Mapping Hepatic Stellate Cell Morphology in Mouse Models of Liver Fibrosis

Fibrosis, or the pathological scarring of tissue, is the final common pathway for nearly all chronic organ diseases and contributes to approximately 40% of deaths in industrialized nations¹. In the liver, fibrosis arises from chronic insults such as viral hepatitis, alcohol use, or increasingly, metabolic dysfunction-associated steatotic liver disease (MASLD), which currently affects 30-40% of adults in the U.S. Chronic insults to the liver and the resulting progressive fibrosis drastically increase the risk of end-stage liver diseases such as cirrhosis, liver failure, and hepatocellular carcinoma^2,3,4,5.

Hepatic stellate cells (HSCs) are central to liver homeostasis and fibrogenesis. In healthy livers, HSCs exist in a quiescent, vitamin A-storing state and contribute to the homeostasis of hepatocytes and, overall, the entire liver^6,7. This is clearly demonstrated by HSC depletion studies, where the removal of nearly all HSCs in healthy liver decreases liver mass and severely limits liver regeneration^7,8. Upon injury, HSCs activate into myofibroblast-like cells that proliferate, migrate to damaged areas, and produce excessive extracellular matrix (ECM), forming fibrotic scars. Activated HSCs in fibrotic livers are also the source of cancer-associated fibroblasts and play tumor-promoting roles in hepatocellular carcinoma^9,10,11,12. Quiescent HSCs possess a unique neuron-like morphology with small cell bodies from which long-range projections extend. HSCs shift to a flat, sheet-like myofibroblast shape when activated in fibrosis¹³. While decades of work have elucidated the signaling pathways controlling ECM production by HSCs, the functional significance of their distinct neuronal morphology, how it is regulated, whether it could facilitate interactions within the HSC niche, and contribute to HSC fibrogenicity and disease remain elusive^14,15.

Current methods to visualize HSCs' morphology are limited to two-dimensional (2D) images of either in vitro cultured cells or in vivo tissue sections, and rarely capture the entirety of HSC morphology with projections extending several cell body lengths in all directions. Moreover, the most commonly used HSC marker, desmin, cannot fully capture the unique neuron-like morphology of the HSCs in immunofluorescence staining, as desmin is a cytoskeleton protein that only labels intermediate filaments within HSCs¹⁶. These methodological challenges thus far hindered the ability to fully capture HSC morphology and their cellular interactions in a three-dimensional (3D) manner in their natural context at single-cell resolution. As a result, fundamental questions regarding how HSC morphology relates to their functional state and interactions with neighboring cell types have remained unresolved.

This article presents an innovative and comprehensive imaging and analysis pipeline to visualize and model HSC morphology in intact liver tissue. This method integrates several key innovations: (1) fluorescent labeling of HSCs using genetic reporter mouse models, (2) optimized in situ liver perfusion and fixation to preserve intact tissue architecture and cellular relations, (3) a modified iDISCO-based tissue clearing protocol to achieve optical transparency¹⁷, (4) high-resolution confocal microscopy for deep-tissue image capture, and (5) a customized computational workflow for 3D image reconstruction and quantitative analysis of individual HSCs. The protocol imposes certain constraints on tissue size and time commitment that need to be taken into consideration. Together, this protocol provides a robust and reproducible framework for capturing the complex 3D structure of HSCs and for mapping their spatial relationships with neighboring vascular, immune, and parenchymal cells. The strategies and pipelines presented here for the liver are readily adaptable to other organs as well^17,18,19.

All mouse experiments used in this study have been approved by the Icahn School of Medicine at Mount Sinai Institutional Animal Care and Use Committee. The reagents and the equipment used are listed in the Table of Materials.

1. Generation of mice with tdTomato (TdT)-labelled HSCs

1. Set up Tg(lox-stop-lox-TdTomato) x Tg(Lrat:Cre) breeders (Jackson Laboratories strain #007914²⁰ and Mutant Mouse Resource & Research Centers (MMRRC) Strain #069595²¹-JAX, respectively).
   NOTE: Expect litter after 3 weeks. Use LratCre females for breeding, as some Cre expression has been detected in the male germline.
2. Collect tails and genotype F1 litter to identify Tg(Lrat:Cre);Tg(LSL-TdT) mice that selectively express the TdTomato fluorophore in hepatic stellate cells.

2. In situ liver perfusion and fixation

1. Pre-surgery set-up
   1. Prepare a ketamine/xylazine mixture for mice anesthesia by mixing 2 mL of ketamine, 1 mL of xylazine, and 7 mL of water.
      NOTE: Ketamine is a controlled substance and requires proper licensure and regulation for laboratory usage.
   2. Prepare the perfusion solution of 4% paraformaldehyde (PFA) in 1x Phosphate Buffered Saline (PBS).
   3. Set up the peristaltic pump by allowing 1x PBS to flow through the tube and set the flow rate at 2 mL/min.
      NOTE: Ensure that there is no air bubble in the tube.
2. Weigh the mouse to determine the amount of the ketamine/xylazine mixture to inject.
3. Inject the ketamine/xylazine mixture intraperitoneally into the mouse.
   NOTE: Inject 200 µL of the mixture for every 20 g of mouse weight.
4. Place the mouse in a container with a lid and wait approximately 20 min for the mouse to be fully anaesthetized.
   NOTE: Pinch the mouse toe to check for the absence of reflex. Ensure that the mouse is still alive when perfusion initiates.
5. Transfer the mouse from the glass container to the surgery platform in the supine position.
   CAUTION: PFA is toxic and volatile; therefore, perform the perfusion procedure in a chemical hood. Collect and dispose of the PFA flow-through properly according to institutional biosafety guidelines.
6. Pinch the palms of the mouse foot with forceps to ensure the mouse does not have any reflex/spasm movements.
7. Stab 4 palms with needles to spread the mouse's abdomen.
8. Spray the abdomen with 70% ethanol to wet the fur.
9. Use forceps to pick up the skin in the middle of the abdomen, align with the hind legs, and start cutting the skin (both layers) to reveal the abdomen.
10. Use PBS-wet cotton swabs to move intestines and other organs to the side and lift the left lobe.
11. Carefully cut the membrane attached between the left lobe and the caudate lobe.
12. Use a suture to tie off at the top base of the left lobe.
13. Cut the left lobe below the suture knot and set the left lobe aside for tissue sample collection and processing, depending on experimental needs (such as directly snap-freezing in liquid nitrogen, embedding in Optical Cutting Temperature (OCT) compound, or formalin fixation).
14. Use PBS-wet cotton swabs to move the liver and reveal the portal vein.
15. Carefully insert the catheter needle into the portal vein at about halfway along the length of the portal vein.
    NOTE: To assist with the catheter insertion, lift the mouse's body by placing a forceps under the body to ensure a flatter surface. Carefully align the catheter needle as parallel as possible with the portal vein before insertion.
16. Remove the catheter needle.
    NOTE: Ensure a backflow of blood through the catheter to indicate successful insertion.
17. Start the perfusion pump at a flow rate of 2 mL/min, connect the tubing with the catheter, and perfuse the liver with 20 mL of 1x PBS.
    NOTE: Ensure that the catheter is completely filled with blood, and the tubing is completely filled with the perfusion buffer, before connecting them to ensure absolutely no air bubbles get introduced into the system.
18. Clamp the pulmonary artery above the liver with a hemostat to stop the 1x PBS from going through the heart.
19. Cut the inferior vena cava to release the pressure buildup.
20. Adjust the pump speed to 5 mL/min.
    NOTE: Allow approximately 15 min to finish the perfusion.
21. Stop the pump to change the reservoir to 4% PFA.
    NOTE: Make sure not to introduce air bubbles into the tube by stopping the pump when there is still liquid in the reservoir.
22. Perfuse the liver with 25 mL of 4% PFA.
23. Once the perfusion is complete, remove the liver from the surrounding organs by cutting the connecting tissues.
24. Place the liver in a 50 mL conical tube, filled with 25 mL of 4% PFA.
25. Put the tube horizontally on a bench rocking platform for 1 h at room temperature at 1.5 rpm, to ensure solution movement.
26. Wash the liver 3 times in 1x PBS on a bench rocking platform for 30 min each at room temperature at 1.5 rpm.
27. Store the liver in 20 mL 1x PBS with 0.02% NaN₃ at 4 °C.
    NOTE: Store the liver long-term at this stage, up to at least one year.

3. iDISCO+-based tissue clearing

NOTE: The original description of the iDISCO+ protocol is accessible at ¹⁷

Day 1: Sample dehydration

1. Cut the liver samples into pieces with sizes approximately 0.5-1 cm³.
   NOTE: Cut different samples into different shapes for easier detection during imaging steps.
2. Prepare 20% (vol/vol), 40%, 60%, and 80% methanol dilutions in water in separate glass bottles.
3. Place liver samples into glass vials with screw tops (e.g., 4-7 mL).
   NOTE: Place up to 6 liver sample pieces in one 7 mL glass vial.
4. Dehydrate the sample using the methanol dilution in H₂O series, 20% (vol/vol), 40%, 60%, and 80% for 1 h each at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.
5. Wash the sample with another 100% methanol on a wheel rotator at 8 rpm, with no centrifugation occurring for 1 h at room temperature.
6. Chill the sample at 4 °C for 30 min to 1 h.
7. Prepare 66% (vol/vol) DCM dilutions in methanol in a glass bottle.
   NOTE: Prepare new dilutions every time. Use glass cylinders to prepare the solution, as DCM can dissolve plastic, such as polystyrene.
8. Incubate samples in 66% DCM/methanol solution overnight on a wheel rotator at 8 rpm at room temperature, with no centrifugation occurring.
   NOTE: Beware of samples floating to the top once the solution is added.

Day 2: Sample bleaching

9. Wash the sample twice with 100% methanol on a wheel rotator at 8 rpm, with no centrifugation occurring for 1 h each time at room temperature.
10. Chill the sample at 4 °C for 30 min to 1 h.
11. Bleach the sample in chilled 5% H₂O₂ in methanol (1 volume of 30% H₂O₂ to 5 volumes of methanol), on a wheel rotator at 8 rpm, with no centrifugation occurring, at 4 °C overnight.
    NOTE: Prepare a new solution every time.

Day 3-5: Sample permeabilization

12. Rehydrate with methanol/H₂O series: 80%, 60%, 40%, 20%, 1x PBS for 1 h at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.
13. Wash the sample twice in PTx.2 (Table 1) for 1 h each at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.
14. Incubate the samples in Permeabilization Solution (Table 1) at 37 °C for 2 days.

Day 5-17: Immunolabeling

15. Block the samples in Blocking Solution (Table 1) for 2 days at 37 °C.
    NOTE: Prepare a fresh blocking solution every time due to the instability of donkey serum. Store leftover blocking solution at 4^{°C for up to 1 month.}
16. Incubate the samples in primary antibodies diluted in PTwH (Table 1) with 5% DMSO and 3% Donkey Serum at 37 °C on a wheel rotator at 5 rpm, with no centrifugation occurring for 4 days.
    NOTE: Start with a 1:500 primary antibody dilution.
17. Wash the samples in PTwH 4-5 times for 1 h each at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.
    NOTE: Optionally, leave the samples to wash overnight.
18. Incubate the samples in secondary antibodies and propidium iodide diluted in PTwH with 3% Donkey Serum at 37 °C on a wheel rotator at 5 rpm, with no centrifugation occurring for 4 days.
    NOTE: Start with a 1:500 secondary antibody dilution. Use a 1:250 propidium iodide dilution from a 5 mg/mL propidium iodide stock solution.
19. Wash the samples in PTwH 4-5 times for 1 h each at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.
20. Leave the sample washing in PTwH overnight, at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.

Day 18: Clearing

21. Dehydrate the sample using the methanol dilution in H₂O series, 20% (vol/vol), 40%, 60%, 80% for 1 h each at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.
22. Wash the sample twice with 100% methanol on rotation for 1 h at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.
23. Incubate the sample in 66% (vol/vol) DCM dilutions in methanol for 3 h at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.
24. Wash the sample twice in 100% DCM for 15 min at room temperature on a wheel rotator at 8 rpm, with no centrifugation occurring.
25. Incubate the sample in DBE overnight at room temperature before imaging.
26. Store the samples long-term in DBE up to 1 year without impacts on the staining signal.

4. Confocal microscopy imaging

1. Place the cleared sample in a 35 mm glass-bottom dish with a #1.5 glass bottom.
2. Fill the dish with ethyl cinnamate to completely submerge the sample.
3. Place the dish on the confocal and adjust the focus to identify the best plane that gives the brightest signal.
4. Acquire a tile image on the brightest plane to capture a large area of the sample, using 1024 x 1024 format and speed 600.
   NOTE: Use a lower objective (10x or 20x) for the tile image to decrease imaging time and file size.
5. Based on the tile image, select regions of interest to collect a higher resolution z-scan.
6. Capture a z-scan through the cleared tissue, covering the whole depth of the tissue at the best resolution of the microscope, using 1024 x 1024 format and speed 600.
   NOTE: Use a 40x objective in oil immersion to give the best resolution for HSC.
7. Save image files in LIF format and view files using Fiji or ImarisViewer 11.0.0.

5. Image analysis for 3D reconstruction and quantitative analysis of individual HSCs

1. Convert image files to IMS format using the ImarisFileConverter 10.2.0.
2. Use the Surface model on Imaris 10.2.0 to detect the object in the images.
3. Use the Create wizard to perform analysis on a selected 3D region of interest, then apply it to the entire image.
   NOTE: Use this to speed up the analysis process.
4. Select the source channel, absolute intensity, and uncheck the smooth button to define the intensity threshold for the whole image.
5. Modify the threshold to capture all the objects in the image.
6. Use the number of voxels to filter out debris objects.
7. Finalize the wizard, press the statistic button to obtain all the statistical measurements.

Successful perfusion and fixing of the mouse liver are represented by a uniform pale, beige-yellow color of the liver, reflecting efficient removal of blood and penetration of the fixative (Figure 1A,B). Improper perfusion results in dark red or brown regions of blood clots, which reduce the optical clarity and lead to a high background signal in these sample regions (Figure 1C).

During tissue clearing, the tissue samples maintain a pale, beige appearance from perfusion through to secondary antibody incubation. Tissue samples transform into a whiter color after secondary antibody incubation compared to the previous steps. Complete optical transparency is achieved when the samples are incubated in DBE (Figure 2). The tissue appears clear, with a hard, glass-like texture, making it easy to manipulate for high-resolution deep-tissue imaging.

Cleared samples stained with antibodies against TdTomato-labeled HSCs maintain strong and stable fluorescence signals under confocal microscopy. Incomplete penetrance of antibodies or incompatible antibodies results in a high background signal in the final images17. Tile scans allow rapid assessment of labeling coverage and capture a large area of the sample, while high-resolution z-stacks acquired with a 40x oil-immersion objective reveal the fine dendritic processes of individual HSCs extending from the cell body (Figure 3 and Figure 4). Multiplexing antibodies of different species allows simultaneous capture of TdT-labeled HSCs and other liver cell types, mapping their physical interaction in 3D space (Figure 3C and Figure 4). F4/80 antibody staining was used to show macrophages interaction with TdT-labeled HSCs (Figure 3C). The sinusoids were stained using CD31 antibody, concurrently with TdT-labeled HSCs, demonstrating the spatial architecture of the liver microenvironment (Figure 4).

The same clearing protocol can be applied to diseased models with no detectable changes in the clearing results. A chronic chemical injury mouse model, with 3 times weekly CCl4 intraperitoneal injection, was used to depict liver fibrosis22. Both control and chronic CCl4 liver samples show strong and stable fluorescence signals under confocal microscopy (Figure 4). Activated HSC in the fibrosis sample exhibited the expected flat, sheet-like myofibroblast morphology. The fibrotic liver also does not affect the process of multiplexing different antibodies, indicated by the clear fluorescence signal of both TdT-labeled HSCs and CD31-labeled endothelial cells in control and chronic CCl4 samples. The results clearly depict HSC interactions with the liver microenvironment (Figure 4).

3D reconstruction of individual HSCs was done using the Surface tool in IMARIS software, showing the high-resolution visualization of HSC morphology. 3D image reconstruction captured individual HSC cellular projections, resembling neurons (Figure 5).

Figure 1: Portal vein cannulation and liver perfusion. (A) Successful insertion of the catheter into the portal vein, showing blood backflow into the catheter as indicated by the white arrow. (B) Well-perfused and fixed liver showing uniform pale, beige-yellow color. (C) Improper perfusion leads to leftover blood and brown regions, indicated by the blue arrows. Please click here to view a larger version of this figure.

Figure 2: Liver tissue samples before and after clearing. Samples after clearing are optically transparent with a hard, glass-like texture. Please click here to view a larger version of this figure.

Figure 3: 3D and tiled images of HSCs and macrophages. (A) 3D image of TdT-labeled HSCs showing complete cell morphology. (B) Tiled images of TdT-labeled HSCs capture a large area of the tissue. (C) 3D image of TdT-labeled HSCs and F4/80-stained macrophages. Zoomed in on selected regions below. Scale bars represent 30 μm (A,C, top), 5 μm (A,C, bottom), 100 μm (B, top), and 50 μm (B, bottom). Please click here to view a larger version of this figure.

Figure 4: 3D images of TdT-labeled HSCs and CD31-stained endothelial cells in healthy and chronic CCl4 mouse models. Scale bars represent 30 μm. Please click here to view a larger version of this figure.

Figure 5: 3D image of TdT-labeled HSCs and the corresponding IMARIS surface segmentation. Different colors represent different objects detected by the IMARIS surface tool. Scale bars represent 30 μm. Please click here to view a larger version of this figure.

Table 1: iDISCO+ based tissue clearing buffers and compositions.

This protocol describes a reproducible and adaptable sample preparation pipeline leading to the three-dimensional visualization and quantitative analysis of HSC morphology in intact liver tissue. The integration of genetic fluorescent cell labeling, in situ liver perfusion, and a modified iDISCO+ clearing method enables imaging of large tissue volumes at single-cell resolution while preserving the spatial architecture of the liver microenvironment.

A critical step for the success of this method is efficient liver perfusion. Incomplete perfusion can trap blood and lipids, leading to poor clearing and increased autofluorescence^23,24,25. Successful insertion of the catheter allows the backflow of blood (Figure 1A). In case of no blood backflow, use a syringe to fill the catheter hub with 1x PBS to avoid air bubbles traveling into the liver during perfusion, which will block the vessels and prevent further perfusion of the liver. Inserting the catheter too far up the portal vein, nearing the lobe, can lead to incomplete perfusion of the entire liver. Inserting the catheter into the portal vein is a tricky step that can lead to accidental puncture of the vein and/or surrounding tissue. If that happened, leave the portal vein catheter in place and quickly insert another catheter in the inferior vena cava, then nick the portal vein to release pressure, and proceed with liver perfusion in the reverse direction (i.e., from inferior vena cava through the liver then out the portal vein). Another key factor is complete submersion of the samples in solution during all the iDISCO+ steps to avoid sample oxidation, leading to a high imaging background¹⁷. During antibody labeling, rotation of the samples is essential to ensure uniform antibody penetration through the dense liver tissue. This method can be used for multiplex imaging to detect different liver cell types of interest (Figure 3C and Figure 4).

Compared with conventional two-dimensional immunofluorescence or section-based imaging, this approach allows direct visualization of the full 3D architecture of HSCs and their physical interactions with neighboring HSCs and other liver cells, confirming autocrine or paracrine signaling. It also overcomes the limitations of cytoskeletal markers like desmin, which incompletely portray HSC morphology in thin sections.

However, the method has several limitations. This method is a several-week-long process that requires advanced planning. The iDISCO process has no intermediate steps to validate the outcome until the final imaging step and has limited stopping points. Antibody compatibility with the clearing solutions is the main bottleneck that leads to low-quality images and the inability to visualize the samples¹⁷. To mitigate this, usage of antibodies with a strong immunofluorescence signal on formalin-fixed, parafilm-embedded (FFPE) or frozen sections (with methanol-based fixation/antigen retrieval) is recommended, but even then, it is a trial-and-error process with most antibodies being incompatible with clearing.

Beyond liver fibrosis, this workflow is broadly applicable to the study of fibroblasts and their interaction partners in other organs such as the lung, kidney, and heart, where these cells harbor distinct morphologies that may contribute to their profibrogenic roles^19,23,26. By enabling accurate 3D morphometric analyses, this platform provides a foundation for exploring how fibroblast structure relates to their function and contributes to tissue homeostasis and disease.