4 in 1: Antibody‐free protocol for isolating the main hepatic cells from healthy and cirrhotic single rat livers

Abstract Liver cells isolated from pre‐clinical models are essential tools for studying liver (patho)physiology, and also for screening new therapeutic options. We aimed at developing a new antibody‐free isolation method able to obtain the four main hepatic cell types (hepatocytes, liver sinusoidal endothelial cells [LSEC], hepatic macrophages [HMΦ] and hepatic stellate cells [HSC]) from a single rat liver. Control and cirrhotic (CCl4 and TAA) rat livers (n = 6) were perfused, digested with collagenase and mechanically disaggregated obtaining a multicellular suspension. Hepatocytes were purified by low revolution centrifugations while non‐parenchymal cells were subjected to differential centrifugation. Two different fractions were obtained: HSC and mixed LSEC + HMΦ. Further LSEC and HMΦ enrichment was achieved by selective adherence time to collagen‐coated substrates. Isolated cells showed high viability (80%‐95%) and purity (>95%) and were characterized as functional: hepatocytes synthetized albumin and urea, LSEC maintained endocytic capacity and in vivo fenestrae distribution, HMΦ increased expression of inflammatory markers in response to LPS and HSC were activated upon in vitro culture. The 4 in 1 protocol allows the simultaneous isolation of highly pure and functional hepatic cell sub‐populations from control or cirrhotic single livers without antibody selection.

approximately account for the 6% of the liver mass and are found in the space of Disse, surrounding LSEC. Major roles of HSC are regulation of the vascular tone and retinoid storage. Besides, upon liver injury, HSC activate and acquire a myofibroblast-like phenotype in which extracellular matrix (ECM) is actively produced and deposed in order to limit the progression of injury and favor tissue regeneration. 11,12 Moreover, many other cell types, either resident (cholangiocytes lining in the biliary duct, cells building the lymphatic vessels, portal fibroblasts, major vessels endothelial cells, liver progenitor stem cells) or transient (immune cells such as T and B lymphocytes, natural killer cells, neutrophils or erythrocytes), are found in the liver constituting the remaining 5% of its mass. 1,[13][14][15][16] The key role of NPC in liver (patho)physiology is getting more notorious. Hence, isolation of primary cells from pre-clinical experimental models is crucial. However, and especially in the field of chronic liver disease (CLD), few reports have described detailed protocols for simultaneous isolation of the main hepatic cell types from cirrhotic rodent models. 17 In 1972, Seglen 18 achieved an important advance by introducing the two-step perfusion technique for isolating rat liver cells. It was not until 2011 when Liu et al 19 described a detailed systematic method for simultaneous isolation of hepatocytes and NPC by means of low-speed centrifugations. Nevertheless, several alternative techniques have been developed to isolate NPC. The most extended being density gradient centrifugation, 20,21 counterflow elutriation, 5,22 fluorescence activated cell sorting (FACS) 17 and immunomagnetic bead isolation also known as MACS. 23 Cell density separation is highly useful to isolate HSC from the NPC suspension. Generally, the densities of LSEC and HMΦ overlap; therefore, further steps are required to purify both cell types. For example, short incubation of these cells on non-coated substrates is sufficient for HMΦ to attach while LSEC can be easily recovered from the suspension. 24,25 Counterflow elutriation has been intensively used for obtaining highly pure LSEC, although it is time-consuming, expensive and requires specialized equipment. The use of FACS and MACS as the gold standard techniques for liver cell isolation has been reasonably questioned as the overall yield is generally low and reliable surface markers are doubtfully available for liver sinusoidal cell populations. Indeed, isolation methods based on "specific" antibody labelling against membrane receptors have two main limitations, especially in disease models: antibody-receptor recognition and binding could trigger intracellular signaling further altering the naïve properties of the isolated cells, and the expression of surface markers may substantially vary due to disease-mediated changes in cell phenotype, thus a particular cell population may be exclusively selected, excluding cells from the same community with slightly different phenotype.
For all the technical requirements and controversies regarding NPC phenotype, the isolation of liver cells based on their characteristics with regard to density and substrate adherence time remain the primary method used to separate these cell populations. Consequently, the aim of this work was to develop a fast, cheap and userfriendly reproducible protocol for the simultaneous isolation of the four main liver cell types from a single rat liver without antibody selection and suitable for different models of CLD. Isolated cells obtained with the herein presented protocol were tested for purity, yield and functionality.

| Animals
Wistar Han rats weighting 300-350 g were used as healthy (control: Ct) animals. Liver cirrhosis (Ch) was induced in 50-75 g Wistar Han rats by chronic inhalation of carbon tetrachloride (Ch-CCl 4 ) three times a week and receiving 0.3 g/L phenobarbital in the drinking water. When rats developed ascites, approximately after 14-16 weeks, toxicants administration was stopped. For thioacetamideinduced cirrhotic animals (Ch-TAA), TAA was dissolved in 0.9% saline approximately 4 h before injection. Cirrhosis induction was evident after 12 weeks of treatment (250 mg/kg of TAA twice a week).
Control and cirrhotic animals (n = 6 per experimental group) were weight and age matched.  Table 1 describes the different buffers required for the protocol.

| Liver perfusion and digestion
Rats were intraperitoneally anesthetized with a combination of 100 mg/kg ketamine and 5 mg/kg midazolam and sprayed with 96% ethanol to set an aseptic environment. A mid-abdominal incision towards the sternum was made and the intestines were displaced to expose the portal vein. 500 μL heparin were injected through the cava vein and the liver was perfused (20G catheter) through the portal vein with pre-warmed Buffer 1 for 10 minutes at a flow rate of 20 mL/min. Simultaneously, the cava vein was cut to allow outflow of the solution. After perfusion, the liver was digested with prewarmed Buffer 2 for 30 minutes at a flow rate of 5 mL/min. The resultant digested liver was excised, cut up and in vitro digestion were performed with Buffer 2 supplemented with 0.01% collagenase A (30% extra collagenase A was added for cirrhotic livers). Disaggregated tissue was filtered using a 100 μm nylon strainer, collected in cold Buffer 3 and centrifuged at 50× g for 5 minutes. The pellet contained hepatocytes, while NPC were found in the supernatant.
Complete protocol for further purification of liver cells is detailed in the following paragraphs and summarized in Figure 1.

| Isolation of primary rat hepatocytes
Pellet containing hepatocytes was rinsed with cold Hanks' balanced salt solution (HBSS) and subsequently centrifuged (50× g, 5 min, 4°C). This washing procedure was performed three times. Hepatocytes above 75% viability (evaluated by trypan blue exclusion) were cultured in medium M1 (Table 2) plated in 0.1 mg/mL collagencoated petri dishes at a density of 109 000 cells/cm 2 and maintained at 37°C in a humidified atmosphere of 5% CO 2 . After 4 hours, cells were rinsed twice with Dulbecco's phosphate-buffered saline (DPBS) and the medium was replaced by medium M2 (Table 2).  F I G U R E 1 Summarized overview of hepatic cells isolation procedure. Control and cirrhotic livers (n = 6) were perfused, digested with collagenase A, and mechanically disaggregated obtaining a multicellular suspension. Hepatocytes were purified by low revolution centrifugations. Non-parenchymal cells were subjected to a differential centrifugation using iodixanol obtaining a pure fraction of HSC and another one of mixed LSEC+HMΦ. Due to the ability of LSEC to specifically adhere to coated substrates, we were able to obtain highly enriched LSEC and HMΦ cultures FERNÁNDEZ-IGLESIAS ET AL.

| Purification of primary non-parenchymal cells
The supernatant containing NPC was centrifuged to eliminate remaining hepatocytes (50× g for 5 minutes at 4°C) and the super-

| Isolation of hepatic stellate cells
HSC-enriched interphase was carefully collected and rinsed with GBSS. After centrifugation at 800× g for 10 minutes at 4°C the cell pellet was resuspended in medium S ( Table 2) and plated on non-coated petri dishes. HSC were maintained at 37°C in a humidified atmosphere of 5% CO 2 O/N.

| Cell yield and viability
Yield and viability of each cell type were evaluated in Ct and cirrhotic animals (Ch-CCl 4 and Ch-TAA) by trypan blue exclusion assessed by two independent researchers.
Yield per gram of tissue was calculated considering liver weight averages of 9, 10 and 13 g for Ct, Ch-CCl 4 and Ch-TAA respectively.

Functional characterization was performed in cells isolated from Ct
and Ch-CCl 4 rats.

| Immunocytofluorescence
Isolated cells were cultured in petri dishes and fixed with 4% paraformaldehyde for 10 minutes, rinsed three times with DPBS

| HSC in vitro activation
Isolated HSC were activated in vitro with consecutive trypsinization.
Once a week, HSC cultured in 25 cm 2 flask were trypsinized using 0.05% EDTA-trypsin until passage 4. HSC protein lysate was obtained in every trypsinization passage. Medium S was replaced thrice a week.

| Statistical analysis
Statistical analysis was performed with SPSS Statistics 19 (Armonk, NY, USA) software for Windows. Results were expressed as mean ± standard error of mean. In order to assess differences between-groups we performed Student's T-test when variables were parametric and Mann-Whitney test for non-parametric variables.
Differences between groups were considered as significant when P-value ≤0.05.

| Hepatic cells isolation: Yield and purity
The viability and yield of hepatic cells isolated from Ct and Ch rats are shown in Table 3. Ct livers yielded more hepatocytes than Ch ones, showing viability around 80%. Isolation of NPC resulted in similar yields of LSEC in all groups, but as expected, the yield of HSC isolated from Ch livers was significantly greater than Ct. HMΦ seemed to be reduced in cirrhotic livers, although this difference was not significant. Viability of NPC in both Ct and Ch rats was ≥95%. No differences in the amount of each type of cells were observed comparing both pre-clinical models of cirrhosis.
Hepatic cells were morphologically characterized in phase-contrast images (Figure 2 and Figure S1). Hepatocytes showed their characteristic mono(bi)-nucleated polygonal shape, freshly isolated LSEC were small and round, HSC exhibited lipid droplets and characteristic stellate shape, and HMΦ are recognized by their stellatemacrophage shape. Additionally, identity and purity of isolated hepatocytes and NPC were evaluated by immunofluorescence of specific proteins for each cell type: albumin (hepatocytes), Reca-1 (LSEC), desmin (HSC) and CD68 (HMΦ) ( Figure 2). As shown in Table 3, purity of Ct or Ch hepatic cells isolation was ≥95%. Importantly, no fluorescence signal was obtained when each hepatic cell subpopulation was incubated with specific antibodies for the other cell types and analyzed using the same experimental parameters (data not shown). of Ct cells, although this difference was not statistically significant ( Figure 3A).

| Liver sinusoidal endothelial cells
Functionality of primary isolated LSEC was analyzed in terms of their endocytic capacity and fenestrae presence. As shown in Figure 3B, LSEC isolated from Ct or Ch rats were able to uptake the fluorescent Ac-LDL. In addition, the endothelium transmembrane perforations, or fenestrae, characteristic of differentiated LSEC were evaluated by SEM. As expected, Ct-LSEC exhibited numerous fenestrae grouped in sieve plates whereas Ch-LSEC showed a typical de-differentiated phenotype with reduced number and diameter of fenestrae ( Figure 3B).

| Hepatic macrophages
Hepatic macrophages, as cells of the immune system, are able to respond to different pro-inflammatory stimuli. As shown in Figure 3C, HMΦ isolated from Ct animals increased the expression of most proinflammatory markers (TNFα, IL-1β, IL-6 and iNOS) in response to LPS without changes in anti-inflammatory markers other than Mrc1.
Regarding cirrhotic animals, all the analyzed markers except Mrc1 were significantly increased in response to LPS. Furthermore, HMΦ isolated from cirrhotic animals exhibited an exacerbated response to LPS when compared to HMΦ isolated from healthy individuals as shown by significant increases in Arg1, IL-10, TNFα, CCL2 and IL-1β.

| Hepatic stellate cells
Assessment of cell activation during in vitro culture was used as marker of HSC functionality. Figure 3D shows the increase in the activation marker α-SMA in Ct and Ch HSC upon trypsinization and during in vitro culture. When comparing Ct vs Ch, basal expression of α-SMA (p0) was significantly increased in cirrhotic animals. Nevertheless, no significant differences were seen in the subsequent overactivated cells (activation passages p1 to p4).

| DISCUSSION
In order to evaluate the pathophysiology or novel treatments for liver diseases, in vivo models have several advantages over cell cul- Although freshly isolated liver cells would be the optimal choice for the above mentioned in vitro models (either for conventional cultures or multi-cell systems), the difficulties and expenses associated with cellular isolation can drive researchers to other alternatives such as the use of immortalized cell lines or primary in vitroexpanded cultures, which may have lost the highly specialized features found in liver cells in vivo. [31][32][33] In this manuscript, we describe and characterize a novel, easy and affordable protocol for the isolation of the four main hepatic cell types from a single rat liver. Importantly, we clearly detail the entire method, and release the complete list of reagents and solutions to allow a proper replication by any researcher from any institution.
One of the advantages of this protocol lies on its independence on antibody-mediated selection, thus avoiding concerns about undesired activation of membrane receptors and subsequent molecular pathways  In addition to healthy livers, we herein validated the protocol in livers from two different models of CLD. It is well known that cirrhotic rats (either CCl 4or TAA-induced) display high amount of ECM, 30,34 which in combination with over-constriction of sinusoids causes an increase in the intra-hepatic vascular resistance (primary cause in the development of portal hypertension). 4,35 Despite the fact that our protocol relies on perfusion of the liver in order to accomplish a proper digestion, we herein show that this protocol is suitable not only for healthy livers (with perfectly arranged sinusoids) but also in models of cirrhosis (distorted hepatic architecture and microcirculation). Interestingly, yields reproduced the expected changes in cell population during cirrhosis, with a significant reduction on hepatocytes (probably due to described parenchymal extinction 36 ), a greater number of HSC 12 and reduced HMΦ infiltration. 37 Regardless, the order of magnitude of the yield for all cell types remained similar, ensuring comparable amounts of starting sample for any experiment to that of control isolations.
As described above, one of the main advantages of freshly isolated liver cells over cell lines is the preservation of their specific phenotype (at least for the first hours in vitro). That is why we validated the phenotype of the four cell types obtained after this procedure, both with molecular markers and functional assays to Finally, the distinctive characteristic of this protocol is that it allows the isolation of the main hepatic cells from just one animal.
This has several implications that represent an advantage to other existing protocols and, in fact, address current needs of the scientific community: 1. Reduction of the number of animals used (ethical improvement) while the required materials are essentially the same to that of a single cell type isolation (economic improvement).
2. Possibility of performing cross-talk experiments or assembling multicellular in vitro culture systems with cells from the same animal (and thus with certainty that they have the same degree of activation/damage or stage of the disease).
3. The 4 in 1 protocol may be used to isolate cells from an animal that has been treated in vivo with a certain therapeutic, thus allowing separate response analysis for each cell type and correlation with in vivo data from the animal (AST, ALT, hemodynamic data among others).
Indeed, only few studies have reported successful separation of the four cell types at once. [19][20][21][22] Comparison of yields and purity with previous bibliography in the field shows similar results to those obtained using the herein proposed protocol. However, the published protocols rely on antibody fractionation using FACS or MACS (with the abovementioned associated limitations), require multiple gradient centrifugation steps (thus with greater duration and cost), and/or rely on the use of pronase, which may be detrimental for hepatocyte yield and viability. Moreover, none of them assessed the suitability for isolation of cells from diseased livers.
In conclusion, we herein describe an antibody-free protocol for simultaneous isolation of hepatocytes, LSEC, HSC and HMΦ with optimal yield, purity and viability, which is suitable for healthy or cirrhotic livers. Its low equipment requirements and quick feasibility make it suitable for laboratories with standard cell-culture facilities, thus encouraging the use of freshly isolated primary cells over suboptimal in vitro models.

ACKNOWLEDG EMENTS
This study was performed at the Esther Koplowitz Center (IDIBAPS).