Dense, continuous membrane labeling and expansion microscopy visualization of ultrastructure in tissues

Abstract

Lipid membranes are key to the nanoscale compartmentalization of biological systems, but fluorescent visualization of them in intact tissues, with nanoscale precision, is challenging to do with high labeling density. Here, we report ultrastructural membrane expansion microscopy (umExM), which combines an innovative membrane label and optimized expansion microscopy protocol, to support dense labeling of membranes in tissues for nanoscale visualization. We validate the high signal-to-background ratio, and uniformity and continuity, of umExM membrane labeling in brain slices, which supports the imaging of membranes and proteins at a resolution of ~60 nm on a confocal microscope. We demonstrate the utility of umExM for the segmentation and tracing of neuronal processes, such as axons, in mouse brain tissue. Combining umExM with optical fluctuation imaging, or iterating the expansion process, yields ~35 nm resolution imaging, pointing towards the potential for electron microscopy resolution visualization of brain membranes on ordinary light microscopes.

Fig. 1. Ultrastructural membrane expansion microscopy (umExM) concept and workflow.

umExM is a modified form of expansion microscopy with a custom-designed amphiphilic membrane labeling probe (termed pGk13a). a Chemical structure of pGk13a. The probe does not contain any fluorophore but has an azide to bind a fluorophore later. b umExM workflow. Blue-colored fine text highlight key differences from ExM and proExM, whereas black fine text highlight the same steps as ExM and proExM. b. i A specimen is perfused and chemically fixed with 4% paraformaldehyde (PFA) + 0.5% calcium chloride (CaCl₂) at 4 °C for 24 hours. The brain is sliced on a vibratome to 100 μm thickness at 0-4 °C. b.ii The specimen is treated with pGk13a (structure is depicted in (a)) at 4 °C overnight (unless otherwise noted, overnight means >16 hours). b. iii The specimen is treated with acrylic acid N-hydroxysuccinimide ester (AX) at 4 °C overnight. b. iv The specimen is embedded in an expandable hydrogel (made with N,N’-Diallyl-L-tartardiamide (DATD) crosslinker) at 4 °C for at least 24 hours. b. v The sample (specimen-embedded hydrogel) is chemically softened with enzymatic cleavage of proteins (i.e., non-specific cleavage with proteinase K) at room temperature (~24 °C) overnight. The probe is not digested during proteinase K treatment since it is composed of D-amino acids. b. vi Then, the sample is treated with 1x phosphate-buffered saline (PBS) to partially expand it. The pGk13a, that is anchored to the gel matrix, is fluorescently labeled via click-chemistry (i.e., DBCO-fluorophore) at room temperature, overnight. b. vii The sample is expanded with water at room temperature for 1.5 hours (exchanging water every 30 minutes).

Fig. 2. Resolution and distortion of umExM.

a Representative (n = 3 cells from one culture) single z-plane structured illumination microscopy (SIM) image of a pre-expanded HEK293 cell expressing mitochondrial matrix-targeted green fluorescent protein (GFP, shown in orange). b Single z-plane confocal image of the same HEK293 cell as in (a), after undergoing the umExM protocol, showing expression of mitochondrial matrix-targeted GFP in the same field of view as shown in (a). GFP, green color. c Single z-plane confocal image of the same umExM-expanded fixed HEK293 cell as in (a), showing pGk13a staining of the membrane in the same field of view as shown in (a). pGk13a, gray color. d Root-mean-square (RMS) length measurement error vs. measurement length, comparing pre-expansion SIM and post-expansion confocal images of cells with mitochondrial matrix-targeted GFP (blue line, mean; shaded area, standard deviation; n = 3 cells). e As in (d) but with post-expansion images showing pGk13a staining of the membrane. f Boxplot showing measured expansion factor as described (n = 4 pairs of landmark points; from 3 fixed brain slices from two mice; median, middle line; 1st quartile, lower box boundary; 3rd quartile, upper box boundary; error bars are the 95% confidence interval; black points, individual data points; used throughout this manuscript unless otherwise noted). g Boxplot showing resolution of post-expansion confocal images (60x, 1.27NA objective) of umExM-processed mouse brain tissue slices showing pGk13a staining of the membrane (n = 5 fixed brain slices from two mice). Scale bars are provided in biological units throughout all figures (i.e., physical size divided by expansion factor): (a–c) 5 μm. Source data are provided as a Source Data file.

Fig. 3. Ultrastructure preservation and continuous labeling of membrane with umExM.

a Representative (n = 5 fixed brain slices from two mice) single z-plane confocal image of expanded Thy1-YFP mouse brain tissue (hippocampus, dentate gyrus) showing pGk13a staining of the membrane. pGk13a staining of the membrane visualized in inverted gray color throughout this figure (dark signals on light background) except for (l). b Magnified view of black boxed region in (a). c As in (a) but imaging of the third ventricle. d As in (a) but imaging of mouse somatosensory cortex layer 6 (L6). e Magnified view of black boxed region in (d). f Representative (n = 2 fixed brain slices from two mice) single z-plane confocal image of expanded Thy1-YFP mouse brain tissue (hippocampus dentate gyrus), that underwent umExM protocol and anti-GFP labeling (here labeling YFP), showing YFP (magenta) and pGk13a staining of the membrane (inverted gray). g Diameter of unmyelinated axons (n = 17 axons from three fixed brain slices from two mice). h As in (f), but imaging of somatosensory cortex L6 that was used for measuring the diameter of myelinated axons. (i) Diameter of myelinated axons (n = 21 axons from two fixed brain slices from two mice). j As in (f) but imaging of the third ventricle that was used for measuring the diameter of cilia. k Diameter of cilia (n = 19 cilia from two fixed brain slices from two mice). l (left) Representative (n = 4 slices of fixed brains from three mice) volume rendering of epithelial cells in the third ventricle from mouse brain tissue, showing pGk13a staining of the membrane. pGk13a staining of the membrane visualized in gray color. (right) Magnified view of yellow boxed region in (left). Yellow arrows indicate putative extracellular vesicles. Serial image sections that were used for the 3D rendering are in Supplementary Fig. 15. m Single z-plane confocal image of expanded mouse brain tissue (third ventricle) processed by umExM, showing pGk5b staining (gray), focusing on the plasma membrane of cilia (i.e., ciliary membrane). n Transverse profile of cilia in the yellow dotted boxed region in (m) after averaging down the long axis of the box and then normalizing to the peak of pGk13a signal. o Boxplot showing the percent continuity of the membrane label (n = 5 separate cilia from two fixed brain slices from one mouse), where we define a gap as a region larger than the resolution of the images (~60 nm, from Fig. 2g), over which the pGk13a signal was two standard deviations below the mean of the intensity of pGk13a along the ciliary membrane. a 5 μm, b 2 μm, c 5 μm, d 5 μm, e 5 μm, f 0.25 μm h, j 1 μm, (l, left) (x); 13.57 μm (y); and 7.5 μm (z) (l, right) 3.76 μm (x); 3.76 μm (y); 1.5 μm (z) (m) 2 μm. Source data are provided as a Source Data file.

Fig. 4. umExM with antibody staining and RNA fluorescence in situ hybridization (FISH).

a Representative (n = 5 slices of fixed brain from two mice) single z-plane confocal image of expanded mouse brain tissue (hippocampus, CA3) after umExM processing with a pre-expansion antibody staining protocol (Supplementary Fig. 20), showing immunostaining with an antibody against the synaptic vesicle protein SV2A. b Magnified view of the yellow box in (a). c Single z-plane confocal image of the specimen of (a), showing pGk13a staining of the same field of view as in (a). pGk13a staining of the membrane visualized in inverted gray color throughout this figure. d Magnified view of the yellow box in (c). e Overlay of (a) and (c). f Magnified view of the yellow box in (e). g Representative (n = 5 slices of fixed brain from two mice) single z-plane confocal image of expanded mouse brain tissue (hippocampus, CA1) after umExM processing with a post-expansion antibody staining protocol (Supplementary Fig. 21), showing immunostaining against the post-synaptic density protein PSD-95. h Magnified view of the yellow box in (g). i Single z-plane confocal image of the specimen of (g), showing pGk13a staining of the same field of view as in (g). j Magnified view of the yellow box in (i). k Overlay of (g) and (i). l magnified view of the yellow box in (c). The examples of PSD95 signals that were aligned with pGk13a signals were pinpointed with yellow arrows. m Representative (n = 3 slices of fixed brain from one mouse) single z-plane confocal image of expanded mouse brain tissue (hippocampus, CA1) after umExM processing with a FISH protocol (Supplementary Fig. 23), showing HCR-FISH targeting ACTB. n Single z-plane confocal image of the specimen of (j), showing pGk13a staining of the same field of view as in (j). o Overlay of (m) and (n). Scale bars: (a–c, g, h, j) 5 μm, (d–f, j–l) 1 μm, (m–o) 20 μm.

Fig. 5. Segmentation ability of umExM.

a.i Single z-plane confocal image of expanded Thy1-YFP mouse brain tissue after umExM processing, showing pGk13a staining of the membrane. a.ii Single z-plane image showing manual segmentation of the cell body in (a.i). a.iii Overlay of (a.i) and (a.ii). (a.iv) Single z-plane confocal image of the specimen of (a.i), showing GFP signal of the same field of view as in (a.i). (a.v) single z-plane image showing manual segmentation of the cell body from (a.iv). (a.vi) overlay of (a.iv) and (a.v). b As in (a), but for segmenting dendrites. c (left) Single z-plane confocal image of expanded Thy1-YFP mouse brain tissue showing pGk13a staining of the membrane. (c.i) Magnified view of the yellow box on the left. c.ii single z-plane image showing manual segmentation of the myelinated axon in (c.i). c.iii overlay of (c.i) and (c.ii). c.iv Single z-plane confocal image of the specimen of (c.i), showing GFP signal of the same field of view as in (c.i). (c.v) Single z-plane image showing manual segmentation of the myelinated axon in (c.iv). (c.vi) Overlay of (c.iv) and (c.v). d As in (c), but for segmenting unmyelinated axons. (e) Rand score of pGk13a signal-guided segmentation of cell body, dendrites, myelinated axon and unmyelinated axons, using anti-GFP signal-guided segmentation as a “ground truth.” (n = 3 cell bodies and n = 3 dendrites from two fixed brain slices from two mice, and n = 5 myelinated axons and n = 5 unmyelinated axons from two fixed brain slices from two mice). Scale bars: (a.i–vi) 5 μm, (b.i–vi) 5 μm, (c) (left) 2 μm; (i–vi) 0.5 μm, (d) (left) 2 μm; (i–vi) 0.5 μm. Source data are provided as a Source Data file.

Fig. 6. Traceability of umExM.

a (pGk13a column) Serial confocal images of expanded Thy1-YFP mouse brain tissue after umExM processing, showing pGk13a staining of the membrane. (GFP column) anti-GFP signal of the same sample in the same field of view. b (left) pGk13a signal-guided manually traced and reconstructed myelinated axon from (a, pGk13a column). (right) As in (left), but with anti-GFP signals. c Rand score (n = 3 myelinated axons from two fixed brain slices from two mice) of pGk13a signal-guided manual tracing of myelinated axons, using anti-GFP signal-guided tracing as a “ground truth.” d As in (a) but with an unmyelinated axon. e As in (b) but for (d). f As in (c) but for unmyelinated axons (n = 3 unmyelinated axons from two fixed brain slices from two mice). g Representative (n = 4 fixed brain slices from two mice) single z-plane confocal image of expanded mouse brain tissue (corpus callosum) after umExM with double gelation processing (Supplementary Fig. 25), showing pGk13a staining of the membrane. The seeding points for manual segmentation are labeled with colors. h Magnified view of the white box in (g). i 3D rendering of 20 manually traced and reconstructed myelinated axons in the corpus callosum. Planes were visualized from raw umExM images that were used for tracing. Scale bars: (a) 0.5 μm, (d) 0.2 μm, (g) 18 μm, (i) 39.25 μm (x); 39.25 μm (y); and 20 μm (z). Source data are provided as a Source Data file.

Fig. 7. Higher resolution umExM.

a Representative (n = 5 fixed brain slices from 2 mice) single z-plane confocal image of post-expansion mouse brain tissue (Somatosensory cortex, L4) that underwent the umExM protocol. Images were taken at 50ms/frame for 20 frames with a confocal microscope with 1.5x optical zoom. pGk13a staining of the membrane visualized in inverted gray color throughout this figure. b Fluctuations in the acquired frames (as in (a)) were resolved with the ‘super-resolution imaging based on autocorrelation with a two-step deconvolution’ (SACD) algorithm. c Boxplot showing resolution of post-expansion confocal images (60x, 1.27NA objective) of umExM + SACD-processed mice brain tissue slices showing pGk13a staining of the membrane (n = 5 fixed brain slices from two mice). d Representative (n = 6 fixed brain slices from one mouse) single z-plane confocal image of post-expansion mouse brain tissue (Somatosensory cortex, L4) after the iterative form of umExM processing (Supplementary Fig. 26), showing pGk13a staining of the membrane. e Magnified view of yellow box in (d). f as in (c) but for the iterative form of umExM (n = 6 fixed brain slices from one mouse). Scale bars: (b) 10 μm, (d, e) 1.5 μm. Source data are provided as a Source Data file.