Scalable tissue labeling and clearing of intact human organs

Three-dimensional (3D) mapping of the human organs at cellular resolution and generating reference maps of organs across ages or diseases represent the common objectives of diverse consortia including Human Biomolecular Atlas Program (HuBMAP)1,2, Human Cell Atlas3, Human Tumour Atlas4 and LungMap5. Traditional histological techniques of slicing, staining, imaging and 3D reconstruction of the cellular details of biological tissue pieces are challenging to scale up to large human organs. For example, mapping one intact adult human brain using such traditional methods took years of tedious work, and the lengthy process resulted in mechanical distortion and accidental loss of sections of tissue6. While improvements are constantly being made in the instruments and methods for iterative image analysis of multimodal imaging of human organs using PET/MRI, these modalities are still limited by low resolution and the inability to probe cellular and molecular parameters.

In recent years, extensive biomedical research in neuroscience7,8,9,10,11,12, development13 and cancer14,15 has substantially benefited from the optical tissue labeling and clearing methods that bypass the major problems of histology. Especially, clearing and imaging of whole adult mouse bodies has opened a holistic examination window into physiological and pathological systems in an unbiased way9,14. However, human organs obviously are much larger in size and, owing to the effects of aging, contain more complex tissue components, which limit the post-staining methods that can be applied to human organs compared with organs from months-to-year-old rodents. Therefore, achieving whole human organ transparency has been challenging, despite numerous trials on slices11,16,17,18,19. Hence, we developed a full pipeline for whole human organ labeling, clearing, imaging and 3D map reconstruction at cellular level, which implemented a new tissue labeling and clearing technology termed SHANEL and an advanced volumetric imaging system using a commercial light-sheet fluorescence microscope (LSFM)20. This pipeline can easily be adopted in laboratories, within routinely available reagents and equipment (Fig. 1).

Fig. 1: Overview of SHANEL pipeline.figure 1

SHANEL is composed of seven parts. After human organs and tissue are collected (Part 1), they are prepared for blood removal and fixation (Part 2). The next step is performed by active pumping (Part 3a) or passive incubation (Part 3b), based on whether the organs are intact or not. Samples are then subjected to permeabilization, delipidation, ECM loosening and labeling with chemical probes. If antibodies compatible with the SHANEL protocol are used, the organ of interest should be dissected into slices less than 1.5 cm thick for immunolabeling (Part 4). Samples are then dehydrated, delipidated and matched for RI, until they become transparent (Part 5). Imaging of whole human organs is performed in an UltraMicroscope Blaze LSFM, and images of tissue slices are captured with an UltraMicroscope II LSFM at high resolution (Part 6). The data are stitched, volume fused, and rendered in a 3D visualization (Part 7).

Development of the SHANEL protocol

Clearing, labeling and imaging of human organs is difficult to implement in practice owing to several challenges. In Table 1, we delineate the key challenges encountered during the time-consuming endeavor of human organ processing and the ways in which these problems could be addressed.

Table 1 Challenges of labeling, clearing and imaging human organsPerfusion and fixation of intact organs

To preserve human organs rapidly and uniformly in a life-like state, we introduced an active perfusion system to deliver 0.01 M phosphate-buffered saline (PBS)/heparin and 4% paraformaldehyde (PFA) solutions into whole organs through the vascular network before the organs are harvested if major blood vessels are available (Fig. 2 and Extended Data Fig. 1a). The advantages of this step include washing out as much remaining blood as possible, preventing clot formation by maintaining circulation of the vascular system and achieving faster tissue fixation than through passive immersion. The organs are dissected carefully to preserve intact anatomical shapes and connect the main arteries with exogenous tubes for later experiments. Alternatively, human samples could be passively fixed in 4% PFA or 10% formalin buffer to covalently crosslink proteins. It is worth mentioning that human organ and tissue donation organizations such as the International Institute for the Advancement of Medicine are reliable resources that provide transplantable organs with intact vascular systems and detailed donor information.

Fig. 2: Blood vessel labeling with dextran solution by active perfusion.figure 2

a, Photo of dissected human kidney with inserted tubing (yellow arrows). Scale bar, 1 cm. b,c, The original tubing was replaced with a PTFE (chemical-resistant, anti-adhesive, biocompatible) tube (b) and tightly fixed by double rope-fastening (c; red rectangle in b). d, The inserted nozzle of the PTFE tube was cut at an angle to ensure smooth insertion into the organ. e, A set of connecting tubing containing a black pump reference tube in the middle, two colorless PVC tubes at both ends, three tube connectors and a pipette tip. Switch between position 1 and position 2 to avoid breakage of the reference tubing. The red arrows indicate the flow circulation direction. f, Blood vessel labeling with dextran solution using a pump perfusion system. g, Rotation of the wheel enables solution flow into the kidney. h, The correct way to fasten the black pump reference tube to the cassette is shown. i, Photos of human kidney after dextran labeling. j, The dextran-labeled human kidney was sealed in a plastic bag.

Organ clearing strategy

Among the hydrophilic reagent-based21,22,23,24, hydrogel-embedding19,25,26,27 and hydrophobic reagent-based9,10,13,14 tissue clearing methods, we chose to work with hydrophobic reagents in the SHANEL pipeline. An important advantage of hydrophobic tissue clearing is sample shrinkage, enabling us to accommodate and image large organs using an LSFM. Ethanol is employed to remove the water inside of human tissue, by increasing ethanol concentration in a stepwise manner. Dichloromethane (DCM) is used to extract the remaining lipids and ethanol, after which the tissue is mostly composed of fixed proteins. In the end, the relatively homogeneous human tissue is rendered transparent by immersing it in BABB solution (benzyl benzoate:benzyl alcohol 2:1, v/v). BABB solution has a refractive index (RI) of 1.56, which is the same as that of the cross-linked proteins.

Imaging system

In general, organs cleared with hydrophobic reagents will shrink ~30% in volume10,20. However, commercially available microscopes such as the LaVision UltraMicroscope II system (chamber size of 72 × 74 × 35 mm, sample traveling range of 10 × 10 × 10 mm in X, Y, Z) or ZEISS Lightsheet 7 (sample size of 10 × 10 × 20 mm) cannot hold large cleared human organs such as an intact eye (size of 30 × 30 × 30 mm). To overcome this limitation, together with Miltenyi Biotec we developed a prototype UltraMicroscope (chamber size of 250 × 90 × 70 mm) (Fig. 3). This microscope is now commercialized by Miltenyi Biotec as UltraMicroscope Blaze (chamber size of 129 × 51 × 64 mm, sample travel range of 50 × 24 × 23 mm)—a fully automated light-sheet microscope for imaging large cleared samples covering the range from an entire mouse to most human organs.

Fig. 3: Organ mounting and imaging with light-sheet microscopy.figure 3

a, Photo of human pancreas before clearing. Scale bar, 1 cm. b, Photo of transparent human pancreas after PI cell nuclei labeling and clearing, showing mesenteric artery (white arrow) and lymph nodes (yellow asterisks). Scale bar, 1 cm. c, The mounting of an organ on a sample holder requires glue and black tape. The red arrows show the boundaries of the moving range of the holder in the Y direction. d, The sample holder was wrapped with four sections of tape, and a drop of glue was placed on top of each section. e, An example of sample positioning with alignment to one end of the holder to cover one sample edge. f, Another example of sample positioning with alignment to the other end to cover the other sample edge. g, The mounted human pancreas was illuminated by light-sheet microscopy.

Permeabilization and decolorization reagents

Efforts to label and clear human brain tissue by screening thousands of delipidation and decoloring chemicals28 or chemicals that increase tissue permeability17,19 have already shown the difficulties related to incomplete tissue transparency, time-consuming procedures and limited antibody penetration. Given that aged human tissue is composed of dense and intricate hydrophobic and hydrophilic molecules, we hypothesized that efficient detergent permeabilization is necessary to render the human tissue accessible to reagents traveling end-to-end through it. Detergents are amphiphilic, possessing both hydrophilic and hydrophobic properties, and form micelles in solution that can interact with molecules of the tissue. Traditional detergents such as ionic sodium dodecyl sulfate (SDS) or non-ionic Triton X-100 (4-(1,1,3,3-tetramethylbutyl) phenyl-polyethylene glycol), carrying typical ‘head-to-tail’ chemical structures, are inefficient at permeabilizing sturdy human tissues because the micelle sizes of these detergents are too big to enter deep inside the tissue. We identified CHAPS (3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate) as a powerful candidate to completely and quickly permeabilize intact human organs, as it forms much smaller micelles with molecular weight of 6,150 Da. CHAPS renders the cellular and extracellular matrix of aged tissue within intact human organs more permeable, thus improving the accessibility of reagents. To remove the red color of remaining blood clots, we screened diverse heme-eluting chemical analogs that are compatible with CHAPS. Analogs bearing ethanolamine structures have been shown to improve decolorization effects in the presence of CHAPS20. Potential candidates such as N-butyldiethanolamine28 and N,N,N′,N′-tetrakis (2-hydroxypropyl) ethylenediamine (Quadrol)24 have been proven to be efficient in decolorizing tissues, but they are at least twice as expensive as N-methyldiethanolamine (NMDEA) (the price of NMDEA is ~30€/L). Considering that intact human organs require large amounts of decolorization chemical agents, we decided to use a NMDEA and CHAPS mixture to achieve cost-efficient permeabilization and decolorization. Still, the volume of the reagents consumed in a single step of handling an intact human brain is ~5–6 L, and the total cost of CHAPS/NMDEA solution is ~1,800€ (ref. 20).

Labeling

Small molecular dyes (<1–2 kDa) or large antibodies (~150 kDa) provide fluorescence contrast of sufficient signal-to-background ratio to identify specific structures of human tissue. DNA or RNA chemical probes show high binding specificities and affinities to the nucleic acid of cells across a wide range of fluorescence spectra (e.g., DAPI, Hoechst in the blue–green range; JO-PRO-1, propidium iodide (PI) in green–red range; TO-PRO3, short-wavelength infrared (SIR) in the infrared range)29. However, primary or secondary antibodies conjugated to chemical fluorochromes cannot penetrate and label more than 1 cm deep into adult human organs due to the probes’ big sizes. Hence, we introduced chemical pretreatments to loosen the cellular and extracellular matrix. First, a solvent mixture of DCM:methanol (2:1, v/v) was used to extract the hydrophobic lipids inside the tissue, which would prevent the free movement of hydrophilic labeling reagents dissolved in buffer30. Second, the tissue was subjected to acetic acid for partial hydrolysis of intertwined collagen by cleavage of the noncovalent intra- and intermolecular bonds31. This process maintains the collagen chains intact, but the cross-links are cleaved. Third, guanidine hydrochloride buffer was employed to extract the proteoglycans of the tissue in dissociative conditions32. After these chemical extraction steps, the tissue matrix became accessible to both small-molecule dyes and large antibodies at a depth range of several centimeters. Cell nuclei in intact human pancreas were perfused and labeled with PI (Fig. 4). Vasculature in multiple human organs was labeled with dextran solution (Fig. 5). SHANEL is also compatible with passive incubation of dye and antibodies to stain PFA-fixed human tissue up to several centimeters size range (Fig. 6). As we have previously demonstrated, 1.5 cm cubic human brain pieces were successfully labeled with primary and secondary antibodies to reveal microglia (e.g., Iba1) and dopaminergic cells (e.g., tyrosine hydroxylase)20. Antibodies conjugated with large protein dyes (e.g., phycoerythrin) also fully labeled more than 1 cm cubic human kidney and lung pieces to reveal cytoplasmic (e.g., catalase), extracellular (e.g., collagen IV) and membranous (e.g., cytokeratin 19) structures of tissue (Extended Data Fig. 2 and Table 2).

Fig. 4: 3D reconstruction of human pancreas.figure 4

a, 3D reconstruction of human pancreas imaged by light-sheet fluorescence microscopy showing the PI-labeled cell nuclei (green) and autofluorescence at 785 nm (magenta). Scale bar, 1 cm. b, Zoomed 3D reconstruction showing the blood vessels and lymph nodes (red rectangle). Scale bars, 1 mm and 200 µm, respectively. c, Higher magnification of the endocrine portion region in the human pancreas is zoomed in using a 12× light-sheet microscopy objective and reconstructed in 3D by Imaris software. Scale bar, 200 µm. d, Various pancreas islets can be easily identified in the 3D-reconstructed endocrine portion region on the basis of their specific shapes (red asterisks). Intralobular ducts (yellow arrows) can be located near the islets. Scale bars, 50 µm, 50 µm and 30 µm, respectively.

Fig. 5: SHANEL of human organs with perfusion of dextran vessel labeling dye.figure 5

a, Human heart after performing the SHANEL protocol. 3D reconstruction shows blood vessels of the coronary artery (red) and myocardium that can be imaged in autofluorescence at 488 nm (AF, green). Scale bars, 1 cm and 2 mm, respectively. b, Human kidney after SHANEL. The structure of glomeruli can be seen in dextran-labeled tissue. TO-PRO3-labeled cell nuclei are shown in blue. Scale bars, 1 cm and 2 mm, respectively. c, One lobe of human lung was cleared and stained with dextran. Bronchus can be seen in AF. Scale bars, 1 cm and 2 mm, respectively. d, Image of human pancreas. Ductal structures are shown with AF, and vasculature is labeled with dextran after SHANEL. Scale bars, 1 cm and 2 mm, respectively. e, Human spleen after SHANEL. The fine structure of the splenic artery is marked with dextran, and the other blood vessel structures are visualized with AF in Imaris. Scale bars, 1 cm and 2 mm, respectively.

Fig. 6: SHANEL of human tissues by passive incubation.figure 6

a, Skull labeled with lectin (blood vessels, yellow) and PI (nuclei, blue); details of blood vessels and cell nuclei can be seen in XY view. Scale bars, 1 mm, 1 mm and 100 µm, respectively. b, Brain slice labeled with lectin (blood vessels) and Neurotrace Nissl stain. Scale bars, 2 mm and 300 µm, respectively. c, Human brain tissue labeled with Iba1 antibody. Penetration of antibody can be verified in the YZ and XZ view cross-sections, details of microglia can be seen in the magnified XY view. Scale bars, 1,000 µm, 700 µm, 500 µm and 40 µm, respectively. d, Lung tissue is labeled with α-SMA antibody, and AF shows morphometry of the bronchial tree and acinar structure. YZ and XZ views, and magnified XY view are also shown. Scale bars, 700 µm, 500 µm, 500 µm and 150 µm, respectively.

Table 2 Dyes and antibodies compatible with SHANEL protocolTable 3 Timing guidelines for treatment of various human organsImage analysis

With the prototype or commercial UltraMicroscope Blaze, it is possible to scan the intact human eye, kidney, thyroid and pancreas by mosaic imaging. The scanning time depends on the size of the sample, the percent overlap between images of the mosaic, the number of scanning channels and the size of the Z step (Extended Data Fig. 3). Consequently, terabytes of large data can be generated from a single organ. Software such as Fiji, Arivis, Imaris and Photoshop can be used to handle the large datasets for 3D reconstruction and movie generation with a standard lab workstation (e.g., >256 gigabytes of RAM and terabytes of storage space) (Extended Data Figs. 46). However, quantitative analysis of such large datasets can be difficult and imprecise using such software, which in general relies on simple strategies such as filter-based normalization, thresholding or watershed algorithms. Recently, deep learning approaches14,20,33 have shown superior performance in quantification of large-scale data in terms of processing accuracy and speed. It is anticipated that 3D human organ mapping could be greatly advanced with a full exploitation of tissue clearing and imaging combined with deep learning technologies.

Applications of SHANEL

Mammalian skeletal bones protect delicate internal organs, contain most of our body’s calcium supply and are especially difficult to label and clear. Expanding deep tissue labeling, clearing and imaging of bones using SHANEL would greatly benefit investigations of the 3D geometric features of bone volume and cells. In addition to soft tissues such as bone marrow, bones contain hard mineral-dense regions that are deposited with calcium-bearing hydroxyapatite crystals embedded in a collagen matrix. The calcium content considerably increases the optical scattering of bone34. EDTA has been demonstrated as an effective decalcification reagent in previous studies35,36. Similarly, Tainaka et al. developed a carbonated hydroxyapatite-based screening system to identify potent decalcification chemicals compatible with tissue clearing and found EDTA combined with imidazole showed superior effects28. Hence, we conducted decalcification of bones using 20% EDTA at 37 °C before SHANEL tissue labeling and clearing, as shown with examples of human skull pieces and pulvinar soft tissue inside the joint cartilage surface surrounded by bones (Fig. 7). As human bones are much thicker and harder than mouse ones, a much longer decalcification time—in the range of weeks to months—is needed to achieve the desired softness. Alternatively, reagents composed of strong, mild or weak acids (e.g., nitric acid, formic acid, hydrochloric acid, chromic acid, etc.) can be used for faster decalcification. For example, 5% nitric acid is an option for rapid decalcification yielding acceptable tissue integrity and antigenicity37,38.

Fig. 7: Passive incubation, clearing and 3D reconstruction of human pulvinar and skull.figure 7

a, Photo of transparent human pulvinar after PI cell nuclei labeling and clearing by passive incubation. Scale bar, 1 cm. b, 3D reconstruction of human pulvinar imaged by light-sheet fluorescence microscopy showing autofluorescence (AF) at 488 nm (gray) and PI-labeled cell nuclei (green). Scale bar, 1 cm. c, XY section view of pulvinar showing the connection between bone tissue and the pulvinar fibrofatty tissue. Scale bar, 150 µm. d, Human skull bone after PI cell nuclei labeling and clearing. Scale bar, 1 cm. e, The human skull bone imaged by light-sheet microscopy (left) and confocal microscopy (right). Scale bars, 500 µm and 200 µm, respectively.

SHANEL tissue clearing technology has been successfully applied to organs from other mammalian species such as pig brain and pancreas, and is compatible with vDISCO immunostaining20. This enables imaging large mammalian organs in which fluorescent proteins such as GFP, YFP, mCherry and tdTomato are expressed. Since transgenic zebrafish, rat, mouse, pig and macaque with fluorescent protein expression in specific organs have been developed, SHANEL could readily be adopted to clear and image organs from diverse organisms. To combine vDISCO immunostaining with SHANEL organ clearing, we recommend that the organs are first actively perfused or passively incubated with the mixture of CHAPS and NMDEA to permeabilize and decolorize tissue. Afterwards, vDISCO immunostaining process is used via perfusion/incubation with the chosen nanobody in a solution of 1.5 vol% goat serum, 0.5 vol% Triton X-100, 0.5 mM of methyl-β-cyclodextrin, 0.2 wt/vol% trans-1-acetyl-4-hydroxy-l-proline and 0.05 wt/vol% sodium azide in PBS. Finally, the organs are ready to be cleared by SHANEL reagents. We anticipate that, after CHAPS and NMDEA treatment, nanobody/antibody immunolabeling and SHANEL tissue clearing could be applied to diverse mammalian species to investigate broad biological questions. In cases where antibody-based tissue labeling fails owing to inability of large antibodies to cross through whole organs, nanobodies that are ten times smaller than conventional antibodies could provide a more viable alternative.

Although the SHANEL tissue labeling and clearing method was developed for intact human organs, it also works for small tissue pieces, for example, on human biopsies20. In general, biopsy samples are small in size and suitable for the application of the passive SHANEL labeling and clearing method.

Comparison with other methods

In the last years, a diverse array of methods for labeling and clearing of human organ pieces has been developed, including CLARITY17,39, OPTIClear16, MASH18, CUBIC28, SWITCH40, SHIELD11 and ELAST19. CLARITY and OPTIClear take months to clear fixed human brain tissue pieces (<5 mm thickness). MASH explored the small-molecule fluorescent dye labeling and clearing of human brain cortex (<5 mm thickness) by modifying the iDISCO protocol. Other methods are applicable to 3D imaging of human myocardial tissue41, lymph node and lung pieces (<1 cm3) (ref. 42). SWITCH allowed multiple rounds of antibody labeling in 100-µm-thick human brain sections. On the basis of the SWITCH method, SHIELD used a tissue transformation strategy to stabilize a 2-mm-thick human brain slice via intramolecular epoxide linkages to prevent degradation. ELAST enabled human brain tissue (<5 mm thickness) antibody labeling and clearing by mechanically stretching tissue–hydrogel hybrids.

For the first time, SHANEL technology achieved the labeling and clearing of intact adult mammalian organs as large as several cubic centimeters, including human brain, pig brain, pig pancreas, human kidney, human thyroid20, human heart, human pancreas, human lung and spleen (Fig. 5). The key step of the SHANEL technology relies on the permeabilization and decolorization by CHAPS/NMDEA solution. The whole process can be conducted by passive incubation or active perfusion, depending on the availability of major vessels for external connection to commercialized pumps. Active perfusion permits scaling up the process to multiple organs, with general lab equipment available in most molecular biology labs, and no special expertise or training required to implement. It is preferred to perform active perfusion for large adult organs, in order to speed up the process to a reasonable period (e.g., 1.5 months for human kidney, 4 months for human brain). The timeline for each step varies case by case depending on the specific organ. Even a pair of kidneys from the same donor could be different in size, hardness, amount of blood clots and pigments accumulation. It is important to ensure that each step—permeabilization and decolorization, delipidation, cellular and extracellular matrix loosening and labeling, dehydration and RI matching—has been fully completed.

Overview of the protocol

SHANEL provides a flexible platform to achieve whole or partial organ labeling, clearing, and imaging at a cellular level in diverse human and animal organs (Fig.1, Extended Data Fig. 7). The pipeline includes the following main stages: (i) preparation of fixed organs by perfusion or passive incubation, depending on the donor (Steps 1–5); (ii) sample pretreatment, including optional steps such as decalcification and blood vessel labeling (Steps 6–10); (iii) labeling with chemical probes or antibodies (Steps 11–18); and (

留言 (0)

沒有登入
gif