Versatile whole-organ/body staining and imaging based on electrolyte-gel properties of biological tissues

Abstract

Whole-organ/body three-dimensional (3D) staining and imaging have been enduring challenges in histology. By dissecting the complex physicochemical environment of the staining system, we developed a highly optimized 3D staining imaging pipeline based on CUBIC. Based on our precise characterization of biological tissues as an electrolyte gel, we experimentally evaluated broad 3D staining conditions by using an artificial tissue-mimicking material. The combination of optimized conditions allows a bottom-up design of a superior 3D staining protocol that can uniformly label whole adult mouse brains, an adult marmoset brain hemisphere, an ~1 cm³ tissue block of a postmortem adult human cerebellum, and an entire infant marmoset body with dozens of antibodies and cell-impermeant nuclear stains. The whole-organ 3D images collected by light-sheet microscopy are used for computational analyses and whole-organ comparison analysis between species. This pipeline, named CUBIC-HistoVIsion, thus offers advanced opportunities for organ- and organism-scale histological analysis of multicellular systems.

Fig. 1. Fixed and delipidated tissue is an electrolyte gel.

a A fixed and delipidated whole mouse brain treated with different chemical conditions. MES: 2-(N-morpholino)ethanesulfonic acid, TEA: triethanolamine. Scale: 5 mm. b Changes in the components inside the tissue during the delipidation procedure. The values were calculated as a ratio to the average of the delipidated samples (day 0) and indicate the means ± SD of the ratios (n = 3 biologically independent samples). c Changes in tissue components before and after delipidation. d Schematic of SAXS measurement and analysis. The relationship between the intensity of the scattered X-rays (I) and the scattering angle (θ) reflects nanometer-order structures inside the sample. The value of I in terms of the width Δr was calculated along the radius line (r) on the 2D intensity plot at a specific angle (φ) [or the circular average (φ = 0–2π)]. The scattering angle (θ) is transformed to the magnitude of the scattering vector q (which is inversely proportional to the size of the inside structure) according to the equation in the panel. L: distance between the sample and the detector, λ: wavelength of the X-ray. e, f I–q plot of the SAXS analysis of the brain slices for different NaCl concentrations, measured at SPring-8 (e) and the Photon Factory of High Energy Accelerator Research Organization (KEK) (f). The values on the y-axis were shifted by multiplying by shift factors (in e, 1000, 500, 80, 30, 10, 2, and 1 for H₂O, PB, PB with 25 mM, 50 mM, 150 mM, 250 mM, and 500 mM NaCl, respectively; in f, 35, 7, 4, 6, and 1 for PB, PB with 25 mM, 50 mM, 150 mM, and 500 mM NaCl, respectively). Lines indicate the slopes of the mass fractal dimension D. g Swelling-shrinkage curves of the delipidated tissue and fixed gelatin gel. The y-axis (S/S0) represents the relative value of the gel area (mean ± SD, n = 3, biologically independent samples).

Fig. 2. Development of a versatile protocol for whole-organ/body dye staining.

a Schematic of the surrogate assay. A cylinder-shaped, fixed gelatin gel containing DNA or rabbit immunoglobulin was used to experimentally evaluate the 3D staining patterns in the interior of the gel (rimmed or gradual patterns). b Representative results of DNA-containing gelatin gel staining with propidium iodide (PI), an ionized, lipophobic, and cell-impermeable nuclear stain. A high concentration of salt modulated the gel–stain interaction, leading to the improvement of the interior staining patterns. Scale: 1 mm. c Representative results of DNA-containing gelatin gel staining with SYTO 16, a lipophilic and cell-permeable nuclear stain. Additional use of ScaleCUBIC-1A (SC-1A) chemicals in high-concentration salt conditions improved the penetration patterns with decreased signal intensities (indicated as “4× enhanced”). Scale: 1 mm. The profiles in b and c show the mean intensity ± SD of the six diameter regions as in a. d, e The results in b and c were replicated by using the fixed and delipidated tissues (cerebellar hemispheres). After 3D staining, the samples were sectioned to evaluate the infiltration of the stain inside the tissue. The sections were also restained with DAPI to check the detectability of the nuclei (post 2D). Scale: 1 mm. f–h The interaction-modulated 3D staining was scaled up to a whole infant marmoset body. f The cleared and PI-stained infant marmoset body. PI staining was performed under a high ion strength condition (see “Methods”). Scale: 1 cm. g The reconstituted whole-body 3D image acquired with a custom-built LSFM (voxel size: 10.3 × 10.3 × 100 μm³) after staining and clearing. White boxes indicate the positions of the reconstituted x–z images in the panels on the right. Scale: 1 cm. h The reconstituted x–z images at the indicated positions in g. L left, R right. Scale: 2 mm.

Fig. 3. Development of a versatile protocol for whole-organ immunostaining.

a, b Representative results of rabbit IgG-containing gelatin gel staining with anti-rabbit secondary IgG (A594). Scale: 1 mm. The profiles show the mean intensity ± SD of the six diameter regions as in Fig. 2a. c–e The results in a and b were replicated with the fixed and delipidated tissues. c, d One-step procedure with mouse anti-NeuN-A488 or unconjugated anti-NeuN with secondary Fab fragment [3D, c] and conventional two-step procedure using secondary anti-mouse IgG after the primary antibody [3D, d] were compared. Three-dimensional staining results were visualized with A488. After 3D staining, the sample was sectioned and restained with rabbit anti-NeuN antibody followed by secondary anti-rabbit IgG (A594) or restained only with secondary anti-mouse IgG (A594) to confirm the detectability of NeuN antigen or the infiltration degree of primary anti-NeuN antibody, respectively (post-2D). Scale: 1 mm. e Collagenase or hyaluronidase treatment was tested by similar staining procedures. Scale: 1 mm. f Summary of essential conditions for efficient 3D staining. g Overview of the clearing, staining and RI matching of the CUBIC-HV protocol (ver. 1.0). The indicated durations are assumed for a whole adult mouse brain. Also see Supplementary Fig. 4a. h Comparison and modification of previous 3D staining protocols based on CUBIC-HV. Adult mouse brain hemispheres were 3D stained with anti-NeuN antibody (Supplementary Fig. 4d). The intensity range was normalized in each image. Scale: 1 mm (for entire sagittal sections) or 50 μm (for enlarged images). i Two-color whole-organ staining and volumetric imaging results of the adult mouse brain with custom-built LSFM. Voxel size: 8.3 × 8.3 × 9 μm³. Three-dimensional images were reconstituted with Imaris software. L left, R right. Scale: 2 mm. j Reconstituted sagittal (y–z) or coronal (x–z) images of the data in i, showing uniform 3D staining and imaging with almost isotropic cellular resolution. The resliced images were prepared with Fiji/ImageJ. Scale: 2 mm (for entire images) or 100 μm (for enlarged images).

Fig. 4. CUBIC-HV is applicable to various antibodies and dyes.

a Results of whole-brain staining and LSFM imaging with various antibodies (yellow) and nuclear stains (SYTOX-G or BOBO-1, blue). The data were reconstructed with Imaris software. The staining, imaging and image processing conditions are summarized in Supplementary Data 2. Voxel size 8.3 × 8.3 × 9 μm³. L: left, R: right. Scale: 2 mm. b Nine sets of the whole-brain imaging data in a (stained with anti-phospho-Nf, calbindin D28K, PV, Sst, ChAT, Dbh, Th, Tph2, or Copeptin antibodies, respectively, with nuclear stains) were registered and aligned to the whole-brain images obtained with anti-NeuN antibody staining and nuclear staining (structural standard). All the transformed data were merged as a pseudo-multitarget whole-brain antibody staining image with Imaris software. The precise registration results are indicated by the brain regions merged across different datasets, such as the paraventricular nucleus (PVN) and the supraoptic nucleus (SON) (NeuN+/copeptin+), the locus coeruleus (LC) in the pons (Dbh+/Th+) and the Purkinje cell layer in the cerebellum (calbindin D28K+/PV+). L left, R right. Scale: 2 mm (whole-brain), 1 mm (reconstituted coronal sections), 0.5 mm (inset).

Fig. 5. CUBIC-HV enables multicolor whole-organ staining and imaging.

a Experimental schematics of multichannel and multimodal whole-brain staining and imaging, shown in Figs. 5 and 6. b–j BOBO-1, PV (Cy3), Sst (A594), and Gad67 (A647) quadruple costaining images of the sole whole adult mouse brain reconstituted with Imaris software. (b-d) Individual antibody staining channels alongside the BOBO-1 channel. Voxel size 8.3 × 8.3 × 9 μm³. Scale: 2 mm. e Merged image of all four channels. Scale: 2 mm. f, g Enlarged horizontal (x–y) images of a representative section. Scale: 0.5 mm. h Reconstituted coronal (x–z) section image at the position indicated in e. Scale: 2 mm. i, j Enlarged images of the indicated positions in h. Scale: 0.1 mm.

Fig. 6. CUBIC-HV enables multimodal whole-organ labeling and imaging.

a-g ChAT (A594) and Dat (A647) double-staining images of the sole Thy1-YFP-H Tg whole adult mouse brain. Voxel size 8.3 × 8.3 × 9 μm³. a Merged image of all three channels. Scale: 2 mm. b Enlarged image at the position indicated in a. Scale: 0.5 mm. c Horizontal view of a representative section. Scale: 0.5 mm. d Enlarged image at the position indicated in c. Neighboring projection tracts labeled with YFP expression or anti-Dat immunolabeling were visualized. Scale: 0.5 mm. e–g Reconstituted sagittal sections at the position indicated in a, visualizing the neighboring projection tracts. Scale: 0.5 mm. Asterisks in e–g indicate nonspecific vascular signals due to insufficient perfusion of the sample.

Fig. 7. CUBIC-HV allows whole-organ cellular circuit analysis.

a Experimental schematic of whole-brain rabies virus (RV) tracing with cell-type immunolabeling. b Whole-brain image of the RV-injected Gad2-Cre adult mouse brain reconstituted with Imaris software. TVA-mCherry and RV-GFP were injected into left M1 region of the cerebral cortex. Voxel size 8.3 × 8.3 × 9 μm³. Scale: 2 mm. c Anti-Sst antibody staining channel was merged with the image in b. Scale: 2 mm. d, e Enlarged horizontal section images at the indicated positions in c, showing the RV-labeled neurons in the local and ipsilateral corticocortical circuits. Scale: 0.5 mm. f Reconstituted sagittal (y–z) section at the position indicated in d. Scale: 0.5 mm. g-j Cropped and enlarged images at the positions indicated in f. Starter neurons (green and magenta) and input neurons (green) with or without an Sst signal could be identified with single-cell resolution. Scale: 0.1 mm.

Fig. 8. CUBIC-HV allows whole-organ cellular function analysis.

a Experimental schematic of whole-brain c-Fos antibody staining in different pharmacological conditions. b Whole-brain images of c-Fos-immunolabeled adult mouse brains with or without administration of NMDA receptor antagonist MK-801. The images were reconstituted with Imaris software. Voxel size 6.5 × 6.5 × 7 μm³. The white box indicates the position of the reconstituted sagittal image in d. Scale: 2 mm. c Resulting images following application of a convolution filter to the images in b. Scale: 2 mm. d Reconstituted sagittal images of brains in b, showing uniform labeling of c-Fos-expressing cells. The c-Fos-expressing cells of the dentate gyrus (DG) region were more obvious in the MK-801 (−) sample (arrowheads in insets). Scale: 1 mm (whole sagittal image) and 0.5 mm (insets).

Fig. 9. CUBIC-HV enables organ-level comparison of rodent and primate brains.

a Experimental schema of the whole-organ comparison analysis of rodent and primate brains. b Comparison of an adult marmoset brain hemisphere with the whole adult mouse brain (only the right hemisphere is depicted), costained with SYTOX-G, GFAP (A594), and α-SMA (A647) antibodies. Note that the orientations were different between samples. Images were reconstituted with Imaris software. Voxel sizes 16.4 × 16.4 × 16 μm³ (for marmoset brain) and 8.3 × 8.3 × 9 μm³ (for mouse brain). MCA middle cerebral artery, PCA posterior cerebral artery, BA basilar artery. Scale: 2 mm. c Reconstituted coronal sections of hemisphere images in b, showing the distributions of corresponding vessel structures and GFAP-positive cells. Note that the different orientations of the sections (z–x and x–z for the marmoset and mouse samples, respectively) could be compared because of the almost isotropic voxel resolution of the z-stack images. ACA anterior cerebral artery, Str striatum, Hp hippocampus. Scale: 2 mm. d Magnified images at the positions indicated in c, showing the similar distributions of vessel-associating GFAP-positive cells in the hippocampus. Voxel size in both images 8.3 × 8.3 × 9 μm³. Scale: 0.5 mm. e Flowchart of staining-based whole-organ imaging and image analysis on CUBIC-HistoVIsion.