Pharmacologic Antagonization of Cannabinoid Receptor 1 Improves Cholestasis in Abcb4-/- Mice

Background & Aims The endocannabinoid system is involved in the modulation of inflammatory, fibrotic, metabolic, and carcinogenesis-associated signaling pathways via cannabinoid receptor (CB)1 and CB2. We hypothesized that the pharmacologic antagonization of CB1 receptor improves cholestasis in Abcb4-/- mice. Methods After weaning, male Abcb4-/- mice were treated orally with rimonabant (a specific antagonist of CB1) or ACEA (an agonist of CB1) until up to 16 weeks of age. Liver tissue and serum were isolated and examined by means of serum analysis, quantitative real time polymerase chain reaction, Western blot, immunohistochemistry, and enzyme function. Untreated Abcb4-/- and Bagg Albino Mouse/c wild-type mice served as controls. Results Cholestasis-induced symptoms such as liver damage, bile duct proliferation, and enhanced circulating bile acids were improved by CB1 antagonization. Rimonabant treatment also improved Phosphoenolpyruvat-Carboxykinase expression and reduced inflammation and the acute-phase response. The carcinogenesis-associated cellular-Jun N-terminal kinase/cellular-JUN and signal transducer and activator of transcription 3 signaling pathways activated in Abcb4-/- mice were reduced to wild-type level by CB1 antagonization. Conclusions We showed a protective effect of oral CB1 antagonization in chronic cholestasis using the established Abcb4-/- model. Our results suggest that pharmacologic antagonization of the CB1 receptor could have a therapeutic benefit in cholestasis-associated metabolic changes, liver damage, inflammation, and carcinogenesis.

E very year, more than 1.2 million people die from complications of cirrhosis, which currently is the 11th most common cause of death worldwide. 1 This number shows the need to establish new and effective therapy options. After removing the cause, the damaged liver usually is able to restore its function, even in the presence of advanced cirrhosis. Until now, however, there has not been an effective pharmacologic antifibrotic therapy for advanced liver injury resulting from chronic cholestasis such as sclerosing cholangitis.
A large number of studies have shown that the endocannabinoid system is a significant mediator of acute and chronic liver disease. The function of the 2 cannabinoid receptors (CB)1 and CB2 has been a subject of scientific research for many years. In addition to their metabolic effects, new insights into their role in the development of liver inflammation and fibrosis in chronic liver damage are of great interest. 2,3 Teixeira-Clerc et al 2 were able to show that CB2 had antiinflammatory and antifibrotic effects in the early stages of liver damage.This protective influence was eliminated by the profibrotic effects of activated CB1 if liver damage persisted. 4 In chronically damaged liver tissue, CB1 is strongly expressed and stimulated by increased release of endocannabinoids by hepatocytes, hepatic stellate cells, and Kupffer cells. It shows its strongest expression in nonparenchymal cells such as inflammatory cells, proliferating cholangiocytes, hepatic stellate cells, and portal myofibroblasts. 2,4 Rimonabant, also called SR141716A, belongs to the active ingredient class of anorectics and is a selective CB1 antagonist. The CB1 blockade in the diseased liver showed positive effects on the development of steatohepatitis, fibrosis, and the metabolic syndrome in several animal models. 2,5,6 Recently, we showed that a global knockout of the Cb1 gene (Cb1 -/-) reduced the expression of the lipid droplet binding protein Perilipin 2 (PLIN2) in the livers of Cb1 -/and hepatitis B surface protein-transgenic mice, which spontaneously develop hepatic steatosis. In addition, the antagonization of CB1 in human cell culture also caused a reduction of PLIN2, a cytoplasmic lipid droplet binding protein involved in the storage of neutral lipids within the lipid droplets. 7 The CB1 receptor is down-regulated during cholestasis. Anandamide, a partial CB1 agonist, suppresses cholangiocyte growth in bile duct ligation (BDL) mice by induction of cholangiocyte apoptosis. 8 Abcb4 -/mice represent a well-characterized model for sclerosing cholangitis beginning with persistent cholestasis that progresses to cirrhosis and liver failure before late childhood. 9 Although knowing the basic genetic defects and the pathology of the disease, being characterized by ductular proliferation in the liver and progressive intrahepatic cholestasis, there is still no successful therapeutic approach. 10 In our present work, the effect of CB1 antagonization on cholestasis and its consequential damage was examined. For this purpose, Abcb4 -/mice were treated with the selective CB1 antagonist rimonabant. Our hypothesis was that the disease progression and consequences of cholestasis can be delayed by CB1 antagonization. Metabolic parameters and specific markers for inflammation, fibrogenesis, and carcinogenesis were examined. Untreated Abcb4 -/mice and mice fed with the CB1 agonist arachidonyl-2-chloroethylamide (ACEA) were included as controls.

Cholestatic Liver Disease
The sterol regulatory element-binding protein-1 (SREBP1) plays a crucial role in the regulation of cholesterol metabolism and fatty acid synthesis. SREBP1 is a downstream effector of CB1, thus contributing to the development of obesity and fatty liver via lipogenesis. 11 It has been shown that cholestasis was associated with reduced messenger RNA (mRNA) expression of Srebp1 and diminished lipogenesis in Abcb4 -/mice. 12,13 Here, we isolated nuclear proteins and analyzed the nuclear amount of the matured transcription factor nuclear SREBP1. Interestingly, we found enhanced amounts of matured nuclear SREBP1 protein in Abcb4 -/mice compared with wild-type (WT) mice, while neither rimonabant nor ACEA altered nuclear SREBP1 significantly ( Figure 1A).
In the course of slowly progressing cholestasis, an increase in connective tissue remodeling was observed in the liver (black arrows, Figure 1B, upper right) of untreated Abcb4 -/mice, and was most evident in the periportal fields ( Figure 1B, upper right). The liver structure and the remodeling of connective tissue in mice treated with rimonabant ( Figure 1B, lower right) was comparable with that of WT mice ( Figure 1B, upper left). Remarkably, pathologic changes in the portal fields of the Abcb4 -/mice also were reduced by treatment with ACEA ( Figure 1B, lower left). Although histologic grading suggested improved scoring in rimonabanttreated animals, statistical significance was not reached.
In the course of hepatocellular damage, accompanying cholestatic disease alanine aminotransferase (ALT) ( Figure 1C) and aspartate aminotransferase (AST) ( Figure 1D) serum values increase. Serum levels of ALT and AST were reduced by treatment with both rimonabant and ACEA. There were no differences in serum ALT and AST levels between rimonabant-and ACEA-treated mice ( Figure 1C and D).
Cytokeratin 19 (CK19) immunostaining (black arrows) visualized a higher number of periportal bile ducts in Abcb4 -/mice and thus showed the highest level of bile duct proliferation in untreated Abcb4 -/mice ( Figure 1E, upper right). Rimonabant treatment and, to a lesser degree, ACEA treatment resulted in reduced bile duct proliferation ( Figure 1E and F).
To determine the degree of cholestasis quantitatively, serum bile acids were analyzed. Cholestasis, particularly because of Abcb4 -/-, leads to a loss of barrier function of bile ducts caused by changes in the tight junctions. This consequently leads to an increase of serum bile acids ( Figure 2A). ACEA and rimonabant tendentially reduced serum bile acid concentrations.
The bile acids transferred into the serum as part of the cholestatic disease showed a clear dominance of the taurine-conjugated bile acids T-u-muricholic acid and T-bmuricholic acid in the serum of Abcb4 -/mice. This predominance of taurine conjugates is typical in mice ( Figure 2B). Interestingly, hepatic bile acid content as well as the hepatic expression of Fxr was not altered significantly, either in Abcb4 -/mice or by treatment with ACEA or rimonabant ( Figure 2C and D).

Hepatic Metabolism
Cytosolic phosphoenolpyruvate carboxykinase (PCK1) catalyzes the conversion of oxaloacetate to phosphoenolpyruvate, the rate-determining step of gluconeogenesis. In the healthy liver, gluconeogenesis takes place mainly in periportal hepatocytes, where a higher expression of PCK1-positive cells can be displayed. A pathologic disturbance of the metabolic zonation in Abcb4 -/mice was detected by immunohistochemical PCK1 staining. The liver tissue of untreated Abcb4 -/mice showed an impaired structure ( Figure 3A, arrows) and zonation ( Figure 3A, arrowheads). Treatment with rimonabant largely preserved a clear zonation, similar to the WT controls. Intriguingly, treatment with ACEA also preserved zonation to a large extent, which might indicate additional physiologic relevance apart from CB1 agonization. The serum glucose levels remained unchanged in all groups (data not shown).
Peroxisome proliferator activated receptor (PPAR)a is involved in the regulation of lipid catabolism and glucose homeostasis. Gene expression of important regulators of glucose metabolism, such as Ppara, Cpt1, and Pck1, were examined by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) at the mRNA level ( Figure 3B and C).
Ppara but not Cpt1 gene expression showed a tendential down-regulation in untreated Abcb4 -/mice compared with WT mice (P ¼ .082). Rimonabant treatment normalized Ppara gene expression (P ¼ .023), but had no effect on Cpt1. ACEA treatment had no effect on Ppara ( Figure 3B). Although Cpt1 was not altered in Abcb4 -/mice, we found reduced expression in ACEA-treated Abcb4 -/mice in comparison with WT mice ( Figure 3B). In summary, zonation of PCK1 and gene expression of Ppara were normalized by rimonabant treatment. Serum glucose levels were largely unaffected by these changes. However, although zonal perturbation occurred, the total amount of hepatic Pck1 mRNA did not change ( Figure 3C).
Because glucose and lipid metabolism are closely linked, the gene expression of Fasn and the transcription factor Pparg were examined, both of which have regulatory functions in lipid metabolism. The hepatic amount of Fasn mRNA was down-regulated in untreated Abcb4 -/mice. Rimonabant treatment tendentially increased Fasn, while ACEA treatment normalized Fasn gene expression to WT levels ( Figure 3D). The gene expression of Pparg was not altered in Abcb4 -/mice compared with WT mice. Rimonabant treatment led to an up-regulation, compared with untreated Abcb4 -/mice ( Figure 3D).
Immunohistochemical staining of fatty acid synthase (FASN) also showed a disturbed zonation ( Figure 3E, arrowheads) and a preserved distribution of hepatocellular FASN protein expression after ACEA and rimonabant treatment.
In summary, a pathologic remodeling of the liver was found in the untreated Abcb4 -/group with dissolved zonation and lobular structure. Analogous to the liver section of the WT group, the rimonabant-treated (and also ACEA-treated) group showed a clear zonation of the liver. Normal liver structure and enzyme values at the WT level were found in both treatment groups.

Hepatic Inflammation
Former studies showed that the alterations in lipid metabolism mediate inflammation, fibrosis, and proliferation in Abcb4 -/mice. 12 Immunostaining showed an infiltration of a higher number of cluster of differentiation antigen 45 (CD45)-positive leukocytes in untreated and ACEA-treated Abcb4 -/mice ( Figure 4A and B). Remarkably, the treatment with rimonabant reduced this cholestasisassociated infiltration of inflammatory cells. Histopathologic examination also suggested a lower hepatic inflammation score in the rimonabant-treated group, but statistical significance was not reached. Likewise, the H&E staining might indicate a moderate infiltration of inflammatory cells in this group ( Figure 1B).
Lipocalin-2 (LCN2, also named neutrophil gelatinaseassociated lipocalin (NGAL)), an important component of the acute-phase reaction, increasingly is expressed in neutrophil granulocytes infiltrating the tissue, but also in hepatocytes activated by proinflammatory stimuli. Although neither hepatic mRNA levels nor protein expression of Lcn2/LCN2 was altered significantly, an increased infiltration of LCN2-positive cells as a characteristic feature of the biliary inflammatory process was shown by immunohistochemistry in the untreated Abcb4 -/group ( Figure 4C-F). Interestingly Lcn2 was increased in the ACEA group in comparison with WT mice, but normalized to WT levels in rimonabant-treated mice ( Figure 4C). The same effects were observed for the infiltration of LCN2-positive cells ( Figure 4F). Moreover, rimonabant reduced the number of infiltrating LCN2-positive cells significantly ( Figure 4F).
Monocyte chemotactic protein 1 (Mcp-1), an inflammatory cytokine, recruits monocytes, T cells, and dendritic cells to the site of inflammation. Untreated Abcb4 -/mice showed up-regulation of Mcp-1. Rimonabant reduced Mcp-1 to WT levels ( Figure 5A). Tumor necrosis factor-a has various functions in liver disease, including attraction and activation of inflammatory cells as well as mediation of hepatotoxicity and regeneration. Tnf-a was induced in untreated Abcb4 -/mice and reduced to WT levels by rimonabant, while ACEA had no effect ( Figure 5B).

Fibrosis
Collagen-1 is the main component of fibrotic tissue. Sirius Red staining indicated pronounced fibrosis in untreated Abcb4 -/mice with deposition of fibrillar collagen around the portal fields and around proliferating bile ducts ( Figure 5A). Histologically, rimonabant-treated mice showed a lower collagen deposition, approximately at WT level, compared with untreated Abcb4 -/mice. In addition, ACEAtreated mice showed increased hepatic collagen deposition, however, this was still lower than the level of collagen expression of untreated Abcb4 -/mice ( Figure 5C).
To analyze the locoregional correlation of collagen-1 and proliferating bile ducts, collagen-1/CK19-co-staining was performed ( Figure 5D). We observed a spatial coherence of proliferating bile ducts and collagen-1 deposits in the untreated Abcb4 -/group. Abcb4 -/mice showed a strong deposition of collagen-1 and increased proliferation of bile ducts (arrowheads, Figure 5D). Moderate bile duct proliferation and collagen-1 deposition were observed in the rimonabant group, comparable with the WT group. Bile duct proliferation and collagen-1 deposition in ACEA-treated mice were similar to untreated Abcb4 -/mice. The CK19-positive cells that did not form bile ducts appeared in the periphery of portal tracts in Abcb4 -/and ACEA mice (arrows, Figure 5D Abcb4 -/and ACEA). Nevertheless, it actually remains speculative whether these cells indicate the initiation of newly forming bile ducts or other associated processes ( Figure 5D).
The cholestasis-dependent increase of hepatic hydroxyproline content in Abcb4 -/mice appeared slightly reduced by ACEA and rimonabant treatment, but did not reach statistical significance ( Figure 5E).
Fibrosis-associated genes such as Timp1 and Mmp2 were subjected to qRT-PCR analysis.
To analyze the proteolytic potential of extracellular matrix (ECM), the gene expression of Mmp2 was analyzed by qRT-PCR. Treatment with rimonabant induced Mmp-2 gene expression in comparison with WT and untreated Abcb4 -/mice, while Timp-1 mRNA levels were similar among all groups ( Figure 5F).
Taken together, fibrosis was clearly induced in Abcb4 -/mice whereas fibrogenesis did not appear to a lesser extend in rimonabant-treated mice. Nevertheless, histopathologic assessment of fibrosis was improved after rimonabant treatment. Interestingly, we observed a pronounced proteolytic potential indicated by Mmp-2 induction in rimonabant-treated mice.

Malignancy-Associated Signaling and Proliferation
Cellular-JUN and signal transducer and activator of transcription 3 (STAT3) are critical regulators of liver cancer development and progression. 14,15 Both pathways also are involved in cholestasis-associated carcinogenesis. 16 To analyze the activation of the c-Jun N-terminal kinase (JNK)/c-JUN signaling pathway, the phosphorylation of c-JUN was examined by Western blot in comparison with the expression of its unphosphorylated form ( Figure 5A).
c-JUN was activated in liver tissue of Abcb4 -/mice while rimonabant treatment resulted in reduced phosphorylation of c-JUN ( Figure 5A), and reduced hepatocellular nuclear translocation ( Figure 5B). STAT3 phosphorylation was not altered significantly between the mouse groups ( Figure 5C). According to the regulatory role of c-JUN in the cell cycle, 17,18 cyclin D1 expression was reduced to WT levels by treatment with rimonabant ( Figure 5D).
In summary, our results show that c-JUN was activated by chronic cholestatic liver damage in Abcb4 knockout mice. The treatment with rimonabant reduced this activation to WT levels, while ACEA had no significant influence. Furthermore, c-JUN-associated proliferation, shown here by Ccnd1 expression, followed the same trend.

Discussion
The common final outcome of chronic liver diseases is the development of liver inflammation, fibrosis, and cirrhosis. 19 Every year, an estimated 1,200,000 people worldwide die from its complications, including portal hypertension and hepatocellular carcinoma, thus demonstrating the need to establish new and effective treatment options for chronic liver diseases. 20 The removal of the cause of liver injury can lead to regeneration of the damaged liver even with advanced cirrhosis. 21 If this is not possible, there is still no effective antifibrotic or anticirrhotic therapy available to date.
Our present work has shown that the administration of the CB1 antagonist rimonabant in Abcb4 -/mice did the following: (1) reduced damage, inflammation, and histopathologic fibrosis of the liver; (2) maintained liver integrity and zonation; and (3) reduced the activation of carcinogenesis-associated signaling pathways.
Thus, modulating the liver endocannabinoid system might be a potential therapeutic option to treat liver injury associated with cholestasis.
In different mouse models such as CCl 4 -, thioacetamide-, and bile duct ligation-induced fibrosis in resected human cirrhotic livers, in cell cultures of human hepatic stellate cells, and hepatic myofibroblasts, the stimulation of CB1 induced profibrotic effects, while the stimulation of CB2 resulted in the opposite outcome. 2,22,23 In diet-induced obese mice, rimonabant had positive effects on liver metabolism and induced a reduction of liver fibrosis. 2,6,24,25 Here, the influence of a pharmacologic modulation of CB1 by the CB1 antagonist rimonabant and ACEA (a CB1 agonist) in Abcb4 -/mice on a Bagg Albino Mouse/c genetic background 26 was examined. Up-regulation of SREBP-1c and Fasn in the signal cascade of activated CB1 contributes to the development of obesity and fatty liver via increased lipogenesis. 11 In bile duct-ligated mice, the CB1 receptor is down-regulated during cholestasis. 8 Accordingly, in Abcb4 -/mice lipogenesis is reduced during cholestasis 13 and these alterations in lipid metabolism mediate inflammation, fibrosis, and proliferation. 12 In our study, we showed an increase of nuclear matured SREBP-1 in untreated Abcb4 -/mice. Neither ACEA-nor rimonabanttreated Abcb4 -/mice showed significant alterations of nuclear SREBP-1.

Damage Caused by Cholestasis
In our study, increased ALT and AST values in untreated Abcb4 -/mice showed increased liver cell damage, which did not occur in rimonabant-treated (and ACEA-treated) mice. The ALT values were reduced significantly under rimonabant, but still higher than the reference values of healthy WT mice (50 U/L). 27 Incomplete ALT normalization could be attributed to the blood concentration of rimonabant being too low owing to the oral dosage form, incomplete absorption in the intestine, or any further metabolization. 23,28 Nevertheless, an optimized pharmacologic antagonization of CB1 might be a promising target to handle liver damage in cholestasis.
The reduction in ALT level in the ACEA group was interesting because, in analogy to other models of chronic liver diseases, the values were expected to worsen. 29 Repeated administration of CB1 agonists, however, could induce CB1 internalization or a reduction of CB1 protein synthesis. 30 Coupling of ACEA to other receptors with protective effects on liver damage was unlikely for a long time because of its previously assumed high specificity for CB1. However, it could be shown that ACEA acts as an agonist of the transient receptor potential vanilloid 1. 31 Activation of transient receptor potential vanilloid 1 by, for example, capsaicin, led to lower lipid droplet formation in the liver of high-fat diet-fed mice. 32 Taking this into account, we speculate that ACEA administration may have a protective effect on some cholestasis-associated liver changes via the agonism at transient receptor potential vanilloid 1, which could explain some of the coherent effects by rimonabant and ACEA that we describe with the current study.
Because of an increased permeability of bile duct epithelium, increased bile acid concentrations can be measured in the blood circulation of Abcb4 -/mice. 33 The conjugation of bile acids with amino acids such as taurine or glycine increases their detergent properties and prevents their precipitation in an acidic environment. It is known that the bile acids in rodents are 80%-90% taurine-conjugated. 34,35 Lower concentration levels of total bile acids in the serum were measured after ACEA and rimonabant treatment compared with the untreated group. Missing statistical significance might be owing to the limited group size and high number of different bile acids measured. Interestingly, ACEA-treated mice reached lower levels than rimonabant-treated mice with regard to inflammation and bile acid concentration. In the analysis of bile acids in the serum, the dominance of taurine-conjugated bile acids, in particular, the tauro-b-muricholic acid, tauro-a-muricholic acid, and tauro-u-muricholic acid, was confirmed in all samples. 9 The reduced parenchymal damage ( Figure 1B and C) indicates that the agonization of the CB1 receptor also might have hepatoprotective effects in Abcb4 -/mice. As described earlier, this may be owing to further interactions of ACEA on hepatic receptors. However, the protective effect of rimonabant in Abcb4 -/mice is congruent with other models of chronic liver damage such as CCl 4 -induced, thioacetamideinduced, and bile duct ligation-induced liver fibrosis. 2

CB1 in View of Hepatic Glucose and Lipid Metabolism
In the context of cholestatic diseases, the liver parenchyma and thus the liver architecture and the sophisticated zonation is disrupted. In the present work, an abolished zonation in untreated Abcb4 -/mice as well as in ACEAtreated Abcb4 -/mice could be shown by immunohistochemical staining for PCK1 and FASN ( Figure 3A and E). Physiologic zonation was regained by treatment with rimonabant.
The endocannabinoid system contributes to the development of steatosis, dyslipidemia, and insulin resistance. 36 PCK1, the rate-determining enzyme of gluconeogenesis, did not show clear staining in the periportal zone, but a diffuse distribution of PCK1-positive hepatocytes. 37 Ghafoory et al 38 showed that the CCl 4 -induced liver damage led to a strong change in the gene expression for enzymes of glucose metabolism in the periportal and perivenous zones. The typical zonal expression of PCK1 was largely preserved through rimonabant treatment ( Figure 3A), which also reflects serious protective effects of rimonabant on metabolism. The gene expression levels of Pck1 did not show any differences between the groups (Figure 2C), and serum glucose concentrations were within the normal range in all groups. Thus, euglycemia probably might be attributed to a compensated metabolism in the stage of liver fibrosis, but not yet cirrhosis or decompensation.
Various models have shown that hepatic CB1 activation increases de novo lipogenesis by activating Srebp1c and Fasn, and at the same time reduces fatty acid oxidation. 11,39 Analogous to a previous study, 11 untreated Abcb4 -/mice showed down-regulation of Fasn expression in our study. The normalization of Fasn expression by rimonabant was associated with a normalization of metabolic processes. Immunohistochemistry for FASN indicated a diffuse distribution of FASN-positive hepatocytes in untreated Abcb4 -/mice and a normalization of the typical zonation in rimonabant-treated mice.
Recently, we showed that the CB1 knockout in vivo and pharmacologic antagonization of CB1 in cell culture decreased PLIN2 expression, which might be an essential step in lipid breakdown. 7 In this study, we show that the pharmacologic modulation of CB1 represents a novel therapeutic approach for the treatment of cholestatic liver injury.

Regulation by PPARs
In the liver, a protective effect of PPARa on the development of NASH and inflammation was shown by downregulation of nuclear factor kappa-light-chain-enhancer of activated B cells, activator protein 1, STATs, and interleukin 6. 40 In the present work, Ppara was tendentially reduced in untreated Abcb4 -/mice compared with WT. In the rimonabant group, normalization of Ppara to WT level was observed, while Ppara expression remained unchanged with ACEA treatment ( Figure 3B). Because there were no differences in the serum glucose concentrations, the cholestatically reduced Ppara expression appears to have no effect on stable hepatic glucose metabolism. Among others, PPARg regulates the differentiation of adipocytes and contributes to lipid accumulation in the liver. These effects are moderated by induction of SREBP-1c, acetyl-CoA carboxylase (ACC), and FASN. 41 Interestingly, Pparg expression was induced significantly in the rimonabant group. The ACEA group, however, showed no differences compared with the untreated Abcb4 -/mice.

Hepatic Inflammation
The enhanced number of infiltrated CD45 þ leukocytes in Abcb4 -/mice was abolished by rimonabant, which indicates an anti-inflammatory effect of pharmacologic CB1 antagonization during cholestasis. Up-regulation of lipocalin 2 during endoplasmic reticulum (ER) stress-induced inflammatory responses protects hepatocytes from being overwhelmed by unfolded protein response upon liver injury. 42 On the other side, LCN2 is secreted into the serum from liver cancer tissue in human beings and mice. 43 Lcn2 increasingly was expressed in the ACEA group, whereas the rimonabant group showed normalized Lcn2 expression on the WT level. A clinical study of 716 patients with cirrhosis showed that LCN2 might be a biomarker of acute-on-chronic liver failure and prognosis in cirrhosis. 44,45 Because LCN2 is a good candidate for hepatocellular carcinoma diagnosis and screening, the reduction of Lcn2 might indicate a beneficial development. Interestingly, the untreated group and the ACEA group did not show up-regulation of interleukin 6, nuclear factor kappa-light-chain-enhancer of activated B cells, I-kB, protein kinase B, or other mediators inducing the aforementioned LCN2 (data not shown). It must be assumed that LCN2 is activated by the damaged hepatocytes via alternative routes. The measurement of proinflammatory markers such as LCN2, tumor necrosis factor-a, and monocyte chemoattractant protein-1 shows the antiinflammatory effect of rimonabant, which correlated with reduced liver damage in our murine model.

Fibrosis
As expected, untreated Abcb4 -/mice showed an increased overall hepatic collagen deposition, which was not altered by treatment with ACEA or rimonabant. Nevertheless, histopathologic assessment of fibrosis was improved after rimonabant treatment. The net deposition of scar tissue depends on the balance between synthesis and degradation. 3,46 The latter reflecting the relative activity of matrix metalloproteinases and their tissue inhibitors of metalloproteinases, which are produced primarily by hepatic stellate cells. The activation of matrix metalloproteinases leads to the dissolution of the deposition of ECM. The activity of the matrix metalloproteinases thus leads to fibrosis regression. 46,47 On gene expression levels, the treatment with rimonabant led to an upregulation of the proteolytic potential, enabling enhanced degradation of ECM in the fibrotic liver. This fact may be reflected by the moderate level of fibrosis in rimonabanttreated animals.

Signal Transduction
In acute and chronic liver damage, JNK1 and its downstream signals are activated and contribute to disease progression. 48 In the present work, Western blot analysis of untreated Abcb4 -/mice showed an increased activation of c-JUN. The hepatoprotective effect of rimonabant treatment was reflected in a lower activation of c-JUN, which was normalized to WT level. In 2001, Gupta et al 49 already showed a direct activation of JNK1 by taurine-conjugated bile acids in a model of rat hepatocytes. C-JUN and cyclin D1 were activated in untreated Abcb4 -/mice ( Figure 6). Rimonabant reduced this activation to WT level. Treatment with ACEA had no effect.

Limitations
A limitation of the study was the way in which medication was administered via the drinking water, which means that different concentrations of active substances cannot be ruled out on an individual basis. Both the amount of substances absorbed via the intestine and the further metabolization remain unknown factors. Alternatives such as intravenous or intra-abdominal injection did not appear physiological. In addition, these procedures might be technically difficult, associated with animal stress, and time consuming.
In summary, the CB1 blockade in Abcb4 -/mice by rimonabant enabled a reduction of liver damage as well as inflammatory and acute phase markers such as LCN2, Mcp1, and Tnf-a. With regard to liver metabolism, treatment with rimonabant resulted in the preservation of the typical liver zonation. Proliferation and carcinogenesis-associated signaling pathways were normalized to WT levels by treatment with rimonabant. We thus conclude that the regulation of the JNK signaling pathway by CB1 modulation might play an important role in cholestatic liver diseases and, if applicable, in hepatic carcinogenesis.

Animal Treatment
The present study was performed with permission of the State of Hesse, Regierungspraesidium Giessen, according to section 8 of the German Law for the protection of animals, and conforms to the National Institutes of Health guide for the care and use of laboratory animals. All experiments were approved by the committee on the ethics of animal experiments of the Regierungspraesidium Giessen, Germany (permit number: V54-19c2015 (1) GI20/10 no. 52/2011). Bagg Albino Mouse/c-Abcb4 -/mice (C.FVB[129P2]-Abcb4 tm1Bor , herein referred to as Abcb4 -/mice) were bred and housed as described previously. 26 Characterization of Abcb4 -/genotype, sample collection, and routine analysis have been described elsewhere. 26 In this project, the population of male Abcb4 -/knockout mice was divided into 3 groups. One group (n ¼ 4) remained untreated and received standard chow. The second group (n ¼ 4) was treated with 1 mg/kg body weight per day ACEA after weaning from the mother in the third week of life. A third group (n ¼ 5) was fed with 1 mg/kg body weight per day rimonabant. Standard chow (R/M-H) supplemented with ACEA and rimonabant was obtained from Sniff (59494; Spezialdiäten GmbH, Soest, Germany). Twenty kilograms of chow was charged with 133 mg rimonabant or 125 mg ACEA.
Mice were killed at the age of 16 weeks. Livers and blood sera were isolated and stored at -80 C. Sixteen-week-old WT mice were used as healthy supercontrols.

Bile Acid Analysis
Bile acids were quantified by Ultra performance liquid chromatography-tandem mass spectrometry, as has been described in the literature. 51

Hydroxyproline Assay
The total hepatic hydroxyproline content was quantified as described previously with minor modifications. 52,53 Briefly, 50 mg mouse liver tissue was hydrolyzed in 1 mL 6 N HCl at 110 C for 14 hours. Hydrolysates were filtered through 45-mm pore filters (Sartorius, Göttingen, Germany). A total of 15 mL of the hydrolysate was dried under nitrogen flow and subsequently redissolved in 50 mL 50% 2propanol. A total of 100 mL of 0.6% chloramine-T (Merck, Darmstadt, Germany) solution was added to the samples and hydroxyproline standard probes (4-hydroxy-L-proline; Sigma-Aldrich, Taufkirchen, Germany) and incubated for 10 minutes at room temperature. Ehrlich's solution (100 mL, 3 g dimethylamino-benzaldehyde [Sigma-Aldrich] in 26 mL 2-propanol þ 8 mL 70% perchloric acid) was added and the samples again were incubated for 45 minutes at 50 C. Absorbance was measured at 570 nm using a microplate reader (Packard BioScience, Meriden, CT). Hydroxyproline levels were calculated against standard curves and expressed as milligrams of hydroxyproline per gram of liver tissue.
Quantitative Real-Time PCR RNA extraction and complementary DNA synthesis as well as qRT-PCR were performed as described previously. 50 Briefly, hepatic RNA was extracted using the RNeasy Mini (QIAGEN, Hilden, Germany), and elimination of genomic DNA was performed with the TURBO DNAfree-Kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's instructions. RNA integrity and purity were analyzed by gel electrophoresis and spectrophotometry and equal amounts of RNA were transcribed into complementary DNA using the iScript complementary DNA Synthesis-Kit (Bio-Rad, Hercules, CA). qPCR was performed using a StepOne-Plus real-time PCR system (Life Technologies, Darmstadt, Germany) and SYBR-Green/ROX dye (Sigma Aldrich, Steinheim, Germany). Primers were purchased by Microsynth (Göttingen, Germany). Individual gene expression was calculated according to the delta delta cycle treshold (DDCt) method. 54

Western Blot
Western blot experiments were performed as described before 55

Triglyceride Measurement
Triglyceride quantification was performed according to the manufacturer's instructions (ab65336; Abcam, Cambridge, MA).