Isolation of primary human liver cells from normal and nonalcoholic steatohepatitis livers

Summary Here, we present a protocol for isolating human hepatocytes and neural progenitor cells from normal and nonalcoholic steatohepatitis livers. We describe steps for perfusion for scaled-up liver cell isolation and optimization of chemical digestion to achieve maximal yield and cell viability. We then detail a liver cell cryopreservation and potential applications, such as the use of human liver cells as a tool to link experimental and translational research.


KEY RESOURCES TABLE
Timing: 30 min for plating, 2-3 weeks for cells to reach 90% confluence (for step 19b).
Timing: 1 h to prepare for control-rate step down with 1-2 h in freezer (for step 19c).
CRITICAL: All liver tissues are obtained in accordance with federal, state, and institutional regulations. Once the donor liver is transferred into the lab, all procedures are then performed in a biological safety cabinet (Figures 2A-2C). PPE, shoes cover, gloves and bouffant cap are required.
1. Turn on biological safety cabinets and run for 10 min before using. Remove all materials and equipment from the cabinet. Clean the interior surfaces of the cabinet with an appropriate disinfectant, following the manufacturer's instructions. Turn on biological safety cabinets and run for 10-30 min before using to purge any remaining contaminants from the air. 2. Place peristaltic pump and filled water bath into the biological safety cabinet. 3. Turn on water bath and warm to 39 C. 4. Prepare perfusion solutions and place them in the water bath to warm up 20 min before starting the perfusion.
CRITICAL: The required volumes of solutions depend on the size of the liver. Dependent on the flow rate, the max flow rate used for 1 perfusion tubing/lobe is 30 mL/min. Only one tubing is usually used for perfusion of the left lobe at the perfusion rate of 30 mL/min for 20 min for max volume of 600 mL.

5.
Load perfusion tubing into the peristaltic pump, with great care to maintain sterility, and place all ends into the Buffer A solution bottle. 6. Turn on the pump to prime the tubing and allow the solution to circulate continuously. 7. Cover a pan of ice with a plastic bag and place it into a second biological safety cabinet. 8. Place a sterile field over the ice pan and then place the liver on top of the sterile field. 9. Place a second sterile field next to the ice pan with all surgical instruments and needed supplies opened and placed on the sterile field aseptically. 10. Don sterile gloves, pick-up the liver and examine for any lacerations, bruising, and remarkable abnormalities.
CRITICAL: If gall bladder is still attached, use a scalpel to remove it ( Figure 2B). Also, remove excess fat, muscle, and diaphragmatic tissues from the liver (Methods video S1).
11. Resect the tissue into three pieces as shown in ( Figure 3A) (Methods video S1).
CRITICAL: Use left lobe for hepatocyte isolation, use central lobe for snap frozen tissue and paraffin blocks, and use right lobe for non-parenchymal cell (hepatic stellate cell, Kupffer cell and endothelial cell) isolation.
12. Insert barbed catheters into the major portal and/or hepatic vessels but do not secure. 13. Connect a 60 mL syringe, filled with cold EMEM medium to the catheters, flush the liver tissue with EMEM medium to determine which vessels provide the most uniform perfusion of the liver, and to remove the residual blood (Methods video S2).
CRITICAL: Multiple vessels can be used but use as few vessels as possible while achieving maximal inflation of liver.
14. Secure the catheters into the selected vessels.

OPEN ACCESS
Note: For this step, barbed luer adaptors can also be used to better secure the tubing in the vessel.
a. Place 6-7 loose loop stiches through the blood vessel wall opening. b. Place the catheter (or luer adaptor) inside the vessel and tighten the suture around the blood vessel. c. Attach the tubing to the catheter. d. Place multiple loops of suture, tightening with a knot each time for the final securing of the perfusion tubing. e. Seal the cut surface of the liver tissue with sutures (size 2-0) or surgical grade glue.
Note: The cut surface must be sealed to achieve full inflation of the liver (Methods videos S3 and S4) CRITICAL: All remaining major vessels on the cut surfaces must be closed with sutures or surgical grade glue.
15. Flush the liver tissue again using the syringe filled with cold EMEM to test for any major leaks from the liver.
CRITICAL: Close any leaks with glue or sutures (Methods videos S3 and S4).
16. Place liver tissue with catheters in a sterile plastic bag. Weigh the liver tissue with the bag closed. 17. Place the bag containing the tissue into the 39 C water bath (Methods video S5), connect to the peristaltic pump with flow rate 20-30 mL/min and perfuse the tissue with Buffer A for 10-15 min ( Figure 3B).
CRITICAL: EGTA chelates calcium that leads to the separation of cell junctions and helps to remove any residual blood. Flow rate is based on the liver tissue size and number of catheters. Left lobe usually needs 500-600 mL of each perfusion buffer for complete digestion.
Note: To ensure constant temperature during digestion, the liver lobe should be placed in a plastic bag and submerged in the water bath alongside the perfusion solutions ( Figure 3B). A key step in the procedure is successful suturing and stabilization of the perfusion setup to avoid dislocation of the perfusion tubing during the perfusion ( Figures 3A and 3B, and Methods videos S1, S2, and S3). Inflation of the whole liver lobe indicates successful insertion of tubing after beginning perfusion with buffer A Note: After perfusion with buffer A, vessels of the liver lobe should be a light brown color due to the EGTA chelating the remaining blood. This step helps to avoid contamination with red blood cells and clotting of hepatic vessels, which may result in incomplete perfusion and digestion. Aspirating buffer A from the bag before changing to the next buffer. Stop the pump while changing between buffers to prevent air. After perfusion with buffer B and/or C, the liver tissue starts fracturing under the capsule. The perfusion should be terminated after separation of liver tissue from the capsule is observed ( Figure 3B).
18. Choose from the following options, depending on the type of cells to be isolated. Human hepatocytes are processed using option a. For non-parenchymal cells follow option b. a. Parenchymal cell isolation Note: For parenchymal cell isolation, it is important to calculate the proper Percoll concentration to be used depending on the pathology of the liver to ensure adequate quantities of live cells are collected ( Figure 3D). When performing the density gradient centrifugation, it is important to proceed with the break off to avoid contamination of live cells with dead cells.
While washing the parenchymal cell population, the cell pellet must be resuspended gently ( Figures 3E-3G).
i. Perfuse the liver tissue with Buffer B for 15-20 min.
CRITICAL: Stop when liver tissue fractures and separates from the liver capsule.
ii. Transfer the liver from the plastic bag into a sterile container containing pre-warmed (37 C) DMEM supplemented with 10% FBS, 25 mM HEPES, 100 nM Dexamethasone, 13 ITS (Insulin-Transferrin-Selenium), 1% Pen/Strep. iii. Dissociate digested liver tissue from the liver capsule with sterile scissors or by sterile gloved hand to release hepatocytes. iv. Filter cell suspension through sterile three-layer nylon mesh-covered funnels to remove cellular debris and clumps of undigested tissue.
CRITICAL: Three-layer nylon mesh from inside to outside size 500 mm, 250 mm, 85 mm, and must be autoclaved before use. Repeat Step 18a(iv) as many times as needed to obtain the maximum number of cells ( Figure 3C).
Note: After the digestion, the liver is carefully removed and minced under sterile conditions ( Figure 3C). The liver cell suspension is then filtered through a 3-layer nylon mesh 500, 250 and 85 mm (outside to inside) to separate undigested tissue remnants, and is washed with warm supplemented DMEM ( Figure 3C and Methods video S6). The final step of the isolation procedure is a density gradient centrifugation including collection and counting of nonparenchymal and parenchymal cell types. vii. Perform low speed centrifugation at 60-80 g for 5 min at 4 C to isolate hepatocytes from other cell types in the suspension ( Figure 3D).

OPEN ACCESS
viii. While waiting during the centrifugation step, weigh the bag with catheters and subtract from the previous weight in step 16 to determine the true liver weight. ix. After centrifugation is complete, aspirate the supernatant leaving the hepatocyte pellets.  iv. Remove the liver tissue from the plastic bag and place in a sterile plastic beaker containing ice-cold EMEM. Then gently cut with sterile scissors to release cells. v. Filter the cell suspension through sterile three-layer nylon mesh-covered funnels to remove cellular debris and clumps of undigested tissues.
CRITICAL: Three-layer nylon mesh from inside to outside size 500 mm, 250 mm, 85 mm, and must be autoclaved before use. Repeat this step as many times as needed to obtain the maximum number of cells ( Figure 3C).
vi. Transfer filtered NPC suspension into 50 mL conical tubes. Remove hepatocytes from the other cell types in the suspension by low-speed centrifugation at 60-80 g for 5 min at 4 C ( Figure 3D).
CRITICAL: Centrifugation speed depends on the fat content of the contaminating hepatocytes. Fattier liver needs higher centrifugation speed.
vii. While waiting during the centrifugation step, weigh the bag with catheters and subtract from the previous weight in step 16 to determine the true liver weight. viii. Decant the supernatant and centrifuge at 935 g for 8 min at 18 C to collect total NPCs ( Figure 3H). ix. Resuspend each pellet in 10 mL EMEM, and bring each tube up to 50 mL with EMEM.
Spin again at 700 g for 8 min ( Figure 3I). x. Discard supernatant, add 10 mL GBSS/B to resuspend the pellet, add 15 mL Nycodenz I, and bring the volume to 50 mL with GBSS/B. Mix well by inversion. This cell suspension will be used in the next step ( Figure 3I).

OPEN ACCESS
CRITICAL: Nycodenz I concentration must be adjusted based on the fat content in the liver. This is a critical step; the final concentration should be 10.4%-11%.
xi. Layer 10 mL Nycodenz II on the bottom of 50 mL tube. Then, layer 30 mL cell mixture from step 18b(x). Finally, slowly layer another 10 mL of GBSS/B onto the cell mixture using a 20 mL syringe with 19-gauge needle held against the inside of the tube wall to avoid mixing the layers. Combine the remaining 20 mL of cell mixture from step 18b(x) to repeat the procedure in step 18b(xi). Centrifuge 2000 g for 20 min at 4 C (Figure 3I).
CRITICAL: Centrifugation without brake is critical.
xii. Under the top layer of clear GBSS/B solution is a white layer. This layer contains enriched hepatic stellate cells (HSCs). Collect this layer by carefully inserting a pipet through the clear GBSS/B layer, and transferring the HSC layer to new 50 mL tubes ( Figure 3I). Step down Program vii. Transfer tubes into liquid nitrogen for long-term storage (vapor phase only).

xiii. The second layer contains LEC and myeloid cells (Kupffer cells (KC) and bone-marrow derived myeloid cells
Note: HSCs can be identified by expression of Vitamin A (which is rapidly photobleached by the UV light) using fluorescent microscopy and immunocytochemistry for expression of GFAP and aSMA (Figure 4). 2 qRT-PCR is another method which provides a sensitive assessment of contamination with specific cell types.

Validation of isolated human cells Hepatocytes
Hepatocytes are identified by their characteristic morphology, multiple nuclei, and expression of Pan-CK, keratin 18 (K18), and HNF4a. Hepatocytes produce albumin, a-macroglobulin, transthyretin (TTR), cytochromes CYP4E3 (the marker of functional detoxification), which can be detected by qRT-PCR, enzymatic reactions, and Western blotting. Among the numerous enzymatic systems involved in hepatocyte metabolism, cytochromes P450 (such as CYP3E1) are upregulated in response to injury or stress. These cytochromes are monoxygenases mainly expressed in the liver. 3 They oxidize a number of compounds and generate cytotoxic or genotoxic metabolites responsible for various conditions, such as AALD. 4

Non-parenchymal fractions
The composition of NPCs is best characterized using flow cytometry (Figures 5A and 5B). When the composition of non-parenchymal cells from normal and diseased donors was compared (vs. commercially available NPCs from Lonza Bioscience; https://bioscience.lonza.com/) similar cellular subsets were detected. As expected, the number of myeloid cells was increased in NPCs from the diseased donor, while the number of CD3 + T cells was decreased. Although LSECs were tested for expression of CD31, the percent of LSECs was low in both populations ( Figure 5C). It is also recommended to test expression of CD146 in LSEC vs. all endothelial cells as expression of CD146 is associated with LSECs. CD146 and others such as Stabilin2 are good markers for LSECs. CD31 also detects endothelial cells from larger blood vessels, and these are not LSECs. Recent studies have suggested that Scavenger receptor type B1(SR-B1) is specifically expressed by LSECs.
HSCs: HSCs are identified by expression of NGFRI, and Desmin. Freshly isolated HSCs from normal livers can also be stained for GFAP using fluorescent microscopy ( Figure 4). Differential analysis

OPEN ACCESS
should be used to rule out contamination with CD31 + LSECs, CD68 + myeloid cells, elastin + activated portal fibroblasts (aPFs) and other cell types 5 ( Figure 4B). HSCs express Vitamin A, independent of whether they are isolated from normal or diseased livers ( Figure 4A). However, following the isolation, expression of Vitamin A is downregulated in culture (starting at z9h after plating) and completely undetectable after 24 h in culture. 6,7 Downregulation of retinoids in human HSCs (compared to mouse HSCs, which retain retinoids in culture) can be explained by the increased content of retinoic acid in rodent diet vs. humans. In addition, Lrat expression (mRNA or protein) is relatively low in human HSCs vs. mouse HSCs. 8 Advantages Large quantities of human liver cells, including parenchymal and non-parenchymal fractions, can be isolated and used immediately or cryopreserved (see Applications). Moreover, non-parenchymal cells can be further fractionated (before or after cryopreservation, using cell sorting 9 or MACS technology 10,11 ).
Large quantities (billions) of parenchymal and non-parenchymal cells can be isolated from the whole liver, resulting in purification of large numbers of highly purified (>90%) and viable (>85%) cells.
Parenchymal (non-steatotic hepatocytes) and non-parenchymal fractions can be cryopreserved, and further analyzed/fractionated after thawing, the composition of viable cells can be determined using trypan blue staining.
Non-parenchymal fraction can be further fractionated using gradient centrifugation method that allows separation of Hepatic Stellate Cells (due to the presence of Vitamin A-expressing liposomes, Figure 4A) from a population enriched in myeloid cells and endothelial cells. Although the expression of Vitamin A is downregulated in activated HSCs, it is not totally suppressed ( Figure 4A), and the amount of remaining Vitamin A + lyposomes is sufficient for HSC isolation using gradient centrifugation from normal or diseased livers. HSCs can be cultured to passage one, analyzed for purity ( Figure 4B) and cryopreserved, thawed and further propagated to passage 3-4. The enriched population of myeloid and endothelial cells can be further purified using MACS beads conjugated anti-CD146 Ab and/or anti-CD31 Ab, the antibodies that are routinely used for isolation of LSECs. Anti-CD11b Ab is used for separation of myeloid cells. (C) The total NPC composition from normal and NASH livers (UCSD) and commercially available NPCs (Lonza) is shown in the pie charts. Cells that were not identified by the described markers were classified as unknown.

OPEN ACCESS
Purified cellular fractions can be cultured using the cell-specific conditions that have been previously developed and are widely used to culture endothelial cells and myeloid cells. 12

Applications
The current protocol provides a comprehensive methodology for the isolation of human parenchymal and non-parenchymal cells from normal and diseased livers. It allows the study of single cell-based gene expression profiles, epigenetic regulatory mechanisms, proteomic and other omics technologies to identify the specific pathways activated in the pathogenesis of NAFL, NASH/ AALD( Figure 4C) compared to normal livers. Isolated cells can be used to perform functional validation of the obtained targets and provide a useful tool to study human liver cells in a wide range of applications, including 2D culture activation, co-culture and functional interactions with other cell types, and 3D cultures of human liver spheroids and ''livers in a dish''.
This protocol can be used for translational research and commercial liver cell isolation. Large quantities of highly purified cells can be used for drug screening, and/or drug validation. On a smaller scale, 2D and 3D in vitro experiments can be designed to compare cells from the same or mismatched (normal vs. diseased) patients, or paremchymal vs. non-parenchymal fractions. 13 These experiments can demonstrate strong patient-specific reproducibility due to the high number of cells available, which can be cryopreserved for ongoing experiments and future use.
Given the high viability and purity of isolated cells, the responses of each cellular population to different stimuli can be studied in vitro using 2D cultures, co-cultures. 13 Responses of hepatocytes, myeloid cells and HSCs isolated from well characterized normal, NAFL, NASH, and AALD livers can be compared. Injury specific phenotypes of the major liver populations can be revealed. Similar studies can be performed using LSECs or other cellular subsets.
The gene regulation (gene expression profile and epigenetic landscape) of freshly isolated normal and steatotic human hepatocytes, myeloid cells, and HSCs can be studied using RNA-Seq or ATAC-Seq, or similar techniques, which allow simultaneous access to the areas of open chromatin within the distal and proximal promoter/regulatory elements of specific genes, and interrogate their potential co-regulation.
The correlation between the gene expression in normal vs. metabolically injured hepatocytes and activated myeloid cells and HSCs/myofibroblasts can be established. The cross talk and signaling pathways between distinct cellular subsets can be outlined.
Freshly isolated, as well as cryopreserved non-parenchymal fraction and parenchymal fractions from normal and NASH/AALD patients can be used for generation of 3D human spheroids. The mismatch of normal vs. diseased parenchymal/non-parenchymal cells can provide a useful insight into the cell interaction and signaling within human spheroids. Human spheroids can become a useful tool for drug screening.
An alternative to 3D liver spheroids is generation of bioprinted organoids that function as ''livers in a dish''. 13,14 Due to the presence of biodegradable resin, liver organoids demonstrate improved liver cell growth and interaction. They remain viable up to 40 days, develop liver-like architecture, and can be used for induction of NASH. Upon exposure to the NASH cocktail, human liver organoids develop steatosis, inflammation, and fibrosis. 3D human organoids can be used for translational research.

LIMITATIONS
The limitations discussed below are not specific to this particular human cell isolation protocol, but are applied to all perfusion/gradient centrifugation-based cell isolation techniques performed on the metabolically injured livers from patients and experimental models of NASH/AALD.
Enzymatic digestion of the whole human liver requires the perfusion of solutions into the liver under specific flow rate. Although unlikely, perfusion can cause undesired tissue damage and stretching, resulting in alterations of the gene expression profiles of isolated cells. This obstacle can be overcome by addition of the RNA polymerase inhibitor flavopiridol to the perfusion solution. 10 Liver cells prepared with or without flavopiridol can be compared in a pilot experiment using qRT-PCR or RNA-Seq to determine the importance of transcription arrest on gene expression profile of specific cellular populations.
Normal hepatocytes can be cryopreserved and used for further experiments after thawing. In general, we do not recommend to cryopreserve steatotic hepatocytes. The fragility of steatotic hepatocytes makes them susceptible to freezing/thawing artifacts. 15 Prolonged culturing can cause plastic activation of hepatocytes leading to their epithelial-to-mesenchymal transition. 17 Cell fate mapping experiments have demonstrated that hepatocytes do not give rise to myofibroblasts in vivo in response to fibrogenic liver injury. [17][18][19] Generation of 3D spheroids may provide an alternative method to investigate hepatocyte responses to stimuli.
Isolation of HSCs from NASH/AALD livers may present difficulties due to contamination with steatotic hepatocytes. 8 Although the perfusion conditions (with pronase/collagenase) are designed to lyse hepatocytes, some small number of fat-loaded hepatocytes with light buoyancy can still be found in the HSC fraction. Culturing of freshly isolated HSCs eliminates damaged hepatocyte contaminations within 3 days.
Prolonged culturing may cause HSC activation. 8 If HSCs need to be amplified for the purpose of large-scale drug screening, it is recommended to grow freshly isolated HSCs until confluent, then harvest and store after passage 1. Unlike other reports 20 (https://genome.ucsc.edu/ENCODE/ protocols/cell/human/Stellate_Crawford_protocol.pdf), we do not recommend amplifying HSCs more than passage 4, 8 as they change their gene expression profile.

TROUBLESHOOTING Problem 1
Liver does not fully inflate. Possible reason: Leak;Perfusion tubing is inserted in a suboptimal location;Catheter is not tightly secured (steps 12, 15).

Potential solution
Push the catheter deeper into the vessel and resecure. Correct leaks by first suturing leaking larger and medium size vessels closed on the cut surface, and then apply surgical glue over sutures. Remove the catheter from the major portal and/or hepatic vessel and re-insert at a different vessel. Apply surgical glue around the catheter suture site or remove catheter, attach to perfusion tube luer adaptor, and re-suture into liver.

Problem 2
Liver does not submerge in water bath. Possible reason: Density of liver is too low (step 17).

Potential solution
Remove air from the sterile plastic bag by submerging bag with liver in the water bath by hand leaving the open end above water.

Potential solution
Re-check enzyme concentrations. Re-check the time perfusion solutions are placed in water bath -too little or too much time in the water bath could decrease enzyme activity. Re-check duration liver is perfused: fibrotic livers require additional digestion time. Ensure water bath is at 39 C when digestion begins, and liver is completely submerged (see troubleshooting guidance for step 17 above)

Potential solution
Make sure three-layer nylon mesh is arranged in the correct size order and filter slowly to prevent overflow of mixture with undigested tissue. Maximum of 1 billion total hepatocytes should be added per centrifuge tube used in hepatocyte wash.
Handle centrifuge tubes gently before aspirating to make sure healthy pelleted cells are not resuspended into wash media and lost in aspirant. Resuspend hepatocyte pellets by swirling gently with media. Do not use serological pipette which will cause mechanical lysing. Add 1 wash cycle to remove dead hepatocytes.

Potential solution
See troubleshooting guidance for steps 18a(i) and 18b(iii) above.

Problem 6
Gradient is disrupted. Possible reason: Layering is performed with too much ejection force or too quickly (step 18b (x, xi)).

Potential solution
Before beginning the layering process, coat the sides of the centrifuge tube with solution already inside. If layering is disrupted, pipette the layered material out, mix the solution inside the tube, and layer again.

Potential solution
See troubleshooting guidance for step 18a(v) above. Avoid pipetting cells with too much force as mechanical stress injures cells. See troubleshooting guidance for step 18a(i) and 18b(iii) above.