Transcription factor Fra-1 induces cholangitis and liver fibrosis

Authors


  • Supported by the Interdisziplinäres Zentrum für Klinische Forschung Erlangen Project A34 (to G.S. and J.Z.), Deutsche Forschungsgemeinschaft Grant FOR 661 (to G.S.) and SFB 423 (to J.Z. and G.S.) and the European Union projects Masterswitch, Adipoa and Kinacept.

  • Potential conflict of interest: Nothing to report.

Abstract

Chronic diseases of the biliary system are common and may cause fibrosis and eventually progression to liver cirrhosis. The aim was to define a new mouse model of a cholangiopathy leading to liver fibrosis in fra-1tg mice. Liver pathology of fra-1tg mice was analyzed in detail by histology and flow cytometry. Transcript levels of fibrosis-related genes and matrix metalloproteinase (MMP) activities were quantified and immunohistochemical analysis additionally applied. The role of the immune system in this model was analyzed by crossing fra-1tg mice with rag2−/− mice. Furthermore, expression of Fra-1 in corresponding human liver diseases was investigated on transcription level and histologically. Fra-1tg mice spontaneously develop biliary fibrosis preceded by ductular proliferation and infiltration of inflammatory cells. Fra-1 protein is present in cholangiocytes and inflammatory cells within the liver. These findings were replicated in human biopsies of patients with advanced liver fibrosis. The inflammatory infiltrate showed a strong increase in activated T cells and decreased natural killer (NK), natural killer T cells (NKT), and B cells in fra-1tg mice as compared to wildtype mice. Moreover, fra-1tg mice develop biliary fibrosis with a time-dependent increase in hepatic collagen content and increase in relative messenger RNA (mRNA) expression of profibrotic genes. Attenuation but not complete prevention of collagen accumulation in liver was observed in the fra-1tg × rag2−/− mice. However, transplantation of fra-1tg bone marrow cells into wildtype mice could not induce disease. Conclusion:Fra-1tg mice spontaneously develop a progressive biliary disease. These mice are an attractive model for the investigation of cholangiopathies and their interaction with the immune system. (HEPATOLOGY 2011;)

The AP-1 transcription factor complex is composed of Jun (c-jun, JunB, JunD) and Fra proteins (c-fos, fosB, fra-1, fra-2), and regulates physiological processes such as stress responses, apoptosis, inflammation, and cancer development.1 Genetic overexpression or deletion of single components of AP-1, however, has revealed the specific involvement of the individual AP-1 members in various disease processes. Fra-1tg mice develop osteosclerosis and have a reduced lifespan, most likely due to progressive destruction of the bone marrow.2

Apart from its effects on bone metabolism, there are several data about the role of Fra-1 in tumor and metastasis development. Overexpression of fra-1 has been reported in several transformed human cell lines3 and possible target genes were also detected.4 Further, there are some data about DNA binding activity of the AP-1 complex in various types of human tumor such hepatocellular carcinoma (HCC), gastric carcinoma, and breast carcinoma.5, 6 A particular involvement of fra-1 in hepatocellular and biliary disorders is not yet known.

Cholangiopathies are a frequent cause of impaired liver function and may progress to liver cirrhosis.7 Several disorders with different etiology, such as primary biliary cirrhosis (PBC), drug-induced cholangiopathy, and graft versus host disease (GVHD) primarily affect the small bile ducts. In contrast, primary sclerosing cholangitis (PSC) mainly involves the large intra- and extrahepatic bile ducts. The pathogenesis of liver fibrosis in these disorders is yet unclear but may involve parenchymal cells such as hepatic stellate cells (HSCs) and cholangiocytes as well as natural killer (NK) cells.

Cholangiocytes are key players in the hepatic response to biliary injury.8 Cholangiocytes respond to various types of injury with proliferation and stimulation of HSC.7 Thus, a common histological finding in the earlier phases of cholangiopathy is proliferation of the small bile ducts. This is often accompanied by an inflammatory infiltrate in the portal tracts. Although the etiology of cholangiopathy varies, the pathogenic processes may share similarities. Inflammation and bile duct proliferation is ultimately followed by a loss of bile ducts and, in the case of chronic cholestatic diseases, by a fibrotic response.9

The exact mechanisms how cholangiocyte injury triggers liver fibrosis are unclear. Several rodent models for cholangiopathy including bile duct-ligation and xenobiotic-administration or spontaneous models have been described.9

Inducible rodent models are indeed helpful for studying the pathways during cholangiopathy development but they cannot reproduce the exact disease course. Spontaneous rodent models are rare. One of the well studied ones is the Mdr2 knockout mouse. The Mdr2 knockout mouse lacks bile phospholipids leading to disruption of bile ducts and, moreover, leakage of bile acids to the portal tract. Further, a strong inflammatory and fibrotic reaction develops and typical cholangitis is evident.10

In this study we describe a new spontaneous model of cholangiopathy associated with bile duct proliferation leading to liver cirrhosis, based on the overexpression of the transcription factor fra-1.

Abbreviations

ALP, alkaline phosphatase; BD, bile duct; CD3, cluster of differentiation 3; ChIP, chromatin Immunoprecipitation; CK19, cytokeratine 19; fra-1tg, fra-1 transgenic mice; GVHD, graft versus host disease; HCC, hepatocellular carcinoma; HSC, hepatic stellate cells; IHC, immunohistochemistry; MMP, matrix metalloproteinase; NK cells, natural killer cells; NKT cells, natural killer T cells; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; PDGF, transforming growth factor; PMNC, polymorphonuclear cells; RT-PCR, reverse-transcription polymerase chain reaction; TIMP-1, tissue inhibitor of metalloproteinases; TGF-β1, transforming growth factor β1.

Materials and Methods

For more details, regarding the following procedures, see the Supporting Information.

Mice and Genotyping.

Mice that constitutively overexpress fra-1 under the major histocompatibility complex class I antigen H2-Kb (H2) promoter (fra-1tg mice, background: C57Bl6) were used.3Fra-1tg × rag 2−/− mice were obtained by cross-breeding.

Human Tissues.

Human liver biopsy specimens were obtained from University Hospitals Graz. Biopsy specimens were registered in the respective biobank and kept anonymous. The research project was authorized by the ethical committee of the Medical University of Graz (Ref. No. 1.0 24/11/2008). The study protocol was in accordance with the ethical guidelines of the Helsinki Declaration.

Histology.

Histology analyses were made on paraffin sections of liver tissue. Hepatic levels of alkaline phosphatase (ALP) activity were analyzed spectrophotometrically.

Hydroxyproline Assay.

Liver collagen content was assessed by analyzing the hydroxyproline content.11 The value of the liver hydroxyproline level was expressed as μg/50mg wet tissue.

Reverse-Transcription Polymerase Chain Reaction (RT-PCR).

Total RNA was extracted from liver tissue and complementary DNA (cDNA) was synthesized using a Reverse Transcription System Kit. Quantitative real-time RT-PCR was performed using LightCycler technology.

Immunohistochemistry (IHC).

Immunohistochemistry was performed on paraffin sections (5 μm thickness). Antibodies used in IHC are provided in the Supporting Information.

Isolation of Blood Cells and Analysis by Flow Cytometry.

Intrahepatic nonparenchymal and blood cells were isolated using 37% Percoll solution and further characterized by fluorescence-activated cell sorting (FACS) analysis.

Chemokine Array.

Expression of chemokines and cytokines was determined in liver tissue of 6-week-old fra-1tg and wildtype mice by RT2 Profiler PCR Array (SABioscience, Germany).

Chromatin Immunoprecipitation (ChIP) Assay.

Bile duct tissue was isolated using a modified protocol from Aviva. ChIP analysis was performed using a commercially available kit from Cell Signaling.

Statistical Analysis.

Data are expressed as the mean ± standard error of the mean (SEM). Group mean values of histological data were compared by paired Student's t test. P-values less than 0.05 were considered significant.

Results

Fra-1tg Mice Develop Progressive Hepatic Portal Inflammation and Ductular Reaction.

As previously described, fra-1 transgenic mice were smaller in size and weight than their wildtype littermates.2 Impressively, fra-1tg mice developed hepatic inflammatory infiltrates mainly confined to the portal tracts (Fig. 1). Infiltration of cells in fra-1tg mice was evident at 6 weeks of age (mean inflammatory area 6.33 ± 1.14 mm2). At 18 weeks we observed bridging inflammatory infiltrates between neighboring portal tracts. However, at 23 weeks of age inflammatory infiltrates covered large areas of the liver of fra-1tg mice (mean inflammatory area fra-1tg 21.08 ± 5.47 mm2 versus wildtype 1.48 ± 0.71 mm2; P < 0.05; Fig. 1B). Even more interesting, we observed a ductular reaction in fra-1tg mice, assessed by staining for cytokeratine 19 (CK19) protein (Fig. 1A).

Figure 1.

Liver morphology of fra-1tg mice versus wildtype mice. Liver morphology of fra-1tg and wildtype mice was examined by conventional histology. Fra-1tg mice showed an inflammatory reaction with a strong increase of inflammatory infiltrates in a time-dependent manner. Formation of new bile ducts (BD) was confirmed and shown by staining with an anti-CK19 antibody. Quantification of inflammatory area is presented as an x-fold increase versus wildtype control (B). Representative images are shown in magnification of 40× for hematoxylin and eosin and 100× for CK19 staining. *P < 0.05 compared to wildtype controls of the corresponding age.

Although we could not detect major changes in the morphology of the larger bile ducts, the small bile ducts were increased in their numbers (Fig. 2). At 10 weeks of age the mean number of bile ducts per portal tract was 1.38 ± 0.05 and 3.24 ± 0.22 for wildtype and fra-1tg mice (mean ± SEM; P < 0.05), respectively. This further increased up to 4.90 ± 1.44 bile ducts in fra-1tg mice at 23 weeks of age, whereas it remained unchanged in wildtype mice (1.32 ± 0.07; P < 0.05).

Figure 2.

Body parameters of fra-1tg and control mice. Body weight, body mass index, liver weight, and bile duct (BD) number and diameter were determined at the age of 6, 10, 18, and 23 weeks. Liver ALP activity was determined at the age of 10 and 23 weeks in fra-1tg (black columns) and wildtype (white columns) mice. Quantification of BD number and average BD area is presented as an x-fold increase versus wildtype mice. *P < 0.05 compared to wildtype controls of the corresponding age.

We next investigated the activity of ALP of fra-1tg and control mice (Fig. 2). ALP is an enzyme presented in bone and liver. As mentioned, fra-1tg mice develop osteosclerosis. For our study we investigated the activity of ALP directly in liver tissue. Increased ALP activity was observed in fra-1tg mice at almost all timepoints, as seen at week 10 (wildtype 1.90 ± 0.74 versus fra-1tg 3.73 ± 1.48; P < 0.05) and 23 (wildtype 0.98 ± 0.14 versus fra-1tg 3.66 ± 1.53; P < 0.05). Thus, specific increase of intrahepatic ALP points to the presence of biliary and liver abnormalities.

Hepatic Up-Regulation of Chemokines in Fra-1tg Mice.

As we observed a strong infiltration of immune cells into the portal tracts, we next asked whether increased chemokine and cytokine expression mediates this cell influx. We found a dramatic induction of distinct chemokines, such as CXCL5 (22-fold), CCL1, CCL5, and CCL7 (about 3.5-fold), CCL-8 (10-fold), and CCL20 (10-fold). We also observed an up-regulation of chemokine receptors such as CCR4 (9-fold) and CCR2 (4-fold) in the liver of fra-1tg mice (Supporting Fig. 1).

Hepatic Infiltration with Activated T Cells and Neutrophils in Fra-1tg Mice.

To determine the composition of the inflammatory infiltrates, we first performed IHC on liver sections of wildtype and fra-1tg mice. The infiltrate in fra-1tg mice is mainly composed of neutrophils and cluster of differentiation 3 (CD3+) T-cells. B cells and macrophages are less frequently found at the site of liver inflammation. This pattern was consistent in fra-1tg mice at all ages investigated (week 6, 10, and 18). Thus, overexpression of fra-1 causes a progressive infiltration of the liver by cells belonging to the innate and adaptive immune system (Supporting Fig. 2).

In order to further define the inflammatory process, we also examined the composition of isolated liver lymphocytes in wildtype and fra-1tg mice by FACS analysis (Fig. 3). The total number of CD3+ T-cells was strongly elevated in 10-, 18-, and 23-week-old fra-1tg mice as compared to wildtype littermates. The fraction of T cells among all infiltrating leukocytes was 73% in 10 weeks old fra-1tg mice and remained constant in 18- (73% of polymorphonuclear cells [PMNCs]) and 23- (75% of PMNCs) week-old fra-1tg mice. Both infiltration of CD4+ T cells and CD8+ cytotoxic T cells were increased in fra-1tg mice. Moreover, we observed a high percentage of CD4+CD25+ T cells in fra-1tg but not wildtype mice. However, these T cells were not regulatory T cells but activated T cells as determined by the absence of Foxp3 expression but expression of the activation marker CD69 (data not shown).

Figure 3.

Characterization of liver and blood PMNC composition. PMNC composition in fra-1tg and wildtype mice was determined by flow cytometry: CD3+ (T cells), CD4+ (T helper cells), CD8+ (T cytotoxic cells, CD19+ (B cells), NK1.1+ (NK cells), NK1.1+CD3+ (NKT cells) were measured. The cell number is presented as percentage of all PMNC or particular PMNC population. *P < 0.05 compared to wildtype controls of the corresponding age.

We also determined the frequency of NK and natural killer T-cells (NKT) in the liver of wildtype and fra-1tg mice, as these cells are crucial for hepatic inflammation.12 Interestingly, the number of NK cells (NK1.1+CD3-) and NKT cells (NK1.1+CD3+) in the liver of fra-1tg mice was dramatically reduced at all examined timepoints (Fig. 3). NK cells displayed only about 2% and NKT cells 2%-5% of all hepatic mononuclear cells in fra-1tg mice, which was about 4-fold and 10-fold less than wildtype mice (all timepoints, P < 0.05 between wildtype and fra-1tg mice).

We then assessed the B cells in the liver of wildtype and fra-1tg mice. We found a strong and consistent reduction of CD19-positive B-cells in fra-1tg mice at all ages investigated (data not shown).

Finally, we analyzed whether the changes in liver lymphocyte composition was a localized or generalized phenotype in fra-1tg mice. Thus, we performed FACS analysis of isolated PMNCs of 23-week-old mice. Similar to liver, the number of CD3+ cells (+32% compared to wildtype mice, P = 0.05) was strongly increased, with a higher level of CD4+ T cells, and a remarkable reduction of CD19+ B cells (−27% of PMNCs) in fra-1tg mice. However, we could not detect any differences in NK and NKT-cell distribution in the peripheral blood between wildtype and fra-1tg mice (Fig. 3).

Fra-1tg Mice Develop Progressive Biliary Cirrhosis.

To determine whether the observed hepatitis in fra-1tg mice leads to fibrosis and liver cirrhosis, we performed Sirius Red staining and analyzed for collagen deposition under brightfield and polarized light. In contrast to wildtype mice, livers of fra-1tg mice displayed significant fibrosis, which progressed in a time-dependent manner. In early stages (week 10), collagen deposition was observed in portal tracts and fibrosis was less evident, whereas fibrosis was widespread and bridged portal fields in older animals (week 23; Fig. 4A). A slight pericellular and perisinusoidal matrix deposition was detected by analysis of sections with polarized light and found in all examined ages of fra-1tg mice but not in controls (data not shown).

Figure 4.

Increased hepatic extracellular matrix content in fra-1tg mice. Accumulation of extracellular matrix in livers of fra-1tg mice was found by staining with Sirius Red and visualized under brightfield light (A). Fra-1tg mice showed increase in matrix tissue assessed by measurement of hydroxyproline content in liver tissue (B). Representative images are shown in magnification of 100×.

Further, Fra-1tg mice demonstrated increased total hepatic collagen accumulation with a steady increase in a time-dependent manner (Fig. 4B). Collagen content was found to be 10-fold higher than in control mice (10 weeks), reaching the highest level at 23 weeks of age. Thus, hepatic inflammation in fra-1tg mice is followed by progressive liver fibrosis.

Up-Regulation of Profibrotic Genes in Livers of Fra-1tg Mice.

To obtain further information about the key matrix molecules involved in the progression of hepatic fibrosis in fra-1tg mice, we analyzed messenger RNA (mRNA) expression of collagen production, profibrogenic, and fibrolytic genes and the time-course of their expression by quantitative real-time PCR (Fig. 5). First, we found a strong induction of procollagen α1 (I), α2 (I), and α1 (III) mRNA expression at all ages in fra-1tg as compared to wildtype mice. Furthermore, the profibrogenic cytokine transforming growth factor β1 (TGF-β1) was strongly induced in fra-1tg mice at early stages of disease (week 10), but declined to wildtype levels at week 18.

Figure 5.

Increased profibrogenic gene expression in fra-1tg mice. Relative hepatic mRNA transcript levels of profibrogenic and collagen genes were determined by real-time PCR at weeks 10, 18, and 23 of age in fra-1tg mice (black columns) and wildtype mice (white columns). Results are expressed as mean ± SEM (n = 4-5/group) and in arbitrary units relative to β-actin mRNA. Data are presented as an x-fold increase versus the corresponding wildtype controls. *P < 0.05 compared to wildtype controls of the corresponding age.

When analyzing the four isoforms of transforming growth factor (PDGF A to D), we found a distinct expression pattern. Expression of PDGF-A and -C was not different in fra-1tg mice as compared to wildtype littermates (Supporting Fig. 3). Interestingly, PDGF-B was induced in transgenic livers at later stages of disease (week 23), whereas the reverse was found for PDGF-D, which was induced early during the disease course (week 10, Fig. 5). We also assessed the expression of matrix metalloproteinases (MMPs), which act as counterregulatory fibrinolytic molecules in liver fibrosis. MMP-2 showed an expression peak at week 10 and a decline of expression thereafter, whereas MMP-9 showed an increase of expression over time, reaching its peak at week 23. Similar to MMP-2, the expression of tissue inhibitor of metalloproteinase (TIMP)-1 reached its peak at week 10 with a 10-fold increase in the liver of fra-1tg mice as compared to wildtype littermates (Supporting Fig. 3).

Colocalization of Fra-1 and Profibrotic Proteins Is Confined to Cholangiocytes in Fra-1tg Mice.

Next we analyzed the localization of Fra-1 protein in the livers of wildtype and fra-1tg mice by IHC. We found Fra-1-positive cells in the liver of transgenic animals. However, positive cells were restricted to specific sites: Cholangiocytes and infiltrating inflammatory cells clearly showed nuclear Fra-1 staining, whereas all other cells were negative (Fig. 6). We also assessed the expression of profibrotic proteins as TGF-β1 and PDGF-D by IHC. As shown in Fig. 6, the profibrotic proteins were expressed in cholangiocytes of fra-1tg but not in wildtype mice. Expression was confined to cholangiocytes of the small as well as the large bile ducts and was virtually absent in other hepatic cell lineages. We could not find expression of TGF-β1 and PDGF-D in other liver compartments.

Figure 6.

Profibrotic proteins are localized to bile ducts in fra-1tg mice. Expression of profibrotic proteins was determined by IHC. Fra-1 protein was only localized to intrahepatic inflammatory cells and bile ducts in fra-1tg livers. Fibrotic markers as PDGF-D and TGF-β1 were clearly present in bile ducts of fra-1tg mice as compared to wildtype controls. Representative images are shown in magnifications of 200× and 400×.

Additionally, we proved the binding of Fra-1 on the promoters of tgfβ1, pdgf-b, and pdgf-d genes in intrahepatic bile ducts. Therefore, we isolated bile ducts and performed a ChIP assay. We demonstrated binding activity of Fra-1 on potential AP-1 binding site of tgfβ1, pdgf-b, and pdgf-d promoters in wildtype and transgenic animals (Supporting Fig. 4).

Fra-1 Expression in Human Samples of Liver Fibrosis.

Next we analyzed the relevance of Fra-1 in samples of human hepatobiliary diseases with evidence of liver fibrosis. We analyzed Fra-1 expression and localization in samples of Wilson disease, focal nodular hyperplasia (FNH), hepatocellular carcinoma (HCC), hepatitis C virus (HCV), nonalcoholic fatty liver disease (NAFLD), PBC, primary sclerosing cholangitis (PSC) patients, and healthy controls. Interestingly, we determined the highest fra-1 mRNA expression in samples of PBC and PSC patients. Expression of fra-1 in liver biopsies of Wilson disease, FNH, HCC, HCV, NAFLD was also evident (Fig. 7A). Immunostaining for Fra-1 showed an evident localization of the transcription factor in inflammatory and bile duct cells in the human biopsies with liver fibrosis, similar to the fra-1tg mice. Healthy controls showed weak staining of inflammatory cells and bile ducts in the portal tracts. Further, we determined the number of Fra-1-positive cells morphometrically. We determined the highest presence of Fra-1-positive inflammatory and bile duct cells in samples of PSC and PBC patients (P < 0.05; Fig. 7A), for which representative images are shown in Fig. 7B.

Figure 7.

Fra-1 localization in human liver tissue. (A) Expression levels of fra-1 were determined in human liver samples of Wilson disease, FNH, HCC, HCV, NAFLD, PBC, PSC patients and healthy controls. Highest expression levels were determined in samples of PBC and PSC patients. mRNA data are presented as an x-fold increase versus the corresponding healthy controls. *P < 0.05 compared to healthy controls. Localization of Fra-1 protein was detected by IHC. Number of Fra-1 positive inflammatory cells is given as a percent of total number of inflammatory cells/portal tract and number of Fra-1-positive BD are shown per portal tract (P < 0.05). (B) Representative images show localization of Fra-1 protein in samples of PBC and PSC patients and healthy controls (magnification 400×).

Involvement of T Cells in the Progression of Liver Fibrosis in Fra-1tg Mice.

As there is a strong infiltration of activated T-cells in the livers of fra-1tg mice, we questioned whether immune cells actually drive hepatic fibrosis in this model. We therefore lethally irradiated wildtype mice and performed an adoptive transfer of bone marrow from fra-1tg mice (data not shown). These chimeric mice did not show any signs of liver inflammation, suggesting that Fra-1 expression in the nonhematopoietic compartment is crucial for development of liver fibrosis. Given that cholangiocytes are the only nonhematopoietic lineage expressing Fra-1, this further supports the notion that Fra-1 expression in cholangiocytes is critical for the liver pathology observed in fra-1tg mice.

We then crossed fra-1tg mice with rag2−/− mice to determine the contribution of lymphocytes to the progression of liver fibrosis. Interestingly, liver pathology was less pronounced in fra-1tg × rag2−/− mice. We could not detect any signs of an inflammatory reaction in the liver of fra-1tg × rag2−/− mice (Fig. 8). Although we could still detect liver fibrosis in fra-1tg × rag2−/− mice with a mean fibrotic area of 2.8 ± 0.5 mm2 as compared to fra-1tg mice (10 weeks, mean fibrotic area 6.0 ± 11.9 mm2), the amount of fibrosis was significantly (P < 0.05) reduced (Supporting Fig. 5). Investigations of mRNA expression of procollagen α1 (I), α2 (I), and α1 (III) in the fra-1tg × rag2−/− determined reduced expression as compared to fra-1tg × rag2+/- mice (Supporting Fig. 5). In addition, the ductular reaction also observed in fra-1tg mice was attenuated in fra-1tg × rag2−/− mice, suggesting that the inflammatory infiltrate participates in liver fibrosis of fra-1tg mice but is not an essential factor for its onset.

Figure 8.

Hematopoietic cell involvement in the development of spontaneous biliary fibrosis in fra-1tg mice. Fra-1tg mice develop spontaneous biliary fibrosis with no obvious commitment of hematopoietic cells, as seen by the transplantation of fra-1tg bone marrow in wildtype recipients. Further, biliary fibrosis was reduced after crossing fra-1tg mice with rag2−/− mice. Representative images are shown (magnification 40×).

Discussion

In this study we demonstrate the involvement of the AP-1 transcription factor Fra-1 in liver injury and fibrosis. Mice overexpressing fra-1 spontaneously develop progressive liver inflammation associated with a ductular reaction resembling cholestatic liver disease, finally causing liver fibrosis.

The clinical spectrum of cholangiopathies is diverse and includes genetic, autoimmune, and acquired diseases of the biliary tree. Despite their heterogeneous etiology, cholangiopathies share a similar clinical disease course. Many cholestatic diseases such as PBC or GVHD primarily involve the small bile ducts. Typically, the biliary epithelial cells respond to injury with proliferation causing the so-called “ductular reaction.” This is not only evident in human disease, but also in animal models of cholestatic liver injury such as the bile duct ligation model.13 However, in most of these disorders the small bile ducts are finally destroyed, causing ductopenia. In our study, we also demonstrated a strong proliferation of small bile ducts arising in fra-1tg mice very early in the disease course. However, we could not detect signs of small bile duct destruction or neoplastic transformation, which appear in certain cholestatic liver diseases. We cannot exclude that such findings would occur at later timepoints, as the lifespan of fra-1tg mice is limited due to progressive bone marrow obliteration caused by osteosclerosis.

Cholangiopathy was associated with progressive liver fibrosis in fra-1tg mice. Progressive liver fibrosis was associated with transient up-regulation of profibrotic cytokines, such as TGF-β and PDGF-D, especially early in the course of disease. These findings were verified by IHC, showing that the expression of these growth factors was mainly confined to the cholangiocytes. Further, Fra-1 directly binds to tgfβ1, pdgf-b, and pdgf-d promoter region as determined by ChIP analysis. Indeed, TGF and PDGF family members are implicated in hepatic fibrosis in animal models of liver fibrosis.14, 15 For instance, hepatic overexpression of TGF-β1 causes progressive liver fibrosis.16 Conversely, inhibition of intracellular signaling molecules of TGF-β1, such as ALK5, protects rats from experimental liver fibrosis.17 Moreover, transgenic expression of PDGF-A can also cause liver fibrosis via induction of TGF-β1. Thus, increased TGF-β1 and PDGF expression might link cholangiopathy in fra-1tg mice to the fibrotic response in the liver.

Interestingly, we could also localize nuclear Fra-1 to the cholangiocytes and to inflammatory infiltrates. Although Fra-1 is expressed in all the different liver cell types at the mRNA level (data not shown), protein synthesis of Fra-1 as determined by IHC is much more restricted. Neither hepatocytes nor other resident liver cells were found to express Fra-1 protein as determined by IHC. Thus, our model points to cholangiocytes as potential drivers of the fibrotic response. Indeed, cholangiocytes can respond to biliary injury with proliferation, secretion of chemokines, and profibrotic growth factors. The latter, in turn, induces activation of hepatic stellate cells, which are thought to be the primary producers of extracellular matrix in fibrotic liver diseases.18

Corresponding to our animal data, we found similar Fra-1 protein localization in human samples with advanced hepatic fibrosis from patients with Wilson disease, FNH, HCC, HCV, NAFLD, PBC, and PSC. Interestingly, samples of PBC and PSC patients showed the highest expression levels and the highest number of positive cells for Fra-1.

Most cholestatic diseases are characterized by a mixed portal inflammatory infiltrate. Its clinical significance, however, is unclear. We also found a strong infiltration of immune cells in the liver in fra-1tg mice. Furthermore, we detected up-regulation of certain chemokines such as CXCL5, CCL-1, CCL-5, CCL-8, and CCL-20 as well as chemokine receptors such as CCR4 and CXCR1 in the livers of fra-1tg mice. Indeed, lymphocytes were shown to be strongly attracted by CCL20.19 Strong up-regulation of CCL20 was also found in mdr2−/− mice, which develop chronic cholangitis.20 Attraction of T and B cells was shown for the chemokines CCL1, CCL5, and CCL8. Release of CCL5 and CCL8 can further promote chemoattraction of eosinophil granulocytes.21, 22 Eosinophil infiltrates were also seen in the liver of fra-1tg mice at age of 6 weeks (data not shown). Further, expression of CXCL5 in eosinophils was reported.23 CXCL5 stimulates the chemotaxis of neutrophils and enhances angiogenesis. In addition, large neutrophil infiltrates were observed in the portal fields of fra-1tg mice, accompanied by a strong increase in bile duct number. Taken together, up-regulation of a certain set of chemokines is likely triggering the influx of inflammatory cells in fra-1tg mice.

In our study we further delineated the phenotype of infiltrating cells by FACS analysis. Interestingly, activated CD4+ T cells expressing CD69 are the dominant phenotype of cells from the adaptive immune system infiltrating the livers of fra-1tg mice, whereas B-cells, NK cells, and NKT cells were drastically reduced to even lower levels than observed in wildtype mice. As these infiltrates also showed fra-1 expression, we aimed to distinguish whether fra-1 expression in cholangiocytes or inflammatory infiltrates is crucial for cholestatic liver disease and fibrosis in fra-1tg mice. Bone marrow chimeras with wildtype recipients and fra-1tg donor bone marrow showed that fra-1 expressed by hematopoietic cells is not sufficient to induce liver disease. Moreover, when fra-1tg mice were crossed with rag2−/− mice, which lack T and B cells, liver infiltrates were completely abolished. However, even in the absence of these inflammatory infiltrates fra-1tg mice developed bile duct proliferation and liver fibrosis, suggesting that lymphocytes may modulate but not initiate cholestatic liver disease and liver fibrosis in this model. This observation fits well with human autoimmune diseases associated with cholestatic hepatitis such as PBC. Although there is strong infiltration of immune cells in the liver in patients with PBC and autoantibodies are formed, immunosuppression does not ameliorate progression to endstage liver disease.24

In summary, we describe a novel model of progressive liver inflammation and liver fibrosis that might be valuable for studying pathogenic mechanisms and drug targets in liver fibrosis.

Acknowledgements

We thank Barbara Happich, Isabell Schmidt, and Cornelia Stoll (University of Erlangen-Nuremberg, Germany) and Eva Lederer (Department of Pathology, Medical University of Graz, Austria) for technical assistance. We thank Erwin Wagner (Institute for Molecular Pathology, Vienna, Austria) for providing the fra-1tg mice.

Ancillary