransmitters, depending on staining specificity, can greatly enrich the connectome data.¹⁰⁴ These qualitative demonstrations were not driven by an explicit hypothesis but suggest how ExM can provide connectome data.

Thus far, quantitative studies using ExM in C. elegans have been mostly cellular rather than neuronal, for example, the ultrastructure organization of P-granules and the dynamics of centrioles. ExM revealed the spatial arrangement of proteins that bind small RNAs in the oocyte P-granule.⁴³ P-granules are membrane-less organelles that are detectable by EM, but their protein ultrastructures cannot be discerned because the proteins cannot be differentially labeled for EM. During centriole elimination in oocytes, ExM has been used to reveal the sequence of ultra-structural changes with high spatiotemporal fidelity, where prior analyses have been limited to serial-section EM or immunohistochemistry for either high spatial or temporal resolution, respectively.¹⁰⁵ Building upon recent advances in the bio-orthogonal click chemistry, multifunctional ExM linkers have been developed to link and label glycans in the ExM gel with high specificity and isotropy, visualizing the distribution of glycans in anatomical context with nanoscale resolution.¹⁰⁶ Applying click chemistry and its compatibility with ExM adds versatility to label almost limitless molecular targets.

2.6.2.

Methods for expansion in C. elegans (ExCEL)

Expansion in C. elegans (ExCel) has primarily been developed by the Boyden laboratory at MIT.^36,37 There are three published protocols for magnifying fixed, whole animals: a standard, an epitope-preserving, and an iterative version (Table 1). The standard protocol allows visualization of fluorescent proteins, which are stable enough to withstand proteinase K digestion. The epitope-preserving protocol uses a gentler digest that allows most endogenous proteins to be labeled by off-the-shelf antibodies. Iterative expansion allows for the highest spatial resolution ($\sim 25\mathrm{nm}$). Imaging on a standard confocal microscope provides the following resolutions and limitations.

Table 1

Summary of expansion microscopy techniques in C. elegans. Table constructed using information from Yu et al.37