Click-ExM enables expansion microscopy for all biomolecules

Expansion microscopy (ExM) allows super-resolution imaging on conventional fluorescence microscopes, but has been limited to proteins and nucleic acids. Here we develop click-ExM, which integrates click labeling into ExM to enable a ‘one-stop-shop’ method for nanoscale imaging of various types of biomolecule. By click labeling with biotin and staining with fluorescently labeled streptavidin, a large range of biomolecules can be imaged by the standard ExM procedure normally used for proteins. Using 18 clickable labels, we demonstrate click-ExM on lipids, glycans, proteins, DNA, RNA and small molecules. We demonstrate that click-ExM is applicable in cell culture systems and for tissue imaging. We further show that click-ExM is compatible with signal-amplification techniques and two-color imaging. Click-ExM thus provides a convenient and versatile method for super-resolution imaging, which may be routinely used for cell and tissue samples. Click-ExM uses click-chemistry-based labeling to increase the versatility of expansion microscopy. Click-ExM enables imaging of numerous classes of biomolecules including lipids, glycans, proteins, DNA, RNA and small molecules.

B y physically expanding proteins or RNA of fixed specimens embedded in a swellable polymer hydrogel, expansion microscopy (ExM) enables nanoscale imaging on conventional diffraction-limited microscopes [1][2][3][4][5][6] . As a critical step of ExM, the biomolecules or the labeled fluorophores need to be covalently anchored into the polymer network during gelation. Furthermore, they have to survive the step of homogenization of the fixed cells (for example, by strong protease digestion), an essential step to ensure isotropic expansion. Although tailored protocols have been developed for ExM imaging of proteins and RNA, other types of biomolecule such as lipids, glycans and small molecules remain challenging to image by ExM.
To expand the applicability of ExM, we herein develop click-ExM, a variant of ExM that combines click labeling and ExM to enable super-resolution imaging of various kinds of biomolecules in a unified procedure (Fig. 1). We demonstrate click-ExM for imaging lipids, glycans, proteins, nucleic acids and small molecules. Furthermore, click-ExM imaging of lipids and proteins is performed on the mouse brain tissue. In addition, signal-amplification techniques can be implemented in click-ExM to overcome the issue of volumetric dilution of fluorescence intensity induced by expansion. Finally, two-color click-ExM imaging is demonstrated.

Development of click-ExM.
To develop a unified protocol for labeling, anchoring and preserving fluorescent signals for different biomolecules, we exploited click labeling, which has emerged as a versatile tool for fluorescence imaging of various kinds of biomolecule 7 . DNA, RNA, proteins, glycans and lipids can all be metabolically labeled with a bioorthogonal or 'clickable' functional group (for example, azide or alkyne), which is subsequently conjugated with fluorescent probes via bioorthogonal chemistry such as Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC or click chemistry) 8,9 and copper-free click chemistry 10 ( Supplementary  Fig. 1). In addition, bioorthogonal functional groups can also be site-specifically incorporated into proteins by using strategies such as genetic code expansion 11 . In click-ExM, the metabolically or genetically labeled biomolecules are reacted with azide-biotin or alkyne-biotin via click chemistry, followed by staining with fluorescently labeled streptavidin ( Fig. 1 and Supplementary Fig. 2). The streptavidin-fluorophore conjugates serve as a trifunctional probe, which binds biotin, anchors into the gel and presents fluorescence. Glutaraldehyde (GA) or the succinimidyl ester of 6-((acryloyl)amino)hexanoic acid (AcX) reacts with the lysine residues of streptavidin and provides the anchoring groups. Up to 14 acryloyl groups were conjugated onto streptavidin by using AcX as the anchoring agent, ensuring efficient incorporation during gelation ( Supplementary Fig. 3a,b). The core structure of streptavidin was resistant to strong protease digestion (that is, by proteinase K) ( Supplementary Fig. 3c), consistent with the previously observed preservation of streptavidin in protein-retention ExM (proExM) 4 . These results indicate that streptavidin can be readily cross-linked into the gel and survive through homogenization and expansion.
The proExM protocol was adapted for click-ExM 4,5 (Extended Data Fig. 1). The expansion factor was determined by two methods. The same cells were imaged both pre-and postexpansion, from which the expansion factor, termed sEF, was calculated either from the transformation matrix (for samples with rigid registration) or distances of paired landmarks (for samples without registration) (Extended Data Fig. 1a,g). Alternatively, the expansion factor was quantified by measuring the gel size before and after expansion, which was termed gEF (Extended Data Fig. 1j). In this work, gEF was kept constant (4.5 ± 0.1) for all click-ExM and proExM experiments.

Click-ExM imaging of lipids.
We first demonstrated click-ExM for imaging lipids, a major biomacromolecule that has not been visualized by ExM. COS-7 cells were treated with alkyne-choline (Alk-Cho) to metabolically label Cho-containing phospholipids, the most abundant phospholipids in cellular membranes 12 (Fig. 2a). After reacting the incorporated Alk-Cho with azide-TAMRA, the membranes could be visualized by confocal microscopy before expansion, but the fluorescence was completely lost after the standard ExM procedure (Extended Data Fig. 2a,b). Membrane staining confirmed that lipids could not be retained through the expansion process in ExM (Extended Data Fig. 2c). We therefore applied click-ExM, in which the Alk-Cho-treated cells were fixed and permeabilized with saponin, followed by click reaction with azide-biotin, staining with streptavidin-Alexa Fluor 555 (AF555), and AcX or GA treatment (Fig. 1). After gelation, digestion and expansion, the AF555 fluorescence was well preserved and the Alk-Cho-incorporated membranes were visualized with super resolution (Fig. 2 and Extended Data Figs. 2d and 3a-i). The click-ExM procedure preserved the fluorescence signal with high efficiency, approximately 60%, which was comparable to proExM (Supplementary Note 1). Furthermore, the lipid structures resolved by click-ExM agreed well with those observed by conventional super-resolution techniques (Supplementary Note 1).
By costaining of the mitochondrial markers, the distribution of Cho-containing phospholipids on the outer membrane of mitochondria was clearly observed by multicolor proExM and click-ExM (Fig. 2b,c and Extended Data Fig. 3a). Similarly, the structures of endoplasmic reticulum and Golgi apparatus could be resolved by click-ExM ( Fig. 2d-g and Extended Data Fig. 3b,c). For click-ExM of Cho-containing phospholipids, sEF determined by imaging the same cells pre-and postexpansion was 4.4, in good agreement with gEF (Extended Data Fig. 3d,e). The root mean square (r.m.s.) error of feature measurements over length scales from 0 to 7.5 μm after click-ExM were within 2% of the measurement distance, which demonstrated high isotropy of click-ExM (Extended Data Fig. 3f).
We then applied click-ExM to rat cardiomyocytes treated with Alk-Cho, in which the network of interconnected and highly dense mitochondria were visualized (Fig. 2h-j and Extended Data Fig. 3g-i). Furthermore, click-ExM was applied to the brain tissue.
Acute mouse brain slices were metabolically labeled with Alk-Cho, and click-ExM revealed membraneous structures of cellular networks, which were continuously traced (Extended Data Fig. 4).
Various lipid reporters including Alk-palmitic acid, Alk-farnesol, Alk-myristic acid and Alk-stearic acid have been developed for labeling fatty acids and prenol lipids [13][14][15][16] . Click-ExM using these lipid reporters enabled super-resolution imaging of the distribution of the corresponding fatty acids or prenol lipids in COS-7 and HeLa cells (Extended Data Fig. 3j). In addition, protein carbonylation by 4-hydroxy-2-nonenal (HNE), a lipid-derived electrophile that reacts with cysteine residues, was visualized by click-ExM by conjugating the resulting aldehyde group on proteins with aminooxy-alkyne 17 (AOyne, Extended Data Fig. 3k). These results demonstrate the broad use of click-ExM in lipid imaging.
Signal amplification for click-ExM. The volumetric dilution of fluorophores in ExM reduces the signal intensity, a disadvantage that can be overcome by signal-amplification techniques. Because amplification by immunosignal hybridization chain reaction (isHCR) has been demonstrated on streptavidin with the biotin-DNA HCR initiator conjugate 18 , click-ExM is compatible with isHCR (Extended Data Fig. 5a). Furthermore, we developed a controllable signal-amplification protocol by exploiting the multivalence of streptavidin (Extended Data Fig. 5b). By synthesizing a biotin trimer (Supplementary Note 2), streptavidin staining could be iteratively performed with the signal gradually increased in each round of amplification (Extended Data Fig. 5c).
Click-ExM imaging of glycans. Next, we applied click-ExM for imaging glycans, another major biomacromolecule not yet amenable to ExM. The sialoglycans on the membrane of HeLa cells were metabolically labeled with azido sialic acid (SiaNAz, Extended Data Fig. 6a) or N-azidoacetylmannosamine (ManNAz) (Fig. 3a), the metabolic precursor of SiaNAz 19,20 . Of note, unprotected azido sugars were used in this work to avoid nonspecific S-glyco-modification induced by per-O-acetylation 21,22 . Click-ExM revealed that the SiaNAz-labeled glycans were widely distributed on the basal plasma membrane (Extended Data Fig. 6b). The sialoglycans on the hollow membrane protrusions such as microvilli were clearly resolved by the nanoscale resolution of click-ExM (Fig. 3b).
These observations were consistent with earlier studies on sialoglycans using localization-based super-resolution microscopy [23][24][25] . Click-ExM imaging of sialoglycans was then performed on rat hippocampal neurons treated with ManNAz. With click-ExM, the tube-like structure of neurites with a diameter of roughly 500 nm was readily and clearly delineated by sialoglycans on the membrane (Fig. 3c, top). By contrast, the tubular structure was unresolvable before expansion (Fig. 3c, bottom). In rat cardiomyocytes that had a relatively low labeling with ManNAz, click-ExM with isHCR amplification revealed sialoglycans on the orderly spaced transverse tubule (T-tubule) network (Fig. 3d, top), with a much better resolution than conventional confocal microscopy 26 (Fig. 3d, bottom and Extended Data Fig. 6c). We then applied click-ExM for imaging O-GlcNAcylation, modification of serine and threonine residues with an N-acetylglucosamine (GlcNAc) monosaccharide, which occurs on various intracellular proteins 27 . HeLa cells were treated with N-azidoacetylgalactosamine (GalNAz) (Fig. 3a), which metabolically labeled both cell-surface glycans and O-GlcNAc 28,29 . After click reaction with alkyne-biotin and staining with streptavidin-dye conjugates, fluorescence was observed in the cytoplasm and nucleus, presumably from the labeled O-GlcNAc ( Fig. 3e and Extended Data Fig. 6d). Nucleoporins, the constituent building blocks of the nuclear pore complex (NPC), are heavily O-GlcNAcylated and can be stained by the O-GlcNAc-recognizing lectin wheat germ agglutinin, which has been used for super-resolution imaging of the central channel of NPC 30 . Heavy GalNAz labeling was colocalized with NPC staining on the nuclear envelop (Extended Data Fig. 6e). Using click-ExM, individual NPC was resolved with a diameter of around 113.5 nm by metabolic O-GlcNAc labeling ( Fig. 3f-h). Moreover, click-ExM of ManNAz-and GalNAz-labeled cells exhibited high isotropy (Extended Data Fig. 6f-k).
Click-ExM imaging of O-GlcNAc could also be performed by using the chemoenzymatic method based on a mutant galactosyltransferase (Y289L GalT1) to label O-GlcNAc with GalNAz 31 (Extended Data Fig. 6l). In addition, click-ExM in combination with chemoenzymatic labeling was generally applicable for nanoscale imaging of other glycans, such as the cancer-associated N-acetyllactosamine (LacNAc) on the surface of HeLa cells 32 (Extended Data Fig. 6m). Of note, the LacNAc level was relatively low and the signal was amplified by the iterative biotin-streptavidin staining method.

Click-ExM imaging of proteins.
To demonstrate the applicability of click-ExM for protein imaging, U2OS and HeLa cells, rat cortical neurons and acute mouse brain slices were treated with azidohomoalanine (AHA) (Fig. 4a), a noncanonical amino acid serving   as a surrogate for methionine, to metabolically label nascent proteins 33 . Click-ExM imaging revealed the distribution of newly synthesized proteins in both cell and tissue samples with high resolution (Fig. 4b and Extended Data Fig. 7a-e). While the streptavidin-based click-ExM serves a unified protocol for various biomolecules, click-ExM could also be performed by direct click labeling with fluorophores for those that can be anchored into the gel, such as proteins (Extended Data Fig. 7b). Furthermore, click-ExM is applicable for imaging specific proteins. Using the genetic code expansion strategy based on the pyrrolysine system, three proteins including geen fluorescent protein (GFP), histone H2B and GTPase KRas were site-specifically incorporated with an alkyne-containing pyrrolysine analog 34 (Extended Data Fig. 7f). Click-ExM resolved the cellular distributions of these proteins (Extended Data Fig. 7g).

Click-ExM imaging of nucleic acids.
For nucleic acid imaging, we used 5-ethynyl-2′-deoxyuridine (EdU) to metabolically label nascent DNA 35 . Under click-ExM, the fine structure of chromatin in cell nuclei was resolved with low distortion (Fig. 4c and Extended Data Fig. 8a-

Fig. 4 |
Click-ExM imaging of proteins, nucleic acids and small molecules. a, Chemical structure of AHA. b, Acute mouse brain slices with a thickness of 150 µm were metabolically labeled with AHA. Postclick-ExM image with an imaging depth of around 36 μm (left) and 3D rendering images of postclick-ExM (middle and right) are shown. Nascent proteins were stained with SA-AF488 (left) or with SA-AF555 (middle and right). Images were acquired using a ×10/0.75-NA air objective (left) or a ×40/1.10-NA water objective (middle and right). The nuclei were stained with Hoechst 33342 (cyan, middle and right). Representative images from three or more slices. c, Chemical structure of EdU, and pre-and postclick-ExM images of nascent DNA (AF488, red hot) in EdU-treated HeLa cells. d, Chemical structure of EU, and pre-and postclick-ExM images of nascent RNA (AF555, red hot) in EU-treated HeLa cells. e, Chemical structure of Az-afatinib, and pre-and postclick-ExM images of HeLa cells treated with Az-afatinib (AF555, red hot). f, Chemical structure of OP-Puro, and pre-and postclick-ExM images of nascent peptides (AF488, red hot) in OP-Puro-treated HeLa cells. g, Chemical structure of Alk-Hoechst, and pre-and postclick-ExM images of DNA (AF488, red hot) in Alk-Hoechst-labeled HeLa cells. AcX was used for anchoring. Scale bars, 100 μm (b left), 10 μm (b middle and right, e,f) and 5 μm (c,d,g). All distances and scale bars correspond to the pre-expansion dimension based on sEF. Representative images from three or more independent experiments. of heterochromatin and euchromatin was differentiated. During S phase, EdU labeling and click-ExM clearly revealed the boundaries of replication domains, which are coreplicating Mb-sized segments of chromosomal DNA (Extended Data Fig. 8d). By using multicolor click-ExM, we showed that the newly synthesized DNA was largely colocalized with histone H3 (Extended Data Fig. 8e). In addition, nascent RNA in HeLa cells was metabolically incorporated with 5-ethynyluridine (EU) 36 and visualized by click-ExM with little distortion ( Fig. 4d and Extended Data Fig. 8f-h). In the nucleus, intense labeling in nucleoli, presumably from preribosomal RNA, and diffused labeling in the nucleoplasm, presumably from messenger RNA, were observed at nanoscale resolution.
Click-ExM imaging of small molecules. In addition to biomacromolecules, we envisioned that click-ExM would enable super-resolution imaging of small molecules, including drugs, metabolites and chemical probes, in the cells. The small molecules should tolerate modification with an azide or alkyne without interfering with their function and cellular localization. Afatinib is an irreversible inhibitor for ErbB family and has been approved by the Food and Drug Administration to treat metastatic nonsmall cell lung cancer with nonresistant epidermal growth factor receptor mutations 37 . We installed an azide at the N-methyl position so that the activity of afatinib was not affected. Click-ExM revealed the distribution of the azide-functionalized afatinib (Az-afatinib) in HeLa cells, and plasma membrane blebbing during afatinib-induced cell apoptosis was observed ( Fig. 4e and Extended Data Fig. 9a-c). Puromycin is an aminonucleoside antibiotic that terminates protein synthesis by covalent conjugation with nascent polypeptide chains. Here, an alkyne-containing puromycin analog, O-propargyl-puromycin 38 (OP-Puro), was used for click-ExM. Intense labeling was observed in various cytoplasmic foci, which probably resulted from rapid degradation of misfolded peptides by proteasome ( Fig. 4f and Extended Data Fig. 9d-f). The widely used DNA stain Hoechst 33258 was derivatized with an alkyne, with which the nuclei were visualized with AF488 by click-ExM ( Fig. 4g and Extended Data Fig. 9g-j). The click-ExM observed distribution of small molecules were confirmed by conventional super-resolution microscopies (Supplementary Note 1).

Two-color click-ExM.
Finally, we demonstrated two-color click-ExM. Lipids and proteins in COS-7 cells were simultaneously labeled with Alk-Cho and AHA, respectively. After sequential click reactions with azide-biotin and alkyne-TAMRA, two-color click-ExM revealed the distributions of Cho-containing phospholipids and nascent proteins (Extended Data Fig. 10a). By pairing azide-biotin/streptavidin with alkyne-digoxigenin (DIG)/anti-DIG antibody, we demonstrated two-color click-ExM imaging of lipids and glycans (Extended Data Fig. 10b). Similarly, azide-FLAG/ anti-FLAG antibody was used for two-color click-ExM imaging of glycans and DNA (Extended Data Fig. 10c).

Discussion
ExM has emerged as an attractive approach to perform superresolution imaging using conventional microscopes. Click-ExM developed in this work provides a unified and convenient protocol for nanoscale imaging of any biomolecule that can be labeled with click chemistry, thus expanding the applicability of ExM to molecules not being imaged before, such as lipids and glycans. Of note, developing ExM protocols compatible with lipid imaging is currently a direction of great interest 39 . Click-ExM exploits the click-labeling method, which is applicable for various biomolecules and is advancing rapidly. By converting the click labels to streptavidin-dye conjugates, click-ExM adapts the well-established proExM workflow (a more detailed discussion is included in Supplementary Note 1). Furthermore, click-ExM is compatible with various signal-amplification techniques. In addition, click-ExM is compatible with multicolor imaging, and a promising future direction is to integrate different click reactions to enable visualization of more than two kinds of biomolecules in the same sample.

online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41592-020-01005-2.  Neuron isolation and culture. Hippocampal and cortical neurons were dissociated from postnatal day 0 rat pups. Briefly, the hippocampus and cortex were dissected from the brain and triturated using a pipette after trypsinization (0.25% (w/v) trypsin-EDTA at 37 °C, 7 min for the cortex and 14 min for the hippocampus). The cells were seeded on 15-mm poly-d-lysine-coated coverslips in 24-well plates and cultured in Neurobasal medium supplemented with 2% (v/v) B-27 supplement, 0.5 mM GlutaMAX, 100 u ml −1 of penicillin and 100 μg ml −1 of streptomycin. One day after plating, 4 μM Ara-C was added to the cultures. The cultures were maintained at 37 °C in 5% (v/v) CO 2 atmosphere, and one-half of medium was replaced with fresh culture medium every 3-4 d. After three washes with PBS, the cells were labeled by incubation with 50 μM alkyne-PEG 4 -biotin, BTTAA-CuSO 4 complex (50 μM CuSO 4 , BTTAA/CuSO 4 6:1, mol/mol) and 2.5 mM sodium ascorbate in PBS for 10 min, and fixed with the same procedures as described in the SiaNAz or ManNAz labeling experiments. Iterative biotin trimer-streptavidin staining was performed for signal amplification (detailed procedures described below).
Metabolic labeling of proteins. The cells were incubated with 2 mM AHA for 1 h (for rat cortical neurons and U2OS cells) or 1 mM AHA for 18 h (for HeLa cells). After three washes with PBS, the cells were fixed, permeabilized and labeled with the same procedures as described in the GalNAz labeling experiments.
Site-specific labeling of proteins. To the plasmid encoding a protein of interest bearing the amber codon (pCDNA3.1-GFP-Y40TAG, pCMV3-KRAS-G21V-I21TAG (a gift from P. R. Chen at Peking University), or pEGFP-H2B-V112TAG) mixed with the plasmid encoding PenK-MbPylRS-tRNA pyl CUA I (CUA, the anticodon of the amber codon; Pyl, pyrrolysine; a gift from P. R. Chen at Peking University) in Opti-MEM (plasmid ratio 1:1) was added X-tremeGENE HP (Roche) at the ratio of 500 ng total plasmids to 1 μl X-tremeGENE HP. The solution was incubated for 15 min and applied to HEK293T or COS-7 cells grown on poly-d-lysine-coated eight-well glass chamber slides (Nunc, 154534) with around 60% confluence. After the cells were cultured for 6 h, the cell culture medium was changed to fresh medium supplemented with 200 μM PenK and the cells were further grown for 24 h. After washing three times with fresh medium (1 h each), the cells were fixed and permeabilized as described in the lipid labeling experiments (for the KRas-expressing cells) or as described in the GalNAz labeling experiments (for the GFP-and H2B-expressing cells). The cells were then click-labeled with azide-PEG 3 -biotin and stained with streptavidin-dye conjugates.

Metabolic labeling of nucleic acids. For DNA labeling, the cells were incubated with 10 μM EdU for 24 h (for U2OS cells) or 100 μM EdU for 12 h (for HeLa cells).
For nascent DNA labeling in the S phase, U2OS cells were labeled with 10 μM EdU for varied durations of time ranging from 15 min to 1 h. For RNA labeling, HeLa cells were labeled with 1 mM EU for 4 h. All the cells were fixed with 4% (w/v) formaldehyde in PBS for 15 min. After three washes with PBS, the cells were permeabilized with 0.1% (v/v) Triton X-100 in PBS for 15 min. The cells were incubated with 100 μM azide-PEG 3 -biotin, 500 μM CuSO 4 , 2 mM THPTA and 5 mM sodium ascorbate in PBS at room temperature for 1 h, followed by five washes with PBS. The cells were incubated with 5 μg ml −1 of streptavidin-dye conjugates in PBS containing 1% (w/v) BSA for 1 h and washed three times with PBS.
Labeling of small molecules. HeLa cells were treated with 10 μM Az-afatinib in DMEM for 12 h. After three washes with PBS, the cells were fixed and permeabilized as described in the lipid labeling experiments and labeled as described in the GalNAz labeling experiments.
For OP-Puro, HeLa cells were treated with 50 μM OP-Puro in DMEM for 2 h. After washing with warm DMEM for 15 min, the cells were fixed, permeabilized as described in the GalNAz labeling experiments and labeled as described in the lipid labeling experiments.
For Alk-Hoechst, HeLa cells were treated with 0.5-20 μM Alk-Hoechst for 45 min and washed three times with warm DMEM for 1 h, followed by labeling using the same procedures as described in the nucleic acid labeling experiments.
Gelation, digestion and expansion. ExM was performed as previously described with some modifications 4,5 . For anchoring with AcX, the cells are incubated in 0.1 mg ml −1 of AcX in PBS overnight at room temperature, followed by washing three times with PBS. For anchoring with GA, the cells are incubated with 0.25% (v/v) GA in PBS for 10 min, followed by washing three times with PBS. The monomer solution (1× PBS, 2 M NaCl, 2.5% (w/v) acrylamide, 0.15% (w/v) N,N′-methylenebisacrylamide and 8.625% (w/v) sodium acrylate) was prepared, frozen in aliquots and thawed before use. Freshly prepared 10% (w/w) N,N,N′, N′-tetramethylethylenediamine (TEMED) and 10% (w/w) ammonium persulfate (APS) were diluted in monomer solution to a final concentration of 0.2% (w/w) to give the gelation solution. The cells were incubated with the gelation solution at 4 °C for 5 min, and then transferred to a humidified 37 °C incubator for 1 h for gelation. The hydrogel was digested in the digestion buffer (50 mM Tris, 1 mM EDTA, 0.1% (v/v) Triton X-100 and 0.8 M guanidine HCl, pH 8.0) containing 8 u ml −1 proteinase K at 37 °C for 4 h for AcX-anchored samples or 2 h for GA-anchored samples. The hydrogel was transferred to deionized water to expand. Water was changed every 20 min until the expansion process was completed. The nuclei were stained with 5 µg ml −1 of Hoechst 33342 to facilitate locating the cells in the hydrogel. isHCR amplification. isHCR amplification was performed as previously described with some modifications 18 . A pair of DNA-fluorophore HCR amplifiers (100 μM each) in 5× SSC buffer were separately snap-cooled at 95 °C for 90 s and then left at room temperature over 30 min. The streptavidin-dye-stained cells were incubated with 0.5 µM DNA-biotin HCR initiator at room temperature for 1 h. The cells were then washed five times with PBS and incubated in amplification buffer (5× SSC buffer, 0.1% (v/v) Tween 20 and 10% (w/v) dextran sulfate in ddH 2 O) containing the pair of DNA-fluorophore HCR amplifiers (150 nM each) overnight at room temperature, followed by five washes with PBS.
The sequences of DNA-fluorophore HCR amplifiers and DNA-biotin HCR initiator: Gelation, digestion and expansion of mouse brain slices. The brain slices were treated with 0.1 mg ml −1 of AcX in a 2-(N-morpholino)ethanesulfonic acid (MES)-buffered saline (100 mM MES and 150 mM NaCl, pH 6.0) for 24 h at room temperature, followed by three washes with PBS. The gelation solution for brain slices was the same as the one for cell samples except for the addition of 0.01% (w/w) 4-hydroxy-TEMPO from a 0.5% (w/w) stock as an inhibitor for polymerization. The brain slices were incubated in gelation solution for 1 h at 4 °C before gelation. The gelation chamber was constructed by placing two stacks of three pieces of no. 1.5 coverglass (high-precision, Fisher Scientific) on a glass slide. The samples were allowed to gel in fresh gelation solution for 2.5 h at 37 °C. Excess gel around the samples was trimmed off by a razor blade, and the samples were digested in digestion buffer with 8 u ml −1 of Proteinase K overnight at room temperature or for 6 h at 37 °C. The hydrogel was transferred into deionized water to expand. The nuclei were stained with 5 µg ml −1 of Hoechst 33342.

Two-color click-ExM imaging.
For two-color imaging of lipids and proteins, COS-7 cells were simultaneously treated with 100 μM Alk-Cho and 1 mM AHA for 12 h. The cells were fixed and permeabilized as described in the lipid labeling experiments. The cells were reacted with azide-PEG 3 -biotin and stained with streptavidin-AF488. After washing three times with 0.1% (w/v) saponin in PBS, the cells were postfixed with 4% (w/v) formaldehyde in PBS for 10 min. The cells were then reacted with alkyne-TAMRA, followed by three washes with 0.2% (v/v) Tween 20 in PBS.
For two-color imaging of lipids and glycans, COS-7 cells incubated with 2 mM ManNAz for 36 h, to which 100 μM Alk-Cho was added. The cells were incubated for another 12 h. After three washes with PBS, the cells were click-labeled with alkyne-DIG at room temperature for 10 min and washed with PBS. The cells were fixed and permeabilized as described in the lipid labeling experiments. The cells were reacted with azide-PEG 3 -biotin and stained with streptavidin-AF555. After postfixing with 4% (w/v) formaldehyde in PBS for 10 min, the cells were immunostained with DyLight 488-conjugated anti-DIG antibody. The cells were washed three times with 0.2% (v/v) Tween 20 in PBS.
For two-color imaging of glycans and DNA, HeLa cells were incubated with 1 mM GalNAz for 24 h, to which 10 μM EdU was added. The cells were incubated for another 24 h. The cells were fixed and permeabilized as described in the GalNAz labeling experiments. The cells were reacted with alkyne-PEG 4 -biotin and stained with streptavidin-AF488. After five washes with 0.2% (v/v) Tween 20 in PBS, the cells were reacted with azide-FLAG and immunostained with anti-FLAG primary antibody and AF555-conjugated secondary antibody. The cells were washed three times with 0.2% (v/v) Tween 20 in PBS.

Quantification of fluorescence retention during click-ExM.
Epifluorescence images of the click-labeled cells were acquired on a Nikon Eclipse Ts2R-FL inverted microscope. Following AcX treatment, click-ExM gelation and digestion, the samples were washed three times with digestion buffer (without Proteinase K). The epifluorescence images were taken again on the same cell regions with identical imaging conditions. For intensity measurements on each sample before and after processing, the same cell regions were segmented and integrated intensity was measured to account for fluorophore dilution.
Microscopy. Imaging acquisition was performed within 24 h after expansion for all samples. To minimize drift of expanded samples during data acquisition, 24 × 50 mm rectangular no. 1.5 coverglass was coated with 0.1 mg ml −1 of poly-d-lysine for 10 min at room temperature and allowed to dry. Excess water was carefully removed using laboratory wipes, and expanded samples were stably attached to the glass surface and maintained immobile during data collection.
Confocal microscopy. Confocal imaging was mainly performed on a Leica SP8X laser scanning confocal system with a HC PL APO CS2 ×63/1.40-NA oil-immersion objective. An HC PL APO CS2 ×40/1.30-NA oil-immersion objective was also used for imaging (Figs. 2b-j and 3c and Extended Data Figs. 3a-c,g-k (except Alk-farnesol group in 3j), 5c and 6e). HC PL APO CS2 ×10/0.40-NA air objective, HC PL APO CS2 ×20/0.75-NA air objective and HC PL APO CS2 ×40/1.10-NA water objective were used for tissue imaging (Fig. 4b and Extended Data Fig. 4). White light laser (470-670 nm) was set to 70% of maximum power, a 405-nm diode laser was used for Hoechst 33342 excitation and 488-nm and 552-nm OPSL lasers were also used for excitation. Some samples were imaged on a Zeiss LSM 700 laser scanning confocal system with a Plan-Apochromat ×63/1.40-NA DIC M27 oil-immersion objective (Extended Data Fig. 6b,d). The pinhole during all the image acquisition was opened at 1 Airy unit. Laser powers and detector gain were optimized for each sample.
Structured illumination microscopy. Lattice structured illumination microscopy imaging was performed on a Zeiss Elyra 7 with a Plan-Apochromat ×63/1.40-NA DIC M27 oil-immersion objective. The system was controlled by the ZEN software (Zeiss, black edition). AF555-labeled samples were excited using a 561-nm laser with laser powers optimized for each sample. Emission was collected through a 570-620-nm filter and a PCO edge sCMOS camera. Image processing was performed by ZEN using automatic parameters.
Airyscan microscopy. A Zeiss LSM 900 with an Airyscan 2 detector was used for image acquisition in combination with a Plan-Apochromat ×63/1.40-NA DIC M27 oil-immersion objective. The system was controlled by the ZEN software (Zeiss, blue edition). AF555-labeled samples were excited using a 561-nm laser with laser powers optimized for each sample. Emission was collected in Airyscan SR detector mode. Image processing was performed by ZEN Airyscan processing using automatic deconvolution parameters.
Stimulated emission depletion (STED) microscopy. STED imaging was performed on a Leica TCS SP8 STED 3X STED microscope with an HCX PL APO CS2 ×100/1.40-NA oil-immersion objective. The system was controlled by the LAS X software (Leica). AF488-labeled samples were excited using a 495-nm laser and a 592-nm STED laser was used for depletion. The pinhole size was set to 0.4 of an Airy unit. Laser powers were optimized for each sample. Emission was collected with a HyD detector between 505 and 581 nm.
Imaging processing and analysis. Images were mainly processed and analyzed by Fiji/ImageJ (https://fiji.sc) 41 . To improve signal visibility, images were only contrasted by changing the minimum and maximum displayed values. Maximal intensity projection of images was used in most cases. All intensity profiles were measured and normalized over a line across the region of interest. For intensity profile in Fig. 3g, the intensity data were normalized and multi-peak fitted to a Gaussian function. For full-width-at-half-maximum (FWHM) measurement in Extended Data Fig. 1e,f, intensity profiles were measured over lines perpendicular to the microtubule orientation. The data were fitted to a Gaussian function and FWHM was calculated from the Gaussian fitting. For size quantification in Fig. 3h,  Fig. 3f was first segmented using automatic Otsu algorithm, and minimum Feret diameter was then measured with 'analyze particles' (size between 0.1-0.2 μm 2 , postexpansion unit). The preclick-ExM image in Fig. 4d was resized to the same size of postclick-ExM image by bicubic interpolation. For intensity quantification in Supplementary Note Fig. 1b, postclick-ExM images were first matched with corresponding preclick-ExM images by cropping and rotation. The images were then segmented using automatic Otsu algorithm, and the integrated intensity was measured in the segmented areas.
The three-dimensional (3D) visualization of images was performed in Imaris (Bitplane) with MIP mode (Figs. 2h, 3b and 4b middle and right and Extended Data Figs. 4b,c and 6d). Images in Extended Data Fig. 6b were deconvolved by the Huygens software (Scientific Volume Imaging) using theoretical point spread functions and water as 'Mounting Medium' . STED images in Supplementary Note Fig. 2 were deconvolved by the Lightning adaptive approach (Leica) using theoretical point spread functions and Glycerol/Water (80:20, Refractive index 1.4429) as 'Mounting Medium' .
Statistical analysis and curve plotting were performed with Prism (GraphPad) and Origin Pro (OriginLab). Chemical structures were drawn using ChemDraw Professional (PerkinElmer). Crystal structure was visualized using PyMOL Molecular Graphics system (Schrödinger, LLC). All graphs were assembled with Adobe Illustrator.
Expansion factor calculation and distortion analysis. gEF was quantified by the sizes of the pre-and postexpanded gels: , where x EF and y EF are the post-/pre-ratios of gel length and width, respectively. For brain slices, the nuclei were stained using Hoechst 33342 and expansion factor was quantified by the distance of same nucleus landmarks before and after expansion with widefield microscopy equipped with low-magnified objectives.
Rigid registration and sEF calculation. Image registration and sEF calculation procedures were adapted from a previous report with some modifications 5 , which were performed using MATLAB R2018a (MathWorks). The pre-expansion images were first stretched by a scaling factor (F esti ) of 3.8-4.8 with bicubic interpolation to approximately match the scale of the postexpansion images. This estimated expansion factor was roughly decided based on the macroscopic measurement of gel sizes. The stretched pre-expansion images were used as the 'fixed image' for rigid registration, and the postexpansion images as the 'moving image' . Rigid registration was performed using the monomodal intensity-based registration with imregtform function. Affine transformation was used in the function to correct the shear deformation of the gel. The x-and y-scaling elements (x corr and y corr ) in the transformation matrix calculated by the imregtform function were used to correct the rough scaling factor (F esti ). The final expansion factor (sEF) was therefore calculated as sEF ¼ F esti´ffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi x corr´ycorr p I .
Nonrigid B-spline registration and distortion analysis. The registered pre-and postexpansion images were then subjected to B-spline nonrigid registration, which was adapted from a previous report with some modifications 5 . To exclude regions with no features, masks of the rigid registered pre-and postexpansion images were generated by Gaussian blur to subtract the background. The displacement field and B-spline registered images were then acquired with the imregdemons function, and the distortion vector fields were plotted with the quiver function. The r.m.s. error was quantified by calculating the difference of distance between each pair of matching features before and after B-spline registration, and plotted as a function of the distance between the matching features.
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
Source data for Figs. 2c,e,g,j and 3d,g,h and Extended Data Figs. 1c,e,f,i,j, 3f, 5c, 6h,k, 7e, 8c,h and 9c,f,j are available with this paper. All other data in this study are available from the corresponding author upon reasonable request.

Code availability
The MATLAB codes and examples for registration and distortion analysis are available from GitHub (https://github.com/Yujie-S/Click-ExM_data_process_ and_example).  . 2). a, Pre-and post-click-ExM images of Alk-Cho-labeled phospholipids (AF555, magenta) in COS-7 cells. Mito-GFP (green) was expressed in the mitochondrial matrix and Tom20 located in the mitochondrial outer membrane was immunostained with AF647 (red). b, Pre-and post-click-ExM images of Alk-Cho-labeled phospholipids (AF555, magenta) and ER expressing Sec61β-GFP (green) in COS-7 cells. c, Pre-and post-click-ExM images of Alk-Cho-labeled phospholipids (AF555, magenta) and Golgi expressing b4Gal-T1-GFP (green) in COS-7 cells. The same post-click-ExM images were shown in Fig. 2. d-f, Nanoscale isotropy of click-ExM for imaging of Alk-Cho-labeled phospholipids. Pre-and post-click-ExM images (d, AF488, red hot) were aligned using rigid registration. Post-click-ExM images before and after non-rigid registration were overlaid and shown in green and magenta, respectively (e). The yellow arrows in e indicate the distortion vector field. From the boxed region in e, the RMS error was quantified as a function of measured distance (f), in which the black line indicates mean value and the grey area indicates ± s.d. g, Pre-and post-click-ExM images of Alk-Cho-labeled phospholipids (AF555, cyan hot) in rat cardiomyocytes. (h) Zoomed-in view of the boxed region from the pre-click-ExM image in g. i, Zoomed-in view of the boxed region from the post-click-ExM image in g. j, Pre-and post-click-ExM images of lipids in cells metabolically labeled with various bioorthgonal reporters including Alk-palmitic acid, Alk-farnesol, Alk-myristic acid and Alk-stearic acid (AF555, cyan hot). COS-7 cells were used for Alk-farnesol labeling and HeLa cells were used for other lipids labeling. k, Pre-and post-click-ExM images of 4-hydroxy-2-nonenal (HNE)-labeled protein carbonylation (AF555, cyan hot) in HeLa cells. Cysteine residues were reacted with HNE through Michael addition, reacted with aminooxy-alkyne (AOyne), and followed with the click-ExM workflow. GA (a-c, g-i) or AcX (d, e, j, k) was used for anchoring. Scale bars: 10 μm (a-e, g, j, k) and 2 μm (h, i). All distances and scale bars correspond to the pre-expansion dimensions based on gEF (a-c, g-k) or sEF (d-f). Analysis of isotropy from ≥3 (d-f) independent experiments. Representative images from ≥3 (g-i, pre-click-ExM images of j) or 2 (a-c, k, post-click-ExM images of j) independent experiments. Fig. 4 | Click-ExM imaging of Cho-containing phospholipids in the mouse brain. a-c, Acute mouse brain slices with a thickness of 150 µm were metabolically labeled with Alk-Cho. Post-click-ExM image with an imaging depth of ~23 μm (a) and 3D rendering images of post-click-ExM (b, c) are shown. Cho-containing phospholipids were stained with SA-AF555 in a and b or with SA-AF488 in c. Images were acquired using a 20×/0.75-NA air objective for a or a 40×/1.10-NA water objective for b and c. The nuclei were stained with Hoechst 33342 (cyan) in b and c. Scale bars: 20 μm. All distances and scale bars correspond to the pre-expansion dimensions based on sEF. Representative images from ≥3 slices. Fig. 5 | Signal amplification by immunosignal hybridization chain reaction (isHCR) or by a biotin trimer. a, Workflow of isHCR for signal amplification. Cells click-labeled with streptavidin-dye were incubated with DNA-biotin HCR initiators, and then incubated with a pair of DNA-fluorophore HCR amplifiers for signal amplification. b, Workflow of iterative signal amplification by using the biotin trimer. c, Fluorescence imaging of HeLa cells with α-tubulin immunostained with AF555, which was amplified by using the biotin trimer. Procedures of iterative signal amplification (top). HeLa cells were immunostained for α-tubulin, and five labeling conditions were used for comparison from left to right: (i) conventional antibody staining with AF555-conjugated secondary antibody, (ii) biotin-conjugated secondary antibody and streptavidin-AF555, (iii)-(v) one, two and four rounds of iterative staining by biotin trimer and streptavidin-AF555. Representative images of each conditions were shown (middle). Histogram shows mean fluorescence intensity quantification in five labeling conditions (n = 5 fields of view per sample, bottom left). Bars represent the mean value and error bars represent the s.d. The P values are listed, and P < 0.05 was considered significant. One-way ANOVA with post hoc Tukey HSD calculator was used for statistical analysis. Magnified image for four rounds signal amplification of immunostained α-tubulin in HeLa cells was shown (bottom right). Scale bars: 50 μm (c, middle) and 10 μm (c, bottom right). Representative images from 2 independent experiments. Fig. 10 | Multicolor click-ExM imaging of different biomolecules. a, Two-color click-ExM imaging of lipids and proteins. COS-7 cells were simultaneously treated with Alk-Cho and AHA. The alkyne-incorporated lipids were reacted with azide-biotin and stained with SA-AF488 (green). The azide-incorporated proteins were reacted with alkyne-TAMRA (magenta). b, Two-color click-ExM imaging of lipids and glycans. COS-7 cells were simultaneously treated with ManNAz and Alk-Cho. The alkyne-incorporated lipids were reacted with azide-biotin and stained with SA-AF555 (magenta). The azide-incorporated glycans were reacted with alkyne-DIG and immunostained with DyLight 488-conjugated anti-DIG antibody (green). c, Two-color click-ExM imaging of glycans and DNA. HeLa cells were simultaneously treated with GalNAz and EdU. The azide-incorporated glycans was reacted with alkyne-biotin and stained with SA-AF488 (green). The alkyne-incorporated nascent DNA was reacted with azide-FLAG and immunostained with anti-FLAG antibody (AF555, magenta). AcX was used for anchoring. Scale bars: 10 μm. All scale bars correspond to the pre-expansion dimensions based on gEF (b) or sEF (a, c). Representative images from ≥3 (pre-click-ExM images of a, c), 2 (pre-click-ExM images of b and post-click-ExM images of a) or 1 (post-click-ExM images of b, c) independent experiments.