Epigallocatechin Gallate Induces Hepatic Stellate Cell Senescence and Attenuates Development of Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) incidence has more than tripled since 1980, with 50,000 new cases predicted annually in the United States (1). HCC is a highly morbid condition with a 5-year survival less than 15%. Surgical resection, liver transplant, and radiofrequency ablation are considered the only curative therapies. However, the majority of patients are diagnosed at an advanced stage, and less than 20% of new cases are considered potentially curable. In addition, HCC is minimally responsive to standard chemotherapeutic agents (2). Until recently, sorafenib was the only drug clinically approved by FDA for treatment of advanced-stage HCC, even though sorafenib has been shown to provide only a modest survival benefit (3, 4).

Given the dismal outlook for HCC therapy, there has been increased interest in chemopreventive strategies. HCC arises from a milieu of chronically inflamed and damaged liver tissue, and thus there is a strong association between cirrhosis and HCC risk (5, 6). Typically, cirrhosis onset occurs years before a diagnosis of HCC, therefore, chemoprevention is highly applicable in this context given the large window of time for intervention.

Although HCC derives from hepatocytes, there is a strong association between liver fibrosis and HCC risk; therefore, stromal cells may play a role in mediating hepatocarcinogenesis (7–9). Hepatic stellate cells (HSC), the primary stromal cell type of the liver, transition to an “activated” myofibroblast-like state upon injury characterized by expression of smooth muscle actin and deposition of extracellular matrix components including collagens and matrix metalloproteinases (MMP). Persistent injury, as in the case of cirrhosis, may result in a state of chronic inflammation and tonic HSC activation, resulting in a pathologic fibrotic response (10). In addition, activated HSCs have also been shown to provide feed-forward stimuli by production of cytokines and chemokines that recruit inflammatory cells to the site of injury, raising concern that persistent HSC activation may support a protumorigenic environment (11). Accordingly, it has been predicted that direct inactivation of HSCs, or senescence, beneficially influences the dynamic balance of inflammation, fibrosis progression, and HCC development, and may therefore provide a useful chemopreventive strategy (12).

Experimental and epidemiologic studies suggest tea polyphenols may have chemopreventive effects in variety of cancers including HCC (13, 14). Green tea is a rich source of natural polyphenols, which are commonly known as catechins, and epigallocatechin gallate (EGCG) is the most abundant catechin in green tea (comprising 50%–80% of the total pool). Several studies have reported the inhibitory effect of EGCG on cancer cell proliferation through inhibition of receptor tyrosine kinases and their downstream signaling pathways (15). EGCG has also been reported to induce cell-cycle arrest through expression of p21, p18, and p53, and inhibition of cyclins and cyclin-dependent kinases (15, 16). Finally, EGCG has been shown to reduce liver inflammation and fibrosis (17). Therefore, we hypothesized that EGCG would have chemopreventive effects in chronic liver injury through its negative regulation of inflammation, fibrosis, and cellular proliferation.

All animal experiments were approved by the Massachusetts General Hospital Institutional Animal Care and Use Committee. Animals were maintained on a 12-hour light/dark cycle at 25°C and fed a standard diet. To induce liver injury, cirrhosis, and HCC, rats were given weekly injections of low dose diethylnitrosamine (DEN; 50 mg/kg, i.p., Sigma-Aldrich) according to our established protocols (18–20). Diethylnitrosamine is a commonly used carcinogen for the liver, and the resultant tumors are known to genetically resemble human HCC tumors with poor prognosis (21, 22). We have also previously shown that diethylnitrosamine-induced liver injury has a similar progression to human disease, albeit on an accelerated course, first with the appearance of fibrosis at 8 weeks, then advanced fibrosis by 12 weeks, and marked liver cirrhosis with HCC at 18 weeks. Male Wistar rats (Charles River Laboratories) were given weekly intraperitoneal injections of 50 mg/kg of diethylnitrosamine for 18 weeks. A subset of rats received EGCG in their drinking water at concentration of 0.02% (0.02 g/100 mL) beginning at the 13th week when fibrosis is established but before the onset of HCC. Treatment was continued for 6 weeks until the study endpoint. On average, rats drink 50 mL of water per day, which means that the average rat in our study consumed 10 mg of EGCG per day. Because the average rat weighed 400 g, our relative dose in this study was approximately 25 mg/kg. Using the calculation, human equivalent dose (HED) (mg/kg) = animal dose (mg/kg) × (animal K_m/human K_m), we calculated the HED at 5.4 mg/kg. Assuming an average human weight of 80 kg, our dose would be approximately 432 mg per day.

Formal-fixed, paraffin embedded tissues samples were sectioned with 5 μm thicknesses and stained with hematoxylin and eosin, Masson trichrome, and antibodies for alpha-smooth muscle actin (α-SMA, ACTA2; 1:100, Sigma-Aldrich) and collagen, type 1, alpha 1 (COL1A1; 1:200, Abcam) according to standard protocols as reported previously (18). A board-certified pathologist scored the fibrosis according to the method of Ishak in a blinded manner.

A cardiac terminal blood withdrawal was performed at the time of sacrifice. Blood was allowed to clot for 2 hours at room temperature before centrifugation at 2,000 rpm for 10 minutes at 4°C. Serum was isolated and stored at −80°C. Biochemical markers of liver injury were measured, including alkaline phosphatase (ALP), alanine transaminase (ALT), aspartate transaminase (AST), total bilirubin (TBIL), and albumin (DRI-CHEM 4000 Analyzer, Heska).

Liver tissue (0.01–0.05 g) was lysed with RIPA Buffer (Sigma-Aldrich) containing protease and phosphatase inhibitors. The isolated lysates were cleared by centrifugation and protein concentrations were measured with the BCA Protein Assay Kit (Thermo Fisher Scientific). Twenty milligram of total protein was separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane. Nonspecific binding was blocked by 10% skim milk in TBS supplemented with 0.1% Tween 20 (TBST). Immunoblots were performed using antibodies raised against α-SMA (1:2,000, Sigma-Aldrich) and β-actin (1:10,000, Abcam). Immune complexes were visualized by enhanced chemiluminescence (Pierce).

Senescence-associated β-galactosidase (SA-β-gal) activity was measured by in situ staining using a chromogenic substrate under acidic conditions (23). Liver sections were rinsed twice in PBS and then incubated at 37°C in a freshly prepared X-gal solution containing 1 mg/mL X-gal (5-bromo-4-chloro-3-indolyl P3-D-galactoside; Sigma-Aldrich), 150 mmol/L sodium phosphate pH 6.0, 5 mmol/L potassium ferrocyanide, 5 mmol/L potassium ferricyanide, 50 mmol/L NaCl, and 2 mmol/L MgCl. Slides were incubated for 16 hours.

TWNT4 cells were cultured in DMEM supplemented with 100 μg/mL streptomycin, 100 units/mL penicillin, and 2 mmol/L l-glutamine (all from Cellgro) and 10% FBS (Atlanta Biologicals). Cells were maintained at 37°C in an atmosphere containing 5% CO₂ and 100% humidity. TWNT4 cells were seeded (10⁴ cells/well) on a 24-well plate for 12 hours to attach. Cells were treated with increasing concentrations of EGCG. After 72 hours, 20-μL MTT solution (0.25 g/mL; Sigma-Aldrich) was added, and cells were incubated at 37°C for 4 hours. Another 100 μL of DMSO (Sigma-Aldrich) was added to each well to dissolve the purple formazon crystals after the removal of the MTT solution, and the plate was gently shaken at room temperature for 20 minutes. The optical density was measured at 570 nm using a Microplate Reader (Molecular Devices Emax).

Total RNA isolation, amplification, and normalization of data were done as described previously (18). Primer and probe sequences for qPCR were purchased from Applied Biosystems Inc.

Images were captured with a Nikon Eclipse Microscope equipped with an Insight CMOS 5.1 digital camera. For quantification, at least 2,000 cells were counted per sample using the cell counter plugin in ImageJ.

Total RNA was extracted from liver tissue using TRizol (Invitrogen) according to the manufacturer's instructions and subsequently treated with DNase I (Promega). Genome-wide gene expression profiling was performed using RatRef-Expression BeadChip Microarrays (Illumina). Scanned data were normalized using cubic spline algorithm, summarized into official gene symbol, and mapped to human orthologous genes using the mapping table provided by the Jackson Laboratory (www.informatics.jax.org). Genes on the microarray were rank-ordered according to differential expression between the experimental conditions. Over- or underrepresentation of each gene signature on the rank-ordered gene list was evaluated on the basis of random permutation test. FDR < 0.25 was regarded as statistically significant. All microarray data analysis was performed using GenePattern Analysis Toolkit (www.genepattern.org) and R statistical computing language (https://www.r-project.org/).

EGCG treatment grossly limited the development of HCC nodules (diethylnitrosamine 18.75 ± 1.76 vs. DEN + EGCG 9.125 ± 1.922; P < 0.001; Fig. 1A and B), and was associated with a significant reduction in the liver/body weight ratio (diethylnitrosamine 6.964% ± 0.5097% vs. DEN + EGCG 5.377% ± 0.4375%; P < 0.05; Fig. 1C). EGCG was well-tolerated and mitigated diethylnitrosamine toxicity–related weight loss (diethylnitrosamine 380.5 ± 13.23 g vs. DEN+EGCG 439.2 ± 15.79 g; P < 0.05; Fig. 1D). Commensurate with these gross findings, EGCG treatment also improved biochemical markers of liver function and injury, including TBIL, ALP, AST, and ALT, which were all decreased (Fig. 1E–H).

To better understand the mechanism of EGCG-mediated HCC chemoprevention, we performed transcriptomic analysis of nontumoral, hepatic tissue from diethylnitrosamine controls and EGCG-treated rats. Hoshida and colleagues previously reported a 186-gene signature derived from cirrhotic liver tissue that can discriminate patients by future HCC risk (24, 25). Diethylnitrosamine treatment resulted in transcriptomic changes that mirror Hoshida's poor prognosis signature. In contrast, EGCG treatment resulted in prevention of this signature toward a “good prognosis” expression profile (Fig. 1I). The significantly upregulated and downregulated gene clusters in EGCG-treated rats are shown in Table 1. Good prognosis genes enriched in EGCG-treated animals involve networks that participate in normal liver function such as organic acid metabolism, lipid and fatty acid metabolism and biosynthesis, amino acid metabolism, catabolic processes, and cellular catabolism. In contrast, the poor prognosis gene networks induced by diethylnitrosamine treatment are primarily involved in wound healing and fibrosis progression, including pathways related to cell adhesion, defense mechanisms, and coagulation. Importantly, these networks were significantly downregulated after EGCG treatment.

EGCG treatment has previously been shown to have antifibrotic properties in rodent models of liver injury. In our study, EGCG treatment for 6 weeks was associated with decreased progression to cirrhosis, evidenced by reduced type I collagen staining in liver tissue (EGCG 2.64 ± 0.40 vs. diethylnitrosamine 15.78 ± 1.326; P < 0.001; Fig. 2A and B). Pathologic scoring of fibrosis using Masson trichrome staining also demonstrated significant improvement in the median Ishak score in EGCG-treated rats compared with untreated diethylnitrosamine controls [4.4, interquartile range (IQR) 2–6 compared with 5.8, IQR 5–6; P < 0.05; Fig. 2C and D].

Upon diethylnitrosamine injury, quiescent HSCs transform to activated α-SMA–positive myofibroblasts that proliferate and promote fibrogenesis. Diethylnitrosamine-injured rats that received 0.02% EGCG in their drinking water had significant reductions in immunostaining of α-SMA on paraffin-embedded liver samples compared with diethylnitrosamine controls, suggesting decreased HSC activation (Fig. 3A and B). This observation was corroborated by Western blot analysis, in which we again observed less α-SMA protein expression in livers isolated from EGCG-treated animals (Fig. 3C). Given the decrease in stromal activation, we investigated whether EGCG induced HSC senescence by measuring HSC SA-β-gal activity (26–28). SA-β-gal activity was increased in EGCG-treated liver tissue as compared with untreated diethylnitrosamine controls, indicating higher levels of senescence (Fig. 3D and E). SA-β-gal–positive cells were specifically located in the fibrotic bands where HSCs are abundant, but not in the surrounding hepatic tissue (Fig. 3D), indicating that EGCG induced senescence in HSCs but not hepatocytes. In addition, proliferating cell nuclear antigen (PCNA), a marker for cellular proliferation, was significantly decreased in EGCG-treated livers compared with untreated diethylnitrosamine controls (Fig. 3F).

A gene signature for activated HSCs was previously established and verified across multiple human datasets and animal models (29, 30). We performed microarray analysis of liver tissue from diethylnitrosamine and DEN+EGCG-treated animals and compared expression profiles with the activated HSC signature. EGCG treatment was associated with HSC inactivation, evidenced by loss of overlap with the HSC activation signature compared with untreated diethylnitrosamine controls (Fig. 3G). Taken together, these findings suggest that EGCG both reduced HSC activation and induced senescence in the setting of diethylnitrosamine liver injury.

To gain more insight into EGCG-induced HSC senescence, we treated TWNT4 immortalized human HSCs in vitro with EGCG. EGCG treatment at concentrations above 20 μmol/L significantly reduced cell proliferation compared with untreated controls by MTT assay. This growth reduction did not appear to be a toxic effect, as cells remained viable in the EGCG treatment groups on microscopic evaluation (Fig. 4A and B). Other general hallmarks of senescence include the absence of proliferative markers such as Ki67, and increased expression of tumor suppressors (P53) and cell-cycle inhibitors (P16 and P21). Consistent with these observations, 20 μmol/L EGCG treatment was also associated with significant reduction (≈1.5-fold) in Ki67 expression (control 48.59% ± 2.523% vs. EGCG 31.24% ± 2.039%; P < 0.001; Fig. 4C and D), and significant increases in cell-cycle inhibitor gene expression, including P16, P21, and P53 (Fig. 4G). Finally, SA-β-gal activity also increased (≈4-fold) in TWNT4 cells after 72 hours of treatment with 20 μmol/L EGCG (control 22.52 ± 1.126 vs. EGCG 62.64 ± 1.255; P < 0.001; Fig. 4E and F). These findings corroborate our in vivo observation that EGCG induces HSC senescence.

Finally, previous reports have identified the presence of a unique secretory phenotype specifically associated with senescence in HSCs, which is characterized by reduced expression of extracellular matrix (ECM) proteins, increased expression of ECM-degrading enzymes, and increased secretion of inflammatory cytokines, referred to as the senescence-associated secretory phenotype (SASP; ref. 31). In our study, compared with untreated cells, EGCG-treated TWNT4 cells exhibited a similar secretory profile upon undergoing senescence in vitro, including: reduced expression of collagen type 1 (COL1A1), collagen type 3 (COL3A1), and fibronectin (FN; Fig. 4H); increased expression of MMP-1 (MMP1) and -3 (MMP3); and increased expression of immune surveillance chemokines, IL6 (IL6) and -8 (IL8; Fig. 4I and J). These findings therefore suggest that EGCG-induced senescence in HSCs also leads to initiation of SASP.

To further demonstrate that EGCG-induced senescence is specific to HSCs, we treated the human HCC cell lines HepG2 and Hep3B with 20 μmol/L EGCG for 72 hours. EGCG treatment had no effect on SA-β-gal activity in HepG2 cells under normal growth conditions (vehicle 1.512% ± 0.19% vs. EGCG 1.231% ± 0.20%; Fig. 5A and B). In addition, no difference in senescence-associated gene expression was observed (Fig. 5C). Similarly, although Hep3B cells have more SA-β-gal activity (<15%) at baseline; no difference was observed with EGCG treatment (Fig. 5D and E). Finally, we did not observe any change in gene expression of the tumor suppressors, P16 or P21, after EGCG treatment (Fig. 5F).

Prior studies have shown a protective effect of the main constituent of green tea, EGCG, against liver injury, oxidative stress, inflammation, and fibrosis in experimental animal models of liver diseases (32–35). In this study, we recapitulate the observed antifibrotic effect in a diethylnitrosamine liver injury model and provide a novel mechanism through stellate cell-specific senescence. In addition, we report that EGCG treatment is also chemopreventive for the development of HCC after diethylnitrosamine injury.

Prior studies evaluating the hepatoprotective effects of EGCG often relied on high drug doses that have not been shown to be safe in humans. High doses of EGCG are hepatotoxic in mice (36), and at least 200 cases of EGCG-related hepatic toxicity have been reported since 1996, likely due to growing consumer interest in herbal supplements in the United States (36, 37). In contrast, the EGCG dose used in this study was designed to be equivalent to a human oral dose of 400 mg per day, which had been demonstrated to be safe and tolerable in a human single-dose ascending study (38). More recent results in humans have demonstrated elevations of ALT and AST after 12 months of green tea extract consumption (1,315 ± 115 mg total catechins including 843 ± 44 mg of EGCG per day) in a placebo-controlled, double-blinded phase II clinical trial assessing breast cancer risk biomarkers in healthy postmenopausal women (39). These elevations occurred in a small proportion of participants and were reversible. Therefore, we believe our results represent the first report showing clinically achievable and safe doses of EGCG can prevent fibrosis progression and prevent HCC development.

Multiple previous studies have demonstrated an association between hepatic fibrosis and HCC risk (18, 40, 41). Although the etiologies of liver disease have specific risks for development of HCC, most patients sequentially develop hepatitis, fibrosis, cirrhosis, and then HCC (42). In 80%–90% of patients, HCCs develops in fibrotic or cirrhotic environment and patients with a high serologic fibrosis index (FIB-4) have an up to 15-fold increased risk for future HCC incidence (43). In addition, one-third of patients with cirrhosis develop HCC within 5 years (42). Fibrogenesis in liver is mediated by HSCs, which deposit ECM molecules and recruit immune cells to the site of injury (44, 45). We previously reported that HSC inactivation prevents fibrosis progression, which is associated with a reduction in HCC development in rodents (18, 46). In this study, EGCG reduced fibrosis progression through induction of senescence in HSCs, which resulted in improved liver function and reduced HCC development

It has previously been shown that senescent HSCs develop a unique secretory phenotype known as the SASP. We observed a similar phenotype in HSCs in vivo and TWNT4 cells in vitro with EGCG treatment, which resulted in reduced proliferation as measured by Ki67 staining and MTT assay, and increased expression of senescence markers including increased SA-β-gal staining, and upregulation of tumor suppressor genes including P16, P21, and P53. TWNT4 cells also increase the expression of immunosurveillance cytokines (IL6 and IL8), which are known to induce recruitment of natural killer cells, regulatory T lymphocytes, and macrophages to remove matrix debris and activated HSC (45, 47).

Interestingly, other reports have shown that the increased SASP from senescent HSC can induce HCC development (48, 49). These studies concluded that while cellular senescence usually provides a barrier to epithelial transformation, senescence in mesenchymal cells, like HSCs, could support cancer growth through the SASP by releasing cytokines like IL6. Most of these studies were performed in the context of obesity-associated HCC in animal models without significant fibrosis. The discrepancy in these studies could be explained by differences in the senescence triggers, the composition of microenvironment, and extent of senescence surveillance in different animal models (48, 50).

Other possible mechanisms could explain the reduction of HCC development after EGCG treatment. This includes reversibility of mutation load that is caused by diethylnitrosamine. In a high-fat diet (HFD) mouse model of HCC, theaphenon E, a standardized green tea extract formulation, blocked HCC formation via prevention of lipid peroxidation–derived mutagenic cyclic DNA-adduct (γ-OHPdG) formation resulting in decreased mutation load particularly G>T, the most common mutation in human HCC (51). Consistent with these results, theaphenon E reduced HCC incidence in xeroderma pigmentosum group A–knockout mice and diethylnitrosamine-injected mice and the reduction in HCC incidence was associated with decreased (γ-OHPdG) levels (52). In addition, using a microarray approach we showed that after EGCG treatment the pathways/genes involved in normal liver metabolic functions such as amino acid and lipid metabolism were markedly upregulated, and at the same time, pathways that are involved in wound healing and immune system response were downregulated. These mechanisms will need to be further evaluated in future studies.

Our data therefore suggests that the cancer prevention effects of EGCG-induced fibrosis regression, including the upregulation of ECM remodeling and inflammatory cytokines that contribute to the clearance of senescent cells and resolution of fibrotic tissue, outweigh any procancer effects of the SASP at least in the setting of cirrhosis. It is also possible that the SASP response needs to be combined with appropriate environmental cofactors, like DCA, which is produced by the gut microbiota in HFD and required for HCC development (48).

Green tea is very popular and mainly consumed in Asia. Several meta-analyses have been performed on Asian populations to investigate the association between green tea consumption and the risk for liver cancer (53–55). While some of these studies have shown no association with cancer risk (56), others have shown a decreased risk of liver cancer with green tea consumption especially in people who drink more green tea for longer periods of time (54–56). Given the slow progression of cirrhosis to HCC development, green tea consumption represents a potentially safe and cheap chemopreventive strategy for HCC.

K.K. Tanabe reports a patent to Prevention of fibrosis and hepatocellular carcinoma issued and research support from Enanta Pharmaceuticals and Zafgen to study drugs to prevent HCC. B.C. Fuchs reports receiving grants from American Institute for Cancer Research during the conduct of the study; grants from Blade Therapeutics, Collagen Medical, and Enanta Pharmaceuticals, and personal fees from Gilead Sciences and Ferring Pharmaceuticals outside the submitted work. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M. Sojoodi, L. Wei, D.J. Erstad, M. Lanuti, K.K. Tanabe, B.C. Fuchs

Development of methodology: M. Sojoodi, L. Wei, K.K. Tanabe, B.C. Fuchs

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Sojoodi, L. Wei, S. Yamada, T. Fujii, R.S. Kim, G.Y. Lauwers, Y. Hoshida, B.C. Fuchs

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Sojoodi, L. Wei, D.J. Erstad, H. Hirschfield, R.S. Kim, Y. Hoshida, K.K. Tanabe, B.C. Fuchs

Writing, review, and/or revision of the manuscript: M. Sojoodi, L. Wei, D.J. Erstad, T. Fujii, Y. Hoshida, K.K. Tanabe, B.C. Fuchs

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Wei, T. Fujii, H. Hirschfield, K.K. Tanabe, B.C. Fuchs

Study supervision: K.K. Tanabe, B.C. Fuchs

This work was supported by an Investigator-Initiated Grant from the American Institute for Cancer Research (to B.C. Fuchs).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.