Mesoscale visualization of three-dimensional microvascular architecture and immunocyte distribution in intact mouse liver lobes

Rational: The complex vascular architecture and diverse immune cells of the liver are critical for maintaining liver homeostasis. However, quantification of the network of liver vasculature and immunocytes at different scales from a single hepatic lobule to an intact liver lobe remains challenging. Methods: Here, we developed a fast and fluorescence-preserving transparency method, denoted liver-CUBIC, for systematic and integrated analysis of the microcirculation and the three-dimensional distribution of dendritic cells (DCs)/macrophages in intact liver lobes. Results: Whole-mount imaging at mesoscale revealed that the hepatic classical lobule preferred the oblate ellipsoid morphology in the mouse liver and exhibited hepatic sinusoids with heterogeneous arrangement and intricate loop structure. Liver fibrosis not only induces sinusoidal density increase but also promotes sinusoidal arrangement with increased sinusoidal branch and loop structure. Meanwhile, we found that CD11c+ DCs followed a lognormal distribution in the hepatic lobules, skewing toward lobular boundary in steady state. CCl4-induced chronic liver injury promoted CD11c+ DC rearrangement at the lobular border before the formation of liver fibrosis. Furthermore, through whole-mount imaging of tumor-immune cell-vascular crosstalk in intact lobes based on liver-CUBIC, we characterized an accumulation of CX3CR1+CCR2+F4/80+ macrophages at metastatic foci in early colorectal liver metastases. Importantly, colorectal cells secrete CCL2 to mobilize CX3CR1+CCR2+F4/80+ macrophages to accumulate at liver micrometastases, and the interruption of CCL2-induced macrophage accumulation inhibits early colonization of metastasis in the liver. Conclusions: Our investigation of the sinusoidal network and DC/macrophage arrangements through the liver-CUBIC approach and whole-mount imaging provide a powerful platform for understanding hepatic circulatory properties and immune surveillance in the liver.

Step-2 The mice were transcardially perfused with approximately 100 mL of lipase saturation solution for 1 h at room temperature. For preparation of the lipase saturation solution, 200 mg lipase (Sigma, cat: L3126) was fully stirred for dissolving in 100 mL PBS within 1 h, and the solution was then centrifuged at 4000 rpm, 10 min, at room temperature. The supernatant was then filtered through with a 70-μm filter before perfusion.
Step-3 The mice were transcardially perfused with 4% PFA for 10 min at room temperature.
Step-4 The harvested organs (e.g., liver, spleen, kidney, and thymus) of mice were immersed in 4% PFA for more than 12 h, at 4 °C.
Step-5 The fixed organs were then washed three times (at least 2 h per wash) with PBS to remove PFA before clearing at room temperature.
Step-6 The livers were immersed in 50% CUBIC-L (mixed with water) for delipidation for more than 6 h under 50-60 rad/s shaking at 37 °C.
Step-7 The livers were immersed in CUBIC-L for 2 days (until the solution became clear) with shaking at 37 °C, and the solution was changed every 24 h.
Step-8 The livers were washed three times (more than 2 h every time) in PBS for volume retraction at room temperature.
Step-9 The livers were immersed in 50% CUBIC-R (mixed with water) for RI matching for more than 6 h at 4 °C.
Step-10 The liver lobes were immersed in CUBIC-R (the pH of CUBIC-R was adjusted to 9.5 with NaOH) for 1 day at 4 °C.

3D immunofluorescent staining protocol of liver-CUBIC
Step-1 The liver after delipidation with the CUBIC-L solution was collected and washed three times (more than 2 h each time) in PBS at room temperature.
Step-2 The liver was immersed in HEPES-TSC buffer (containing 10 mM HEPES, 200 mM NaCl, 0.5% casein, and 10% Triton X-100) with shaking for 1.5 h. At the same time, the antibody solution (containing primary antibodies and secondary antibodies) was diluted in HEPES-TSC with shaking for 1.5 h at room temperature.
Step 3 The liver was immersed in the antibody dilution for 2 days at room temperature and 1 day with shaking at 4 °C for staining.
Step-4 The liver was washed twice in 0.1 M PB with 0.1% TritonX-100 for 30 min at room temperature.
Step-5 The liver was washed twice in 0.1 M PB for 1 h at room temperature.
Step-6 The liver was postfixed with 1% (w/v) PFA (pH 7.4) in PB overnight at room temperature.
Step-7 The liver was washed twice in 0.1 M PB (more than 1 h each time) at room temperature.
Step-8 The liver was immersed in 50% and 100% CUBIC-R for RI matching at room temperature.

3D immunofluorescent staining protocol of iDISCO
Step-1 The fixed liver was washed three times (more than 30 min each time) in PBS at room temperature.
Step-3 The liver was incubated in 33% methanol/66% DCM with shaking overnight at room temperature.
Step-4 The liver was washed twice in 100% methanol (more than 30 min each time) with shaking at room temperature.
Step-5 The liver was immersed in 5% H2O2 in methanol for 12 h at 4 °C.
Step-15 The liver was incubated in 33% methanol/66% DCM with shaking for 3 h at room temperature.
Step-16 The liver was immersed in DiBenzyl Ether (sigma, 108014) for RI matching.

Confocal imaging
For imaging of the cleared liver lobe, an LSM 780 NLO confocal microscope (Zeiss, Germany) equipped with a 10× water immersion objective (N.A. 0.45) and 32 anode Hybrid-GaAsP detector was used. First, the 50% CUBIC-R solution containing 2% agarose (defined as CUBIC-agarose solution) was heated in a microwave oven until the liquid boiled and the agarose was dissolved completely. Next, 5 mL of CUBIC-agarose solution was poured into a petri dish (55 mm in diameter and 17 mm in depth). When the gel temperature decreased to approximately 37 °C , the RI-matched liver lobe was placed into the CUBIC-agarose solution. The CUBIC-agarose solution did not need to flood the sample. When the CUBIC-agarose solution solidified, the transparent sample was embedded in a petri dish and then covered with 3-5 mL of CUBIC-R solution for imaging with an LSM780 microscope equipped with a 10× objective.
For imaging and segmentation of the HV, PV, and HA, hepatic lobules, DCs/macrophages, and liver micrometastasis, 3D imaging of an intact liver lobe was performed at 26.3-µm z-steps in a field of approximately 10  10  2.3 mm 3 , and the imaging conditions were adjusted according to the fluorescence intensity of different samples. The microscope zoom was 1×. For segmentation and reconstruction of the hepatic sinusoid structure, imaging was performed at 2-µm z-steps in a field larger than 1.2 × 1.2 × 1.2 mm 3 . The microscope zoom was set to 0.7×. Spectral imaging and linear unmixing were used to distinguish autofluorescence, vascular tissue and immune cells.
To compare the retention of the fluorescence signal (YFP and RFP) after liver clearing between liver-CUBIC and CUBIC, imaging was performed at 5-µm z-steps in a field of 1024 × 1024 pixels, and the imaging conditions were kept consistent for different samples without spectral imaging and linear unmixing. The microscope zoom was set to 1×. The mean fluorescence intensity of YFP and RFP cells in the liver lobes was further investigated and calculated using Imaris software. For imaging of immunostained liver sections, an LSM 710 confocal microscope (Zeiss, Germany) equipped with a 20× dry immersion objective (N.A. 0.8) was used. Images were acquired over a z-range deeper than 20 µm in 3-µm z-steps. The excitation wavelengths were 405 nm for DAPI and BV421, 488 nm for GFP and YFP, 561 nm for tdTomato, PE and Alexa Fluor 594 (AF594), and 633 nm for AF647 and APC.

Vessel and hepatic lobule segmentation
All segmentation of hepatic vessels and lobules was performed using Imaris software (Bitplane, Belfast, UK). In brief, the AF647-CD31Ab-labeled vessels were first automatically segmented and reconstituted using the "Surfaces" module with "absolute intensity" thresholding. Based on the existing vascular architecture of the HV, PV, HA, and CV obtained by micro-CT, the liver blood vessels are composed of three components, namely, the HV, the PV, and the HA. In the branched structure of large blood vessels, the HV has the largest average diameter, followed by the PV, and the diameter of the HA is significantly smaller than that of the liver venous system. In addition, the liver CV is located in the center of the liver lobular structure and is the branch of the HV. Thus, according to these features, the HV, PV, and HA structures were further manually segmented using "Mask Properties" in "Surface". The hepatic lobule was manually segmented according to the distribution of the liver PV and HV (the PV was distributed at the edge of the hepatic lobule, and the branch of the HV was located in the center of the hepatic lobule) using the "Surfaces" module in Imaris.

Quantification of the CD11c + cell spatial distribution in the hepatic lobules
We analyzed the distribution of CD11c + cells in 97 hepatic lobules in the caudate lobe of CD11c-Venus mice. First, we analyzed the distance between CD11c + cells and the CV of the hepatic lobule using Imaris software; the cells less than 20 μm from the CV were defined as CD11c + cells distributed around the hepatic CV. Quantification of the proportions of these cells was then performed using GraphPad Prism. In addition, we calculated the spatial distribution index of the remaining CD11c + cells in the hepatic lobules. As shown in Figure 5, the distribution index was calculated as = 2 where D1 is the distance between CD11c + cells and the central vessel of hepatic lobules and D2 is the distance between CD11c + cells and the boundary of hepatic lobules.
According to the calculated distribution index, we calculated the logarithm of the distribution index and the probability of CD11c + cells and simulated the distribution of CD11c + cells through exponential growth, Gaussian and lognormal models using GraphPad Prism. The distribution of CD11c + cells under different bin values (bin = 0.02, 0.04, 0.06, 0.08, and 0.1) was verified.

Topological analysis of the hepatic sinusoids
To quantitatively analyze the topology of the hepatic lobules, the cleared lobes were imaged in a volume (X-Y-Z) larger than 1.2× 1.2 × 1.2 mm 3 at a resolution of 1.19 × 1.19 × 2 μm 3 /voxel size (X-Y-Z). The imaging data were first processed with a "Gaussian filter" using Imaris, and the lobule regions were then manually segmented based on the location of the CV, sinusoids and PTs. The sinusoids were then automatically segmented and processed through binarization using Imaris with "background subtraction" thresholding. The binarized sinusoid data were then processed with "AutoSkeleton" in Amira software (Thermo Fisher) to extract the sinusoid network parameters in the lobules. For additional network analysis, the exported parameters were processed and analyzed using Network Analyzer (Cytoscape 3.7.2).

Flow cytometry
The mice were first perfused with PBS to remove blood from the liver. The livers were then collected, cut into pieces, and subjected to enzymatic digestion at 37 °C with 0.5 mg/mL collagenase type IV (Worthington, LS004188) and 0.1 mg/mL Liberase TM TL

Statistical analysis
GraphPad Prism was used for the statistical analyses. The error bars denote the SDs or SEMs. Unpaired t test or Mann-Whitney U test was used for comparisons of groups.
Significant differences between the groups are indicated as follows: ns for no significant difference, * for P < 0.05; ** for P < 0.01; *** for P < 0.001; and **** for P < 0.0001.  cell at each depth. These data show that the average RFP intensity of each cell in the liver-CUBIC group was 1.98-3.14-fold higher than that in the CUBIC group, and the average YFP intensity of each cell in the liver-CUBIC group was 1.27-1.91-fold higher than that in the CUBIC group. The data were collected from 9 image regions. (C-D) The total RFP and YFP fluorescence intensity of cells in the liver-CUBIC group was 2.55-to 4.45-fold and 1.57-to 4.03-fold higher than that in the CUBIC group, respectively (n=9 measurements per group). The error bars denote the SEMs; Mann-Whitney U test.