The etiology of liver damage imparts cytokines transforming growth factor β1 or interleukin-13 as driving forces in fibrogenesis


  • Potential conflict of interest: Nothing to report.


It is unknown whether transforming growth factor β1 (TGF-β1) signaling uniformly participates in fibrogenic chronic liver diseases, irrespective of the underlying origin, or if other cytokines such as interleukin (IL)-13 share in fibrogenesis (e.g., due to regulatory effects on type I pro-collagen expression). TGF-β1 signaling events were scored in 396 liver tissue samples from patients with diverse chronic liver diseases, including hepatitis B virus (HBV), hepatitis C virus (HCV), Schistosoma japonicum infection, and steatosis/steatohepatitis. Phospho-Smad2 staining correlated significantly with fibrotic stage in patients with HBV infection (n = 112, P < 0.001) and steatosis/steatohepatitis (n = 120, P < 0.01), but not in patients with HCV infection (n = 77, P > 0.05). In tissue with HBx protein expression, phospho-Smad2 was detectable, suggesting a functional link between viral protein expression and TGF-β1 signaling. For IL-13, immunostaining correlated with fibrotic stage in patients with HCV infection and steatosis/steatohepatitis. IL-13 protein was more abundant in liver tissue lysates from three HCV patients compared with controls, as were IL-13 serum levels in 68 patients with chronic HCV infection compared with 20 healthy volunteers (72.87 ± 26.38 versus 45.41 ± 3.73, P < 0.001). Immunohistochemistry results suggest that IL-13–mediated liver fibrogenesis may take place in the absence of phospho–signal transducer and activator of transcription protein 6 signaling. In a subgroup of patients with advanced liver fibrosis (stage ≥3), neither TGF-β nor IL-13 signaling was detectable. Conclusion: Depending on the cause of liver damage, a predominance of TGF-β or IL-13 signaling is found. TGF-β1 predominance is detected in HBV-related liver fibrogenesis and IL-13 predominance in chronic HCV infection. In some instances, the underlying fibrogenic mediator remains enigmatic. (HEPATOLOGY 2009.)

Liver fibrosis and its end stage, cirrhosis, represent the final common pathways of chronic liver diseases (CLDs).1 Cytokine-mediated hepatic stellate cell (HSC) activation is recognized as a common key event during liver fibrogenesis.2 After liver damage, parenchymal cells (hepatocytes) and mesenchymal cells (Kupffer cells, endothelial cells, pit cells, HSCs) release cytokines, which participate in HSC activation and extracellular matrix synthesis.3, 4 However, it is still unclear whether profibrogenic master cytokines exert their activities in all CLDs. Alternatively, divergent pathways may predominate in certain etiologies.

Transforming growth factor β1 (TGF-β1) is considered the key profibrogenic cytokine in liver disease,1, 5 with participation in most critical events during liver fibrogenesis. For example, TGF-β1 mediates HSC activation, hepatocyte apoptosis, and extracellular matrix formation and induces expression of other profibrogenic mediators such as tissue inhibitor of metalloproteinase-1 and connective tissue growth factor (CTGF).1, 5, 6 More recently, TGF-β1 was found to play a pivotal role in epithelial-to-mesenchymal transition of hepatocytes.7–9 Clinical studies revealed elevated TGF-β1 serum levels in patients with chronic hepatitis B virus (HBV)/hepatitis C virus (HCV) infections or alcoholic liver disease.10–12 Different signaling pathways have been described for TGF-β1, the most important being the canonical Smad pathway.13 Smad proteins are divided into three functional classes: receptor-regulated (Smad1, 2, 3, 5, and 8), common mediator (Co-Smad4), and inhibitor (Smad6 and 7).13 Blocking TGF-β1 production and Smad-dependent signaling proved to be successful therapeutic strategies in experimental liver fibrosis models.14

Although TGF-β1 is commonly regarded as a profibrogenic mediator in the liver, some studies display paradoxical or contradictory data. A clinical study reported that elevated TGF-β1 levels are associated with protection from liver fibrosis in patients infected with Schistosoma mansoni.15 Another study indicated that TGF-β1 was not biologically relevant in a S. japonicum infection–associated fibrosis mouse model.16 Thus, it remains to be answered whether there are exceptions to TGF-β's profibrogenic role.

Some studies have linked the cytokine interleukin (IL)-13 to fibroproliferative diseases of diverse organs.17–19 IL-13 is a Th2 cytokine that shares many functional activities with IL-4, given that both cytokines engage with the same IL-4 receptor α chain/signal transducer and activator of transcription protein 6 (STAT6) pathway.17–19 Recent studies of transgenic and knockout mice have demonstrated that IL-13, but not IL-4, participates in S. mansoni–induced liver fibrosis and tissue scar formation.17–20 In the lung, IL-13 may selectively stimulate and activate TGF-β1,21 indicating cooperative activities of both cytokines. On the other hand, a schistosomiasis-based liver fibrotic mouse model demonstrated that IL-13 may also exert its profibrogenic role via a TGF-β1–independent pathway.22

The present study was initiated to address the question of whether TGF-β expression/signaling differs in liver diseases, primarily those incited by schistosomiasis compared with other CLDs. Second, the role of IL-13 was elucidated in CLD patients with different etiologies. Our results suggest that both TGF-β1 and IL-13 are associated with fibrosing CLD; however, their impact differs depending on the cause of liver damage.


CLD, chronic liver disease; CTGF, connective tissue growth factor; HBV, hepatitis B virus; HCV, hepatitis C virus; HSC, hepatic stellate cell; IFN-γ, interferon γ; IL, interleukin; mRNA, messenger RNA; α-SMA, α-smooth muscle actin; STAT6, signal transducer and activator of transcription protein 6; TGF-β1, transforming growth factor β 1; VEGF, vascular endothelial growth factor.

Materials and Methods


Recombinant human TGF-β1 and IL-13 were obtained from R&D (Minneapolis, MN). Dulbecco's modified Eagle's medium, penicillin/streptomycin, and L-glutamine were purchased from Cambrex (Verviers, Belgium). Fetal bovine serum was obtained from Invitrogen (Karlsruhe, Germany). Leflunomide was obtained from Sigma-Aldrich (St. Louis, MO).

The following antibodies were used: polyclonal rabbit anti–phospho-Smad2 (ser465/467, 3101) and monoclonal rabbit anti–phospho-STAT6 (Tyr641, 9364) (Cell Signaling Technology, Frankfurt, Germany); polyclonal goat anti-CTGF (sc-14939), IL-13 (sc-1292), polyclonal rabbit anti–glyceraldehyde 3-phosphate dehydrogenase (sc-25778), vascular endothelial growth factor (sc-152), TGF-β1 (sc-146), Egr-1 (sc-110), and mouse anti–cytokeratin-8 (sc-8020) (Santa Cruz Biotechnology, Santa Cruz, CA); polyclonal rabbit anti-tubulin (Ab-11270) and monoclonal mouse anti-HBx (Ab-39716) (Abcam, London, UK); monoclonal mouse anti–collagen type I (C2456) and β-actin (A5441) (Sigma-Aldrich); α-smooth muscle actin (M0851) (Dako, Hamburg, Germany); and CD68 (Zymed Lab, San Francisco, CA).

Liver Tissues.

Three hundred ninety-six liver tissues with different CLDs, including (1) chronic HBV (n = 112), (2) chronic HCV (n = 77), (3) S. japonicum infection (n = 24), (4) S. japonicum and HBV coinfection (n = 59), (5) S. japonicum and HCV coinfection (n = 4), and (6) steatosis/steatohepatitis (n = 120, including alcoholic fatty liver disease and alcoholic steatohepatitis [n = 68] and nonalcoholic fatty liver disease and nonalcoholic steatohepatitis [n = 52]) were collected from 1995 to 2007 at five institutions (Institute of Infectious Diseases, First Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, China; Department of Pathology, Faculty of Medicine at Mannheim, University of Heidelberg, Germany; Department of Medicine and Laboratory of Alcohol Research, Liver Diseases and Nutrition, Salem Medical Center, University of Heidelberg, Heidelberg, Germany; Institute of Clinical Pharmocology, University of Bern, Bern, Switzerland; Department of Internal Medicine II, University of the Saarland, Homburg, Germany). Five nonfibrotic liver tissues with gallstones were used as controls. Wedge liver tissues with S. japonicum infection were collected from those patients who accepted surgery between 1995 and 2004. The remaining liver samples were obtained via liver biopsy from 2000 to 2007. The basic characteristics of these patients are shown in Supporting Table 1.

The study protocol involving human samples (tissues, serum) was approved by the local ethics committees, and every patient provided written informed consent. The study protocol adhered to the ethical guidelines of the 1975 Declaration of Helsinki.

Serum IL-13 Analyses in HCV Patients and Healthy Volunteers.

Eighty-eight serum samples, 68 from patients with chronic hepatitis C and 20 from healthy volunteers, were collected. The basic characteristics of these specimens are shown in Supporting Table 2. Serum IL-13 levels were determined via Quantikine enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN).


HSC isolation and culture were performed as described.23 To obtain quiescent and transdifferentiated HSCs, medium containing 10% fetal bovine serum was changed daily up to days 2 and 8. For cytokine treatment studies, HSCs were cultured in Dulbecco's modified Eagle's medium with 0.2% fetal bovine serum for 12 hours before challenge. The clonal cell line CFSC-2G (provided by P. Greenwel, Bronx, NY) was cultured as described.24


Liver tissue was fixed in 4% formaldehyde and embedded in paraffin. Four-micrometer sections were stained with hematoxylin-eosin, picro-sirius red, and Masson trichrome. Scheuer criteria were applied to determine inflammatory grade and fibrotic stage.25

Immunohistochemistry and Immunofluorescence Staining.

Sections were deparaffinized in serial ethanol dilutions. After a phosphate-buffered saline wash, sections were transferred into 10 mM sodium citrate buffer (pH 6.0), and antigen unmasking was performed in a microwave. After cooling down, sections were incubated in peroxidase blocking reagent (Dako) for 1 hour and with primary antibody overnight at 4°C. For immunohistochemistry, EnVision peroxidase (Dako) was applied for 1 hour at room temperature after a phosphate-buffered saline wash. Sections were developed with diaminobenzidine for 5 minutes. For immunofluorescence, secondary antibodies were Alexa488-rabbit anti-mouse immunoglobulin G, Alexa555-donkey anti-rabbit immunoglobulin G and Alexa633-donkey anti-rabbit immunoglobulin G (Molecular Probes/Invitrogen, Karlsruhe, Germany). Nuclei were stained with SYTOX Green (Molecular Probes/Invitrogen). Samples were mounted using DakoCytomation Fluorescent Mounting Medium (Dako). Confocal microscopy was performed as described.7

For semiquantitative analysis, immunohistochemical scores were calculated as follows: positive cell counts, grades 0-4 (0, no positive cells; 1, <25% positive cells; 2, 25%-50% positive cells; 3, 50%-75% positive cells; 4, >75% positive cells); intensity of positive staining, grades 1-3 (1, weak positive staining, usually yellow; 2, strong positive staining, usually brown; 3, very strong positive staining, usually deep brown to black). The final immune staining score was calculated as number × intensity.

For qualitative and quantitative evaluation of histology and immunohistochemistry, one pathologist blinded to sample identity assessed inflammatory grade, fibrotic stage, and immune score according to the above-mentioned criteria.

Western Blot Analysis.

Radio immunoprecipitation assay buffer (1× Tris-buffer saline, 1% Nonidet P40, 0.5% sodium deoxycholate, and 0.1% sodium dodecyl sulfate) was used to lyse liver tissue and cultured cells. Forty micrograms of tissue protein and 20 μg of cell protein were separated via sodium dodecyl sulfate–polyacrylamide gel electrophoresis (4%-12% Bis-Tris or 8% Tris-glycine gels; NuPAGE, Invitrogen) and transferred to nitrocellulose membranes (Pierce, Rockford, IL). Nonspecific binding was blocked with 5% milk in TBST before adding primary antibodies. Horseradish peroxidase–linked anti-rabbit, anti-mouse, or anti-goat antibodies (Santa Cruz) were used as secondary antibodies. The membrane was developed with Supersignal Ultra (Pierce).

Statistical Analysis.

Results are summarized as the mean ± standard deviation and depicted graphically as the mean ± standard error of the mean. To determine differences between groups for not normally distributed data, medians were tested via nonparametric analysis of variance. The degree of association between nonparametric or ordinal variables was assessed using the Spearman nonparametric correlation. Parametric data were analyzed using the Student t test (two-tailed). A P value of less than 0.05 was considered significant.


C-Terminal Smad2 Phosphorylation in Liver Tissue with Different Chronic Diseases.

To visualize activated TGF-β signaling in chronically damaged liver, we stained for phospho-Smad2. Liver tissues from five patients with absence of fibrotic lesions were negative for phospho-Smad2 and served as controls. In contrast, a prominent staining was observed in tissue samples obtained from patients with CLD and divergent underlying etiologies. Nuclei of hepatocytes, bile duct epithelial cells, inflammatory cells, Kupffer cells, and HSCs were immunopositive (Supporting Fig. 1). As expected, the distribution and/or number of phospho-Smad2 positive cells as well as the intensity of staining varied in dependence of disease etiology. Among six different kinds of CLD investigated, the strongest phospho-Smad2 staining was detected in chronic HBV-infected patients. Phospho-Smad2 immunostaining was semiquantitatively scored, yielding values of ≥5 in 69% (77/112) and >8 (44/112) in 39% of HBV-infected patients (maximum score = 12) (Table 1).

Table 1. phospho-Smad2 and IL-13 Staining in Chronic Liver Diseases
 Immunohistochemistry Score, n (%)
phospho-Smad2 (n = 396)   
HBV (n = 112)35 (31%)33 (30%)44 (39%)
HCV (n = 77)42 (55%)35 (45%) 
Steatosis/steatohepatitis (n = 120)67 (56%)52 (43%)1 (1%)
S. japonicum (n = 24)11 (46%)12 (50%)1 (4%)
S. japonicum + HBV (n = 59)40 (68%)17 (29%)2 (3%)
S. Japonicum + HCV (n = 4)4 (100%)  
IL-13 (n = 289)   
HBV (n = 63)59 (94%)4 (6%) 
HCV (n = 77)20 (26%)52 (67%)5 (7%)
Steatosis/steatohepatitis (n = 89)52 (58%)30 (34%)7 (8%)
S. japonicum (n = 20)14 (70%)6 (30%) 
S. japonicum + HBV (n = 36)17 (47%)19 (53%) 
S. Japonicum + HCV (4) 4 (100%) 

Compared with chronic HBV-infected patients, the phospho-Smad2 staining score was lower in other diseased liver tissue (Supporting Fig. 2). In chronic HCV infection, 45% (35/77) of patients were strongly positive (staining score ≥5), with the remainder having a weak staining or lacking any signals (steatosis/steatohepatitis, 44%

equation image

positive; S. japonicum infection, 54%

equation image

positive; HBV and S. japonicum coinfection, 32%

equation image

positive) (Table 1).

Phospho-Smad2 Correlates with Inflammatory Grade and Fibrotic Stage in Liver Diseases Due to Chronic HBV Infection or Steatosis/Steatohepatitis.

Coefficiency values were calculated for phospho-Smad2 immunostaining scores versus inflammatory grades/fibrotic stages to assess whether TGF-β/Smad signaling correlates with liver disease severity. The correlation was found to be significant in tissue with chronic HBV infection (P < 0.001) (Fig. 1A) and steatosis/steatohepatitis (P < 0.05) (Fig. 1B), but not in those with S. japonicum infection (P > 0.05) (Fig. 1C). In liver tissue with chronic HCV infection, a significant correlation was determined for phospho-Smad2 staining score versus inflammatory grade (P < 0.05) (Fig. 1D) but not fibrotic stage (P > 0.05) (Fig. 1D). For liver tissue coinfected with HBV and S. japonicum, we determined an inverse correlation between phospho-Smad2 score and inflammatory grade/fibrotic stage (P < 0.05, P = 0.05, respectively) (Fig. 1E).

Figure 1.

Correlation analysis between phospho-Smad2 immunostaining scores and the severity of diseases (inflammatory grade and fibrotic stage) in different chronic liver disease entities. Coefficiency values were calculated using Spearman nonparametric correlation. Top panels depict phospho-Smad2 immunostaining scores for different inflammatory grades, bottom panels for different fibrotic stages. (A) HBV. (B) Steatosis/steatohepatitis. (C) S. japonicum. (D) HCV. (E) S. japonicum/HBV coinfection.

Furthermore, we compared phospho-Smad2 immunostaining scores in tissues with advanced inflammatory grade/fibrotic stage (≥3) with those exhibiting lower inflammatory grade/fibrotic stage (<3). The results showed that phospho-Smad2 immunostaining scores significantly increased in liver tissue with advanced fibrosis and/or inflammation from patients with chronic HBV infection and in those with profound fibrosis with underlying steatosis/steatohepatitis (Supporting Table 3).

These data indicate that there is a stringent link between TGF-β1/Smad signaling and fibrotic stage in patients with chronic HBV infection or steatosis/hepatitis. They also indicate that this link is not present in patients with chronic HCV or S. japonicum infection.

HBx Protein Expression and Smad2 Activation Correlates in Chronic HBV-Infected Liver.

Most chronic HBV-infected liver tissues stained intensely positive for phospho-Smad2; however, in HBV tissues coinfected with S. japonicum phospho-Smad2 staining, intensity was significantly low (Table 1, Supporting Table 2). It is known that HBx protein trans-activates the TGF-β1 promoter and interacts with Smad4, thereby amplifying TGF-β1 signaling in vitro and in HBx transgenic mice.26, 27 S. mansoni infection inhibits HBV replication in HBV transgenic mice.28 We therefore hypothesized that HBV itself, via the HBx protein, mediates Smad2 phosphorylation in HBV-infected liver cells and that S. japonicum infection may counteract this response via lowered HBx protein expression in coinfected patients. We used immunohistochemistry to determine HBx protein expression in liver tissue of patients with advanced fibrosis stage (≥3); this resulted in diagnoses of chronic HBV infection (n = 10) or combined HBV/S. japonicum infection (n = 10). All HBV-infected liver tissues displayed strong HBx protein expression in hepatocytes and nonparenchymal liver cells (Fig. 2A,B) with nuclear and cytoplasmic localization (Fig. 2A,B). Confocal laser scanning microscopy revealed colocalization of HBx protein and phospho-Smad2 in the nuclear compartment of hepatocytes and nonparenchymal liver cells (Fig. 2A). In contrast, the HBx protein immunostaining score was significantly lower in patients coinfected with HBV/S. japonicum (P < 0.01, Fig. 2C). Concomitant with reduced HBx protein expression, weak phospho-Smad2 staining was detectable in liver tissue coinfected with HBV/S. japonicum (Fig. 2B).

Figure 2.

HBx protein influences Smad2 phosphorylation in liver tissue with chronic HBV infection. (A) Immunofluorescence costaining for phospho-Smad2 (red) and HBx protein (green) in liver tissue with chronic HBV infection is shown. Colocation of phospho-Smad2 and HBx protein is present in nuclei of hepatocytes (white arrows). Sytox green was used to stain nuclei (blue). (B) Immunohistochemical staining was used to analyze phospho-Smad2 (brown) and HBx protein (brown) expression in two representative HBV patients with or without S. japonicum coinfection. (C) Twenty specimens were randomly chosen from advanced fibrotic liver tissues (stage ≥3) with isolated chronic HBV or HBV/S. japonicum coinfection (10 samples per group). HBx immunostaining score was determined as mentioned in Materials and Methods.

Although these results do not confirm HBx-dependent regulation of TGF-β1 signaling in human liver cells, the data strongly suggest that S. japonicum and HBV infections have opposing impacts on Smad2 activation in liver cells.

TGF-β1 and IL-13 Effects in Schistosomiasis-Infected Patients.

In control liver tissue and tissue obtained from four schistosomiasis patients with advanced fibrosis (stage ≥3), TGF-β1 and IL-13 protein expression was quantified by immunoblotting of tissue lysates (Fig. 3A). In sample Sj-101, large amounts of TGF-β1 protein was detected, while lysates Sj-103 and Sj-104 exhibited increased IL-13 protein (Fig. 3A). These results again underscore the notion that TGF-β1/Smad signaling is not a prerequisite for fibrosis in patients with chronic schistosomiasis. In addition, we stained liver tissues from 24 patients with isolated S. japonicum infection for α-smooth muscle actin (α-SMA) expression. Enhanced α-SMA staining was present irrespective of phospho-Smad2 positivity (Fig. 3B), consistent with the correlation data presented on fibrosis scores.

Figure 3.

There are some fibrotic liver tissues with S. japonicum infection in the absence of TGF-β1 and phospho-Smad2. (A) Western blot analysis for TGF-β1 and IL-13 protein expression was performed with liver tissue lysates from four patients with S. japonicum infection and one control without liver fibrosis. (B) Immunohistochemical staining was used to score phospho-Smad2 and α-SMA (both brown) in S. japonicum infected liver tissue. Representative images from two patients are shown.

IL-13 but Not TGF-β1 Induces CTGF, α-SMA, Egr-1, and Vascular Endothelial Growth Factor Expression in Primary Cultured HSCs.

It is well recognized that reducing IL-13 levels by means of soluble IL-13Rα2-Fc markedly retards liver fibrogenesis in schistosomiasis-based mouse models.29 However, the molecular mechanism of profibrogenic IL-13 effects in liver cells is still enigmatic. To address this question, we measured collagen type I, CTGF, α-SMA, Egr-1, and vascular endothelial growth factor (VEGF) protein expression in primary HSCs and CFSC-2G cell lines in the absence and presence of IL-13. Similar to TGF-β1, IL-13 induced collagen type I protein expression in quiescent HSCs (Fig. 4A). Contrary to TGF-β1, IL-13 also induced expression of CTGF and Egr-1 in primary quiescent and activated HSCs, as well as α-SMA and VEGF in activated HSCs (Fig. 4B-D). In CFSC-2G, a rat HSC cell line derived from a cirrhotic liver, both TGF-β1 and IL-13 induced α-SMA protein expression. IL-13 had no effect on CTGF protein expression alone, but further enhanced it in concert with TGF-β1 (Fig. 4E). These data indicate that IL-13 participates in HSC activation and may aggravate fibrogenesis in cooperation with TGF-β1.

Figure 4.

IL-13 up-regulates fibrotic proteins in rat HSCs and CFSC-2G cell lines. (A) Collagen type I expression was quantified via western blotting in quiescent HSCs (day 3) in a time course experiment following IL-13 (50 ng/mL) stimulation. (B-E) CTGF, α-SMA, Egr-1, and VEGF protein expression following TGF-β1/IL-13 challenge in quiescent (day 3), activated (day 9) HSCs and CFSC-2G cell lines was quantified via immunoblotting. Glyceraldehyde 3-phosphate dehydrogenase/β-actin was used as a loading control. Semiquantitative values of band intensities normalized for loading controls are provided (only showing the bars with significant induction). *P < 0.05, **P < 0.01 compared with untreated samples.

Phospho-STAT6 Is a Marker for IL-13 Signaling In Vitro but Does Not Correlate with IL-13 Expression In Vivo.

STAT6 phosphorylation is thought to be a key event in IL-13–dependent signal transduction.19 To address the question whether IL-13 induces profibrogenic proteins via phospho-STAT6, we preincubated the CFSC-2G cell line with leflunomide, a specific phospho-STAT6 inhibitor, for 6 hours before IL-13 administration. Western blot analysis revealed that 50 μM of leflunomide significantly decreased IL-13–dependent phospho-STAT6 generation but did not reduce IL-13–dependent α-SMA expression in the CFSC-2G cell line (Fig. 5A). These data indicate that the IL-13–dependent effect does not require STAT6 phosphorylation.

Figure 5.

Phospho-STAT6 is only detected in a minority of patients with IL-13 expression and liver fibrosis. (A) Leflunomide, a specific STAT6 phosphorylation inhibitor, was used for 6 hours in CFSC-2G cells before IL-13 treatment. Western blot analysis was used to measure α-SMA and phospho-STAT6 expression. β-Tubulin was used as a loading control. Semiquantitative values of band intensities normalized for β-tubulin are provided. *P < 0.05 compared with untreated samples. (B) Immunohistochemical staining for IL-13 and phospho-STAT6 (both brown) expression in 120 liver tissues with different CLDs. Two representative liver tissues with HBV/S. japonicum coinfection are shown in the top and middle panels (insets in the top panel show negative controls). Positive phospho-STAT6 staining was present in the cytoplasm (highlighted in top right frame). Graphs in the bottom panel show the number for specimen with phospho-STAT6 positive/negative staining out of 120 liver tissues (IL-13 positive).

We next wished to determine the abundance and localization of phospho-STAT6 in fibrotic livers. We assessed STAT6 tyrosine-641 phosphorylation via immunohistochemistry in 120 liver samples (HBV [n = 40], HCV [n =20], alcoholic steatohepatitis [n =20], S. japonicum [n =20], and HBV/S. japonicum coinfection [n =20]). Contrary to our expectations, phospho-STAT6–positive staining was only detected in five patients (S. japonicum [n = 1], HBV/S. japonicum coinfection [n = 2], HCV [n = 1], and alcoholic steatohepatitis [n = 1]) (Fig. 5B). Although five samples displayed both IL-13 and phospho-STAT6 positivity adjacent to inflammatory and fibrotic areas (Fig. 5B, top panel), the remaining 115 samples showed IL-13 positivity in the absence of phospho-STAT6 staining (Fig. 5B, lower panel). Phospho-STAT6 staining was mostly detected in the cytoplasm and in damaged nuclei of hepatocytes (Fig. 5B, small frame in right panel). These results suggest that phospho-STAT6 positivity is not a useful marker for IL-13 signaling in liver fibrosis.

IL-13 Expression in Different CLDs.

Cell types from different origins synthesize IL-13 under pathological conditions.17 Immunohistochemistry analysis revealed that the following cell types are IL-13–immunopositive in chronically damaged liver: hepatocytes, sinusoid cells (HSCs and Kupffer cells), and inflammatory cells (Supporting Fig. 3). Like phospho-Smad2 in damaged liver, the distribution and number of IL-13–positive cells and staining intensities varied in dependence of disease etiologies. Among six different kinds of CLD investigated, intense IL-13 staining was present in liver tissues with HCV and steatosis/steatohepatitis (Table 1 and Supporting Fig. 4). In chronic HCV-infected patients the IL-13 staining score was ≥5 in 74% (57/77) and even >8 in 7% (5/77) of cases, while it was ≥5 in 42% (37/89) and >8 in 8% (7/89) of patients with steatosis/steatohepatitis. In HBV/S. japonicum–coinfected liver samples, IL-13–positive staining (immunostaining score ≥5) was detected in 53% (19/36) of patients; a score of >8 was not reached. Only 30% (6/20) of liver tissues from patients with S. japonicum infection were IL-13–immunopositive. In chronic HBV-infected patients, weak or absent IL-13 staining was found in 93% (59/63) of liver samples, suggesting that IL-13 is not critically involved in chronic HBV-related fibrogenesis (Table 1).

IL-13 Expression Correlates with Inflammatory Grade and Fibrotic Stage in Chronic HCV and HBV Infection and Steatosis/Steatohepatitis.

To test for an association between IL-13 expression and severity of liver disease, coefficiency values were calculated for IL-13 staining scores and inflammatory grade/fibrotic stages. A strong correlation was present in chronic HBV-infected patients (P < 0.001) (Fig. 6A), although the IL-13 score in most chronic HBV-infected liver tissues was lower. This suggests that IL-13 plays a role in chronic HBV infections that progress to fibrosis. A significant correlation between IL-13 staining score and inflammatory grade/fibrotic stage was also found in chronic HCV-infected tissue (P < 0.01) (Fig. 6B) and steatosis/steatohepatitis (P < 0.05) (Fig. 6C). In contrast, there was no significant correlation between IL-13 staining and fibrotic stage in S. japonicum–infected patients and those with HBV coinfection (P > 0.05) (Fig. 6D,E).

Figure 6.

Correlation analyses between IL-13 and the severity of disease (inflammatory grade and fibrotic stage) in different chronic liver diseases. Coefficiency values were calculated using Spearman nonparametric correlation. Top panels depict IL-13 immunostaining scores for different inflammatory grades, bottom panels for different fibrotic stages. (A) HBV. (B) HCV. (C) Steatosis/steatohepatitis. (D) S. japonicum. (E) S. japonicum/HBV coinfection.

We also tested for an association between IL-13 staining and advanced inflammatory grade/fibrotic stages (stage ≥3) as well as lower inflammatory grade/fibrotic stages (stage <3). The results show that the IL-13 score increases with advanced fibrosis and/or inflammation in livers from patients with chronic HBV or HCV infection and steatosis/steatohepatitis (Supporting Table 4). Although no significant correlation between IL-13 score and fibrotic degree was apparent, IL-13 scores were increased in liver tissues when advanced inflammation occurred due to schistosomiasis-related CLDs (Supporting Table 4).

To further confirm IL-13 expression in chronic HCV infection, we measured serum IL-13 levels via enzyme-linked immunosorbent assay in 68 HCV patients. Compared with 20 healthy volunteers, significantly elevated IL-13 serum levels were detected in HCV patients (72.87 ± 26.38 versus 45.41 ± 3.73; P < 0.001) (Fig. 7A). Alternatively, we detected IL-13 protein expression via immunostaining in liver tissue lysates from three HCV patients and one control and found that IL-13 protein was more abundant in the three HCV patients (Fig. 7B). These data suggest that IL-13 may play a functional role in fibrogenesis related to chronic HCV infection or steatosis/steatohepatitis.

Figure 7.

IL-13 expression in patients infected with chronic HCV. (A) Serum IL-13 levels were significantly higher in chronic HCV-infected patients compared with healthy volunteers. (B) Western blot analyses for IL-13 protein expression was performed with liver tissue lysates from three patients with HCV infection and one control without liver fibrosis. β-Actin was used as a loading control.

In a Subgroup of Patients with Fibrotic Liver Disease, Neither TGF-β1 nor IL-13 Signaling Is Activated.

Liver tissues from 76 patients with advanced fibrosis (stage ≥3) were examined for TGF-β1/Smad2 and IL-13 expression/signaling (Table 2). Positivity for phospho-Smad2/IL-13 detection was defined as staining scores ≥ 5 (maximum score = 12). Twenty-three percent of samples showed predominant TGF-β1/Smad2 positivity, whereas 26% were IL-13–positive. Thirty percent of liver tissues displayed both TGF-β1/Smad2 and IL-13 staining. In 21.3% of liver tissues, neither TGF-β1/Smad2 nor IL-13 staining was apparent. These results leave open the possibilities that (1) signaling events are temporarily shut off in the course of disease or (2) yet unclassified profibrogenic mediators act in these cases and drive liver fibrogenesis.

Table 2. phospho-Smad2 and IL-13 Expression in Different CLDs with Advanced Fibrosis (Stage ≥3)
  1. Abbreviation: IHC, immunohistochemistry.

p-Smad2 IHC score ≥5++
IL-13 IHC score ≥5++
HBV (n = 24)17 (71%) 4 (17%)3 (12%)
HCV (n = 32) 13 (41%)17 (53%)2 (6%)
Alcoholic steatohepatitis (n = 11)1 (9%)3 (27%)3 (27%)4 (37%)
S. japonicum (n = 12)4 (33%)2 (17%)3 (25%)3 (25%)
S. japonicum + HBV (n = 29)3 (10%)10 (35%)5 (17%)11 (38%)
Total (n = 76)25 (23%)28 (26%)32 (30%)23 (21%)

In HBV-associated advanced fibrotic livers, 71% of samples were phospho-Smad2–positive, whereas 17% were positive for both phospho-Smad2 and IL-13 (Table 2). By contrast, in HCV-related advanced fibrotic livers, 41% of samples displayed IL-13 positivity, whereas 53% were positive for both phospho-Smad2 and IL-13 (Table 2). In the latter group, isolated positivity for phospho-Smad2 was not observed. These data suggest that IL-13 participates in the development of HCV-associated advanced liver fibrosis in a considerable portion of patients.


A systematic analysis of TGF-β1 signaling in liver tissue from patients with diverse CLD was the primary goal of the present study with the hypothesis that this cytokine does not uniformly act in fibrogenesis. Analyses were performed with a large liver tissue collection from patients with different inciting damages, all resulting in CLDs. The following insights on TGF-β1 and IL-13 signaling and their roles in liver fibrosis formation were obtained: (1) TGF-β1/Smad2 signaling in liver fibrogenesis is not a generalized feature and detected in an etiology-dependent manner; (2) TGF-β1/Smad2 signaling is a common feature in HBV-associated liver fibrosis and is similarly activated in steatosis/steatohepatitis; (3) IL-13 expression patterns suggest a contributive role of this Th2 cytokine in HCV-related liver fibrosis and steatosis/steatohepatitis; (4) STAT6 phosphorylation is not an absolute requirement for IL-13 activities; and (5) in some cases, both cytokine signaling events, TGF-β1 and IL-13, are absent.

How can these results be reconciled with the general view of liver fibrogeneis? In chronic HBV-infected patients, we observed a significant correlation between phospho-Smad2 immunohistochemistry score and the severity of liver disease, regarding both inflammatory grade and fibrotic stage in liver tissue. Consistent with other reports,10 we also detected elevated active TGF-β1 levels in serum from these patients (data not shown). Colocalization studies of phospho-Smad2 and HBx protein expression imply that HBx protein is promoting enhanced TGF-β1/Smad2 signaling in these patients. This hypothesis is supported by findings that HBx activates transcription of the TGF-β1 gene promoter in hepatoma cells; HBx protein enhances TGF-β1 transcriptional activity by stabilizing Smad4-containing complexes with components of the basic transcriptional machinery, and HBx facilitates nuclear translocation of Smad proteins.26, 27 A similar expression pattern for TGF-β1 and HBx protein has been reported for HBx transgenic mice.26, 27

We determined Smad2 phosphorylation as a means of assessing activated TGF-β signaling in liver tissue, mainly because of the assay robustness and the superior quality of staining results. In addition to Smad2, TGF-β1 may signal via Smad3 and Smad1,13 and visualizing Smad2 phosphorylation does not provide a complete view on TGF-β1 downstream targets.30 On the other hand, Smad2 and Smad3 proteins are mostly activated to a similar extent when the ALK5 receptor is switched on through ligand occupation. Notably, compared with the intense phospho-Smad2 staining in HBV patients, we found a significant lower phospho-Smad2 immunohistochemistry score in HBV patients coinfected with S. japonicum. It is known from HBV transgenic mice that schistosomiasis displays antiviral effects and inhibits HBV replication concomitant with the intrahepatic induction of nitric oxide and Th1-type cytokines.28 The S. mansoni–dependent antiviral effect was partially blocked by genetically deleting interferon γ (IFN-γ), indicating that IFN-γ, likely via nitric oxide, mediates most of this antiviral activity.28 Consistent with these findings, our presented data are in line with the interpretation that infection by S. japonicum reduces HBx protein expression in HBV patients, altogether resulting in decreased Smad2 phosphorylation.

In patients with steatosis/steatohepatitis (including alcoholic steatohepatitis), we also determined a significant correlation of Smad2 phosphorylation with fibrotic stage and inflammatory grade. An association between serum procollagen-III-N-propeptide and TGF-β1 has been reported in 61 patients with alcoholic liver disease.12 In vitro, ethanol and acetaldehyde induce TGF-β1 expression in HSCs, resulting in α2 (I) collagen gene promoter activation and up-regulated protein expression.31 Furthermore, acetaldehyde induces TGF-β1 receptor type II expression and secretion of TGF-β1, which is required for TGF-β1 signaling in HSCs.32 Because these effects occur in a dose-dependent fashion, enhanced TGF-β1 signaling may be anticipated in advanced disease stages. Taken together, the phospho-Smad2 score may be suitable as a prognostic/diagnostic marker to determine disease activity.

In chronic HCV-infected liver tissue, no correlation of phospho-Smad2 signaling with fibrotic stage was found, whereas a significant correlation with inflammatory grade was evident in the same samples, although the phospho-Smad2 immune staining score did not markedly increase in liver tissue with advanced inflammation/fibrosis (grade/stage ≥3) compared with those showing less severity (grade/stage <3). These results are consistent with results from a clinical study on 35 patients suffering from chronic HCV infection in which the mean levels of TGF-β1 messenger RNA (mRNA) were 200-fold higher compared with controls, whereas no correlation was found between TGF-β1 mRNA and histological indices of liver fibrosis.33 The finding that TGF-β and downstream signals (e.g., phospho-Smad2) do not correlate with fibrotic stage in HCV-infected patients, however, does not preclude per se the importance of TGF-β in HCV-mediated fibrogenesis. A recent study from Matsuzaki's group showed that chronic inflammation associated with HCV infection shifts hepatocytic TGF-β signaling from tumor suppression to fibrogenesis.34 Controversial reports on the relationship between HCV infection and TGF-β1/Smad signaling exist in the literature. Taniguchi and colleagues35 showed that HCV core protein up-regulates TGF-β1 mRNA levels in human hepatoma cells by interfering in the transcriptional regulation and association with the TGF-β1 promoter at neleotides −376 to −331 bp. Bataller and coworkers36 found that core and NS3-NS5 protein of HCV induce the secretion of active TGF-β1 and expression of procollagen α1(I) in quiescent rat HSCs. On the other hand, Murata et al.37 found that TGF-β1 suppresses viral RNA and protein expression of the HCV replicon and that the antiviral effect of TGF-β1 is associated with cellular growth arrest in a Smad signaling-dependent manner in the HCV replicon cell line MH14. Furthermore, NS5A protein of HCV abrogates phosphorylation of Smad2 and heterodimerization of Smad3 and Smad4 in hepatoma cell lines.38 Our immunohistochemical data imply that such complex interactions between TGF-β/Smad signaling and HCV are not the driving force for liver fibrogenesis in chronic HCV-infected patients.

In schistosomiasis-associated liver fibrosis, TGF-β1 is also not the main regulator of collagen expression and accumulation.15, 16, 22 In S. mansoni–infected baboons, the Schistosoma-soluble egg antigen induces the synthesis of cytokines such as IL-4, IL-5, IL-10, IL-12, and TGF-β1. Among those, only TGF-β1 remains elevated during chronicity of infection; however, TGF-β1 correlates with diminished hepatic granuloma size. This implies participation of TGF-β1 in the down-regulation of disease severity.39 An epidemiologic study analyzed the association between fibrosis degree and cytokine levels, including IFN-γ, tumor necrosis factor-α, TGF-β1, IL-4, IL-10, and IL-13 in an S. mansoni endemic area in Brazil. Multivariate analysis revealed that only IL-13 levels strongly correlate with fibrosis, whereas an elevated TGF-β1 level is associated with protection against fibrosis.15 Our data are consistent with the above findings and show that (1) there is no correlation between phospho-Smad2 and extent of fibrosis in S. japonicum–infected human livers; (2) advanced fibrosis still exists in schistosomiasis without significant TGF-β/Smad2 signaling; and (3) a significant negative correlation is evident between phospho-Smad2 staining and fibrotic stage/inflammatory grade in patients coinfected with S. japonicum and HBV. These results support the notion that TGF-β plays an antifibrogenic role in schistosomiasis.

Regarding the question of why and how TGF-β plays either profibrotic or antifibrotic roles in different settings of liver disease, we should consider the answer provided by T. Wynn. Wynn stated that the TGF-β1 source is crucial to differing outcomes18—for example, macrophage-derived TGF-β being profibrotic21—whereas T cell–derived TGF-β may be suppressive to fibrogenesis.40 Further evidence will be needed to prove this statement.

In schistosomiasis-based multiorgan fibrosis, IL-13 has been identified as a profibrogenic mediator.17–20, 29, 41 IL-13 is a dominant Th2-type cytokine that is present at all stages of S. mansoni infection.18 IL-13–deficient mice display reduced liver fibrosis and enhanced survival following S. mansoni infection.20 Moreover, treatment of S. mansoni–infected mice with a soluble IL-13Rα2-Fc decoy receptor results in a significant antifibrotic effect and decreased collagen synthesis in the liver.41, 42 Another study reported that in CLD, schistosomiasis-based fibrosis correlates with high IL-13 and low IFN-γ/IL-10 levels.41, 43 Notably, IL-13–driven liver fibrosis was unaffected in Smad3- and TGF-β–deficient mice, indicating that TGF-β1/Smad3 signaling is not a prerequisite for the profibrogenic activities of IL-13.22 Consistent with these reports, we found enhanced IL-13 protein expression in 2/4 S. japonicum infected livers with severe fibrosis, whereas TGF-β1 protein levels were low or undetectable.

To further elucidate profibrogenic IL-13 effects, we incubated rat HSCs and observed that IL-13 simulates collagen type I expression, similar to TGF-β1. IL-13 but not TGF-β1 increased Egr-1 and α-SMA expression in quiescent and activated HSCs as well as CTGF and VEGF only in activated HSCs. In the myofibroblastic CFSC-2G cell line, IL-13 enhanced TGF-β1–induced CTGF protein expression. The above results indicate that IL-13 exerts its fibrogenic role via TGF-β1–dependent as well as TGF-β1–independent pathways and thus may have a more general profibrotic role that is not restricted to schistosomiasis-associated liver fibrosis. Elevated IL-13 expression is also present in liver tissues with chronic HCV infection or steatosis/steatohepatitis with levels exceeding the ones detected in patients suffering from S. japonicum infection. IL-13 correlated with fibrotic stage/inflammatory grade in patients with chronic HBV or HCV infection and steatosis/steatohepatitis. None of the chronic HCV-associated liver tissues with advanced fibrotic stages displayed predominant TGF-β1/Smad signaling, while 93.75% of these samples exhibited high IL-13 expression levels. Furthermore, we found more abundant IL-13 protein expression in liver tissue lysata compared with controls and significantly elevated IL-13 serum levels in HCV patients compared with healthy volunteers. All these findings underline the notion of IL-13–mediated HCV-dependent liver fibrosis. In the present study, we did not find a correlation between IL-13 and inflammatory grade/fibrotic stage in schistosomiasis-based disease, including isolated S. japonicum infection or with HBV coinfection. Possible explanations include the possibility that of the collected schistosomiasis-infected liver tissues derived from a subgroup of patients, all underwent splenectomy due to splenomegaly. At this stage, most patients present with advanced fibrotic stage (69% of specimens displayed fibrotic stage ≥3). In addition, the number of patients belonging to this group was rather low and limited the power of the statistical analyses.

Sugimoto and colleagues44 observed that IL-13 induces phosphorylation of Stat6 via the IL-13Rα1 with 100-fold induction of collagen I mRNA expression in LI90 cells derived from a hepatic mesenchymal tumor with similarities to human HSCs.44 In our hand, significant STAT6 phosphorylation was present upon IL-13 stimulation in HSCs and CFSCs. Thus, we hypothesized STAT6 as a marker of IL-13 signaling and performed phospho-STAT6 staining with 120 fibrotic livers that were IL-13–positive. However, only five of these tissues stained positive for phospho-STAT6. In vitro, IL-13–induced α-SMA protein expression was not decreased in CFSCs after blocking STAT6 phosphorylation with leflunomide, a specific inhibitor of this step. These data suggest that IL-13 does not require STAT6 for signaling in the majority of cases.

It is expected that in addition to TGF-β1 and IL-13, additional profibrotic mediators in CLDs exist. For example, cationic amino acid transporter-2 was recently identified as an important regulator of liver fibrosis.45 Consistent with the assumption of further critical participants, a share of 23% patients with advanced liver fibrosis neither displayed TGF-β1/Smad signaling nor IL-13 expression.


We thank Mrs. Alexandra Müller for isolating rat HSCs, Mrs. Carolin Stump for real-time polymerase chain reaction, Mr. Si-Wei Chen for statistics, and Mrs. Yan-Min Zhang for performing enzyme-linked immunosorbent assays. We are grateful to Dr. P. Greenwel for the CFSC-2G cell line.