www.nature.com/scientificreports OPEN Hepatic stellate cells secrete Ccl5 to induce hepatocyte steatosis Received: 14 November 2017 Accepted: 17 April 2018 Published: xx xx xxxx Byeong-Moo Kim1, Ahmed Maher Abdelfattah1, Robin Vasan2, Bryan C. Fuchs2,3 & MichaelY. Choi1,4 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. Hepatic stellate cells (HSCs) are considered to play a central role in hepatic fibrogenesis1–4. Since fibrosis is the common pathway that ultimately leads to cirrhotic organ failure or hepatocellular carcinoma that occurs in progressive liver diseases, understanding the biology of HSCs is pivotal for all liver diseases that can lead to hepatic fibrosis. This paradigm is particularly important in non-alcoholic fatty liver disease (NAFLD) which has become the most significant hepatic illness in the world. Just in the United States NAFLD affects over 64 million people by a recent estimate5, and it will soon become the most common cause of liver transplant and hepatocellular carcinoma6,7. Although several drugs are undergoing clinical trials there are still no well-established pharmacologic agents to treat NAFLD, and the main mode of management is through reducing risk factors such as obesity, hyperlipidemia, and diabetes8. Hence, elucidating the HSC pathobiology in NAFLD is crucial to identify novel treatment options. Although fibrosis that leads to end-stage organ failure or cancer is an important feature of NAFLD, the disease does not start with fibrosis. Instead, 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 cirrhosis9,10. HSCs are known to contribute to fibrosis, but much less is known about their function during NAFLD’s earlier stages, in steatosis and inflammation without fibrosis. One of the difficulties in studying HSCs in the context of other cell types during the early stages of NAFLD is that they are relatively hard to identify in vivo to study their role and that insightful in vitro assays that model HSC’s interaction with other cell types are difficult to develop. In response, we have devised an ex vivo assay that cocultures healthy hepatocytes with primary HSCs from mouse models of steatohepatitis to study their interaction. The assay greatly enhances our ability to define HSC function and the mechanism of its action during fatty inflammation of the liver by identifying HSC-secreted mediators that have profound effects on nearby hepatocytes. In healthy humans and mice, HSCs are quiescent, residing near sinusoids in the space of Disse. However, once the liver is afflicted with a chronic illness such as NAFLD/NASH, quiescent HSCs become transdifferentiated to give rise to myofibroblasts11. These activated HSCs, termed myofibroblasts, are known to express cytokines, chemokines, extracellular matrix proteins, and other genes that contribute to hepatic fibrogenesis12. Our data indicate that chemokine (C-C motif) ligand 5 (Ccl5, a.k.a. Rantes) is one of the HSC-secreted mediators in NASH 1Department of Medicine, Gastrointestinal Unit, Massachusetts General Hospital and Harvard Medical School, Boston, MA, 02114, USA. 2Department of Surgery, Massachusetts General Hospital Cancer Center and Harvard Medical School, Boston, MA, 02114, USA. 3Division of Surgical Oncology, Massachusetts General Hospital Cancer Center and Harvard Medical School, Boston, MA, 02114, USA. 4Harvard Stem Cell Institute, Cambridge, MA, 02138, USA. Correspondence and requests for materials should be addressed to M.Y.C. (email: mchoi@mgh.harvard.edu) SCientifiC Reports | (2018) 8:7499 | DOI:10.1038/s41598-018-25699-9 1 www.nature.com/scientificreports/ Figure 1.  Mouse model of steatohepatitis using choline-deficient, L-amino acid-defined, high fat diet. (A–C) After three weeks of CDAHFD, mice were both grossly and microscopically affected with steatohepatitis that resembles early NASH. (D) Hepatic stellate cells harvested from CDAHFD fed mice retained retinoid positive lipid droplets that fluoresce with ultraviolet light. Nearly 99% of the isolated hepatic stellate cell population fluoresced with ultraviolet light based on FACS analysis. (E) The harvested hepatic stellate cell population did not express albumin or F4/80 based on immunofluorescence. CDAHFD, L-amino acid-defined, high fat diet; NASH, non-alcoholic steatohepatitis; UV, ultraviolet. that directly induce steatosis and pro-inflammatory factors in initially healthy hepatocytes. Another group already demonstrated that human HSCs express CCL5 when challenged with TNF-alpha, IL-1beta, or CD40L in vitro13, but our work suggests for the first time that HSCs express Ccl5 in early NASH to directly induce steatosis and upregulation of pro-inflammatory factors in hepatocytes. Although Ccl5 is also expressed by other liver cell types14–17 and is better known for causing hepatic fibrosis through the recruitment of immune cells14,15, its pro-steatotic action in early NAFLD/NASH has not been recognized until now. Similarly, the essential function of HSCs in hepatic fibrogenesis is well accepted, but their pro-steatotic role, however minor, in directly influencing hepatocytes in early NAFLD/NASH has not been clear. Using a mouse model of steatohepatitis, our result suggests that HSCs have a function in inducing steatosis and inflammation, not just fibrosis. 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 carcinoma18. 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 genes1–3, 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 weeks19. 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 protocol20. 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). SCientifiC Reports | (2018) 8:7499 | DOI:10.1038/s41598-018-25699-9 2 www.nature.com/scientificreports/ 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 days21,22. 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 vivo23. 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 investigated14,15. 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 SCientifiC Reports | (2018) 8:7499 | DOI:10.1038/s41598-018-25699-9 3 www.nature.com/scientificreports/ Figure 2.  Hepatic stellate cells from mice with steatohepatitis induced by CDAHFD are primed to become activated. (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. SCientifiC Reports | (2018) 8:7499 | DOI:10.1038/s41598-018-25699-9 4 www.nature.com/scientificreports/ Figure 3.  Hepatic stellate cells isolated from mice with steatohepatitis induce steatosis and inflammation in cocultured hepatocytes. (A) Diagram of the experiment showing coculture of healthy hepatocytes with hepatic stellate cells from either control mice or from mice with CDAHFD induced steatohepatitis. (B) Top row: Healthy hepatocytes cocultured with HSCs from control mice showed very few lipid droplets stained with Bodipy. Bottom row: Initially healthy hepatocytes cocultured with HSCs from mice with CDAHFD induced steatohepatitis developed significantly more lipid droplets. (C) Induction of fatty accumulation in hepatocytes can be achieved when the conditioned media from HSCs from mice with CDAHFD induced steatohepatitis is applied to healthy hepatocytes. (D) Top row: When quiescent HSC conditioned media was applied to healthy hepatocytes, steatosis was not induced. Bottom row: Conditioned media of HSCs from mice with CDAHFD induced steatohepatitis induced steatosis in initially healthy hepatocytes. Lipid droplets stained with Bodipy. (E) Initially healthy hepatocytes cocultured with HSCs from mice with CDAHFD induced steatohepatitis expressed higher levels of several inflammatory cytokines and chemoattractants than when they were cocultured with control HSCs, measured with qPCR. Data are presented as mean +/− SD (*P < 0.05). NASH, non-alcoholic steatohepatitis; CDAHFD, choline-deficient L-amino acid defined high fat diet; Hep, hepatocyte; HSC, hepatic stellate cell. SCientifiC Reports | (2018) 8:7499 | DOI:10.1038/s41598-018-25699-9 5 www.nature.com/scientificreports/ Figure 4.  Ccl5 is secreted by hepatic stellate cells isolated from mice with CDAHFD induced steatohepatitis and induces hepatocyte steatosis. (A,B) Expression of Ccl5 and Acta2 co-localized to hepatic stellate cells in both mouse steatohepatitis and human NASH assessed by immunofluorescence (7 out of 11 human samples tested positive). (C) Hepatic stellate cells isolated from mice with steatohepatitis continued to express Ccl5 at a greater level than initially healthy hepatic stellate cells, even after further activation on dish, measured with qPCR. (D) Recombinant Ccl5 protein induced steatosis in freshly isolated primary mouse hepatocytes, observed with Bodipy staining. (E) Hepatocytes that became steatotic with recombinant Ccl5 upregulated pro-inflammatory cytokines and chemokines, measured with qPCR. (F,G) The conditioned media from Ccl5 overexpressing hepatic stellate cells induced steatosis, detected with Bodipy stain, and caused increased expression of several inflammatory cytokines and chemoattractants in primary hepatocytes, measured with qPCR. (H) Hepatic stellate cells overexpressing Ccl5 also had increased expression of other pro-inflammatory mediators besides Neutralizing Ccl5 iCncHl5S, CmceoasnudrietdiowneitdhmqPedCiRa .wAitlhl daabtaloacrkeinprgeasnentitbeoddaysrmedeuance+d /s−teaStDos(is*Pin