Profibrogenic transforming growth factor-β/activin receptor–like kinase 5 signaling via connective tissue growth factor expression in hepatocytes


  • Hong-Lei Weng,

    1. Molecular Alcohol Research in Gastroenterology, Department of Medicine II, Faculty of Medicine at Mannheim, University of Heidelberg, Mannheim, Germany
    2. Institute of Infectious Diseases, First Affiliated Hospital, School of Medicine, Zhejiang University, China
    Search for more papers by this author
  • Loredana Ciuclan,

    1. Molecular Alcohol Research in Gastroenterology, Department of Medicine II, Faculty of Medicine at Mannheim, University of Heidelberg, Mannheim, Germany
    Search for more papers by this author
  • Yan Liu,

    1. Molecular Alcohol Research in Gastroenterology, Department of Medicine II, Faculty of Medicine at Mannheim, University of Heidelberg, Mannheim, Germany
    Search for more papers by this author
  • Jafar Hamzavi,

    1. Molecular Alcohol Research in Gastroenterology, Department of Medicine II, Faculty of Medicine at Mannheim, University of Heidelberg, Mannheim, Germany
    Search for more papers by this author
  • Patricio Godoy,

    1. Molecular Alcohol Research in Gastroenterology, Department of Medicine II, Faculty of Medicine at Mannheim, University of Heidelberg, Mannheim, Germany
    Search for more papers by this author
  • Haristi Gaitantzi,

    1. Molecular Alcohol Research in Gastroenterology, Department of Medicine II, Faculty of Medicine at Mannheim, University of Heidelberg, Mannheim, Germany
    Search for more papers by this author
  • Stefan Kanzler,

    1. Medical Clinic, University Hospital Mainz, Mainz, Germany
    Search for more papers by this author
  • Rainer Heuchel,

    1. Ludwig Institute for Cancer Research, Uppsala, Sweden
    Search for more papers by this author
  • Uwe Ueberham,

    1. Paul-Flechsig Institute for Brain Research, Department of Neuroanatomy, University of Leipzig, Germany
    Search for more papers by this author
  • Rolf Gebhardt,

    1. Institute for Biochemistry, University of Leipzig, Germany
    Search for more papers by this author
  • Katja Breitkopf,

    1. Molecular Alcohol Research in Gastroenterology, Department of Medicine II, Faculty of Medicine at Mannheim, University of Heidelberg, Mannheim, Germany
    Search for more papers by this author
  • Steven Dooley

    Corresponding author
    1. Molecular Alcohol Research in Gastroenterology, Department of Medicine II, Faculty of Medicine at Mannheim, University of Heidelberg, Mannheim, Germany
    • Molecular Alcohol Research in Gastroenterology, Department of Medicine II, Faculty of Medicine at Mannheim, University of Heidelberg, Theodor-Kutzer Ufer 1-3, 68135 Mannheim, Germany
    Search for more papers by this author
    • fax: 0049-621-383-1467

  • Potential conflict of interest: Nothing to report.


Connective tissue growth factor (CTGF) is important for transforming growth factor-β (TGF-β)–induced liver fibrogenesis. Hepatic stellate cells have been recognized as its major cellular source in the liver. Here we demonstrate the induction of CTGF expression in hepatocytes of damaged livers and identify a molecular mechanism responsible for it. CTGF expression was found by immunohistochemistry in bile duct epithelial cells, hepatic stellate cells, and hepatocytes in fibrotic liver tissue from patients with chronic hepatitis B infection. Similarly, CTGF expression was induced in hepatocytes of carbon tetrachloride–treated mice. CTGF expression and secretion were detected spontaneously in a medium of hepatocytes after 3 days of culture, which was enhanced by stimulation with TGF-β. TGF-β–induced CTGF expression was mediated through the activin receptor–like kinase 5 (ALK5)/Smad3 pathway, whereas activin receptor–like kinase 1 activation antagonized this effect. CTGF expression in the liver tissue of TGF-β transgenic mice correlated with serum TGF-β levels. Smad7 overexpression in cultured hepatocytes abrogated TGF-β–dependent and intrinsic CTGF expression, indicating that TGF-β signaling was required. In line with these data, hepatocyte-specific transgenic Smad7 reduced CTGF expression in carbon tetrachloride–treated animals, whereas in Smad7 knockout mice, it was enhanced. Furthermore, an interferon gamma treatment of patients with chronic hepatitis B virus infection induced Smad7 expression in hepatocytes, leading to decreased CTGF expression and fibrogenesis. Conclusion: Our data provide evidence for the profibrogenic activity of TGF-β directed to hepatocytes and mediated via the up-regulation of CTGF. We identify ALK5-dependent Smad3 signaling as the responsible pathway inducing CTGF expression, which can be hindered by an activated activin receptor–like kinase 1 pathway and completely inhibited by TGF-β antagonist Smad7. (HEPATOLOGY 2007.)

Upon chronic damage from, for example, a hepatitis virus infection or alcohol abuse, the wound healing response of the liver derails, and excessive deposition of extracellular matrix (ECM) leads to organ fibrosis with increasing loss of liver function.1 A wound healing process is initiated by a complex network of cytokines, including profibrotic proteins transforming growth factor-β (TGF-β) and connective tissue growth factor (CTGF).2

TGF-β signaling is required for the activation of hepatic stellate cells (HSCs) and is therefore sometimes termed the master cytokine of liver fibrogenesis.1, 3 TGF-β signals through transmembrane receptors that stimulate cytoplasmic Smad proteins, which modulate the transcription of target genes, including those encoding ECM, such as procollagen-I and procollagen-III.4 By functional criteria, Smad proteins are subdivided into 3 classes: receptor Smads (R-Smads), common mediator Smads, and inhibitory Smads. Smad7, an inhibitory Smad, blocks R-Smad phosphorylation and subsequent downstream events by forming a stable complex with the activated TGF-β type I receptor.5

CTGF was discovered more than a decade ago as a mitogen secreted from human endothelial cells.6 In fibroblasts, CTGF is induced by TGF-β and thus is considered a downstream mediator of some TGF-β effects.7, 8 The promoter of CTGF contains a functional SMAD binding site, and TGF-β induction of CTGF in dermal fibroblasts is R-Smad–dependent.9 CTGF is overexpressed in fibrotic lesions of diverse organs and tissue types, and the degree of overexpression correlates with the severity of disease.8, 10–12 The induction of liver-specific CTGF expression by TGF-β has been assigned to HSCs, leading to increased migration, proliferation, and collagen expression of these cells.13 Furthermore, silencing CTGF expression with CTGF small interfering RNA (siRNA) delivered into the portal vein of rat livers prevented carbon tetrachloride (CCl4)–induced fibrosis by inhibiting TGF-β induction and HSC activation.14 A role of hepatocytes for liver damage–dependent CTGF expression has not been identified so far.

Inhibitory Smad7 is very efficient for blunting TGF-β effects in general and has been previously used by us and others to intervene fibrogenesis in chronic diseases of the liver, kidneys, lungs, and skin.15–18 Interferon gamma (IFN-γ) is also recognized as a cytokine eliciting effects antagonistic to TGF-β, and antifibrotic activity has been reported in a variety of organs.19, 20 The incubation of human proximal tubular epithelial cells with IFN-γ inhibited TGF-β–mediated α-smooth muscle actin expression and blunted TGF-β–induced fibronectin and plasminogen activator inhibitor-1 (PAI-1) expression.21 Adeno-associated virus–mediated overexpression of human IFN-γ inhibited the activation of isolated HSCs in CCl4-treated rats.22 Our own studies indicate that IFN-γ decreases TGF-β–dependent activation of R-Smad signaling by inducing Smad7 expression in quiescent and activated HSCs.23 In line with this, IFN-γ displayed a therapeutic effect in patients with chronic hepatitis B virus (HBV)24 and hepatitis C virus infections.25

In this study, we show for the first time that in a damaged liver, TGF-β provides profibrogenic signals in hepatocytes via activin receptor–like kinase 5 (ALK5)/R-Smad–dependent induction of CTGF expression, a mechanism appearing similarly in animal models and human patients. Furthermore, we show that the abrogation of TGF-β signaling by Smad7 or IFN-γ decreases the expression of profibrogenic CTGF.


ALK1, activin receptor–like kinase 1; ALK5, activin receptor–like kinase 5; CCl4, carbon tetrachloride; cDNA, complementary DNA; CRP, C-reactive protein; CTGF, connective tissue growth factor; ECM, extracellular matrix; HBV, hepatitis B virus; HSC, hepatic stellate cell; IFN-γ, interferon gamma; mRNA, messenger RNA; PAI-1, plasminogen activator inhibitor-1; PCR, polymerase chain reaction; R-Smad, receptor Smad; RT-PCR, reverse-transcription polymerase chain reaction; S7ΔE1, deletion of ExonI of Smad7; siRNA, small interfering RNA; S7tg, Smad7 transgenic mice; TGF-β, transforming growth factor-β.

Materials and Methods


Human recombinant TGF-β and human recombinant IFN-γ were from R&D (Minneapolis, MN). The antibodies were as follows: goat polyclonal anti-CTGF (sc-14939, Santa Cruz Biotechnology, Santa Cruz, CA), rabbit polyclonal phospho-Smad2 (Ser465/467) and phospho-Smad1/5/8 (3101 and 9511, Cell Signaling Technology, Frankfurt, Germany), rabbit polyclonal Smad2 and Smad3 (51–1300 and 51–1500, Zymed Laboratory, Berlin, Germany), rabbit polyclonal Smad7 and albumin (24477 and 19196, Abcam, Ltd., Cambridge, United Kingdom), and mouse monoclonal β-actin (A5441, Sigma-Aldrich, Missouri). siRNA for Smad2 and Smad3 was from Qiagen (Hilden, Germany). Williams' medium E, insulin, and dexamethasone were obtained from Sigma. Dulbecco's modified Eagle's medium, penicillin/streptomycin, and L-glutamine were purchased from Cambrex (Verviers, Belgium). Fetal bovine serum and Lipofectamine 2000 were from Invitrogen (Karlsruhe, Germany).


Hepatitis B surface antigen–positive patients with biopsy-proven hepatic fibrosis (stages 2–4 according to the Scheuer criterion26) were treated with 50 μg (106 units) of recombinant human IFN-γ 1b (Shanghai Clonbiotech Co., Ltd., Shanghai, China) on a daily basis for 3 months and on alternate days for the subsequent 6 months. This clinical study was approved by the local ethics committee, and further details have been described.24 We selected 10 patients with a serious fibrosis degree (stage between 3 and 4) from the IFN-γ–treated group to measure CTGF, Smad7, and Smad3 expression in liver tissue before and after IFN-γ treatment.


Primary hepatocytes were isolated from livers of male C57/BL-6 mice (100–150 g) with collagenase perfusion as described.27 Cells were plated on collagen-coated 6-well plates at a density of 3 × 105 cells/well in Williams' medium E supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 1% penicillin/streptomycin, and 100 nM dexamethasone. The cells were incubated in 5% CO2 at 37°C to facilitate attachment, and the medium was changed after 4 hours with serum-free Williams' medium E supplemented with 2 mM L-glutamine, 1% penicillin/streptomycin, and 100 nM dexamethasone. After overnight incubation in 5% CO2 at 37°C, the medium was changed with serum-free Williams' medium E supplemented with 2 mM L-glutamine and 1% penicillin/streptomycin. Recombinant TGF-β1 and/or IFN-γ were added to the serum-free culture medium for the indicated times.

Transgenic and Knockout Mice.

The generation of double-transgenic mice with conditional liver-specific expression of TGF-β1 was described.28 Briefly, for the construction of the TGF-β1 expression vector, mutated porcine TGF-β1 minimal complementary DNA (cDNA; kindly provided by Dr. Nashreen Khalil, Manitoba Institute of Cell Biology, Winnipeg, Canada) and pBI-5 (CVU89934; GenBank at NCBI, Bethesda, MD) were used. Transgenic mouse lines were generated by pronuclear injection with standard techniques. Positive animals were identified by polymerase chain reaction (PCR) and Southern blotting.

Smad7 exon I–deleted mice were generated as previously described.29 Briefly, the Smad7 locus was isolated from a 129Sv genomic library.30 The targeting vector contained a fragment of the Smad7 promoter and 5′ untranslated leader sequence, followed by a PGKneobpA expression cassette replacing the translated part of exon 1 and the exon I/intron I boundary, a 1-kilobase HindIII-NotI genomic 3′ fragment, and a herpes simplex virus thymidine kinase expression cassette in pBluescript SK+. Homologous recombination events were screened by Southern blotting, and mice were genotyped. Heterozygous mutant mice were backcrossed 5 generations onto CD-1 mice obtained from Charles River Laboratories. Animals were used between 4 and 8 months of age.

A Flag-tagged mouse Smad7 cDNA fragment was generated by BamHI/XhoI restriction digestion of pcDNA3.1-Flag-Smad7 plasmid (kindly provided by P. ten Dijke, Leiden University, the Netherlands). Sticky ends were blunted and cloned into the SmaI restriction site of the 8.2-kilobase U2 plasmid, which comprises the C-reactive protein (CRP)–promoter sequences, the 5′-cap site, the CRP protein coding sequences from exon1 and exon2, the intron 1 sequences, and the 3′-polyadenylation signal, as described.31 Correct cloning was confirmed by DNA sequencing.

Animal experiments were performed in accordance with the European Council Directive of November 24, 1986 (86/609/EEC), and were approved by the local authorities. All efforts were made to minimize the number of animals and their suffering.

Chronic Liver Injury.

We studied liver fibrosis in CRP-Smad7 transgenic mice (n = 10), CD-1 Smad7 knockout mice (n = 10), and FVB wild-type mice (n = 10) after the intraperitoneal injection of CCl4 (0.5 mL/kg, diluted 1:4 in mineral oil) twice weekly. Sham-treated mice (n = 3 of each) received mineral oil. After 8 weeks, animals were killed 24 hours after the last CCl4 injection. Liver samples from several lobes were either fixed in buffered formalin or snap-frozen in liquid nitrogen.

Preparation of the Cell Lysates and Western Blot.

Lysate preparations (radio immunoprecipitation assay buffer; 1× Tris-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% sodium dodecyl sulfate) and western blot analyses were performed as described.32 Protein (20 μg) was separated by sodium dodecyl sulfate/4%–12% polyacrylamide gel electrophoresis (4%–12% Bis-Tris Gel, NuPAGE, Invitrogen) and transferred to nitrocellulose membranes (Pierce, Rockford, IL). Horseradish peroxidase–linked anti-goat (Santa Cruz; sc-2033) or anti-rabbit antibodies (DAKO Cytomation/P0217) were used as secondary antibodies. The membrane was developed with Supersignal Ultra (Pierce). Individual protein bands were quantified by densitometry with a Lumi-Imager (Roche) and normalized for β-actin.

Adenoviral Infections.

AdALK1siRNA, AdSmad2, AdSmad3, and AdSmad7 were kindly provided by C. Heldin (Ludwig Institute for Cancer Research, Uppsala, Sweden), and infections were performed as described.33 The infectivity was determined with the Rapid Titre Kit from BD Bioscience, and 50–100 infectious units per cell (50–100 moi) of the single virus clones were used. Generally, more than 90% of hepatocytes were infected. Before infection, cells were cultured for 2 hours in Williams' medium E with 5% fetal bovine serum. Then, the medium was changed to Williams' medium E without serum, and 16 hours later, hepatocytes were stimulated as indicated.

Knockdown of Activin Receptor–Like Kinase 1 (ALK1), Smad2, and Smad3 by siRNA.

siRNA interfering ALK1 was made through the cloning of GATCCCCCACGGCTCCCTCTATGACTTTCAAGAGAA-GTCATAGAGGGAGCCGTGTTTTTGGAAA and complementary oligonucleotides derived from mouse ALK1 in pSuper34 and then was used for adenovirus generation and knockdown experiments as described.35

siRNA knockdown of Smad2, Smad3, and Smad7 was performed with HP Genome Wide siRNA duplexes (Qiagen). Briefly, cells were seeded at a density of 50.000/cm2 in collagen-coated plates as described,27 allowed to attach for 4 hours, and then washed 3 times with a Hank's BSS buffer. Chemically synthesized siRNAs at a final concentration of 25 nM were transfected with Lipofectamine 2000 according to the manufacturer's instructions in Williams' E medium supplemented with 100 nM dexamethasone and 5 mM L-glutamine. The oligos used were siRNA-Smad2 SI00195216, siRNA-Smad3-1 SI00210259, siRNA-Smad3-4 SI00210280, and AllStars Negative Control siRNA 1027287. Twenty-four hours after transfection, the medium was changed, and the TGF-β stimulations proceeded. The knockdown of target genes was evaluated with western blots.

Real-Time Reverse-Transcription Polymerase Chain Reaction (RT-PCR) Analysis of the CTGF Expression.

The total RNA was purified from hepatocyte monolayer cultures with the High Pure RNA isolation kit (Roche Diagnostic GmbH, Mannheim, Germany) according to the manufacturer's protocol, and the concentration was measured spectrophotometrically. cDNA was synthesized from 1 μg of RNA with the Transcriptor First Strand cDNA synthesis kit (Roche Diagnostic). Quantitative real-time RT-PCR was performed on the ABI-Prism 7700 sequence detection system (Applied Biosystems) with the TaqMan universal PCR master mix No AmpErase UNG (part no. 4324018). The following gene expression assays were used: CTGF (part no. Mm00515790_g1) and peptidylprolyl isomerase A (housekeeping gene; part. no. Mm02342429_g1). All reagents were purchased from Applied Biosystems. Samples were run in triplicate under the following conditions: initial denaturation for 2 minutes at 50°C and for 10 minutes at 95°C followed by 40 cycles of 15 s at 95°C and 1 minute at 60°C. The levels of gene expression in each sample were determined with the comparative cycle threshold method.


Liver tissues were fixed in 4% neutralized formalin and embedded in paraffin. Four-micrometer sections were stained for 1 hour at room temperature. After they were washed, EnVision peroxidase (DAKO) was applied for 1 hour at room temperature. Sections were developed with diaminobenzidine for 5 minutes.

For semiquantitative analysis, from every specimen, 10 fields (magnification, ×200) were selected randomly before and after IFN-γ treatment. Positive cells were counted by a pathologist, who was blind for the sequence of treatment, under a light microscope. The evaluation criteria can be described as follows. The positive cell numbers were divided into 5 grades: (0) no positive cells, (1) less than 25% positive cells, (2) between 25% and 50% positive cells, (3) between 50% and 75% positive cells, and (4) more than 75% positive cells.

Statistical Analyses and Expression of Results.

The data presented are representative for at least 3 experiments. The results are expressed as the means ± the standard deviation, and a statistical analysis was carried out with the Student t test and Spearman's rank correlation coefficient for paired data; a P value less than 0.05 was considered to be significant.


CTGF Expression in Hepatocytes.

Paradis and colleagues36 reported that HSCs are the major cellular source of CTGF in rat and human livers during experimental liver fibrogenesis. Later, Sedlaczek et al.37 found that not only HSCs but also activated bile duct epithelial cells provide CTGF. Consistent with those previous publications, the immunohistochemical staining of livers from patients with a chronic HBV infection showed intense CTGF staining in bile duct epithelial cells and HSCs (Fig. 1). Remarkably, we in addition found strong CTGF-positive staining in hepatocytes of the investigated patients, whereas hepatocytes of a normal liver were negative.

Figure 1.

Expression of CTGF in liver cells: immunohistochemical staining for CTGF in bile duct epithelial cells, HSCs, and hepatocytes in liver tissue from a representative patient with chronic HBV infection and in hepatocytes of TGF-β transgenic mice versus wild-type mice.

In line with our data, CTGF was detectable in the medium of primary hepatocytes after 3 days in culture (Fig. 2B), and this indicated that CTGF is expressed and secreted in this cell type in vitro.

Figure 2.

TGF-β induces the expression of CTGF in hepatocytes. (A) CTGF mRNA expression was detected by real-time PCR in primary cultured hepatocytes with or without TGF-β stimulation (5 ng/mL) at different times (0.5, 2, 4, 8, and 24 hours). (B) CTGF protein expression was detected by a western blot in cell lysate and a medium of primary cultured hepatocytes at different times (0.5, 3, 6, and 24 hours) of stimulation. β-Actin was used as a loading control. (C) CTGF expression was detected in a medium of cultured hepatocytes after 1 day and 2 days of stimulation with TGF-β. (D) Real-time RT-PCR for CTGF mRNA expression using whole liver total RNA preparations was used to compare wild-type and TGF-β transgenic mice.

TGF-β Induces the Expression of CTGF in Hepatocytes.

TGF-β activity is strongly induced during chronic liver damage, and previous studies implied a link between TGF-β and CTGF during the activation of HSCs.2 We here show that TGF-β rapidly induces CTGF messenger RNA (mRNA) expression in hepatocytes with a maximal effect after 2 hours (Fig. 2A). CTGF protein levels continuously increased up to 24 hours after TGF-β administration (Fig. 2B). In accordance with this, CTGF secretion from hepatocyte cultures was enhanced and occurred at earlier times in the presence of TGF-β in comparison with untreated cells (Fig. 2C).

In agreement with the data for cultured mouse hepatocytes, TGF-β transgenic mice displayed CTGF expression in hepatocytes (Fig. 1), and the staining quantity strongly correlated with the serum TGF-β concentrations (P < 0.01; Table 1). In real-time RT-PCR, the fold change of CTGF in livers of TGF-β transgenic mice was 15.65 versus that of control mice (Fig. 2D; P < 0.05).

Table 1. Serum TGF-β Concentrations and Liver CTGF Expression in TGF-β Transgenic Mice
NumberSexSerum TGF-β Concentration (ng/mL)CTGF-Positive Hepatocytes
  1. The Spearman rank correlation coefficient between the serum TGF-β concentration and CTGF-positive hepatocytes in liver tissue is 0.605 (P < 0.01).


TGF-β Induces the Expression of CTGF in Hepatocytes via the ALK5 Pathway.

Dependent on the cell type, TGF-β mediates Smad signals via ALK1 or ALk5.35, 38 In HSCs, TGF-β induces the expression of the inhibitor of differentiation (Id1) through the ALK1 pathway but not the ALK5 pathway, whereas the induction of PAI-1 expression is mediated via ALK5.39 To determine the TGF-β–dependent signaling pathway that mediates the induction of CTGF in hepatocytes, we separately used SB431542 to block ALK5 or AdsiALK1 to knock down ALK1 expression prior to the TGF-β treatment. Alternatively, AdcaALK5 and AdcaALK1, leading to a high expression of constitutively active receptors, were used to enhance those pathways.

AdcaALK5 but not AdcaALK1 increased CTGF mRNA expression in hepatocytes, whereas the ALK5 inhibitor SB431542 was inhibitory for CTGF induction after 2 hours of treatment with TGF-β (Fig. 3A). siRNA-dependent knockdown of ALK1 (Supplementary Fig. 1) significantly knocked down TGF-β/ALK1–dependent Smad1 phosphorylation (Fig. 3B) but did not influence TGF-β–dependent CTGF mRNA expression. Interestingly, knocking down ALK1 expression enhanced culture-dependent CTGF expression in untreated hepatocytes (Fig. 3A), and this pointed to autocrine stimulation of hepatocytes via the ALK5 pathway and crosstalk between both receptors. Furthermore, knocking down ALK1 increased both the intrinsic and TGF-β–dependent phosphorylation of Smad2, confirming its negative role for ALK5 signaling. In line with this, knocking down ALK1 led to increased CTGF expression (Fig. 3C), whereas SB431542 was inhibitory for phospho-Smad2 and CTGF expression.

Figure 3.

TGF-β induces the expression of CTGF in hepatocytes through the ALK5 pathway. To inhibit ALK5, hepatocytes were pretreated with 5 μmol/mL SB431542, and to block ALK1, they were infected with recombinant adenoviruses encoding siRNAs for ALK1 before stimulation with TGF-β. AdLacZ was used as a negative control for virus infection. (A) Real-time RT-PCR CTGF mRNA analysis after 2 hours of TGF-β stimulation, (B) ALK1-dependent Smad1 phosphorylation after 2 hours of TGF-β treatment, and (C) CTGF protein expression after 24 hours of TGF-β stimulation. β-Actin was used as a loading control. Representative western blots are shown.

We conclude that TGF-β induces the expression of CTGF in hepatocytes through the ALK5 signaling pathway but not the ALK1 signaling pathway and that culture-dependent CTGF induction in hepatocytes results from autocrine TGF-β stimulation.

TGF-β–Induced Expression of CTGF in Hepatocytes Is Smad3-Dependent.

To further characterize the TGF-β signaling pathway leading to CTGF expression, mouse hepatocytes were infected with adenoviruses encoding Smad2 or Smad3 expression cassettes and stimulated for 2 hours with 5 ng/mL TGF-β 24 hours later. The overexpression of Smad3, but not Smad2, enhanced CTGF protein expression with or without the TGF-β treatment (Fig. 4A). Alternatively, siRNAs directed to Smad2 and Smad3 were transfected into hepatocytes to separately block Smad2/Smad3-dependent signaling (Fig. 4B). After 24 hours of TGF-β stimulation, CTGF protein expression was decreased by Smad3 knockdown but not Smad2 knockdown (Fig. 4C). Finally, we infected hepatocytes with AdcaALK5 and measured the CTGF expression with or without knockdown of Smad2 or Smad3. Constitutively active ALK5 significantly enhanced CTGF protein expression (Fig. 4D) in line with the RNA data (Fig. 3A). CaALK5-dependent CTGF expression was decreased by the knockdown of Smad3, whereas decreasing the Smad2 expression had only a minor effect, if any. (Fig. 4D). These results are consistent with a previous report on fibroblasts9 and indicate that Smad3 is required for the TGF-β–induced expression of CTGF in hepatocytes.

Figure 4.

TGF-β–induced expression of CTGF in hepatocytes is Smad3-dependent. (A) Primary cultured hepatocytes were infected with AdSmad2 or AdSmad3, and this led to an overexpression of Smad2/3. After an overnight culture, the hepatocytes were treated with 5 ng/mL TGF-β for 2 hours, and the CTGF protein expression was measured by a western blot. (B-D) 25 nM siRNAs for Smad2 and/or Smad3 were transfected into primary cultured hepatocytes 24 hours before TGF-β stimulation. Smad2, Smad3, and CTGF expression were examined by western blots, and β-actin was used as a loading control. (C) Scrambled oligonucleotides for Smad2 (lane 1) and Smad3 (lane 2) were used as siRNA knockdown controls. (D) caALK5 (adenoviral infection)-dependent CTGF expression was examined after the knockdown of Smad2 and/or Smad3.

It is noteworthy that Smad3 overexpression increased endogenous Smad2 expression (Fig. 4A), whereas knocking down Smad3 decreased Smad2 levels in hepatocytes (Fig. 4B); this indicated a synergistic effect, whose biological meaning remains to be elucidated.

Smad7 Inhibits TGF-β–Mediated CTGF Expression in Hepatocytes.

In a previous study,17 we blunted HSC activation and reduced fibrogenesis in bile duct ligated rats by the expression of the TGF-β antagonist Smad7. To test its impact on CTGF expression in hepatocytes, we infected primary cultured hepatocytes with AdSmad7 prior to the TGF-β treatment. Smad7 reduces TGF-β–induced mRNA and protein expression of CTGF in hepatocytes (Fig. 5A,B). To examine the Smad7 effect on CTGF expression in vivo, we investigated CTGF expression in CCl4-treated FVB mice, Smad7 transgenic mice (described in the Material and Methods section), and such with a disrupted Smad7 gene.40 The control FVB mice displayed very weak CTGF staining in hepatocytes (Fig. 5C). After 8 weeks of the CCl4 treatment, inflammation and fibrosis were obvious, and CTGF expression was induced, especially in and next to fibrotic lesions (Fig. 5C). CTGF-positive staining was found in different cell types, including hepatocytes. CTGF expression was enhanced in the livers and hepatocytes of mice bearing a disrupted Smad7 gene, whereas the CCl4-dependent induction of CTGF expression was abrogated in transgenic mice with Smad7 overexpression in hepatocytes (Fig. 5C). These data are summarized by quantitative immunohistochemistry in Fig. 5D and indicate that Smad7 expression in hepatocytes reduces CTGF expression in mice with CCl4-dependent chronic liver damage.

Figure 5.

Smad7 inhibits TGF-β–induced CTGF expression in hepatocytes. (A) Primary cultured hepatocytes were infected with AdSmad7 before TGF-β stimulation. After 24 hours, the total RNA was extracted, and the CTGF mRNA expression was measured by real-time PCR. AdLacZ was used as negative control. (B) CTGF protein expression was measured by a western blot with or without AdSmad7 infection as indicated. β-Actin was used as a loading control. (C) CTGF expression in liver tissue was detected by immunohistochemical staining in 8-week–CCl4-treated or untreated wild-type mice, as well as a S7ΔE1 and S7tg mice. (D) Semiquantitative analysis of CTGF expression in wild-type, S7ΔE1, and S7tg mice (n = 10).

IFN-γ Inhibits TGF-β–Induced CTGF Expression in Hepatocytes.

Recently, we showed that IFN-γ interfered with profibrogenic TGF-β signaling by blunting the activation of HSCs23 and decreasing liver damage and fibrosis in patients with chronic hepatitis B infection.24 Here we analyze its impact on CTGF expression in hepatocytes. Different doses of IFN-γ (5, 50, and 500 ng/mL) with or without TGF-β were examined, and a dose-dependent reduction of CTGF protein expression was found. In parallel, Smad7 was up-regulated by IFN-γ in hepatocytes (Fig. 6A). To provide further evidence for a direct link between IFN-γ, Smad7, and CTGF expression, Smad7 siRNA knockdown experiments were performed. Under these conditions, IFN-γ could not blunt TGF-β–induced CTGF expression (Fig. 6B), and this indicates that Smad7 is required and represents indeed the mediator of the CTGF inhibitory IFN-γ effect. We investigated CTGF expression in 10 patients with chronic HBV infection before and after 9 months of the IFN-γ treatment. Strong CTGF expression was found in most liver cells, including hepatocytes, in liver tissue with serious fibrosis and inflammation (Fig. 7A), whereas CTGF expression in hepatocytes was significantly reduced after 9 months of the IFN-γ treatment, as determined by semiquantitative immunohistochemistry (Fig. 7B).

Figure 6.

IFN-γ inhibits TGF-β–induced CTGF expression in hepatocytes. (A) CTGF and Smad7 expression were measured by a western blot in primary cultured hepatocytes after stimulation with 5 ng/mL TGF-β and/or different doses (5, 50, and 500 ng/mL) of IFN-γ. β-Actin was used as a loading control. (B) 50 nM siRNAs for Smad7 were transfected into primary cultured hepatocytes 12 hours before TGF-β and IFN-γ stimulation. CTGF protein expression was detected by a western blot, and β-actin was used as a loading control.

Figure 7.

IFN-γ induces Smad7 and inhibits Smad3 expression in hepatocytes from patients with chronic HBV infection. (A) CTGF, Smad7, and Smad3 expression in biopsied liver tissues from 1 representative patient with chronic HBV infection before and after 9 months of IFN-γ treatment. (B) Semiquantitative analysis of CTGF, Smad7, and Smad3 expression from 10 patients with chronic HBV before and after 9 months of IFN-γ treatment. The following semiquantitative system was used to analyze CTGF expression:

  • IPositive cell numbers (grades: 0-4): 0, no positive cells; 1, less than 25% positive cells; 2, between 25% and 50% positive cells; 3, between 50% and 75% positive cells; and 4, more than 75% positive cells.
  • IIIntensity of positive staining (grades: 1-3): 1, positive staining is weak, usually yellow; 2, positive staining is strong, usually brown; and 3, positive staining is very strong, usually deep brown to black.

The final score is the number of positive cells times the intensity. This is due to the fact that the major differences between the untreated and IFN-γ–treated patients were related to the staining intensity instead of the positive cell numbers.

In parallel with CTGF, Smad3 expression was overrepresented in untreated HBV patients, whereas the IFN-γ treatment resulted in decreased Smad3. In accordance with the hepatocyte cell culture data, Smad7 expression was induced in hepatocytes of these patients by the IFN-γ treatment (Fig. 7A). A morphometric analysis based on immunohistochemistry is presented as Fig. 7B.

Our data suggest that the hepatocyte-specific expression of CTGF during liver damage in either animal models or human patients is mediated by TGF-β via ALK5/Smad3 signaling and can be abrogated by the direct blocking of this pathway by Smad7 or by indirect blocking by IFN-γ treatment, the latter probably mediating its effect at least partially by up-regulating Smad7 in this cell type.


Excessive production, deposition, and contraction of ECM in fibroproliferative diseases are under intense investigation because they represent 1 of the largest groups of disorders. Currently, the only redress for patients with fibrosis is organ transplantation, which, however, is limited by an insufficient supply of organs, and patients often die while waiting to receive suitable organs. Thus, understanding the molecular mechanisms of fibrogenesis is required to determine new targets for efficient drug intervention.

In this report, we describe new findings regarding the interconnection between TGF-β and CTGF in liver fibrogenesis.

Upon liver damage of any etiology, TGF-β signal transduction is rapidly induced and assists in attracting neutrophils, macrophages, and fibroblasts, which in turn release more of this cytokine. Accordingly, TGF-β is consistently present in wound fluid throughout repair processes.1 In chronic disease, constitutive TGF-β activity leads to continuation of the fibrotic response and to disease progression.

Beside TGF-β, CTGF, a member of the CCN family of matricellular proteins, has been identified as a highly profibrogenic molecule. CTGF expression is induced in fibrotic lesions, including the liver,36, 37, 41, 42 and promotes fibroblast proliferation, matrix production, granulation tissue formation,6, 43, 44 cell adhesion and migration in a variety of cell types,45–47 and collagen matrix contraction in fibroblasts.43 Most of these actions represent features associated with liver fibrogenesis and HSC activation.

Positive CTGF immunostaining is found in most human liver biopsies with significant fibrosis. In situ hybridization showed CTGF mRNA expression in spindle cells in both fibrous septa and sinusoidal lining and was originally assigned to HSCs.36 In this cell type, CTGF mRNA was increased during activation, and a direct stimulating effect of recombinant CTGF on collagen gene transcription was demonstrated.48, 49 In a rat model of biliary fibrosis, it was found that proliferating bile duct epithelial cells are an additional source of CTGF in the liver.37 We show now that hepatocytes express and secrete significant amounts of CTGF. Furthermore, we show hepatocyte-specific CTGF expression in CCl4-damaged mouse livers and in patients with liver fibrosis from chronic HBV infection.

In CCl4-dependent rat liver fibrosis, siRNA injections (0.1 mg/kg) targeting CTGF reduced the expression of CTGF protein in the liver for 3 days.14 The repeated administration of CTGF siRNA for 4 or 6 weeks markedly attenuated fibrogenesis. The number of activated HSCs, serum procollagen type III, hepatic hydroxyproline, and liver fibrosis staging were significantly decreased.

CTGF mediates some of the effects of TGF-β on fibroblast proliferation, adhesion, and ECM production, including collagen and fibronectin,8, 47, 50–53 thus representing a downstream mediator of TGF-β signaling.

In HSCs, TGF-β stimulation induced CTGF mRNA and protein expression after 1 hour with a peak after 4 hours, and this was attributed to mRNA stabilization due to the inhibition of RNA-degrading enzymes.13

In chronically damaged livers of human patients or rodents, extensive TGF-β signaling can be observed in hepatocytes by nuclear staining of tissue sections with phospho-Smad antibodies (Yoshida et al.54 and our own unpublished results). We show here that treating hepatocytes with TGF-β induces CTGF expression, which peaks after 2 hours at the RNA level; this indicates a direct transcriptional effect. CTGF protein expression with subsequent secretion is induced at 3 hours, peaking after 24 hours. A dominant role of TGF-β for hepatocyte CTGF expression was further confirmed in the livers of transgenic mice expressing active TGF-β.

In most cell types, TGF-β signals via the TGF-β type I receptor TβRI, which is also known as ALK-5, and the downstream mediators R-Smad2/3. In endothelial cells, it has been demonstrated that TGF-β can also bind to and signal through the activation of ALK-135, 38, 55 via R-Smads1/5. We have recently shown that both receptors are used by TGF-β in HSCs39 and now similarly show the activation of both pathways in hepatocytes. Experiments with constitutively active mutants of both receptors, Smad overexpression, and siRNA knockdown suggest that ALK5-Smad2/3 signaling, but not ALK1-Smad1/5/8 signaling, is required for CTGF induction. This was strengthened by the fact that a compound inhibitor, SB431542, directed to ALK5, blunted TGF-β–dependent CTGF expression in hepatocytes. We also found that ALK1 activity antagonizes ALK5 signaling, and consequently, the knockdown of endogenous ALK1 increased basal and TGF-β–induced CTGF mRNA expression and simultaneously enhanced basal and TGF-β–mediated ALK5 signaling in general, as shown by increased phosphorylation of Smad2. Similar data have been reported previously for endothelial cells, in which TGF-β discriminates proangiogenic and antiangiogenic properties by activating ALK1 or ALK5, respectively. Although the ALK5 pathway led to the inhibition of cell migration and proliferation, the ALK1 pathway induced endothelial cell migration and proliferation, and this indicated a fine balance between ALK5 and ALK1 signaling.38 Furthermore, in this cell type, ALK5 kinase activity is required for full ALK1 activation, and after recruiting ALK5 into the receptor complex, ALK1 directly antagonizes ALK5/Smad signaling.35 Our findings correlate well with those of Goumans and colleagues,35 who showed that ALK1 directly antagonizes ALK5 signaling in endothelial cells. It has been delineated that the 2 pathways are fine-tuning the activation state of endothelium by mediating opposite cellular responses, such as the induction of cell migration and proliferation via ALK1 and the inhibition of these activities via ALK5.38 Although our results clearly demonstrate that basal ALK1 and ALK5 signaling occurs and up-regulation of CTGF by TGF-β is limited to some extent by ALK1, the precise function of the 2 parallel pathways in hepatocytes remains to be investigated.

Our experiments further display that Smad3, not Smad2, is the predominant mediator of CTGF expression downstream of ALK5 activation. This gives further support to the role of Smad3 in the fibrotic response, as it was reported previously that animals lacking Smad3 show accelerated wound healing, reduced granulation tissue formation, increased epithelialization, and reduced inflammation.56 Furthermore, elevated levels of activated nuclear Smad3 were documented in several models of fibroblast acquisition, including bleomycin-induced fibrosis, HSC activation, and the leading edge of scleroderma lesions.57–59 Interestingly, we found that Smad3 overexpression increased endogenous Smad2 expression, whereas knocking down Smad3 decreased Smad2 levels in hepatocytes. The detailed biological meaning remains to be elucidated in the future.

Even if there seems to be a predominant role for Smad3, we have to mention that we did not yet analyze and therefore cannot exclude the impact of Smad-independent TGF-β signaling on CTGF expression in hepatocytes, as it has been described to occur in mesangial cells.60 The question of how CTGF mediates its functions, especially during the pathobiochemistry of chronic liver disease, is still under investigation. At least some effects are provided through integrins, heparin sulfate–containing proteoglycans, and the low-density lipoprotein receptor–related protein.42, 46, 52, 61–63 However, identifying a specific CTGF receptor and determining how intracellular signals initiated by CTGF are mediated and if hepatocyte function is controlled via CTGF are subjects of current research.

There is some evidence that CTGF may enhance the ability of TGF-β to bind to its receptors at low TGF-β concentrations.64 The application of CTGF or TGF-β alone to mice causes a transient wound healing effect,12, 65 whereas simultaneous subcutaneous coinjection induces a sustained fibrotic response.66 In line with this, in mesangial cells, CTGF induces TGF-β–inducible early-response gene expression, a known repressor of Smad 7 transcription, thus abrogating negative feedback regulation and allowing continued activation of the TGF-β signaling pathway. As a result, expression of the TGF-β–responsive genes PAI-1 and collagen type III was maximally stimulated by TGF-β and CTGF versus TGF-β alone, whereas CTGF alone had no significant effect.67

Blunting TGF-β signaling by the ectopic expression of Smad7 abrogated intrinsic and TGF-β–stimulated CTGF expression in hepatocytes in culture, and this suggests that autocrine TGF-β signaling is responsible for up-regulated CTGF expression during culture. In vivo, after CCl4 intoxication, CTGF expression was strongly enhanced in livers of mice with a disrupted Smad7 gene in comparison with controls and was reduced in mice with hepatocyte-specific overexpression of a Smad7 transgene.

IFN-γ, a pleiotropic cytokine released by T cells and natural killer cells, plays fundamental roles in innate and acquired immune responses68 and occurs immediately after injury and suppresses collagen synthesis,69, 70 thus antagonizing fibrogenesis. In a fibrosarcoma-derived U4A cell line, it was shown that IFN-γ induced Smad7.71 Although this mechanism does not seem to be generally applicable to normal fibroblasts,69 we have recently shown that a similar TGF-β antagonistic route occurs in activated HSCs, in which IFN-γ/Jak/Stat signaling induced Smad7 expression, blunted profibrogenic TGF-β signaling, and reduced type I collagen expression.23 In line with this, the IFN-γ treatment of patients with liver fibrosis from chronic HBV infection displayed antifibrotic effects.24

In this report, we show that antifibrotic effects of IFN-γ, beside HSCs, are directed to hepatocytes because CTGF expression is reduced by IFN-γ treatment in primary cultured mouse and human hepatocytes and in patients with liver fibrosis from chronic HBV infection. We further show that IFN-γ induces Smad7 expression in cultured mouse hepatocytes and in those of chronic HBV patients, whereas Smad3 availability is abrogated in parallel. This points to TGF-β signaling as a primary target of IFN-γ effects directed to CTGF expression in hepatocytes, although from the present data a direct signaling of IFN-γ downstream mediators to the CTGF promoter is not excluded.

TGF-β has long been known to induce matrix synthesis and contraction by fibroblasts. Nevertheless, the precise contribution of this protein to fibrotic disease is still unclear. Although HSCs are highlighted as a master cell type, becoming activated and displaying a myofibroblast phenotype depending on TGF-β, the present findings highlighting TGF-β–dependent CTGF induction in hepatocytes suggest that this cell type actively contributes to fibrogenesis and represents a major cellular target for profibrogenic TGF-β effects (Fig. 8).

Figure 8.

Model for TGF-β–induced CTGF expression in hepatocytes and fibrogenesis. TGF-β induces CTGF expression in hepatocytes through the ALK5/Smad3 pathway, whereas ALK1 activation antagonizes this effect. IFN-γ strongly induces Smad7 overexpression in hepatocytes and thereby abrogates TGF-β–dependent CTGF expression. CTGF secretion by hepatocytes is significantly enhanced by TGF-β treatment, and the secreted CTGF represents a possibly potent profibrogenic mediator during chronic liver damage (for example, by generating myofibroblasts from HSCs).


We appreciate the excellent technical assistance from A. Müller and C. Stump. We are grateful to Dr. Peter ten Dijke and Dr. Carl Henrik Heldin for the diverse reagents of the TGF-β signaling pathway.