Lack of interleukin-6/glycoprotein 130/signal transducers and activators of transcription-3 signaling in hepatocytes predisposes to liver steatosis and injury in mice


  • Potential conflict of interest: Nothing to report.


A deregulated cytokine balance is involved in triggering the sequence from steatosis to nonalcoholic steatohepatitis, ultimately leading to liver fibrosis and cancer. To better define the role of proinflammatory interleukin-6 (IL-6)-type cytokines in hepatocytes we investigated the role of IL-6 and its shared receptor, glycoprotein 130 (gp130), in a mouse model of steatohepatitis. IL-6−/− mice were fed a choline-deficient, ethionine-supplemented (CDE) diet. Conditional gp130 knockout and knockin mice were used to achieve hepatocyte-specific deletion of gp130 (gp130Δhepa), gp130-dependent rat sarcoma (Ras)-(gp130ΔhepaRas), and signal transducers and activators of transcription (STAT)-(gp130ΔhepaSTAT) activation. CDE-treated IL-6−/− mice showed a significant hepatic steatosis at 2 weeks after feeding. The mice rapidly developed elevated fasting blood glucose, insulin serum levels, and transaminases. To better define IL-6-dependent intracellular pathways, specifically in hepatocytes, we next treated gp130Δhepa mice with a CDE diet. These animals also developed a marked steatosis with hyperglycemia and displayed elevated insulin serum levels. Additionally, gp130Δhepa animals showed an imbalanced inflammatory response with increased hepatic tumor necrosis factor-alpha and decreased adiponectin messenger RNA levels. Dissecting the hepatocyte-specific gp130-dependent pathways revealed a similar disease phenotype in gp130ΔhepaSTAT mice, whereas gp130ΔhepaRas animals were protected. In CDE-treated mice lack of gp130-STAT3 signaling was associated with immune-cell-infiltration, jun kinase-activation, a blunted acute-phase-response, and elevated transaminases. Furthermore, gp130Δhepa and gp130ΔhepaSTAT mice showed beginning signs of liver fibrosis compared to gp130ΔhepaRas mice and controls. Conclusion:During CDE treatment mice lacking IL-6 and gp130-STAT signaling in hepatocytes are prone to hepatic metabolic changes and inflammation. This ultimately leads to progressive steatohepatitis with signs of liver remodeling. Thus, the presented model allows one to further dissect the role of IL-6/gp130-type signaling in hepatocytes during fatty liver degeneration to define new therapeutic targets in metabolic liver diseases. (HEPATOLOGY 2010.)

The metabolic syndrome associated with hepatic steatosis, ultimately leading to steatohepatitis and hepatic fibrosis, is one of the fastest-growing medical problems in Western industrial countries.1 Simple fat storage in hepatocytes alone is not associated with disease progression. However, addition of an inflammatory element triggers liver remodeling and thus promotes liver fibrosis and cancer.2 Steatohepatitis affects 10%-20% of the population in developed Western countries, whereas the prevalence in obese patients is up to 55%-70%.3 The pathogenesis of nonalcoholic steatohepatitis (NASH) includes the intracellular accumulation of triglycerides, the development of insulin resistance, and the subsequent infiltration of inflammatory cells,4 which can eventually damage the liver. Therefore, understanding the basic mechanisms involved in switching from fat accumulation to steatohepatitis and further disease progression is important to define future treatment options.

Cytokines and growth factors have been described to be involved in the pathogenesis of steatohepatitis and NASH.5 Besides tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6) as a major proinflammatory cytokine has been implicated in modulating hepatic steatosis and other pathophysiological changes leading to insulin resistance and NASH development.6 Patients with NASH display elevated hepatic IL-6 levels and a positive correlation between IL-6 expression and the degree of inflammation and fibrosis has been shown.7 Overexpression of IL-6 in mice increases energy expenditure, leading to higher insulin levels and a pronounced proinflammatory state, but was insufficient to cause any signs of hepatic steatosis by itself.8 Additionally, the IL-6-dependent signal transducers and activators of transcription-s (STAT3) factor has recently been shown to be involved in the regulation of gluconeogenesis and provided protection against insulin resistance.9 However, a true causal relationship between IL-6-dependent signaling and disease progression in specific target cells has not been clearly demonstrated yet. Especially the contribution of hepatocytes versus inflammatory cells of the immune system is unclear.

IL-6 is a member of the IL-6 cytokine family, which uses glycoprotein 130 (gp130) as a common transdomain receptor triggering distinct intracellular pathways, e.g., STAT3/1 and rat sarcoma (Ras) activation.10 To investigate the role of IL-6 and gp130-dependent pathways in hepatocytes during dietary-induced hepatic steatosis with progression to steatohepatitis, we used hepatocyte-specific conditional gp130-knockout/knockin mice in a liver injury (choline-deficient ethionine-supplemented, CDE) model. Using the CDE model we demonstrate that IL-6 by way of gp130-STAT3 activates intracellular mechanisms in hepatocytes, which protect from hepatic fat accumulation, the development of sustained elevated blood glucose levels and insulinemia, and the induction of hepatic inflammation. Thus, this pathway prevents the occurrence and progression of fatty liver degeneration.


αSMA, alpha smooth muscle actin; APR, acute phase response; CDE, choline-deficient ethionine-supplemented; CK-19, cytokeratin 19; FAS, fatty acid synthase; gp130, glycoprotein 130; IL-6, interleukin-6; JNK, jun kinase; MCD, methionine choline deficient; NASH, nonalcoholic steatohepatitis; NF-κB, nuclear factor kappa light chain enhancer of activated B-cells; PPAR, peroxisome-proliferator-activated receptor; Ras, rat sarcoma; SAA, serum-amyloid-A; SOCS, suppressors of cytokine signaling; SREBP, sterol regulatory binding protein; STAT, signal transducers and activators of transcription; TNF-α, tumor necrosis factor-alpha.

Materials and Methods

Mouse Experiments.

Mice were housed in 12-hour light/dark cycles with free access to food and water and were treated in accordance with the criteria of the German administrative panel on laboratory animal care. Both knockout and wildtype animals showed a food intake of 5-6 g per day without differences between chow or dietary treatment. At least five mice were treated and analyzed in parallel. All experiments were repeated twice. The generation of gp130 mouse constructs is described in detail as Supporting Information.

For CDE treatment, mice were fed a choline-deficient chow (MP Biomedicals, Cat. No. 960210, Solon, OH). Drinking water was supplemented with 0.15% ethionine (DL-Ethionine-Sulfone, E-8251, Sigma-Aldrich, Hannover, Germany) as described.11 To determine dosing, different ethionine concentrations were applied.


Cryopreserved liver tissues were fixed in 4% paraformaldehyde (Roth, Karlsruhe, Germany), washed in phosphate-buffered saline (PBS) containing 0.02% sodium-azide (Roth), and blocked using 2% bovine serum albumin (BSA) in PBS-azide.

CD4(+) cells were detected using a directly fluorescein isothiocyanate (FITC)-labeled antibody (clone GK 1.5, Molecular Probes/Invitrogen, Karlsruhe, Germany). For the detection of CD11b (BD, Heidelberg, Germany) cells and of collagen (Santa Cruz, Heidelberg, Germany) we used secondary antibody detection systems. For anti-CK-19 immunofluorescence (Santa Cruz, Heidelberg, Germany) tissues were fixed in ice-cold acetone and blocked with a 1:5 dilution of blocking reagent (Promega, Mannheim, Germany) and detected by way of an antirabbit Alexa488-labeled secondary antibody (Molecular Probes/Invitrogen).


All numerical results are expressed as mean ± standard error (SE) and represent data from at least five animals per timepoint. All significant P-values were measured by Student's t test. A value of P < 0.05 was considered significant (*P < 0.05, **P < 0.01, ***P < 0.001).

Further Applied Methods.

sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), western blot, quantitative real-time polymerase chain reaction (PCR), insulin enzyme-linked immunosorbent assay (ELISA), the determination of hepatic triglyceride levels, and general histological procedures are described in the Supporting Information.


IL-6 Protects from Steatohepatitis During CDE Treatment.

IL-6 family members are involved in the regulation of hepatic metabolic activities and have implications for the development of dietary and inflammatory liver diseases. To directly investigate the participation of IL-6 on hepatic metabolism we first applied IL-6-knockout mice to a CDE diet. Unexpectedly, 2 weeks after CDE treatment we could detect a massive microvesicular steatosis in IL-6−/− mice (Fig. 1A) that was confirmed by oil-red staining (Fig. 1B) and by determining hepatic triglyceride levels (Fig. 1C). This effect was even aggravated under continued CDE feeding for 12 weeks (Fig. 1A-C). Moreover, it was associated with higher transaminases (Fig. 1D) and the infiltration of inflammatory cells (Supporting Fig. 4A), indicating the development of steatohepatitis. Additionally, IL-6−/− mice displayed significantly higher fasting blood glucose levels compared to controls (Fig. 1E), suggesting an impaired glucose tolerance.

Figure 1.

IL-6 knockout mice develop steatohepatitis after CDE feeding. (A) Hematoxylin and eosin (H&E) stainings were performed on liver sections derived from IL-6 knockout (IL-6−/−) and control mice 2 weeks and 12 weeks after CDE feeding. Marked liver steatosis was only detected in livers of IL-6−/− mice. (B) Oil-red stainings of liver sections are depicted before and after CDE or chow treatment. (C) Hepatic triglyceride levels were determined before, 2, and 12 weeks after CDE-feeding as indicated. Significantly higher hepatic triglyceride levels were evident in CDE-treated IL-6−/− animals (n = 5, **P < 0.01, ***P < 0.001). (D,E) Alanine aminotransferase (ALT) (D) and fasting blood glucose (E) levels were determined in IL-6−/− and control mice before, 2, and 12 weeks after CDE diet. Significantly higher ALT and fasting blood glucose levels were found in IL-6−/− mice after CDE treatment (n = 5, **P < 0.01, ***P < 0.001).

CDE Treatment Induces Hepatic Steatosis and Inflammation in Hepatocyte-Specific Conditional gp130 Knockout Mice.

As our initial findings suggested that IL-6 protects from insulin resistance and steatohepatitis after CDE feeding, we aimed to further investigate the contribution of IL-6-dependent pathways in hepatocytes. We therefore used conditional knockout mice carrying a hepatocyte-specific deletion of gp130, the common IL-6 family signal transducer12 (gp130Δhepa). Those mice and their respective controls (gp130loxP/loxP) were fed the CDE diet, which can induce variable signs of hepatic fat storage in wildtype mice.11

In our study gp130Δhepa mice developed progressive fat accumulation and inflammation (Fig. 2A; Supporting Fig. 4A), whereas controls did not show any significant steatosis until the 12th week of feeding. However, steatosis became clearly evident in wildtype mice at later timepoints (16 weeks of CDE-treatment; see Supporting Fig. 1). Fat storage was further confirmed by oil-red staining of intracellular fat (Fig. 2B) and by determining the hepatic triglyceride content, which was significantly higher in gp130Δhepa mice after 12 weeks of CDE treatment (Fig. 2C).

Figure 2.

Progression of hepatic steatosis in CDE-treated animals is associated with lack of gp130 expression in hepatocytes. (A) Gp130loxP/loxP and gp130Δhepa mice were fed with CDE or chow diet for 12 weeks. H&E stainings of liver sections were performed. Strong macrovesicular steatosis was only observed in the gp130Δhepa group fed with the CDE diet. The timepoint before starting CDE feeding is indicated as a control. (B) Oil-red stainings of liver sections are depicted before and after CDE or chow treatment. They revealed strong fat accumulation specifically in livers of gp130Δhepa mice 12 weeks after feeding. (C) Hepatic triglyceride levels were determined before and 12 weeks after feeding as indicated. Significantly higher hepatic triglyceride levels were evident in CDE-treated gp130Δhepa mice (n = 5, **P < 0.01).

Increased Hepatic Inflammatory Response and Elevated Blood Glucose Level After gp130 Depletion.

To test whether hepatic fat accumulation triggers liver injury and steatohepatitis, we determined transaminase levels after CDE treatment. Significantly higher transaminases were found in the gp130Δhepa group compared to controls (Fig. 3A). On a molecular level, messenger RNA (mRNA) expression of the proinflammatory cytokines IL-6 and TNF-α was elevated specifically in the gp130Δhepa CDE group (Fig. 3B,C). In contrast, the protective adipokine adiponectin was increased in gp130loxP/loxP and suppressed in gp130Δhepa mice (Fig. 3D).

Figure 3.

gp130Δhepa animals develop steatohepatitis after CDE feeding. (A) ALT serum levels were determined before or 4, 8, and 12 weeks after feeding chow or CDE diet. A significantly higher increase in ALT levels was found in the CDE group (n = 5, *P < 0.05). (B-D) IL-6 (B), TNF-α (C), and adiponectin (D) liver mRNA levels were determined by quantitative real-time PCR before and 12 weeks after feeding as indicated. Significant differences were found 12 weeks after CDE feeding in gp130Δhepa mice (n = 5, *P < 0.05, **P < 0.01).

We now aimed to investigate if mice with evidence of steatohepatitis also showed metabolic alterations. Gp130Δhepa mice displayed significantly increased fasting blood glucose levels (Fig. 4A) following CDE feeding. Insulin plasma levels (Fig. 4B) were strongly elevated in those animals, suggesting that gp130Δhepa mice developed insulin resistance. In order to evaluate whether this leads to an impaired glucose tolerance, fasting naïve and CDE-treated mice were challenged with glucose. Although non-CDE-treated mice did not differ, we observed higher fasting blood glucose and significantly higher blood glucose levels 240 minutes after stimulation (Fig. 4C).

Figure 4.

CDE treatment induces metabolic changes in gp130Δhepa mice. (A,B) Fasting blood glucose (A) and insulin plasma (B) levels were determined before and 12 weeks after feeding as indicated. Higher fasting glucose and insulin levels were detected in gp130Δhepa mice 12 weeks after CDE treatment (n = 5, *P < 0.05, **P < 0.01). (C) 1.5 g/kg bodyweight of 20% glucose was injected intraperitoneally after 12 weeks of chow CDE feeding. Blood glucose levels were determined before and 240 minutes after glucose injection (n = 5, *P < 0.05).

Gp130-Dependent Prevention of Fatty Liver Development and Inflammation Is Regulated by Way of gp130-STAT3-Induced Pathways.

As gp130 activates distinct intracellular pathways, we intended to define the cascade that mediates hepatocellular protection from steatohepatitis in the CDE model. In earlier experiments we generated hepatocyte-specific hybrid gp130 mice, which either specifically activate only gp130-dependent Ras (gp130ΔhepaSTAT) or STAT (gp130ΔhepaRas) signaling13 (Fig. 5A).

Figure 5.

Dissection of gp130-dependent signaling in hepatocytes. (A) A schematic cartoon of gp130-dependent signaling is depicted (left). After gp130 dimerization, STAT and ras/MAPK-signaling cascades are activated. By combining conditional knockout and knockin technology in hepatocytes, gp130 is completely deleted (gp130Δhepa) or only the STAT (gp130ΔhepaRas) or ras/MAPK (gp130ΔhepaSTAT) pathway can be activated. (B) H&E stainings of liver sections after 12 weeks of chow or CDE diet feeding. Extensive macrovesicular steatosis was only detected in gp130Δhepa and gp130ΔhepaSTAT CDE-fed mice. (C) Significantly higher hepatic triglyceride levels were found in gp130Δhepa and gp130ΔhepaSTAT mice 12 weeks after CDE treatment (n = 5, *P < 0.05). (D) ALT serum levels were determined at different timepoints after CDE treatment. Significantly higher ALT levels were evident in gp130Δhepa and gp130ΔhepaSTAT mice 12 weeks after CDE feeding (n = 5, *P < 0.05). (E) Fasting blood glucose levels are depicted in mice treated for 4 and 12 weeks with a CDE diet. Significantly higher glucose levels were found in gp130Δhepa and gp130ΔhepaSTAT mice (n = 5, *P < 0.05).

The data obtained with gp130-pathway-specific conditional knockout mice revealed that the occurrence and progression of hepatic steatosis and inflammation after CDE treatment were associated with an impaired activation of the gp130-STAT pathway in hepatocytes. Steatohepatitis as evidenced by histological fat accumulation (Fig. 5B), an increased hepatocyte triglyceride content (Fig. 5C), and transaminases (Fig. 5D) after 12 weeks of CDE feeding were only found in gp130Δhepa and gp130ΔhepaSTAT animals, but not in the gp130ΔhepaRas and the control group. These findings were also associated with metabolic changes consisting of elevated fasting blood glucose levels (Fig. 5E) in gp130Δhepa and gp130ΔhepaSTAT during CDE treatment (4 and 12 weeks).

Lack of STAT3 Activation and Acute-Phase Gene Induction in Hepatocytes Inversely Correlates with the Regulation of Lipid Metabolism-Related Genes.

Our analysis suggested that the liver phenotype in the CDE model correlated with lack of gp130-STAT3-dependent signaling in hepatocytes. Thus, we then analyzed the regulation of distinct targets of this pathway. Hepatic IL-6 mRNA expression was moderately induced in all treatment groups (Fig. 6A). Next, activation/phosphorylation of STAT proteins was analyzed by western blotting (Fig. 6B,C). Although a strong STAT3 phosphorylation was found in livers of wildtype (gp130loxP/loxP) mice 12 weeks after CDE feeding, STAT3 activation was clearly diminished in gp130Δhepa mice and almost absent in gp130ΔhepaSTAT animals. Interestingly, the p-STAT3 signal was significantly more enhanced in livers of gp130ΔhepaRas mice (Fig. 6B). In contrast, STAT1-phosphorylation did not differ between naïve and CDE-treated gp130loxP/loxP, gp130Δhepa, and gp130ΔhepaRas mice, whereas a lower STAT1 phosphorylation was observed in the gp130ΔhepaSTAT group (Fig. 6C).

Figure 6.

Lack of acute phase response (APR) induction and differential gene expression in gp130Δhepa and gp130ΔhepaSTAT mice. (A) IL-6−/− liver mRNA levels were determined by quantitative real-time PCR before and 12 weeks after feeding as indicated. Significant differences were found 12 weeks after CDE-feeding in gp130Δhepa mice (n = 5, *P < 0.05, **P < 0.01). (B,C) Western blot analysis was performed with liver cell extracts before and 12 weeks after CDE feeding. Membranes were probed with an anti-phospho-STAT3 (B), anti-STAT1, and anti-phospho-STAT1 antibodies (C). Analysis 12 weeks after CDE treatment revealed significantly lower STAT3 phosphorylation in livers of gp130Δhepa and gp130ΔhepaSTAT mice. (D,E) Liver mRNA levels of SAA (D) and SOCS3 (E) (fold-induction) were determined by quantitative real-time PCR before and 12 weeks after CDE treatment. SAA and SOCS3 induction was significantly upregulated in livers of wildtype and gp130ΔhepaRas mice after 12 weeks of CDE diet (n = 5, *P < 0.05, **P < 0.01, ***P < 0.001).

Expression of serum-amyloid-A (SAA), the major acute phase gene in mice, was significantly increased in livers of gp130loxP/loxP mice 12 weeks after CDE feeding (Fig. 6D). In contrast, in gp130Δhepa and gp130ΔhepaSTAT mice SAA induction was strongly diminished. However, a more than 2-fold stronger SAA induction was evident in gp130ΔhepaRas livers compared to gp130loxP/loxP control mice. Suppressors of cytokine signaling-3 (SOCS3) as a STAT3 target gene and a critical intracellular regulator of the IL-6-induced inflammatory response was only up-regulated in gp130loxP/loxP and gp130ΔhepaRas livers (Fig. 6E).

To better characterize the inflammatory response in these animals the hepatic expression of TNF-α and adiponectin was analyzed. An inverse regulation of these cytokines was found. Higher TNF-α levels were evident in mice with enhanced steatosis (gp130Δhepa and ΔSTAT), whereas adiponectin expression was elevated in livers of control and gp130ΔhepaRas mice (Fig. 7A,B).

Figure 7.

Differential expression of lipid metabolism related genes. (A,B) TNF-α (A) and adiponectin (B) liver mRNA levels (fold-induction) were determined by quantitative real-time PCR before and 12 weeks after CDE diet. Significant differences are indicated (n = 5, *P < 0.05, **P < 0.01). (C,D) SREBP-1 (C) and FAS (D) liver mRNA levels (fold-induction) were determined by quantitative real-time PCR before and 12 weeks after CDE-feeding. Significant changes are indicated (n = 5, *P < 0.05, **P < 0.01). (E) Western blot analysis was performed with liver cell extracts before and 12 weeks after CDE feeding. JNK activation was detected with a phospho-JNK-specific antibody. Stronger JNK phosphorylation was found in livers of gp130Δhepa and gp130ΔhepaSTAT mice 12 weeks after CDE feeding. The specific signals for phospho-p54 and phospho-p46 are depicted (ns, nonspecific). Membranes were reprobed with an anti-GAPDH antibody as a loading control.

Next we investigated whether the differences in acute-phase regulation and the observed metabolic changes had an impact on the expression of regulatory genes involved in lipid metabolism. The expression levels of sterol regulatory element-binding protein (SREBP-1) and fatty acid synthase (FAS) were significantly up-regulated in gp130Δhepa and gp130ΔhepaSTAT livers 12 weeks after CDE treatment compared to gp130loxP/loxP mice (Fig. 7C,D). Interestingly, these differences were more pronounced in the gp130ΔhepaRas and gp130ΔhepaSTAT mice, as livers of gp130ΔhepaSTAT showed strongly induced and those of gp130ΔhepaRas reduced SREBP-1 and FAS levels.

Metabolic stress usually causes enhanced jun kinase (JNK) activation in hepatocytes, which is known as a critical regulator of hepatic metabolism and insulin resistance.14 We therefore performed western blot analysis, showing significantly increased JNK phosphorylation in gp130Δhepa and an even stronger signal in gp130ΔhepaSTAT mice (Fig. 7E).

Occurrence of Steatohepatitis Is Associated with More Immune Cell Infiltration and Fibrosis.

As gp130Δhepa and gp130ΔhepaSTAT mice displayed higher proinflammatory cytokine expression and signs of steatohepatitis, we next studied if these mechanisms had an impact on immune cell infiltration. After 12 weeks of CDE feeding immunohistochemistry revealed significantly higher numbers of CD4(+) cells specifically in gp130Δhepa and gp130ΔhepaSTAT livers (Fig. 8A; Supporting Fig. 4B). Additionally, we detected significantly higher numbers of CD11b(+) cells in these two mouse strains (Fig. 8B; Supporting Fig. 4C).

Figure 8.

Deletion of gp130-STAT signaling in hepatocytes is associated with an altered inflammatory response and liver remodeling after CDE feeding. (A,B) Liver sections were stained with anti-CD4 (A) and CD11b (B) antibodies before and 12 weeks after CDE feeding. A quantitative analysis of positive stained cells per view field in a 10-fold magnification is shown (n = 5, *P < 0.05, **P < 0.01). At least 10 view fields of four livers/group were analyzed. (C,D) Liver sections were stained with anti-αSMA (C) and anti-α1collagen (D) antibodies before and 12 weeks after CDE feeding. Representative stainings are shown. Stronger signals were evident in livers of gp130Δhepa and gp130ΔhepaSTAT mice 12 weeks after CDE diet. (E,F) Liver sections were stained for Sirius red (E) or cytokeratin 19 (CK-19, F) before and 12 weeks after CDE diet. Representative stainings are shown. Positive signals were only evident in livers of gp130Δhepa and gp130ΔhepaSTAT mice 12 weeks after CDE feeding.

Finally, we studied whether the differences in fat accumulation, metabolic changes, and acute-phase regulation had an impact on liver remodeling in gp130Δhepa and gp130ΔhepaSTAT mice. Immunostaining for alpha-smooth muscle actin (αSMA) revealed an activation of hepatic stellate cells (Fig. 8C) only within these two groups. These results were further confirmed by collagen immunohistochemistry (Fig. 8D) and Sirius red stainings (Fig. 8E), which demonstrated enhanced accumulation of collagen fibers in livers of gp130Δhepa and gp130ΔhepaSTAT mice. Additionally, CK-19 staining was performed to detect biliary epithelial and stem cells. Ongoing chronic liver injury due to CDE treatment in gp130Δhepa and gp130ΔhepaSTAT livers revealed proliferation of biliary epithelial cells and activation of the intrahepatic stem cell compartment (Fig. 8F). A quantification of the immunohistochemical analyzes is provided as Supporting Fig. 3A-D.


Steatohepatitis in animal models is frequently induced by a methionine-choline-deficient (MCD) diet, high fat, or a high sucrose diet.15 Those nutritional insults induce reactive oxygen species, activate nuclear receptors, and dysregulate lipid peroxidation. As a consequence, accumulation of fatty acids in hepatocytes is found, resulting in insulin resistance of the liver and the production of proinflammatory mediators. These events stimulate the recruitment of immune cells to the liver, tissue damage, and progression of the disease with signs partly resembling NASH in humans.

In the present study we used a choline-deficient diet supplemented with ethionine in the drinking water, which has been described as a model of chronic liver injury. Fatty liver degeneration may occur, depending on the experimental conditions,11 and the genetic background as shown for other liver injury models.16 In our analysis we thus detected significant steatosis after 16 weeks in wildtype mice (Supporting Fig. 1). However, we observed a remarkable but rather unexpected strong fat accumulation in IL-6−/− and hepatocyte-specific gp130 knockout mice. We were therefore interested to investigate the relevance of these findings of chronic liver inflammation in the CDE-treated animals, which primarily do not develop a classical steatosis or NASH.

By using IL-6−/− mice the essential role of this cytokine for pathogenesis in the CDE model was uncovered (Fig. 1), a finding that prompted us to further dissect IL-6-dependent signaling pathways in hepatocytes. The striking phenotype in IL-6−/− mice could be confirmed in hepatocyte-specific gp130-deleted animals. However, as gp130 is also expressed in other cells, e.g., nonparenchymal cells, this finding does not rule out that other cells also may contribute to this phenotype.

Interestingly, CDE feeding of gp130Δhepa mice triggered early triglyceride accumulation that was associated with significantly higher fasting blood glucose levels and a strong increase in serum insulin. These findings suggest the development of insulin resistance, which likely is limited to the liver. This was confirmed by higher blood glucose levels in gp130Δhepa mice in a glucose tolerance test (Fig. 4C).

In parallel, hepatic inflammation and higher transaminases were observed in gp130-deficient mice, which was associated with the inverse regulation of hepatic TNF-α and adiponectin expression. TNF-α is not only involved in the recruitment of inflammatory cells, but also promotes nuclear factor kappa light chain enhancer of activated B-cells (NF-κB)-dependent inhibition of lipid catabolism and β-oxidation,17 whereas adiponectin can prevent NASH development by inhibiting the deleterious effects of NF-κB activation. These results demonstrate that the lack of gp130-dependent signaling specifically in hepatocytes redirects the CDE model from chronic liver injury into a model of fatty liver degeneration. This reprogrammed model now shows signs of fatty liver disease progression, as found when mice are fed a hypercaloric diet.

By using hepatocyte-specific knockout animals that either activate gp130-dependent STAT- or Ras-dependent signaling, we aimed to define the intracellular pathway, which is responsible for this phenotype. Strong STAT3 activation, which was evident in control livers after 12 weeks of CDE feeding, was even more pronounced in gp130ΔhepaRas mice, as they showed enhanced STAT3 phosphorylation and SAA expression. This observation was specific for STAT3, as we found no major regulation for STAT1.

In agreement with Ueki et al.,18 we found an inverse correlation of STAT3 activation with distinct genes involved in lipid metabolism (Fig. 7C,D). Thus, our results provide further evidence that STAT3 not only regulates genes involved in gluconeogenesis, as described,9 but also genes controlling lipid metabolism. Interestingly, nuclear receptors peroxisome-proliferator-activated receptors (PPAR)α and γ were not significantly regulated in our model (Supporting Fig. 2).

Our results indicate that gp130-dependent STAT3 signaling protects hepatocytes from metabolic insults during CDE-feeding. Lack of hepatic STAT3 activation was previously reported to be associated with insulin resistance.9 Those experiments were undertaken using hepatocyte-specific STAT3 knockout (STAT3Δhepa) animals, which have the disadvantage that after gp130 stimulation STAT1 is activated and thus an IFN-like response is triggered in hepatocytes.19 Therefore, our present results better define this mechanism, as specifically the IL-6/gp130-STAT3 axis in hepatocytes is important to prevent metabolic changes and inflammation. In additional studies nutritional insults triggered hormonal feedback mechanisms in the brain, which increased insulin expression and resulted in stronger hepatic STAT3 activation.20 Therefore, our data suggest that the gp130/STAT3-pathway can protect from insulin resistance not only during metabolic insults, but also during inflammatory injury of the liver, and thus seems to be of general relevance.

This is in contrast to a recent report published by Sabio et al.21 Those authors used a high-fat diet model and show that IL-6 produced in adipocytes triggers insulin resistance in hepatocytes by way of SOCS3. At present there is no obvious experimental evidence that might explain these opposing results. However, at least in our CDE model no obvious fat accumulation was evident in adipocytes. Moreover, we only found a modest SOCS3 induction in gp130ΔhepaRas mice, the group with the highest level of STAT3 activation, which is protected from the occurrence of metabolic adaptations. Together, this suggests that SOCS3 does not seem to have a fundamental role in CDE-mediated liver pathology. Therefore, one explanation could be that the feedback mechanisms between brain, fat tissue, and liver might determine which role IL-6/gp130- and SOCS3-dependent signaling in hepatocytes and other cells can play in fatty liver degeneration.

Another pathway leading to JNK activation has been described in many models to mediate inflammation and insulin resistance.14 Its inhibition, most prominently of JNK1, has been shown to be beneficial for the prevention of experimental type-2 diabetes and steatohepatitis.22, 23 In the CDE model, enhanced JNK phosphorylation, especially of JNK1, was found in gp130Δhepa and more prominent in gp130ΔhepaSTAT animals. Therefore, these results indicate that JNK may play an important role for pathogenesis in the CDE model.

In the CDE model fatty liver degeneration and increased inflammation also triggered the recruitment of immune cells to the liver. Especially CD11b-expressing cells, namely, monocytes and macrophages, as well as CD4-positive cells were found in gp130Δhepa and gp130ΔhepaSTAT livers. The inflammatory response mediated by these cells and very likely also other liver nonparenchymal cells triggered hepatic stellate cell activation, and thus deposition of collagen, eventually leading to liver fibrosis. Additionally, the CDE diet has also been used in the past to study the occurrence of intrahepatic regenerative cells, namely, oval cells, during the development of chronic liver injury. Stronger activation of these cells was influenced by gp130-STAT3, as blocking of this pathway in hepatocytes also provoked a stronger oval cell response as part of an adaptive response to tissue injury.

In summary, we established a mouse model of fatty liver disease leading to hepatic inflammation. Interestingly, the crosstalk between gp130-STAT3 and insulin signaling is essential to determine the progression of fatty liver disease during CDE feeding. This new model will allow further definition of the protective gp130-STAT pathway for insulin resistance in hepatocytes, and thus help to find potential new treatment options for fatty liver disease in the future.