Hepatic stellate cells secrete Ccl5 to induce hepatocyte steatosis

Non-alcoholic fatty liver disease (NAFLD) encompasses a wide spectrum of disease severity, starting from pure steatosis, leading to fatty inflammation labeled as non-alcoholic steatohepatitis (NASH), and finally fibrosis leading to cirrhosis. Activated hepatic stellate cells (HSCs) are known to contribute to fibrosis, but less is known about their function during NAFLD’s early stages prior to fibrosis. We developed an ex vivo assay that cocultures primary HSCs from mouse models of liver disease with healthy hepatocytes to study their interaction. Our data indicate that chemokine Ccl5 is one of the HSC-secreted mediators in early NASH in humans and in mice fed with choline-deficient, L-amino acid defined, high fat diet. Furthermore, Ccl5 directly induces steatosis and pro-inflammatory factors in healthy hepatocytes through the receptor Ccr5. Although Ccl5 is already known to be secreted by many liver cell types including HSCs and its pro-fibrotic role well characterized, its pro-steatotic action has not been recognized until now. Similarly, the function of HSCs in fibrogenesis is widely accepted, but their pro-steatotic role has been unclear. Our result suggests that in early NASH, HSCs secrete Ccl5 which contributes to a broad array of mechanisms by which hepatic steatosis and inflammation are achieved.


Results
Mice fed a choline-deficient, L-amino acid-defined, high fat diet for three weeks develop steatohepatitis. NAFLD starts as simple fatty liver that can progress to steatohepatitis, fibrosis, and ultimately cirrhosis and hepatocellular carcinoma 18 . Although HSCs are known to have a critical function in the development of hepatic fibrosis by secreting extracellular matrix proteins and expressing other pro-fibrotic genes 1-3 , less is known about their action during the initial, fatty liver and the early inflammatory phases of NASH. To study their function during this early stage of the disease, we isolated HSCs from a mouse model of steatohepatitis relying on choline-deficient, L-amino acid-defined, high fat diet (CDAHFD). These mice normally develop severe steatosis with moderate inflammation but no fibrosis within 3 weeks, mild fibrosis within 6 weeks, moderate fibrosis in 9 weeks, and moderate to severe fibrosis in 12 weeks 19 . Thus, mice fed with CDAHFD for three weeks have fatty and inflamed liver that resembles the histology of early NASH prior to fibrosis. Grossly, at three weeks, their skin became oily, evidenced by the greasy coat that sticks together in clumps (Fig. 1A). Internally, their livers were lighter in color, demonstrating hepatic fat (Fig. 1B). Microscopically, they showed several large and small lipid droplets in the cytoplasm of hepatocytes, confirming moderate to severe steatosis with moderate inflammation but no fibrosis (Fig. 1C). To study the role of HSCs at this stage of liver disease, we isolated them by optimizing and improving on an established protocol 20 . Each 8-week-old female C57BL/6 mouse fed with either normal diet or CDAHFD for three weeks yielded 0.5-1 million HSCs. This population of cells was 98.9% pure for positive fluorescence with ultraviolet light (Fig. 1D), a signal specific for retinoid filled lipid droplets present in quiescent HSCs. Furthermore, absence of cells expressing albumin or F4/80 in this population confirmed almost no contamination of hepatocytes or macrophages, respectively (Fig. 1E). Primary HSCs isolated from mice with non-fibrotic steatohepatitis more quickly become activated in culture. Quiescent HSCs are known to become activated spontaneously to myofibroblasts if they are cultured on plastic dish for 7-10 days 21,22 . This phenomenon is likely driven by the high degree of stiffness inherent to plastic which acts as the activating signal to HSCs, and this same mechanism of action may also take place in the stiffening hepatic tissue as fibrosis progresses which itself becomes one of many activating signals in vivo 23 . HSCs harvested from liver with steatohepatitis without fibrosis induced by three weeks of CDAHFD morphologically resembled quiescent HSCs from healthy mice, retaining UV-fluorescing lipid droplets, but they were slightly larger in size (Figs 1D and 2A). More significantly, they were functionally distinct. Their morphologic transdifferentiation to activated HSCs occurred more quickly in culture, requiring only 3-6 days, instead of 7-10 days. As opposed to the usual seven days, by the third day in culture, HSCs isolated from CDAHFD induced non-fibrotic steatohepatitis increased in size with the spread morphology (Fig. 2B). Furthermore, by day 3.5 in culture, these HSCs had heightened expression of activation markers alpha smooth muscle actin (Acta2) and alpha-1 type I collagen (Col1a1) covering the expanded cytosol (Fig. 2B). This elevated expression of activation markers was maintained even after HSCs became fully activated on plastic dish (Fig. 2C). Finally, we analyzed the expression of several cytokines and chemokines involved in inflammation and recruitment of immune cells, and HSCs from steatohepatitic liver showed significantly higher expression of tumor necrosis factor-alpha (Tnf), cluster of differentiation 14 (Cd14), interleukin 6 (Il-6), and chemokine (C-C motif) ligand 5 (Ccl5) and 8 (Ccl8) by day 3.5 in culture compared to HSCs from healthy mice (Fig. 2D). These results indicate that although HSCs isolated from the liver with CDAHFD induced steatohepatitis resembling early NASH have a quiescent morphology with retinoid positive lipid droplets, they are intrinsically different from quiescent HSCs and primed to become fully activated.
Primary HSCs isolated from non-fibrotic steatohepatitic liver induce steatosis and inflammation in healthy hepatocytes. Since HSCs from steatohepatitic liver are more susceptible to become activated, phenotypically distinct from quiescent HSCs, we tested their ability to influence nearby hepatocytes. When we cocultured primed HSCs from three weeks of CDAHFD with hepatocytes isolated from mice eating standard diet, initially healthy hepatocytes became steatotic with numerous lipid droplets within three days (Fig. 3A,B). This result was obtained while HSC-hepatocyte coculture was performed using a Transwell that physically separates the two cell types while secreted proteins and small molecules can freely traverse this barrier, suggesting that HSCs are inducing hepatocyte steatosis not through cell-cell contact but by secreting mediators. We verified this phenomenon when steatosis was induced in healthy hepatocytes by applying the conditioned media of primed HSCs harvested from mice challenged with three weeks of CDAHFD (Fig. 3C,D). Finally, we evaluated the expression levels of various pro-inflammatory cytokines and chemokines in steatotic hepatocytes cocultured with primed HSCs. The result shows that these hepatocytes were not only steatotic but also expressed many pro-inflammatory genes that further activate HSCs and attract immune cells (Fig. 3E).
Hepatocyte steatosis is induced by Ccl5 secreted from HSCs. Since hepatocyte steatosis did not require cell-cell contact with HSCs originating from steatohepatitic liver, steatosis inducing signals are likely transmitted through secreted mediators in a paracrine manner. To identify the secreted factors responsible for inducing steatosis in nearby hepatocytes, we performed a cytokine array blot of the conditioned media from HSCs in culture. One of the highly up-secreted proteins was Ccl5 (Supp. Fig. 1), which was already demonstrated to be upregulated at the transcriptional level (Fig. 2D), and enzyme-linked immunosorbent assay (ELISA) confirmed the up-secretion of Ccl5 from these HSCs (Supp. Fig. 4). Finally, the upregulation of Ccl5 in HSCs from CDAHFD induced steatohepatitis in mice and early NASH in humans was further verified by immunofluorescence which showed co-localization of Ccl5 with Acta2, an activated HSC marker in the liver (Fig. 4A,B). Moreover, these HSCs continued to express Ccl5 at a greater level than initially normal HSCs, even after further activation on dish (Fig. 4C). Although Ccl5 is a relatively well-investigated chemokine, known to promote hepatic fibrosis, its possible pro-steatotic effect has not been investigated 14,15 . To test this hypothesis, we investigated whether Ccl5 directly causes steatosis in hepatocytes or is simply an upregulated gene with an unrelated function. First, we applied recombinant Ccl5 protein at various concentrations to nontransformed mouse hepatocyte cell line AML12 in culture. Indeed, purified Ccl5 protein induced steatosis in AML12 cells starting at 1 ng/ml, but more robustly with 50 ng/ml and 100 ng/ml (Supp. Fig. 2). Ccl5 also caused steatosis in primary mouse hepatocytes at these concentrations, seen with Bodipy stain (Fig. 4D). Furthermore, these hepatocytes that became steatotic with recombinant Ccl5 upregulated pro-inflammatory cytokines and chemokines of their own (Fig. 4E), demonstrating Ccl5's pro-steatotic and pro-inflammatory effects. Interestingly, applying recombinant Ccl5 to hepatocytes caused those hepatocytes to upregulate Ccl5 themselves in a feed-forward manner. To further understand this HSC-hepatocyte interaction involving Ccl5, we cloned the full length Ccl5 gene and overexpressed it in HSCs (Supp. Fig. 3). We collected the conditioned media from these Ccl5 expressing HSCs, and applied it on healthy hepatocytes. As expected, these hepatocytes also formed lipid droplets, detected by Bodipy stain (Fig. 4F). Furthermore, these same hepatocytes increased expression of various pro-inflammatory cytokines and chemokines but in greater folds than when recombinant Ccl5 was directly applied (Fig. 4G). To investigate the reason for this greater induction of pro-inflammatory cytokines and chemokines in hepatocytes by Ccl5 overexpressing HSCs, we checked the expression levels of other cytokines and chemokines besides Ccl5 in these cells. It turns out, the HSCs overexpressing Ccl5 also had increased expression of other pro-inflammatory mediators besides Ccl5, suggesting that Ccl5 induces either an autocrine signaling or purely an intracellular signaling cascade that leads to upregulation of genes such as Tnf and Il-6 ( Fig. 4H). Pro-inflammatory cytokines such as Tnf, Il-6, and unidentified factors upregulated by Ccl5 in HSCs are likely being secreted and further inducing certain pro-inflammatory gene expression in nearby hepatocytes. Lastly, the pro-steatotic effect of Ccl5 secreted by HSCs was further evidenced when a Ccl5 neutralizing antibody applied to CDAHFD-HSC conditioned media attenuated its induction of hepatocyte steatosis (Fig. 4I, Supp. Fig. 5). We also confirmed that (A) Hepatocytes from mice with CDAHFD induced steatohepatitis were steatotic while hepatic stellate cells were only slightly larger in size compared to those from healthy mice. After eight days of culture, hepatic stellate cells from early NASH became much larger in size compared to control stellate cells. (B) Hepatic stellate cells from mice with CDAHFD induced steatohepatitis more quickly increased in size and in expression of activation markers Colla1 and Acta2, visualized here just three days after the initial cell isolation. (C) Hepatic stellate cells from mice with CDAHFD induced steatohepatitis maintained higher expression of Col1a1 and Acta2 than those from control mice even after both populations were fully activated on plastic dish. Expression levels measured with qPCR. (D) Hepatic stellate cells from mice with CDAHFD induced steatohepatitis had elevated expression of several pro-inflammatory cytokines compared to those from control mice. Expression levels measured with qPCR. All data are presented as mean +/− SD (*P < 0.05). NASH, non-alcoholic steatohepatitis; CDAHFD, choline-deficient L-amino acid defined high fat diet; HSC, hepatic stellate cell; qHSC, quiescent hepatic stellate cell; acHSC, activated hepatic stellate cell; Col1a1, alpha-1 type I collagen; Acta2, alpha smooth muscle actin; ns, not significant.  the source of Ccl5 is HSCs, not hepatocytes, in these assays by demonstrating the lack of Ccl5 immunofluorescence signal from hepatocyte culture (Supp. Fig. 6).

Blocking the action of Ccl5 using an inhibitor of Ccr5 decreases hepatocyte steatosis in vitro.
Ccl5 is known to interact with the membrane C-C chemokine receptors including type 1 (Ccr1), type 3 (Ccr3), and type 5 (Ccr5) 14,15,24 . To determine expression levels of Ccr1, Ccr3, and Ccr5, we performed quantitative PCR (qPCR)of these three genes in primary mouse hepatocytes. The result showed that hepatocytes express Ccr5 significantly more than Ccr1 or Ccr3 (Fig. 5A). To test whether Ccl5 expressed by HSCs signal through Ccr5 on hepatocytes to induce steatosis and to upregulate pro-inflammatory cytokines, we applied both recombinant Ccl5 and the Ccr5-specific inhibitor Maraviroc on hepatocytes. As suspected, pro-steatotic effect of Ccl5 was attenuated with Maraviroc (Fig. 5B,C).

Discussion
Our study accomplishes two goals that should further facilitate the study of HSC biology in the setting of NASH. First, knowing that studying HSC's interaction with nearby hepatocytes in their microenvironment during liver diseases would be facilitated by having a reliable ex vivo system that uses primary cells, we developed our coculture system using a mouse model of steatohepatitis histologically resembling early NASH. Using primary HSCs was important because malignant or immortalized cell lines often do not reflect the molecular physiology and the measurable behavior of primary cells 25 . Also, the mouse as a species provides a vast array of genetic, diet, chemical, and anatomic models of liver diseases. Although there is no perfect animal model that completely recapitulates human NASH, we chose to utilize the CDAHFD model because it establishes severe steatosis with moderate inflammation in three weeks which eventually leads to fibrosis in subsequent weeks, the stepwise progression seen in humans 19 . Our coculture system allowed us to better define the function and its mechanism of HSCs in affecting nearby hepatocytes during steatohepatitis development. Several points should be noted here. Although we studied the function of diseased HSC's influence on healthy hepatocyte, the system also can be applied to investigate diseased hepatocyte's action on healthy HSC. In fact, other important cell types such as macrophages and sinusoidal endothelial cells could potentially be used as long as a highly pure population can be isolated and cultured. Moreover, although we have not tried, it is possible to coculture three or more cell types together to study the interactions among them. The cell type directly harvested from a diseased liver should more accurately reflect its diseased phenotype, which is the strength of our ex vivo system. However, it must be emphasized that terminally differentiated primary cells often cannot be maintained in culture for a prolonged period without changes in their phenotype. This is the reason why our coculture assays in this study did not extend beyond three days in culture. We believe that any terminally differentiated primary cells cultured ex vivo for more than 5-7 days lose their original phenotype. Furthermore, the insight we gain from studying primary cells in a diseased context is only as good as the animal model. Nevertheless, our coculture system is a useful tool to better define the role of hepatic cell types and their interactions in the setting of various liver diseases.
Second, using the coculture assay we developed, we wanted to better understand the function of HSCs in a mouse model of non-fibrotic steatohepatitis that may simulate early NASH prior to the onset of fibrosis. Our data demonstrate that HSCs directly induce steatosis in healthy hepatocytes by secreting mediators that act in a paracrine fashion. One of those mediators that have a direct steatotic effect on hepatocytes is Ccl5 which seems to exert its effect through Ccr5. This unexpected result suggests that HSCs have pathophysiologic functions beyond laying down excessive amount of extracellular matrix to cause fibrosis during the mid to late stages of NASH. We should caution, however, that this new finding does not necessarily indicate that the pro-steatotic role of HSCs or Ccl5 during the early stage NASH is the major mechanism by which liver becomes steatotic. For example, it is known that Ccl5 is secreted by other cell types such as macrophages, platelets, endothelial, T cells, and hepatocytes [14][15][16][17] . Hence, hepatocytes likely become steatotic through multiple molecular signals and cell types, and the proportion of the phenotype contributed by HSCs through Ccl5 may be minor. Interestingly, supporting the mild pro-steatotic function of Ccl5, Ccl5 knockout mice demonstrate a trend toward lower hepatic triglyceride levels compare to their wildtype littermates 15 . Finally, although hepatocytes are generally considered the main driver of inflammation that signals to other cell types including HSCs, our result suggests that a significant amount of pro-inflammatory signals may reciprocate back to hepatocytes from HSCs to create a vicious circle of reinforcing inflammatory signals 12 . Although not directly tested in this study, such feed-forward escalation in hepatic inflammation likely involves other cell types as well.
In summary, our coculture assay using primary cells from the liver is a versatile method by which the interaction between isolatable cell types from any mouse model of liver disease can be researched. The system attempts to faithfully recreate the microenvironment of cells within the liver by culturing freshly harvested primary cells for a short duration, lasting no more than 5-7 days. We applied the method to coculture diseased HSCs from steatohepatitic liver with healthy hepatocytes for three days, and discovered that HSCs can directly induce hepatocytes to become steatotic by secreting the chemokine Ccl5. Although this result does not preclude the possibility of other secreted factors or cell types causing similar phenotype, the data reveals a novel function of Ccl5 and HSCs. Our coculture system has the potential to uncover other therapeutically relevant functions of HSCs and other liver cell types, which is urgently needed given the rising incidence of NASH and the lack of FDA-approved therapies. Hepatic stellate cells and hepatocyte isolation. Hepatocytes were isolated from C57BL/6 mice fed the control diet and hepatic stellate cells isolated from mice fed the control diet or the CDAHFD by enzymatic digestion and Percoll density gradient centrifugation with modifications. The portal vein was perfused in situ with 30 mL of HBSS (without Ca2 + and Mg2 + ) and 30 mL of 0.05% collagenase B (Roche Diagnostics, Indianapolis, IN, USA), respectively, at 37 °C with a flow rate of 6 ml/min. After perfusion, the partially digested liver was excised and incubated with 20 mL of 0.05% collagenase and DNase I 10 μg/mL, (Roche Diagnostics, Indianapolis, IN), at 37 °C for 30 minutes. The tissue was passed through a 70 µm nylon mesh to remove undigested materials and suspended in washing buffer (PBS containing DNase I). Hepatocytes were separated from the non-parenchymal cells and debris by centrifugation at 4 °C in the following sequence: twice for 5 minutes at 50 g, and twice again for 5 minutes at 20 g. The supernatant was collected for stellate cell isolation and the hepatocytes present in the pellet were re-suspended in DMEM. Primary mouse HSCs were purified from the remainder of non-parenchymal cells. Cells were centrifuged at 635 g for 10 minutes and resuspended in washing buffer followed by pass through a 70 µm nylon mesh. The pellets were resuspended in 10 ml of 35% Percoll (GE Healthcare, Pittsburgh, PA, USA) with an overlay of 1 ml PBS. After centrifugation at 1130 g for 30 minutes, HSCs are in the layer located between the PBS and 35% Percoll. Cells were counted using a hemocytometer (Neubauer chamber) and 0.4% trypan blue (Sigma-Aldrich, St. Louis, MO, USA). The purity was determined by UV sorting with BD FACS Aria II SORP Cell Sorter (BD Biosciences, San Jose, CA, USA).

Methods
Cell culture. Isolated hepatocytes were re-suspended in DMEM medium containing 10% FBS, plated onto collagen-coated six-well plates at a density of 5 × 10 5 cells/well in 1.5 ml culture medium, and cultured for 4 hours. The medium was then changed into serum-free medium and the cells were cocultured with hepatic stellate cells and loaded onto cell-culture inserts of 3 µm pore size (Corning, Corning, NY, USA Immunofluorescence. Cells were fixed with 4% paraformaldehyde for 10 minutes at room temperature and Fluorescence-Activated Cell Sorting. Cell sorting was done using a BD FACS Aria II SORP Cell Sorter (BD Biosciences, Franklin Lakes, NJ, USA). The pellet was resolved in 4 °C Hank's complete and filtered using 40 μm nylon gaze. The sorting of the HSC required excitation via UV laser and measuring the emission in the Indo-1 channel based on a 505 nm long pass filter. The sample loading port was set to 4 °C and 300 rpm. We used a 100 μm nozzle and a pressure of 20 psi. HBSS with calcium or magnesium was used as sheath fluid. The flow rate was set to 5000 events per second and the threshold was adjusted to 5000. The collection device was set to 4 °C. The collection tube was a 5 mL glass tube that contained 1 mL of Hank's BSS without calcium or magnesium, 10 mM HEPES, and 20% of fetal bovine serum (FBS).
Enzyme-linked immunosorbent assay. Isolated HSCs from normal and CDAHFD treated mice were cultured in six-well plates at a density of 4 × 10 5 cells/well for 72 h. Supernatants were collected, and an ELISA for Ccl5 (Abcam, Cambridge, MA) was performed according to the manufacturer's instructions. Results are expressed as Ccl5 secretion for 72 h.
Neutralization assay. Neutralizing antibody to Ccl5 and normal goat immunoglobulin G control were procured (R&D Systems, Minneapolis, MN) and used (2 μg/mL) for the neutralization assay. Antibodies were incubated with conditioned media for 1 hour at 37 °C before applying it to primary hepatocytes.

Statistical analysis.
All the experiments were performed in triplicate and repeated at least two times. Data were expressed as mean +/− SD. Statistical analysis was performed using a Student t test for unpaired data to compare the values between the two groups and one-way analysis of variance among multiple groups. Differences in values were considered significant at P < 0.05.