Potential conflict of interest: Nothing to report.
Growth hormone (GH) resistance and low serum levels of insulinlike growth factor 1 (IGF-1) are common features in human liver fibrosis and cirrhosis. Signal transducer and activator of transcription 5 (STAT5) controls several vital functions in the liver, including GH-mediated transcription of IGF-1. To investigate the role of STAT5 in liver fibrogenesis, we specifically deleted the Stat5a/b locus both in hepatocytes and cholangiocytes in the multidrug resistance gene 2 knockout (Mdr2−/−) mouse model of cholestasis. Double knockout mice develop an early and severe liver fibrosis phenotype, accompanied by perturbed expression of key regulators of bile acid homeostasis. Deletion of Stat5 resulted in GH resistance, and IGF-1 levels in serum were undetectable. We could observe reduced expression of important hepatoprotective genes, such as epidermal growth factor receptor (Egfr), hepatocyte nuclear factor 6 (Hnf6), prolactin receptor (Prlr), and leukemia inhibitory factor receptor (Lifr) as well as increased numbers of apoptotic hepatocytes. Conclusion: Our data suggest that loss of STAT5 sensitizes hepatocytes to bile acid–induced damage and apoptosis caused by disruption of GH-induced transcription of Igf-1 and down-regulation of hepatoprotective genes. These findings could contribute to the understanding of liver fibrosis and future treatment strategies for liver fibrosis. (HEPATOLOGY 2010.)
Liver fibrosis is a consequence of chronic liver damage and constitutes a major worldwide health care burden. Major causes of liver fibrosis are alcohol abuse, nonalcoholic fatty liver disease, infection by hepatitis B and hepatitis C viruses, metabolic disorders, and cholangiopathies.1, 2 The canonical process of liver fibrosis starts with hepatocyte damage, resulting in chronic inflammation and activation of hepatic stellate cells. Activated hepatic stellate cells undergo proliferation and transdifferentiate to myofibroblasts.3 Myofibroblasts express α-smooth muscle actin (αSMA) and secrete numerous profibrogenic cytokines, such as transforming growth factor beta (TGFβ) and platelet-derived growth factor beta (PDGFβ). Ultimately, they synthesize and deposit fibrillar collagen, which results in an excess of extracellular matrix.4
Signal transducer and activator of transcription 5 (STAT5) is activated in the liver in response to growth hormone (GH) and controls the transcription of the insulinlike growth factor 1 (Igf-1) gene.5 Mice lacking STAT5 in hepatocytes have low levels of circulating IGF-1 and develop GH resistance,6 two conditions that are often associated with human liver cirrhosis.7, 8 Furthermore, treatment with IGF-1 has been shown to improve liver fibrosis in experimental models.9, 10
To investigate the role of the GH-STAT5-IGF-1 axis in liver fibrosis, we used the Mdr2−/− (multidrug resistance gene 2 knockout) mouse model of cholestatic liver disease.11 Tissue-specific deletion of STAT5 in hepatocytes and cholangiocytes in these mice (Mdr2−/−; STAT5Δhep) resulted in severe liver fibrosis at 3 to 8 weeks of age, GH resistance, and lack of IGF-1 in the serum. Analysis of Mdr2−/−; STAT5Δhep mice revealed a significant down-regulation of hepatoprotective factors, perturbed bile acid homeostasis, and increased apoptosis of hepatocytes. Our findings suggest that functional GH–STAT5–IGF-1 signaling protects mice from cholestasis-induced liver fibrosis.
Mice carrying hepatocyte-specific and cholangiocyte-specific deletion of Stat5 (STAT5;AlfpCre, referred to as STAT5Δhep) and Mdr2−/− mice have been described elsewhere.11, 12 We obtained double-knockout mice (Mdr2−/−;STAT5Δhep) by respective intercrossing and maintained animals on a mixed genetic background (129Sv/Balb/c/C57BL/6). Mdr2+/− mice are disease-free and were used as controls. For experimental procedures we used 8-week-old male mice. Mice were kept at the Decentralized Biomedical Facilities, Medical University of Vienna, under standardized conditions, and all animal experiments were carried out according to an ethical animal license protocol and contract approved by the Medical University of Vienna and Austrian Federal Ministry of Education, Arts and Culture authorities.
Quantitative Reverse Transcription Polymerase Chain Reaction and Microarray Analysis.
For quantitative reverse transcription polymerase chain reaction (qRT-PCR), 1 μg total RNA from male control and mutant animals (n = 6/genotype) was reverse-transcribed into complementary DNA, using a commercially available kit (Applied Biosystems). Quantitative RT-PCR was performed using the SYBR green method. Messenger RNA (mRNA) levels were normalized for murine glyceraldehyde 3-phosphate dehydrogenase, and relative mRNA abundance was calculated using the ΔCt (threshold concentration) method. Primer sequences are listed in Supporting Table 1.
For microarray analysis, total RNA from four male control or mutant mice was pooled and hybridized to GeneChip Mouse Gene 1.0 ST arrays (Affymetrix, Santa Clara, CA). We analyzed for coordinate changes in functional gene sets using gene set enrichment analysis software (GSEA)13 and visualized significant x-fold up-regulation or down-regulation with a Microsoft Excel macro.
Histology and Immunohistochemistry.
Formalin-fixed, paraffin-embedded liver sections were stained with hematoxylin-eosin (HE), chromotrope aniline blue (CAB) and periodic acid-Schiff (PAS) using standard procedures. We performed immunohistochemistry, using the following antibodies: STAT5 (sc-835; Santa Cruz Biotechnology, Santa Cruz, CA), cleaved caspase 3 (#9661S; Cell Signaling Technology, Danvers, MA), αSMA (Clone 1A4; Sigma, Steinheim, Germany), Ki67 (NCL-Ki67-P; Novocastra Laboratories, Newcastle, UK) and cytokeratin 19 (monoclonal rat-anti-Troma-III antibody developed by Rolf Kemler). Ki67-positive hepatocytes were quantified using HistoQuest software (TissueGnostics GmbH, Vienna, Austria; www.tissuegnostics.com).
Western Blot Analysis.
We probed blots with antibodies against cytokeratin 19 (monoclonal rat-anti-Troma-III antibody), beta-Actin (#A5316; Sigma, Steinheim, Germany), HSC70 (sc-7298; Santa Cruz Biotechnology, Santa Cruz, CA), cleaved caspase 3 (#9661S), pY-STAT5 (#71-6900; Invitrogen, Carlsbad, CA), STAT5b (polyclonal rabbit-anti-mouse, raised against epitope aa775-788), STAT3 (#06-596; BD Biosciences, Franklin Lakes, NJ), pY-STAT3 (#9131), TGF-β (#3709; Cell Signaling, Beverly, MA), sodium taurocholate cotransport polypeptide, and organic anion transporting polypeptide 1 (both gifts from Bruno Stieger, Switzerland).
Serum IGF-1 and GH levels were determined using enzyme-linked immunosorbent assay (ELISA) kits (IBL, Hamburg, Germany, and Millipore, Billerica, MA, respectively). Serum levels of alanine aminotransferase (ALT), alkaline phosphatase (ALP), and bilirubin were measured using Reflotron-based test strips (Roche Applied Science, Mannheim, Germany). For quantification of serum bile acid content, we used a commercial 3α-HO-steroid-Dehydrogenase assay (Ecoline S+; DiaSys, Holzheim, Germany).
Caspase 3 Activity Measurement.
For quantification of apoptosis, liver protein lysates were incubated with a fluorogenic Caspase 3 substrate (Ac-DEVD-AFC; Alexis Biochemicals, San Diego, CA). Caspase 3 activity was determined by fluorometry.
Hepatic hydroxyproline content was assessed as described.14
Primary Hepatocyte Treatment.
Primary hepatocytes were isolated by standard liver perfusion and cultivated as described.15 After 24 hours in culture, we treated the hepatocytes for 4 hours with 100 μM DCA or 100 μM glycochenodeoxycholic acid in the absence or presence of recombinant murine IGF-1 (50 μg/mL).
Chromatin immunoprecipitation analysis was performed with livers from untreated control mice and GH-injected control and Stat5Δhep mice (2 μg/g body weight, 30 minutes, n = 4/group) using a STAT5 antibody (#sc-835x; Santa Cruz Biotechnology, Santa Cruz, CA) (see Supporting Methods).
Results are presented as mean ± standard error of the mean. Statistical analyses were performed by analysis of variance followed by the Dunn's or Tukey's post-hoc tests using SigmaStat software (San José, CA). Data were considered statistically significant: *P < 0.05; **P < 0.01; ***P < 0.001
Phenotypical Effects of Hepatic STAT5 Deletion in Mdr2−/− Knockout Mice.
To test whether STAT5 plays a role in cholestasis-induced liver fibrosis, we conditionally deleted Stat5 in hepatocytes and cholangiocytes (STAT5Δhep)16, 17 in the Mdr2−/− mouse model of cholestatic liver disease.11 Deletion of Stat5 in hepatocytes and cholangiocytes was confirmed by immunohistochemical staining and western blot analysis (Supporting Fig. 1A, B). Double-knockout mice (Mdr2−/−;STAT5Δhep) were smaller, appeared sick as early as 3 weeks of age, and showed significantly reduced survival compared with controls (Fig. 1A, B). Mdr2−/−;STAT5Δhep livers were enlarged, and the liver mass to body mass ratio of these animals was significantly increased (Fig. 1C). Mdr2−/− livers showed increased STAT5 phosphorylation (Supporting Fig. 1B), suggesting a compensatory role of activated STAT5 in response to cholestatic liver injury. Levels of phosphorylated STAT3 were unchanged in control, Mdr2−/−, and Stat5Δhep livers but up-regulated in Mdr2−/−;Stat5Δhep livers (Supporting Fig. 1B). STAT5-deficient livers (STAT5Δhep and Mdr2−/−;STAT5Δhep) had a yellow appearance because of elevated lipid levels as described.6 (Supporting Fig. 2A). In addition to macroscopic liver abnormalities, serum ALP and ALT were elevated in Mdr2−/−;STAT5Δhep mice (Fig. 1D). Bilirubin and bile acid serum levels were also significantly increased (Fig. 1D). Messenger RNA and protein levels of sodium taurocholate cotransport polypeptide and organic anion transporting polypeptide 1, two basolateral bile acid importers regulated by STAT5,18, 19 were strongly decreased in mice lacking hepatic STAT5 (Supporting Fig. 2B, C). Analysis of other components of bile acid homeostasis showed decreased mRNA levels of farnesoid X receptor (FXR), a major bile acid sensor, and small heterodimer partner-1 (SHP1), a negative regulator of bile acid biosynthesis,20 and unchanged levels of the cytochrome p450 family member 7A1 (Supporting Fig. 2C). Taken together, Mdr2−/−;STAT5Δhep mice display liver damage and perturbed regulation of bile acid homeostasis.
Deletion of Hepatic STAT5 Exacerbates Liver Fibrosis in Mdr2−/− Mice.
Histopathological examination revealed a disturbed liver architecture in Mdr2−/−;STAT5Δhep mice in contrast to control, Mdr2−/− and STAT5Δhep littermates (Fig. 2A). CAB staining demonstrated prominent periportal and pericellular collagen depositions in Mdr2−/−;STAT5Δhep livers (Fig. 2B). Heterogeneous glycogen staining in these liver sections by PAS showed nonfunctional hepatocyte areas only in compound animals (Fig. 2C). Increased bile duct proliferation in compound mice could be observed by immunohistochemistry and confirmed by western blot against cytokeratin 19 (Fig. 2D, F). Quantitative RT-PCR analysis showed an up-regulation of fibrogenesis hallmark genes collagen (Col)1 and 3, α-Sma, Pdgfβ, TgfβR1, PdgfβR1, Timp1 and 2, Mmp 2, 3, 13-15 in Mdr2−/−;STAT5Δhep livers (Fig. 2E and Supporting Fig. 2D). Tgfβ was slightly upregulated on the mRNA level in Mdr2−/− and Mdr2−/−;Stat5Δhep livers, whereas protein levels were unchanged in the four genotypes (Supporting Fig. 2D, 2E). Hydroxyproline content was significantly increased in Mdr2−/−; STAT5Δhep livers (Supporting Fig. 3). Collectively, these data suggest that deletion of hepatic STAT5 accelerates and aggravates the development of liver fibrosis in Mdr2−/− mice.
Disruption of GH-STAT5-IGF-1 Signaling in STAT5Δhep and Mdr2−/−STAT5Δhep Livers.
To gain better understanding of the molecular mechanisms underlying the disease, we applied global gene expression analysis in 8-week-old male mice (Supporting Table 3). GSEA revealed perturbed GH-IGF-1 signaling in STAT5Δhep and Mdr2−/−; STAT5Δhep livers. Most notably, expression of Igf-1 was dramatically reduced (Fig. 3A). Insulin-like growth factor binding protein acid labile subunit (Als), which is required for the formation of higher-order complexes between IGF-1 and its binding proteins21 was down-regulated as well. We confirmed reduced Igf-1 and Als mRNA transcripts by qRT-PCR (Fig. 3B). Serum levels of IGF-1 were slightly increased in Mdr2−/− mice, greatly reduced in STAT5Δhep mice, and below detection levels in Mdr2−/−; STAT5Δhep mice. In contrast, GH levels were strongly elevated in STAT5Δhep and Mdr2−/−;STAT5Δhep animals, as shown by ELISA (Fig. 3C). These results demonstrate that STAT5Δhep and Mdr2−/−;STAT5Δhep mice develop GH resistance, which is manifested by low systemic IGF-1 levels and high serum levels of GH.
Down-regulation of Hepatoprotective Genes and Increased Apoptosis in Mdr2−/−; STAT5Δhep Mice.
Egfr, prolactin receptor (Prlr), the STAT5 target gene hepatocyte nuclear factor 6 (Hnf6),22 and leukemia inhibitory factor receptor (Lifr) have been shown to protect hepatocytes from experimentally induced liver damage.23–26 These genes were strongly down-regulated in STAT5Δhep and Mdr2−/−;STAT5Δhep mice, as shown by gene expression analysis (Fig. 4A). Reduced mRNA levels of Egfr, Hnf6, Lifr, and Prlr were confirmed by qRT-PCR (Fig. 4B). In addition to the Igf-1 and Hnf6 genes, we could determine significant STAT5 binding to the promoter regions of Prlr and Lifr by using chromatin immunoprecipitation analysis and weak binding to the Egfr promoter (Fig. 4C). Absence of EGFR protein was further confirmed by immunohistochemistry and western blot analysis (Supporting Fig. 5A,B). These data show that, besides Igf-1, transcription of additional hepatoprotective factors is directly and indirectly impaired on Stat5 deletion.
Because Mdr2−/−;STAT5Δhep mice show severe liver degeneration, we investigated whether hepatocyte proliferation and apoptosis were affected. Hepatocyte proliferation was not diminished in Mdr2−/−;STAT5Δhep mice but increased (Fig. 5A; Supporting Fig. 4). However, GSEA revealed that pro-apoptotic genes were up-regulated. Several members of the caspase (Casp) family of proteases, namely Casp12, Casp4, Casp1, and Casp6, as well as tumor necrosis factor–related apoptosis-inducing ligand and Fas ligand, were increased on the mRNA level (Supporting Fig. 6). Moreover, we detected Casp3 cleavage product by Western blot analysis (Fig. 5B) and significantly higher Casp3 activity in liver protein lysates (Fig. 5C) of Mdr2−/−;STAT5Δhep mice. Immunohistochemical staining with an antibody against cleaved Casp3 confirmed the presence of apoptotic hepatocytes in Mdr2−/−; STAT5Δhep livers (Fig. 5D). Furthermore, when treated with bile acids, primary hepatocytes from Stat5Δhep but not from control mice showed a significant increase in Casp3 activity. IGF-1 treatment of Stat5Δhep hepatocytes reversed Casp3 activation to normal levels (Fig. 5E).
These data suggest that STAT5-deficient hepatocytes are prone to bile-acid–induced apoptosis because of lack of IGF-1 and reduced expression of hepatoprotective genes.
Two common features observed in human fibrosis and cirrhosis are low levels of circulating IGF-18 and GH resistance.7 GH signaling through STAT5 regulates a variety of important metabolic and biological functions in the liver,6, 12, 27 including transcription of Igf-1.5, 28 Treatment of patients with cirrhosis with exogenous GH does not significantly increase basal IGF-1 serum levels,29, 30 suggesting an involvement of aberrant signal transduction via STAT5 in liver fibrogenesis. Lack of STAT5 was recently shown to contribute to chemically induced liver fibrosis by up-regulation of TGF-β and activation of STAT3.31 Although STAT3 activation was increased in Mdr2−/−;Stat5Δhep livers, we did not observe significant changes of TGF-β protein expression between the four genotypes. Additionally, genetic deletion of Stat3 in livers of Mdr2−/− mice promotes liver fibrosis (Mair et al., manuscript submitted for publication). These different studies suggest that molecular events contributing to liver fibrosis on loss of STAT5 could depend on the use of a chemical or a cholestatic model of liver fibrosis.
In this work we demonstrate that loss of hepatic Stat5 disrupts the GH-STAT5-IGF-1 axis and severely aggravates liver fibrosis in a genetic mouse model of cholestasis. Although commonly used genetic models of liver fibrosis show only limited or transient fibrosis,11, 32–34 Mdr2−/−;STAT5Δhep mice are characterized by intense collagen deposition between liver lobules (bridging fibrosis) and in pericellular spaces, accompanied by high serum levels of bile acids and bilirubin. Moreover, fibrosis is manifest as early as at weaning age, and mice have a significantly shortened live span. Several factors may contribute to the pathogenesis of this phenotype: (1) down-regulation of hepatoprotective genes, (2) impaired bile acid homeostasis, and (3) increased hepatocyte apoptosis.
First, global gene expression analysis has shown that important genes for proper function and protection of hepatocytes are down-regulated in diseased mice. Hnf6, Prlr, Lifr, and Egfr have all been implicated in the protection of hepatocytes from experimentally induced liver damage.23–26Igf-1 and Hnf6 are known to be controlled by STAT5.5, 22, 28 In addition, our experiments suggest that STAT5 regulates the transcription of Prlr and Lifr. Although we could not detect significant binding of STAT5 to two regions of the Egfr promoter, Egfr expression was shown to be down-regulated in growth hormone receptor–deficient mice.35 The lack of these factors could render Mdr2−/−;STAT5Δhep livers more sensitive to bile acid toxicity, cholestatic liver inflammation, and liver fibrosis.
Second, FXR and small heterodimer partner protect hepatocytes from cholestatic injury by transcriptional repression of cytochrome p450 family member 7A,36, 37 which constitutes the rate-limiting enzyme in bile acid synthesis.38 Transcript levels of Fxr and Shp1 were significantly decreased in STAT5Δhep and Mdr2−/−;STAT5Δhep livers. Consequently, Mdr2−/−;STAT5Δhep mice could not counteract high bile acid levels by down-regulation of CYP7A1 (Supporting Fig. 2C) Additionally, the basolateral bile acid transporters sodium taurocholate cotransport polypeptide and organic anion transporting polypeptide 1, both transcriptional targets of STAT5,18, 19 were down-regulated not only in Mdr2−/−;STAT5Δhep but also in STAT5Δhep livers. This further emphasizes the role of STAT5 in the regulation of hepatic bile acid transport.
Third, the main consequences of hepatic Stat5 deletion in Mdr2−/− livers result from disrupted GH-mediated Igf-1 transcription. IGF-1 improves chemically induced liver fibrosis in rats9 and prevents apoptosis in a number of cell types, such as fibroblasts expressing c-myc39 or heart muscle cells.40 Blocking IGF-1 signal transduction in primary rat hepatocyte cultures using an IGF-1R blocking antibody or small interfering RNA for Igf-1 enhances apoptosis induced by glycochenodeoxycholate or tauro-CDC.41 Although necrosis is the predominant form of cell death in rodent cholestasis models,42 Mdr2−/−;STAT5Δhep hepatocytes seem to become increasingly apoptotic, most likely because of the lack of hepatoprotective factors, especially IGF-1. In primary mouse hepatocytes, inhibition of the IGF-1 signal results in reduced growth and survival43 and increased apoptosis because of bile acid toxicity.44 Accordingly, we demonstrated that Mdr2−/−;STAT5Δhep hepatocytes show a significantly increased apoptotic response under cholestatic conditions in vivo. In addition, we could show that primary hepatocytes lacking STAT5 are more sensitive to bile acid–induced apoptosis in vitro. Furthermore, supplementation of IGF-1 prevented these cells from bile acid–induced apoptosis.
Taken together, our results further emphasize the importance of GH–STAT5–IGF-1 signal transduction in the prevention of degenerative liver diseases.
The authors thank Dr. Ilan Stein for providing Mdr2−/− mice and Lothar Hennighausen for Stat5flox mice; Michaela Schlederer, Deeba Khan, and Vukoslav Komnenovic for technical assistance; Michael Freissmuth, Peter Fickert, and Martin Holzenberger for helpful discussions; and Dagmar Stoiber for critical reading of the manuscript.