Potential conflict of interest: Nothing to report.
Many patients with chronic hepatitis caused by hepatitis C virus (HCV) infection develop liver fibrosis with high risk for hepatocellular carcinoma (HCC), but the mechanism underling this process is unclear. Conversely, transforming growth factor beta (TGF-β) activates not only TGF-β type I receptor (TβRI) but also c-Jun N-terminal kinase (JNK), which convert the mediator Smad3 into two distinctive phosphoisoforms: C-terminally phosphorylated Smad3 (pSmad3C) and linker-phosphorylated Smad3 (pSmad3L). Whereas the TβRI/pSmad3C pathway suppresses epithelial cell growth by upregulating p21WAF1 transcription, JNK/pSmad3L-mediated signaling promotes extracellular matrix deposition, partly, by upregulating plasminogen activator inhibitor 1 (PAI-1). We studied the domain-specific Smad3 phosphorylation in biopsy specimens representing chronic hepatitis, cirrhosis, or HCC from 100 patients chronically infected with HCV, and correlated Smad3 phosphorylation with clinical course. As HCV-infected livers progressed from chronic hepatitis through cirrhosis to HCC, hepatocytic pSmad3L/PAI-1 increased with fibrotic stage and necroinflammatory grade, and pSmad3C/p21WAF1 decreased. Of 14 patients with chronic hepatitis C with strong hepatocytic pSmad3L positivity, 8 developed HCC within 12 years; only 1 of 12 showing little pSmad3L positivity developed HCC. We further sought molecular mechanisms in vitro. JNK activation by the pro-inflammatory cytokine interleukin-1β stimulated the pSmad3L/PAI-1 pathway in facilitating hepatocytic invasion, in the meantime reducing TGF-β-dependent tumor-suppressive activity by the pSmad3C/p21WAF1 pathway. Conclusion: These results indicate that chronic inflammation associated with HCV infection shifts hepatocytic TGF-β signaling from tumor-suppression to fibrogenesis, accelerating liver fibrosis and increasing risk for HCC. (HEPATOLOGY 2007;46:48–57.)
Over 170 million people are infected with hepatitis C virus (HCV) worldwide, resulting in a large disease burden and significant mortality.1 Because HCV is rarely cleared in the acute phase of infection, most patients become chronically infected and develop chronic hepatitis.1 Some of these patients sustain liver fibrosis as a result of chronic liver damage in conjunction with accumulation of extracellular matrix (ECM) proteins. Such accumulation is characteristic of most types of chronic liver disease.2, 3 Accumulation of ECM proteins distorts hepatic architecture by forming a fibrous scar. Ultimately, nodules of regenerating hepatocytes become enclosed by scar tissue, which defines cirrhosis. Cirrhosis is not only the end-stage of progressive fibrosis but also a prerequisite for development of HCV-related hepatocellular carcinoma (HCC),4 which occurs at a yearly rate of 6% in Japan. HCV infection leading to chronic inflammation apparently promotes both fibrosis and preneoplastic changes in the liver.5
Transforming growth factor beta (TGF-β), which can potently inhibit epithelial cell growth to act as a tumor suppressor,6 is also a key regulator of epithelial-to-mesenchymal transition (EMT) in cell phenotypes.7 EMT not only underlies epithelial degeneration and fibrogenesis in chronic degenerative disorders, but also endows dedifferentiated malignant epithelial cells with mesenchymal, migratory, and proteolytic properties that are required for local tumor invasiveness.8 Inhibition of a pro-inflammatory cytokine interleukin-1β (IL-1β) at initiation of EMT has been found to attenuate fibrogesis,9 suggesting a causative link between chronic inflammation and EMT.
Progress over the past 10 years has disclosed important details of how TGF-β elicits its responses. The main downstream signaling pathway for TGF-β involves the Smad proteins.10, 11 Although several studies of EMT have suggested that the process involves Smad-independent pathways,12 recent studies using Smad3 knockout mice have indicated that signaling through the Smad3-dependent pathway is required for injury-dependent multistage transition of an epithelial cell to a mesenchymal phenotype.13 We therefore have focused on Smad3 signaling,14, 15 and have recently reported different roles of Smad3 phosphoisoform-mediated signaling in epithelial cells and mesenchymal cells.16, 17 Thus, TGF-β activates not only TGF-β type I receptor (TβRI) but also c-Jun N-terminal kinase (JNK), converting Smad3 into two distinctive phosphoisoforms: C-terminally phosphorylated Smad3 (pSmad3C) and linker-phosphorylated Smad3 (pSmad3L).16 The TβRI/pSmad3C pathway inhibits growth of epithelial cells including hepatocytes, whereas JNK/pSmad3L-mediated signaling promotes ECM deposition by activated mesenchymal cells such as hepatic stellate cells (HSCs).17 According to these phosphorylation-defined activities, we sought the molecular mechanisms by which Smad3 phosphoisoforms govern progression from chronic hepatitis C through cirrhosis to HCC.
Between 1992 and 2005, 1,387 patients with chronic HCV infection underwent liver biopsy at the Departments of Gastroenterology and Hepatology of Kansai Medical University Hospital. In alphabetical order, we chose 20 cases per stage (F1 to 4) and also 20 HCC cases from the patients with HCV-related chronic liver disease. All patients were positive for anti-HCV antibody, as assessed by a second-generation enzyme immunoassay (Ortho Diagnostics, Tokyo, Japan), and/or for HCV-RNA, as detected by an Amplicor HCV assay (version 1; Roche, Tokyo, Japan), but were negative for hepatitis B surface antigen (Abbott Laboratories, North Chicago, IL). Pathology records and histological slides were reviewed independently by two pathologists with special interest in liver diseases (Y.U. and N.S.). Necroinflammatory activity and fibrotic stage were graded histologically according to Desmet's classification.18 The histological grade of HCC and the classification of primary tumor regional lymph nodes, distant metastasis (pTNM) were determined according to the criteria of the International Working Party19 and the criteria of the International Union Against Cancer and the American Joint Committee on Cancer,20 respectively. We counted and scored pSmad3C/L positivity in the nuclei of hepatocytes adjacent to collagen fibers in portal tracts. Hepatocytic Smad3 phosphorylation in HCV-infected livers was scored as follows: 0, no phosphorylation; 1, <25% Smad3 phosphorylation; 2, 25% to 50% Smad3 phosphorylation; 3, 50% to 75% Smad3 phosphorylation; 4, >75% Smad3 phosphorylation. As controls, we used uninvolved normal liver tissue from a patient with a metastatic liver tumor. Written informed consent was obtained from each patient according to the Helsinki Declaration. We also obtained approval for this study from the institutional ethics committee.
Follow-up and Detection of HCC.
Twenty-six patients with chronic hepatitis C with grade F2 to 3 fibrosis, who underwent liver biopsy between 1992 and 1994, were enrolled in a program for early diagnosis of HCC. During the surveillance period, the patients were followed up with abdominal ultrasonography and determination of α-fetoprotein level every 3 to 6 months.
Domain-Specific Antibodies (Abs) Against the Phosphorylated Smad3.
Two polyclonal anti-phospho-Smad3 sera, α pSmad3L (Ser 208/213) and α pSmad3C (Ser 423/425), were raised against the phosphorylated linker and C-terminal regions of Smad3 by immunization of rabbits with synthetic peptides.14 Relevant antisera were affinity-purified using the phosphorylated peptides.
Immunohistochemical and Immunofluorescence Analyses.
Immunohistochemical analyses were performed as described previously.17 Primary antibodies used in this study included mouse monoclonal anti-αSMA Ab (1.7 μg/ml; DAKO, Glostrup, Denmark), mouse monoclonal anti-PAI-1 Ab (0.4 μg/ml; Santa Cruz Biotechnology, Santa Cruz, CA), mouse monoclonal anti-p21WAF1 Ab (0.5 μg/ml; DAKO), in addition to rabbit polyclonal anti-pSmad3L (2 μg/ml), and rabbit polyclonal anti-pSmad3C (0.5 μg/ml) described above.
For double-labeling immunofluorescence analyses, sections exposed to a pair of primary Abs (rabbit plus mouse) were then incubated in a 1:500 dilution of goat anti-rabbit IgG conjugated with a red fluorophore (Alexa Fluor 594; Molecular Probes, Eugene, OR) and goat anti-mouse IgG conjugated with a green fluorophore (Alexa Fluor 488; Molecular Probes). Images were obtained with a fluorescence microscope (Carl Zeiss Microimaging, Oberkochen, Germany).
Immunoprecipitation and Immunoblotting.
Cultured rat hepatocytes (Clone 9 cells)21 were starved for 15 hours in serum-free medium, and were incubated for 8 hours in the absence or presence of 10 μM JNK inhibitor SP600125 (Calbiochem, San Diego, CA). The cells were then treated with 20 pM TGF-β (R&D Systems, Minneapolis, MN) and/or 400 pM IL-1β (R&D Systems) for 30 minutes. Cultured cells, or frozen tissues representing either HCC or underlying liver diseases, were extracted with cell lysis buffer. Cell extracts were subjected to immunoprecipitation with anti-Smad3 Ab (BD Bioscience, San Jose, CA), followed by adsorption to protein G-Sepharose (Amersham Pharmacia Biotech, Buckinghamshire, UK). Phosphorylation of Smad3 was analyzed using rabbit polyclonal anti-pSmad3L Ab and anti-pSmad3C Ab as described previously.15
Isolation of RNA, reverse transcription, and PCR for TβRII, Smad2, Smad4, PAI-1, p21WAF1, and GAPDH was performed as described previously.22
Membranes with 8-μm pores covered with Matrigel (BD Biosciences, Bedford, MA) on the upper surface were coated with type I collagen on the lower side. Infiltrating cells were counted in 5 regions selected at random as described previously.15
[3H] Thymidine Incorporation.
DNA synthesis was measured by incorporation of 1 μCi/ml [3H] thymidine (Amersham Pharmacia Biotech) into 5% trichloroacetic acid-precipitable material after a 4-hour pulse as described previously.23
All P values were based on two-tailed statistical analysis. The Mann-Whitney U test was used to identify significant differences in hepatocytic Smad3 phosphorylation among grades of necroinflammatory activity as well as fibrotic stages. A P value below 0.05 was considered to indicate significance. Kaplan-Meier analysis and log-rank test were performed to determine cumulative incidence of HCC in patients with abundant (scores 3 to 4) and sparse (scores 0 to 2) Smad3L phosphorylation.
Hepatocytes Showing pSmad3L-Mediated Signaling Adjacent to Collagen Fibers in Portal Tracts.
We initially analyzed mutations of TGF-β type II receptor, Smad2, and Smad4 genes in 10 HCC and 6 cirrhotic liver samples, finding no mutation in any liver samples.
Figure 1A presents distribution of pSmad3L and pSmad3C in chronic hepatitis C specimens, which showed moderate fibrosis and severe necroinflammatory activity. HSCs are currently recognized as a key element in development of liver fibrosis.3 As a result of chronic liver damage, HSCs undergo progressive activation to myofibroblast-like cells, which are characterized by synthesis of a large amount of ECM components. During the transdifferentiation process of the cultured HSCs, pSmad3C-mediated signal decreased whereas the pSmad3L pathway predominated, stimulating ECM deposition.14 These past observations fully support the current finding of pSmad3L, not pSmad3C, in nuclei of α-smooth muscle actin (αSMA) immunoreactive myofibroblasts in the portal tracts (Fig. 1A, portal tract portions in α pSmad3L and α pSmad3C columns; Fig. 1B, portal tract portions in α pSmad3L and αSMA columns).
Besides myofibroblasts, hepatocytes in HCV-infected livers exhibited phosphorylation at Smad3L. pSmad3L was observed particularly in groups of hepatocytes adjacent to collagen fibers in portal tracts (Fig. 1A and B, liver lobule portion in α pSmad3L column). Thus, hepatocytes were regulated by the same pSmad3L pathway during chronic inflammation as in myofibroblasts. Extent of phosphorylation at Smad3L decreased in the hepatocytes distant from portal tracts. Distribution of pSmad3C presented a sharp contrast to that of pSmad3L (Fig. 1A, liver lobule portion in α pSmad3C column): pSmad3C was predominantly located in hepatocytic nuclei distant from portal tracts.
Hepatic Fibro-carcinogenesis: Reciprocal Change in pSmad3L and pSmad3C Pathways.
We next sought to discriminate between tumor-suppressive and fibrogenic Smad3 signaling by staining sections using anti-pSmad3L Ab and anti-pSmad3C Ab, paired with sections stained for plasminogen activator inhibitor 1 (PAI-1) and p21WAF1, respectively. By upregulating p21WAF1 transcription, pSmad3C participates in tumor-suppression,24 whereas the pSmad3L-mediated signaling promotes ECM deposition, partly, by upregulating PAI-1 transcription.25
In specimens from a patient with chronic hepatitis C, the above distribution of pSmad3L fit well with the pattern shown by PAI-1 immunolabeling (Fig. 1C, chronic hepatitis panels in α pSmad3L and α PAI-1 columns), with both strongly apparent in groups of hepatocytes adjacent to collagen fibers in portal tracts. Hepatocytes rather than other cells thus appeared the cells primarily responsible for PAI-1 expression associated with chronic inflammation. Amounts of linker phosphorylation and PAI-1 staining increased further in hepatocytes in cirrhosis (Fig. 1C, cirrhosis panels in α pSmad3L and α PAI-1 columns). The spatial distribution profiles differed between chronic hepatitis and cirrhosis. In particular, pSmad3L and PAI-1 were observed in essentially all hepatocytes in cirrhotic liver. Moreover, linker phophorylation and PAI-1 staining increased further as chronic liver disease progressed to HCC (Fig. 1C, HCC panels in α pSmad3L and α PAI-1 columns).
Similarly to pSmad3C distribution, hepatocytes showed increased p21WAF1 staining in chronic hepatitis C specimens (Fig. 1D, chronic hepatitis panels in α pSmad3C and α p21WAF1 columns). In contrast to intense staining for pSmad3L and PAI-1, pSmad3C and p21WAF1 staining decreased in hepatocytic nuclei in cirrhotic liver (Fig. 1D, cirrhosis panels in α pSmad3C and α p21WAF1 columns), and it was sparse in HCC (Fig. 1D, HCC panels in α pSmad3C and α p21WAF1 columns).
Double immunofluorescence studies in chronic hepatitis C and HCC specimens confirmed that pSmad3L and pSmad3C were co-localized in PAI-1- and p21WAF1-immunoreactive hepatocytes and cancer cells, respectively (Fig. 1E,F). We then quantified extent of phosphorylation at Smad3L and Smad3C by immunoblotting. Remarkable upregulation of pSmad3L was seen whereas pSmad3C gradually decreased with progression of hepatic fibro-carcinogenesis (Fig. 1G).
Taken together, the fibrogenic pSmad3L/PAI-1 pathway in hepatocytes came to predominate whereas the tumor-suppressive pSmad3C/p21WAF1 pathway became quiescent as chronic hepatitis progressed to cirrhosis and then HCC.
Increased Extent of Hepatocytic Phosphorylation at Smad3L in Proportion to Fibrotic Stage and Necroinflammatory Activity in HCV-Related Chronic Liver Disorders.
Histological evaluation, which is useful in assessing the stage of chronic hepatitis C and has significant prognostic and therapeutic implications,1 includes separate considerations of fibrosis and necroinflammatory activity.18 In addition, we semiquantitatively scored hepatocytic Smad3 phosphorylation from 0 to 4 (Fig. 2A). Table 1 shows clinical backgrounds and positivities for pSmad3C and pSmad3L of HCV-related chronic liver diseases. We selected the patients with early HCC. In highly fibrotic livers (F3 to 4), phosphorylation at Smad3L was significantly greater than that in livers with mild fibrosis (F1 to 2) (Fig. 2B). Moreover, Smad3L phosphorylation in HCC was greater than that in cirrhotic liver (F4). In contrast, Smad3C showed less phosphorylation in cirrhotic liver and HCC than that in livers with grade F1 to 2 fibrosis (Fig. 2C). These findings indicated a close relationship between domain-specific phosphorylation of Smad3 and carcinogenic state of livers with advanced fibrosis.
Table 1. Clinicopathologic Features and Smad3L/C Phosphorylation in Specimens from Patients with HCV-Related Chronic Liver Disease
We also examined the relationship between pSmad3L and necroinflammatory activity during hepatocytic fibrocarcinogenesis. Among 60 HCV-infected livers with stages F1 to F3 fibrosis, Smad3L phosphorylation in hepatocytic nuclei was greater in highly active hepatitis (A2 to 3) than that in less active hepatitis (A0 to 1; Fig. 2D). Collectively, the extent of Smad3L phosphorylation increased in proportion to fibrotic stage, necroinflammatory activity, and HCC development in HCV-related chronic liver disorders.
Interleukin 1β (IL-1β)-Associated Shift of TGF-β Signaling from the Tumor-Suppressive pSmad3C/ p21WAF1 Pathway to the JNK/pSmad3L/PAI Pathway in Cultured Rat Hepatocytes.
In chronic hepatitis C specimens, Smad3 was phosphorylated at the linker region mainly in groups of hepatocytic nuclei adjacent to collagen fibers in portal tracts (Fig. 1A and B, α pSmad3L panel). During inflammation, IL-1β acts as a key cytokine to mediate the immune response.26 In addition, serum level of IL-1β is elevated in patients with chronic hepatitis C.27 In particular, IL-1β as well as TGF-β have been shown to be released from infiltrating macrophages in portal tracts during chronic inflammation.28 These findings suggest that elevated IL-1β might alter hepatocytic TGF-β signaling in chronic hepatitis C.
We investigated this hypothesis using rat cultured hepatocytes.21 Additional treatment of IL-β with TGF-β increased hepatocytic phosphorylation at Smad3L (Fig. 3A, α pSmad3L panel), PAI-1 transcription (Fig. 3B, PAI-1 panel), and invasion (Fig. 3C) above those of TGF-β or IL-1β treatment alone, suggesting that shared molecules activated by TGF-β and IL-1β signals might activate pSmad3L/PAI-1 pathway. In this respect, we previously reported that Smad3L could serve as a substrate for JNK.15 Accordingly, we investigated whether JNK activity affected these effects. Pretreatment of the hepatocytes with a JNK inhibitor SP600125 reduced the subsequent increase in pSmad3L, PAI-1 transcription, and hepatocytic invasion triggered by TGF-β and/or IL-1β stimulation, suggesting a direct role of the JNK/pSmad3L/PAI-1 pathway in facilitating cell invasion in response to TGF-β and IL-1β stimulation.
Conversely, treatment with TGF-β plus IL-1β did not induce as much pSmad3C (Fig. 3A, α pSmad3C panel) and p21WAF1 transcription (Fig. 3B, p21WAF1 panel) as TGF-β treatment alone. IL-1β partially blocked the antiproliferative effect of TGF-β (Fig. 3D). In addition, SP600125 blocked the inhibitory effect of IL-1β on TGF-β-induced Smad3C phosphorylation and p21WAF1 transcription. Moreover, pretreatment with SP600125 significantly enhanced the antiproliferative effect of TGF-β and lowered cell proliferation on IL-1β stimulation. These results indicated that the JNK pathway antagonized the antiproliferative effect of TGF-β via the pSmad3C/p21WAF1 pathway in the cultured hepatocytes. Not only IL-1β but also tumor necrosis factor alpha as another Th1 cytokine shifted from a tumor-suppressive pSmad3C pathway to a fibrogenic pSmad3L/PAI-1 pathway (data not shown).
Hepatocytes Strongly Positive for pSmad3L in Chronic Hepatitis C and Subsequent Advancement to HCC.
Finally we investigated whether pSmad3L-positive hepatocytes in chronic hepatitis C would progress to HCC. We reassessed the patients with chronic hepatitis C with stage F2 to F3 fibrosis. The relationship between pSmad3/L positivity and clinicopathological characteristics of the patients is shown in Table 2. HCC was detected within 12 years in 8 of 14 patients with ample Smad3L phosphorylation (scores 3-4), whereas only 1 of 12 patients with little Smad3L phosphorylation (scores 0-2) developed HCC (Fig. 4). Thus, occurrence of HCC within 12 years was much more frequent in patients with abundant Smad3L phosphorylation in hepatocytes than that in those showing sparse Smad3L phosphorylation (log rank = 0.008). Our semiquantitative analyses have the potential to provide data that hepatocytic pSmad3L in HCV-infected livers could be clinically useful biomarker to predict risk of HCC occurrence in the future.
Table 2. Clinicopathologic Features, Smad3L/C Phosphorylation, and HCC Incidence in Specimens from Patients with HCV-Related Chronic Hepatitis (F2 to F3)
Chronic inflammation associated with persistent HCV infection is clearly the primary inducer of liver fibrosis and cancer. However, little information is available on the molecular mechanisms by which chronic inflammation causes progressive liver fibrosis and ultimately HCC. Although myofibroblasts transdifferentiated from HSCs entrapped in the stroma have historically been considered the primary cells involved in development of liver fibrosis,3 possible direct involvement of hepatocytes in fibrosis has not been examined. In parallel with emergence of the EMT paradigm in fibrosis and carcinogenesis, a large body of work has established roles for epithelial cells as important mediators of progressive fibrosis and carcinogenesis.7 During progression of HCV-related chronic liver disorders, our current data indicated that hepatocytes affected by chronic inflammation undergo transition from the tumor-suppressive pSmad3C pathway, which is characteristic of mature epithelial cells, to the JNK/pSmad3L pathway, which appears to favor the state of flux shown by myofibroblasts, accelerating liver fibrosis while increasing risk of cancer (Fig. 5).
Constitutively phosphorylated Smad3L was observed in premalignant lesions including cirrhosis as well as in HCC (Fig. 1C). Because JNK was found to be phosphorylated constitutively and activated in the early stages of HCC,29 constitutive Smad3L phosphorylation in cirrhosis and HCC could be a direct consequence of JNK signaling. JNK acts as an important regulator of TGF-β signaling by increasing the basal amount of pSmad3L available for fibrogenic action in hepatocytic nuclei, in the meantime shutting down TGF-β-dependent tumor-suppressive activity by pSmad3C (Fig. 5). Loss of an epithelial homeostasis and acquisition of a migratory, mesenchymal phenotype are essential for tumor invasion. In this context, JNK activity is important for pSmad3L-dependent signaling that might interact with other oncogenic pathways, especially activator protein 1, to maintain a mesenchymal phenotype of hepatocytes.12 This mechanism could explain why several studies indicate a correlation between Smad3 and invasiveness as well as metastatic potential in human cancers.22
Exploring the shift from epithelial to mesenchymal TGF-β signaling during human hepatic fibro-carcinogenesis appears to offer new direction for diagnostic and therapeutic approaches. This new paradigm of the JNK/pSmad3L-mediated signaling in chronic liver disorders should prove useful in future investigations of other cancers associated with chronic inflammation, such as gastric cancer.30 From the viewpoint of TGF-β signaling, a key therapeutic aim in chronic liver disorders would be restoration of the lost tumor-suppressor function observed in normal hepatocytes at the expense of effects promoting hepatic fibro-carcinogenesis. Our current model suggests that specific inhibitors of the JNK/pSmad3L pathway could suppress progression of both liver fibrosis and cancer. In molecularly targeted therapy for human fibrosis and cancer, pSmad3L and pSmad3C could be assessed as biomarkers to evaluate the likely benefit from specific inhibition of the JNK/pSmad3L pathway.