Five‐in‐One: Simultaneous isolation of multiple major liver cell types from livers of normal and NASH mice

Abstract NASH is a chronic liver disease that affects 3%–6% of individuals and requires urgent therapeutic developments. Isolating the key cell types in the liver is a necessary step towards understanding their function and roles in disease pathogenesis. However, traditional isolation methods through gradient centrifugation can only collect one or a few cell types simultaneously and pose technical difficulties when applied to NASH livers. Taking advantage of identified cell surface markers from liver single‐cell RNAseq, here we established the combination of gradient centrifugation and antibody‐based cell sorting techniques to isolate five key liver cell types (hepatocytes, endothelial cells, stellate cells, macrophages and other immune cells) from a single mouse liver. This method yielded high purity of each cell type from healthy and NASH livers. Our five‐in‐one protocol simultaneously isolates key liver cell types with high purity under normal and NASH conditions, enabling for systematic and accurate exploratory experiments such as RNA sequencing.


| INTRODUC TI ON
Non-alcoholic steatohepatitis (NASH) is a chronic liver disease that affects 3%-6% of individuals in the general population and has no available FDA-approved medicines. 1 Although the 'two-hit' theory is well acknowledged in the field, 2 novel pathway discovery in NASH pathogenesis is in high demand. A recent review emphasized the contribution of hepatocyte-macrophage-stellate cell crosstalk in fibrogenesis, 3 and another report recognized hepatic stellate cells (HSCs) as a hub of intracellular signalling via stellakine secretion. 4 Unfortunately, the detailed signalling within a specific cell type and complex crosstalk among non-parenchymal cell (NPC) types are still largely unknown. Consequently, further understanding of how different cell types contribute to NASH pathogenesis is urgent for the identification of novel therapeutic targets.
To date, several methods have been established for liver cell subtype isolation, most in normal livers. 5,6 The key step is density gradient centrifugation after a two-step collagenase perfusion. 7 However, due to the density similarity between liver endothelial cells (LECs) and hepatic stellate cells (HSCs), this method alone failed to achieve satisfying purity. Furthermore, due to the physiological and biological changes of diseased liver cells, especially HSCs, the isolation of cells from normal and disease livers using the reported protocol is challenging, and isolated cell types are limited. 8,9 Fernández-Iglesias and colleagues described a high-purity method in isolating hepatocytes, LECs, HSCs and macrophages simultaneously in healthy and cirrhotic rat livers via centrifugation and sequential plating on coated dishes. 10 This method is applicable for in vitro study but not for experiments geared towards understanding in vivo biology.
Currently, no methods have been reported to isolate both parenchymal and non-parenchymal cell types in NASH livers for in vivo biology study.
Here, utilizing single-cell RNA sequencing-based cell surface markers, we describe a technique that can simultaneously isolate five key cell types in one liver from healthy and NASH mouse models with high purity, enabling the accuracy of downstream mechanistic studies within or among cell types.

| Liver perfusion and digestion
Mice were anaesthetized and perfused with the Liver Perfusion Buffer (Gibco) and then digestion buffer (HBSS containing 5 mM CaCl2, 10 mM HEPES and 0.2 mg/ml collagenase I & II (Liberase TM, Roche)). After digestion, the liver was excised and maintained in the Hepatocyte Wash Media (Gibco) on ice.

| Isolation of hepatocytes and nonparenchymal cells
Hepatocytes were released from liver lobes and then filtered through 70μm cell strainers (Corning). After centrifugation and wash, hepatocytes were resuspended in 40% (chow) or 30% (NASH) Percoll (Sigma) in Williams E Media (Gibco) and centrifuged to collect live hepatocytes on the bottom ( Figure 1A).
The remaining liver lobes were minced and further digested by 2.5 mg/mL collagenase D (Roche) and 100 ng/ml DNAse I (Sigma) at 37°C for 20-30 min. After centrifugation and wash, NPC pellets were loaded on top of Percoll gradient layers (10mL 15% Percoll on the top and 10mL 40% Percoll on the bottom) for further centrifugation. NPC layer in the middle was collected ( Figure 1A).

| NPC cell sorting and imaging
NPCs were incubated with Fc-receptor blocking antibody first, then subsequently stained with antibodies anti-CD31, anti-CD45, anti-F4/80 and anti-CD140b (Table S1) for sorting. Sorts were performed on BD FACSAria TM Fusion. An unstained hepatocyte sample was run to minimize hepatocyte contamination in the collected fractions.

| Statistical analysis
Data were normalized from three independent experiments. All data were analysed by GraphPad Prism software.
For more methods, please refer to Supplementary Material.

| Mice on HFHFD and HFHFD+CCL4 developed NASH phenotype
As shown in Figure S1, the two NASH models developed NASH histology hallmarks, including liver steatosis, inflammation and fibrosis ( Figure S1).

| Yield and viability
As shown in the FACS plots (Figure 1), different cell types were isolated from both chow and NASH livers. Using this method, we were able to isolate hepatocytes >90% viability from mice on chow and >75% viability from mice on NASH models. The viability of NPCs was consistently >99% across all isolated cell types from chow and NASH models. Cell yields were variable among different cell types (Table S2).

| High purity of isolated cells from normal and NASH livers
To test the purity of yield, we performed gene expression and morphological studies on isolated cells (

| DISCUSS ION
Though NASH affects 3-6% of individuals in the United States, deep understanding of its pathogenesis is limited. Liver cell type isolation has been achieved by multiple methods, but none of these techniques were applied to NASH livers. Here, we report a Five-in-One method that can simultaneously isolate five cell types in healthy and NASH livers.
We could not isolate HSCs by autofluorescence alone or auto-fluorescence+/CD45-( Figure S4), a method suggested by Tacke and Schwabe groups. 8,9 This is likely due to the disparity of isolation

ACK N OWLED G EM ENT
We thank Sean D'Italia and Dominic Vergata for their help in sample harvesting.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest to disclose. All authors are employees and shareholders of Regeneron Pharmaceuticals.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

Ye Zhou
https://orcid.org/0000-0003-3861-3120 F I G U R E 2 Purity and morphology evaluation on liver cells collected from mice on chow and HFHFD NASH model. Relative gene expression of (A) Uox (Hepatocyte marker), (B) Adgre1 (macrophage marker), (C) Ltb (TBNK marker), (D) Cyp4b1 (endothelial cell markers) and (E) Pdgfrb & (F) Col3a1 (HSC markers) in isolated cell types. Data were normalized by setting the strongest expressed cell type as 100%. Endothelial cells are marked in yellow, macrophages (Mφ) are in cyan, non-macrophage immune cells (TBNK) are in red, and HSCs are in green. (G) Bright field (left) and Nile Red staining (right) reveal morphology changes and lipid accumulation in steatotic hepatocytes. H, Albumin (red) and DAPI (Blue) staining of primary hepatocytes shows the purity of isolated hepatocytes. Arrows point to non-hepatocyte cells with negative albumin signals. I, Images of single NPC cells captured from flow cytometry display morphologies and clear cell surface marker staining. Data represent mean ± SEM. n = 3 independent experiments. Scale bar is 50 μm in (H) and 7 μm in (I).