IFN-γ/STAT1 acts as a proinflammatory signal in T cell-mediated hepatitis via induction of multiple chemokines and adhesion molecules: a critical role of IRF-1

Barbara Jaruga, Feng Hong, Won-Ho Kim, Bin Gao

Abstract

We have previously shown that IFN-γ/STAT1 plays an essential role in concanavalin A (ConA)-induced T cell hepatitis via activation of apoptotic signaling pathways. Here we demonstrate that IFN-γ/STAT1 also plays a crucial role in leukocyte infiltration into the liver in T cell hepatitis. After injection of ConA, leukocytes were significantly infiltrated into the liver, which was suppressed in IFN-γ−/− and STAT1−/− mice. Disruption of the IFN regulatory factor-1 (IRF-1) gene, a downstream target of IFN-γ/STAT1, abolished ConA-induced liver injury and suppressed leukocyte infiltration into the liver. Additionally, ConA injection induced expression of a wide variety of chemokines and adhesion molecules in the liver. Among them, expression of ICAM-1, VCAM-1, monokine induced by IFN-γ (Mig), CC chemokine ligand-20, epithelial cell-derived neutrophil-activating peptide (ENA)-78, IFN-inducible T cell-α chemoattractant (I-TAC), and IFN-inducible protein-10 (IP-10) was markedly attenuated in IFN-γ−/−, STAT1−/−, and IRF-1−/− mice. In primary mouse hepatocytes, Kupffer cells, and endothelial cells, in vitro treatment with IFN-γ activated STAT1, STAT3, and IRF-1, and induced expression of VCAM-1, ICAM-1, Mig, ENA-78, I-TAC, and IP-10 mRNA. Induction of these chemokines and adhesion molecules was markedly diminished in STAT1−/− and IRF-1−/− hepatic cells compared with wild-type hepatic cells. These findings suggest that in addition to induction of apoptosis, previously well documented, IFN-γ also stimulated hepatocytes, sinusoidal endothelial cells, and Kupffer cells partly via an STAT1/IRF-1-dependent mechanism to produce multiple chemokines and adhesive molecules responsible for promoting infiltration of leukocytes and, ultimately, resulting in hepatitis.

  • concanavalin A
  • interferon-γ
  • liver injury
  • Kupffer cells
  • sinusoid epithelial cells

alcohol consumption and hepatitis viral infection are two dominant causes of chronic liver disease, affecting billions of people worldwide. The molecular and cellular mechanisms underlying the pathogenesis of alcoholic liver disease and hepatitis infection are not fully understood. Increasing evidence suggests that elevated cytokines play important roles in the pathogenesis of liver diseases (1, 12, 19, 27, 50). For example, it has been reported that serum levels of IFN-γ are significantly elevated in alcoholic hepatitis (27), hepatitis B infection (14, 24), and hepatitis C infection (30, 42, 46). High levels of IFN-γ mRNA expression are also detected in the livers of patients with chronic hepatitis C infection (30), and in hepatic infiltrating T lymphocytes and peripheral T lymphocytes (42). It has been well documented that elevated IFN-γ is involved in front-line defenses against viral infection through induction of antiviral proteins and modulation of immune responses (40). However, the role of IFN-γ in liver damage is less clear.

IFN-γ activity is mediated through activation of the JAK-STAT signaling pathway. On IFN-γ binding, tyrosine kinases (JAK1 and JAK2) associated with the IFN-γ receptor (IFNGR1 and IFNGR2) are activated, leading to phosphorylation and activation of STAT1. Activated STAT1 dimerizes and translocates into the nuclei to activate transcription of a number of genes, including IFN regulatory factor-1 (IRF-1). IRF-1 is a transcription factor that controls transcription of many antiviral and antiapoptotic genes (20, 39). IFN-γ activates other STATs, such as STAT3, STAT4, STAT5, and STAT6 as well; however, the functions of these STATs when activated by IFN-γ are less clear. IFN-γ−/− and STAT1−/− mice have been shown to be resistant to concanavalin A (ConA)- or LPS/d-galactosamine-induced liver injury (13, 18, 45, 48). Disruption of the IRF-1 gene, a downstream target of IFN-γ/STAT1, protects against mortality associated with injection of LPS, ConA, or P. Berghei ANKA infection (44). These results indicate that IFN-γ and IRF-1 are both involved in experimentally induced liver injury. Furthermore, IFN-γ has been reported to directly induce cellular apoptosis in cultured hepatocytes in an IRF-1-dependent manner (17). Therefore, it is believed that the detrimental effects of IFN-γ/STAT1 in liver injury is mediated partly through induction of the proapoptotic IRF-1 gene and, consequently, induction of hepatocyte apoptosis and hepatocellular damage. Here, we further demonstrate that the IFN-γ/STAT1/IRF-1 pathway plays an important role in the infiltration of leukocytes in the ConA-mediated hepatitis model. As shown in this study, IFN-γ not only targeted hepatocytes, but also sinusoidal endothelial and Kupffer cells via an STAT1/IRF-1-dependent mechanism, resulting in the production of chemokines [such as monokine induced by IFN-γ (Mig), epithelial neutrophil-activating peptide (ENA)-78, IFN-inducible T cell-α chemoattractant (I-TAC), and IFN-inducible protein-10 (IP-10)] and adhesion molecules (such as ICAM-1 and VCAM-1), which may play critical roles in the infiltration of leukocytes into the liver.

MATERIALS AND METHODS

Materials.

Anti-STAT1, anti-phospho-STAT1 (Tyr701), anti-phospho-STAT3 (Tyr705), and anti-STAT3 antibodies were obtained from Cell Signaling Technology (Beverly, MA). Anti-IRF-1 antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-Mig antibody was obtained from R&D Systems (Minneapolis, MN).

Mouse models of hepatitis induced by injection of ConA.

Seven- to eight-wk old male IFN-γ −/− mice (C57BL/6 background), IRF-1−/− mice (C57BL/6 background), and male C57BL/6J mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Seven- to eight-wk-old male 129/SvEv background STAT1−/− mice and 129SvEv control mice were purchased from Taconic Farms (Germantown, NY). Preliminary data showed that C57BL/6J mice exhibited more susceptibility to ConA-induced T cell hepatitis than 129/SVEv mice. Therefore, IFN-γ−/−, IRF-1−/−, and C57BL/6J mice were injected intravenously with ConA (12 μg/g). STAT1−/− mice and 129SvEv mice were injected intravenously with ConA (22 μg/g). The mouse livers were then collected for detection of neutrophil and eosinophil infiltration, hemotoxylin and eosin staining, and RT-PCR analysis at various time points post-ConA injection.

Isolation and culture of primary mouse hepatocytes, sinusoidal endothelial cells, and Kupffer cells.

Control and knockout mice weighing 20–25 g were anesthetized with pentobarbital sodium (30 mg/kg ip), and the portal vein was cannulated under aseptic conditions. The liver was subsequently perfused with EGTA solution (in mM: 5.4 KCl, 0.44 KH2PO4, 140 NaCl, 0.34 Na2HPO4, 0.5 EGTA, and 25 Tricine, pH 7.2), and DMEM (GIBCO-BRL, Manasses, VA) and digested with 0.075% collagenase solution. The isolated mouse hepatocytes were then cultured in Hepato-ZYME-SFM medium containing 5% fetal bovine serum (GIBCO-BRL) in rat tail collagen coated plates for 2 h, and then changed to serum-free DMEM medium for 16 h, followed by treatment with IFN-γ for various time periods.

Sinusoidal endothelial cells and Kupffer cells were isolated by collagenase perfusion and differential centrifugation in Percoll (Sigma, St. Louis, MO) as previously described (51). The viability of isolated sinusoidal endothelial cells was >95% in all isolations as determined by the trypan blue exclusion test. The purity of sinusoidal endothelial cells as examined by phase-contrast microscopy was 90.7 ± 5.2%. Cells were cultured at 37°C in RPMI 1640 medium supplemented with fetal calf serum (20%), l-glutamine (2 mM), gentamycin (100 μg/ml), and dexamethasone (1 μM) in a humidified atmosphere (100%) containing 5% CO2-95% air. Cultured cells were identified as liver endothelial cells, given immunological evidence of the von Willebrand factor. Kupffer cells were cultured at 37°C in RPMI 1640 medium supplemented with 20% fetal calf serum and gentamycin (100 μg/ml) in a humidified atmosphere (100%) containing 5% CO2-95% air. Both sinusoidal endothelial cells and Kupffer cells were cultured overnight in serum-containing medium, and then replaced with serum-free medium for 4 h, followed by stimulation with IFN-γ for various time periods.

Western blot analysis.

Cells were lysed in lysis buffer (in mM: 30 Tris, pH 7.5, 150 sodium chloride, 1 PMSF, and 1 sodium orthovanadate, plus 1% Nonidet P-40 and 10% glycerol) for 20 min at 4°C, vortexed, and then centrifuged at 16,000 rpm for 10 min at 4°C. Tissues were homogenized in lysis buffer at 4°C, vortexed, and spun at 16,000 rpm for 10 min at 4°C. The supernatants were mixed in Laemmli loading buffer, boiled for 4 min, and then subjected to SDS-PAGE. After electrophoresis, proteins were transferred onto nitrocellulose membranes and blotted against primary antibodies for 16 h. Membranes were washed with TPBS [0.05% (vol/vol) Tween 20 in phosphate-buffered saline (pH 7.4)] and incubated with a 1:4,000 dilution of horseradish peroxidase-conjugated secondary antibodies for 45 min. Protein bands were visualized by enhanced chemiluminescence reaction (Amersham Pharmacia Biotech, Piscataway, NJ).

Determination of liver injury.

Liver injury was determined by either hematoxylin and eosin staining of liver sections or by analysis of serum aminotransferase activities. For hemotoxylin and eosin staining, livers were fixed with 10% formalin/phosphate-buffered saline for 24 h, sliced, and then stained with hematoxylin-eosin. Serum levels of alanine aminotransaminase (ALT) and asparate aminotransferase (AST) were quantified by measuring plasma enzyme activities of ALT and AST utilizing a kit from Drew Scientific (Cumbria, UK).

Immunohistochemistry.

Liver sections were immunostained for neutrophils as described previously using anti-neutrophil MPO antibody (Lab Vision, Fremont, CA) (15). The number of neutrophils in the liver sections was counted in 10 randomly chosen visual fields (magnification, ×200), and the average from 10 selected microscopic fields was calculated.

Assay of hepatic eosinophil peroxidase activity.

Hepatic eosinophil peroxidase (EPO) activity was measured as described previously (41). EPO enzyme activity in hepatic tissues was calculated by subtracting the mean background optical density and expressed as change OD 490 nm/min.

RT-PCR.

RT-PCR was performed as described previously (37). The sequences of the primers used in the study are listed in Table 1. The β-actin gene was amplified as an internal control. PCR using RNA without reverse transcription did not yield amplicons, indicating a lack of genomic DNA contamination. The PCR bands were scanned by using Storm PhosphoImager (Molecular Dynamics, Sunnyvale, CA) and quantified by using ImageQuant software (Molecular Dynamics) and normalized to β-actin mRNA levels at each time point. Fold induction is the relative induction compared with untreated wild-type control.

View this table:
Table 1.

PCR primers for murine chemokines and adhesion molecules

Statistical analysis.

For comparing values obtained in three or more groups, one-factor ANOVA was used, followed by Tukey's post hoc test. Statistical significance was taken at the P < 0.05 level.

RESULTS

IRF-1 plays a critical role in ConA injection-mediated liver injury.

Senaldi et al. (44) reported that IRF-1−/− mice showed less mortality than control mice after injection of a lethal dose of ConA. However, the role of IRF-1 in ConA-induced liver injury has not been investigated. We (13) and others (47) have demonstrated that IRF-1 activation correlated with liver injury in ConA-mediated hepatitis, suggesting that IRF-1 may play an important role in T cell-mediated hepatitis. To test this hypothesis, we compared ConA-induced liver injury in IRF-1−/− mice and control wild-type mice. As shown in Fig. 1A, ConA-induced liver injury (elevated ALT levels) was markedly attenuated in IRF-1−/− mice compared with wild-type mice. Liver histology revealed massive necrosis and hepatic infiltration of leukocytes in wild-type mice but not in IRF-1−/− mice (Fig. 1B).

Fig. 1.

IFN regulatory factor (IRF-1) plays an essential role in concanavalin A (ConA)-induced T cell hepatitis. A: wild-type and IRF-1−/− mice (5/group) were administered ConA (12 μg/g iv). After 9 and 24 h, mice were killed, and the serum was collected to measure alanine aminotransaminase (ALT) activity. Values are means ± SE at each time point from 5 mice. **P < 0.001 compared with corresponding ConA-treated wild-type mice. B: photomicrographs of representative mouse livers from 24-h ConA-treated wild-type and IRF-1−/− mice with hemotoxylin and eosin staining are shown (original magnification, ×200).

IFN-γ, STAT1, and IRF-1 play critical roles in ConA injection-mediated infiltration of neutrophils and eosinophils in the liver.

To understand the mechanisms underlying the detrimental effects of the IFN-γ/STAT1/IRF-1 pathway in T cell hepatitis, we compared ConA-induced infiltration of neutrophils and eosinophils between wild-type mouse livers and knockout (IFN-γ−/−, STAT1−/−, and IRF-1−/−) mouse livers. As shown in Fig. 2A, control mouse livers contained very low numbers of neutrophils. After ConA injection, neutrophils significantly infiltrated the liver in wild-type mice with peak effect at 9 h. In IFN-γ−/−, STAT1−/−, and IRF-1−/− mice, hepatic infiltration induced by ConA was markedly attenuated. The suppression of neutrophil infiltration was more prominent in IFN-γ−/− mice compared with STAT1−/− and IRF-1−/− mice.

Fig. 2.

IFN-γ, STAT1, and IRF-1 play important roles in the infiltration of neutrophils and eosinophils in ConA-induced T cell hepatitis. IFN-γ−/− mice, IRF-1−/− mice, and their wild-type C57BL/6J control mice were administered ConA (12 μg/g iv), and STAT1−/− mice and their wild-type 129SvEv control mice were injected with ConA (22 μg/g iv). A: after 9 and 24 h of ConA administration, livers were collected and immunostained with anti-MPO antibodies for detection of neutrophils. B: eosinophils were detected by measuring hepatic eosinophil peroxidase (EPO) activities. In A, 10 fields were randomly selected and positive cells were counted. Values in A and B are means ± SE at each time point from 4–5 mice. *P < 0.05, **P < 0.001 compared with corresponding ConA-treated wild-type control mice. OD, optical density.

Infiltration of eosinophils was determined by measuring hepatic EPO activity. As shown in Fig. 2B, control mouse livers contained low levels of EPO activity, which was significantly elevated 9 and 24 h post-ConA injection. Elevation in EPO activity was significantly suppressed in IFN-γ−/−, IRF-1−/−, and STAT1−/− mice compared with their corresponding control mice.

ConA-mediated induction of VCAM-1, IACM-1, Mig, CCL-20, ENA-78, I-TAC, and IP-10 expression in the liver is impaired in IFN-γ−/−, STAT1−/−, and IRF-1−/− mice.

The above data clearly showed that IFN-γ, STAT1, and IRF-1 were involved in the infiltration of leukocytes in ConA-mediated hepatitis. To understand the mechanisms underlying their involvement in leukocyte infiltration, expression of a variety of chemokines and adhesion molecules was compared between wild-type mice and IFN-γ−/− mice. As shown in Fig. 3A, ConA injection significantly induced expression of numerous adhesion molecules and chemokines in the livers of wild-type mice, with peak effect between 3 and 9 h after injection. Among them, induction of VCAM-1, ICAM-1, Mig, CCL-20, ENA-78, I-TAC, and IP-10 was markedly suppressed in IFN-γ−/− mice (Fig. 3, A and D), whereas induction of MCP-1, Rantes, CCL-1, CCL-6, CCL-17, BCA-1, BRAK, and PF-4 was enhanced in IFN-γ−/− mice. These findings suggest that IFN-γ induces VCAM-1, ICAM-1, Mig, CCL-20, ENA-78, I-TAC, and IP-10 expression, but may suppress induction of other chemokines in T cell-mediated hepatitis.

Fig. 3.

IFN-γ, STAT1, and IRF-1 play important roles in the induction of chemokines and adhesion molecules in ConA-induced T cell hepatitis. IFN-γ−/− mice, IRF-1−/− mice, and their wild-type C57BL/6J control mice were administered ConA (12 μg/g iv), and STAT1−/− mice and their wild-type 129SvEv control mice were injected with ConA (22 μg/g iv). After various time periods, livers were collected and used to isolate total RNA, which was subjected to RT-PCR analyses [25 cycles for IFN-inducible T cell-α chemoattractant (I-TAC) and IFN-inducible protein-10 (IP-10) and 30 cycles for other genes] of various chemokines and adhesion molecules. Liver tissues were also subjected to Western blot analysis of monokine induced by IFN-γ (Mig) and β-actin. AC represent 3 independent experiments. D: expression of VCAM-1, ICAM-1, CCL-20, epithelial neutrophil-activating peptide (ENA)-78, I-TAC, IP-10 in A was quantified by PhosphorImager analysis and normalized to the same number cycles (25 or 30 cycles) of RT-PCR amplification of β-actin mRNA levels at each time point. Expression of Mig protein was quantified by PhosphorImager and normalized to β-actin protein in Western blot analysis. Fold induction is the relative induction compared with untreated wild-type control. Values are means ± SE from 3 independent experiments at each time point. *P < 0.01, #P < 0.05, compared with corresponding ConA-treated wild-type controls. PECAM, platelet endothelial cell adhesion molecule; BCA, B-cell-attracting chemokine; SDF, stromal cell-derived factor; BRAK, chemokine ligand 14; KC, cytokine-induced neutrophil chemoattractant; PF, placental factor.

To further determine whether IFN-γ-mediated induction of adhesion molecules and chemokines was STAT1- and IRF-1-dependent, we additionally compared expression of these factors between wild-type and STAT1−/− or IRF-1−/− mice. As shown in Fig. 3, B and C, ConA injection-mediated induction of VCAM-1, ICAM-1, Mig, CCL-20, ENA-78, I-TAC, and IP-10 was also significantly suppressed in STAT1−/− and IRF-1−/− mice compared with wild-type control mice. However, the magnitude of the downregulation observed in STAT1−/− and IRF-1−/− mice relative to wild-type mice was much less compared with the downregulation observed between IFN-γ−/− mice and wild-type mice (Fig. 3).

IFN-γ activates STAT1, STAT3, and IRF-1 in hepatocytes, sinusoidal endothelial cells, and Kupffer cells.

The above findings suggest that IFN-γ is important for the induction of a variety of adhesion molecules and chemokines in ConA-mediated hepatitis. To further understand which hepatic cell types are targeted by IFN-γ, hepatocytes, sinusoidal endothelial cells, and Kupffer cells were isolated and stimulated with IFN-γ in vitro. As shown in Fig. 4, IFN-γ treatment significantly induced STAT1 and STAT3 activation, and induced expression of IRF-1 protein in primary mouse hepatocytes, which is consistent with our previous report (13). IFN-γ also activated STAT1 and STAT3, and induced expression of IRF-1 in both sinusoidal endothelial cells and Kupffer cells (Fig. 4).

Fig. 4.

IFN-γ activates STAT1 and STAT3 and induces IRF-1 protein expression in hepatocytes, sinusoidal endothelial cells, and Kupffer cells. Primary mouse hepatocytes, sinusoidal endothelial cells, and Kupffer cells were isolated and cultured in serum-containing medium overnight, followed by culturing in serum-free medium for 4 h, and then stimulated with IFN-γ (10 ng/ml). After various time periods, cells were harvested, and total protein extracts were prepared for Western blot analysis using antibodies as indicated. Data are representative of 2 independent experiments.

IFN-γ induces expression of chemokines and adhesion molecules in hepatocytes, sinusoidal endothelial cells, and Kupffer cells via STAT1- and IRF-dependent mechanisms.

The effects of IFN-γ on the induction of chemokines and adhesion molecules in hepatocytes, sinusoidal endothelial cells, and Kupffer cells were examined. As shown in Fig. 5A, IFN-γ treatment induced expression of VCAM-1, ICAM-1, Mig, ENA-78, I-TAC, and IP-10, but not CCL-20 in primary mouse hepatocytes. Induction of these chemokines and adhesion molecules by IFN-γ was markedly diminished in STAT1−/− and IRF-1−/− mouse hepatocytes. Suppression of IFN-γ-induced I-TAC and Mig expression in IRF-1−/− mice relative to wild-type mice was less evident compared with the difference between STAT1−/− mice and wild-type mice.

Fig. 5.

IFN-γ induces expression of VCAM-1, ICAM-1, Mig, I-TAC, and IP-10 mRNA in hepatocytes, sinusoidal endothelial cells, and Kupffer Cells via STAT1- and IRF-1-dependent mechanisms. Hepatocytes, sinusoidal endothelial cells, and Kupffer cells were isolated from STAT1−/− mice and their wild-type 129SvEv control mice and from IRF-1−/− mice and their wild-type C57BL/6 control mice. These cells were cultured in serum-containing medium overnight, followed by culturing in serum-free medium for 4 h, and then stimulated with IFN-γ (10 ng/ml). After various time periods, total RNA was isolated and subjected to RT-PCR analyses of various chemokines and adhesion molecules as indicated. Data are representative of 2 independent experiments.

In wild-type mouse Kupffer cells, IFN-γ treatment induced expression of VCAM-1, ICAM-1, Mig, ENA-78, and I-TAC, whereas in STAT1−/− and IRF-1−/− mouse Kupffer cells, induction of these genes was diminished. IFN-γ treatment weakly induced expression of IP-10 mRNA and failed to induce expression of CCL-20 mRNA in Kupffer cells. Similarly, IFN-γ treatment induced significant expression of VCAM-1, ICAM-1, Mig, ENA-78, I-TAC, IP-10 mRNAs in sinusoidal endothelial cells, and induction of these genes was suppressed in STAT1−/− and IRF-1−/− mouse sinusoidal endothelial cells. Densitometric analyses revealed that IFN-γ induction of VCAM-1, ICAM-1, Mig, ENA78, and I-TAC was significantly reduced in STAT1−/− and IRF-1−/− Kupffer cells and endothelial cells compared with their wild-type control cells (P < 0.05 or P < 0.01). IFN-γ induction of IP-10 was also significantly diminished in STAT1−/− endothelial cells and Kupffer cells, and IRF-1−/− endothelial cells compared with their wild-type control cells (P < 0.05 or P < 0.01).

DISCUSSION

Previous studies (13, 47) suggest that IFN-γ/STAT1 contributes to ConA-induced T cell hepatitis via induction of proapoptotic genes such as IRF-1. In the present paper, we provide evidence demonstrating that the IFN-γ/STAT1/IRF-1 pathway also acts as a proinflammatory signal in T cell mediated hepatitis. Our findings suggest that ConA stimulates natural killer T cells and other cells to produce IFN-γ, which then targets hepatocytes, sinusoidal endothelial cells, and Kupffer cells via activation of STAT1, STAT3, and IRF-1. Consequently, activation of STAT1 and IRF-1 stimulates expression of VCAM-1, ICAM-1, Mig, ENA-78, I-TAC, and IP-10, which along with other chemokines, attracts neutrophils and eosinophils into the liver, resulting in hepatitis. Although we have not examined the T cell influx in STAT1−/− and IRF-1−/− mice, it is likely that T cell infiltration is also attenuated in these mice compared with control mice, because the critical role of neutrophils in T cell infiltration in this model has been reported (3) and neutrophil infiltration was reduced in STAT1−/− and IRF-1−/− mice.

Both TNF-α and IFN-γ have been shown to play an essential role in the development and progression of ConA-induced T cell hepatitis (13, 21, 22, 45). This is probably because TNF-α and IFN-γ synergistically induce expression of several chemokines and adhesion molecules (33, 35). Depletion of either of them results in a marked reduction in ConA-induced liver injury and inflammation (13, 21, 22, 45). Here we showed that expression of VCAM-1, ICAM-1, Mig, CCL-20, ENA-78, and IP-10 was markedly attenuated in IFN-γ−/− mice compared with wild-type mice. This downregulation was also observed in STAT1−/− and IRF-1−/− mice compared with wild-type mice, but was less profound compared with the difference observed between IFN-γ−/− mice and wild-type mice (Fig. 3). Collectively, these findings suggest that IFN-γ plays a pivotal role in the induction of VCAM-1, ICAM-1, Mig, CCL-20, ENA-78, and IP-10 in the liver during ConA-induced T cell hepatitis partly through STAT1- and IRF-1-dependent mechanisms. Furthermore, we provide in vitro evidence suggesting that IFN-γ induces VCAM-1, ICAM-1, Mig, ENA-78, I-TAC, and IP-10 expression in hepatocytes, sinusoidal endothelial cells, and Kupffer cells via STAT1- and IRF-1-dependent mechanisms. IFN-γ activation of STATs and induction of several chemokines (such as Mig and IP-10) in primary mouse, rat, and human hepatocytes has previously been well documented (13, 34, 36, 37). We further showed here that IFN-γ activated STAT1 and STAT3 and induced expression of several chemokines and adhesion molecules in primary mouse hepatocytes as well as in sinusoidal endothelial cells and Kupffer cells. This induction was significantly attenuated in primary STAT1−/− and IRF-1−/− cells, suggesting that IFN-γ-mediated induction of chemokines and adhesion molecules through STAT1- and IRF-1-dependent mechanisms (Figs. 4 and 5). IRF-1 is a transcription factor that binds to IRF-1 response elements in the promoter regions of genes to stimulate genes transcription (20, 39). IRF-1 response elements have been identified in the promoters of VACM-1 (23, 33), ICAM-1 (35), Mig (35), IP-10 (32), I-TAC (10), which may provide a molecular basis for IFN-γ-induced gene expression in primary hepatocytes, endothelial cells, and Kupffer cells via an IRF-dependent mechanism. Interestingly, IFN-γ induction of several chemokines and adhesion molecules was not completely abolished in STAT1−/− and IRF-1−/− cells (Fig. 5). This is probably because other STATs (such as STAT3 and STAT5) or other IRF transcription factors activated by IFN-γ may also be involved in IFN-γ induction of these chemokines and adhesion molecules.

The roles of IFN-γ-controlled adhesion molecules (ICAM-1 and VCAM-1) and chemokines (Mig, CCL-20, ENA-78, and IP-10) in hepatic inflammation and injury have previously been investigated in several models of liver injury. For example, neutrophil infiltration and neutrophil-mediated liver injury caused by bile duct ligation were significantly attenuated in ICAM-1−/− mice compared with wild-type mice (9). Liver regeneration and leukocyte infiltration after partial hepatectomy were also significantly attenuated in ICAM-1−/− mice relative to control mice (43). These findings suggest that ICAM-1 plays an important role in leukocyte infiltration, induction of liver injury, and promotion of liver regeneration in these models. However, the roles of VCAM-1 and ICAM-1 in ConA-induced liver injury have been controversial. Pretreatment with anti-VCAM-1 or anti-ICAM-1 monoclonal antibodies were shown to attenuate ConA-induced liver injury (29, 52), but other reports failed to confirm these findings (26, 53). These discordant findings may partly be due to the different pretreatment regimens used in these studies (26, 29, 52, 53). Mig (CXCL9), IP-10 (CXCL10), and I-TAC (CXCL11) belong to the CXC chemokine receptor (CXCR)3 family of chemokines that bind to CXCR3 receptors, and are potent chemoattractants for alloantigen-primed T cells (8). Several studies reported that expression of these chemokines are elevated in several animal models of liver injury (4, 36) and in human liver diseases (2, 11, 28, 31, 49). However, the effects of CXCR3 chemokines in liver injury remain obscure. Kakimi et al. (16) reported that blocking Mig and IP-10 protected against liver injury in viral hepatitis B transgenic mice by reducing the recruitment of host-derived mononuclear cells into the livers, suggesting that Mig and IP-10 are detrimental in liver injury in this model. However, other studies (4) have suggested that IP-10 plays a protective role against liver injury in murine models of liver injury induced by acetaminophen. ENA-78 is a 78 amino acid 8-kDa protein belonging to the CXC chemokine family and has neutrophil-activating and chemoattracting properties. Recent data (5, 6) suggested that ENA-78 contributes to hepatic neutrophil influx and liver injury but also promotes liver regeneration after partial hepatectomy via stimulation of hepatocyte proliferation. Taken together, VCAM-1, ICAM-1, Mig, CCL-20, ENA-78, and IP-10 likely play important roles in hepatic neutrophil influx in ConA-induced T cell hepatitis. The downregulation of these factors in IFN-γ−/−, STAT1−/−, and IRF-1−/− mice after injection of ConA (Fig. 3) may contribute to decreased infiltration of neutrophils and eosinophils in these mice relative to wild-type mice (Fig. 2). The observed decreased infiltration of neutrophils and eosinophils may partly contribute to lesser liver injuries in IFN-γ−/−, STAT1−/−, and IRF-1−/− mice post-ConA injection compared with wild-type mice, because both neutrophils and eosinophils have been shown to play essential roles in ConA-induced T cell hepatitis (3, 25).

In summary, in addition to its proapoptotic action, the IFN-γ/STAT1/IRF-1 pathway also acts as a proinflammtory signal in the liver via induction of VCAM-1, ICAM-1, Mig, CCL-20, ENA-78, and IP-10. Elevations of these chemokines and adhesion molecules have been reported in chronic liver disease (2, 11, 28, 31, 49), which is likely mediated, in part, by activation of the IFN-γ/STAT1 signal pathway, because chronic liver disease is associated with high levels of IFN-γ (14, 24, 27, 30, 42, 46) and STAT1 (7, 38). Therefore, the IFN-γ/STAT1/IRF-1 pathway may be a potential therapeutic anti-inflammatory target to treat human liver disease.

Footnotes

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

REFERENCES

View Abstract