|Home | About | Journals | Submit | Contact Us | Français|
T cell–mediated fulminant hepatitis is a life-threatening event for which the underlying mechanism is not fully understood. Injection of concanavalin A (Con A) into mice recapitulates the histological and pathological sequelae of T cell–mediated hepatitis. In this model, both signal transducer and activator of transcription factor 1 (STAT1) and STAT3 are activated in the liver. Disruption of the STAT1 gene by way of genetic knockout attenuates liver injury, suppresses CD4+ and NK T cell activation, and downregulates expression of proapoptotic interferon regulatory factor-1 protein and suppressor of cytokine signaling-1 (SOCS1), but enhances STAT3 activation and STAT3-controlled antiapoptotic signals. Studies from IFN-γ–deficient mice indicate that IFN-γ not only is the major cytokine responsible for STAT1 activation but also partially accounts for STAT3 activation. Moreover, downregulation of STAT3 activation in IL-6–deficient mice is associated with decreased STAT3-controlled antiapoptotic signals and expression of SOCS3, but upregulation of STAT1 activation and STAT1-induced proapoptotic signals and exacerbation of liver injury. Taken together, these findings suggest that STAT1 plays a harmful role in Con A–mediated hepatitis by activation of CD4+ and NK T cells and directly inducing hepatocyte death, whereas STAT3 protects against liver injury by suppression of IFN-γ signaling and induction of antiapoptotic protein Bcl-XL. STAT1 and STAT3 in hepatocytes also negatively regulate one another through the induction of SOCS.
T cell–mediated immune responses play a central role in hepatocellular injury induced by autoimmune hepatitis, viral infection, alcohol consumption, and hepatotoxins (1–5). Activated T cells are detected in a variety of human liver diseases. For example, CD4+ T cells are the predominant population of T cells infiltrating into the liver in human autoimmune liver disease (6). In chronic active hepatitis B or C, both CD4+ T cells and CD8+ T cells have been implicated in progression of liver injury and viral clearance (7–10). Additionally, infiltrating CD4+ and CD8+ lymphocytes in the liver significantly correlate with regenerating nodules, intralobular inflammation, and central sclerosis in alcoholic liver disease (5). T cell–mediated hepatitis can be induced in rodents by injection of the T cell mitogenic plant lectin concanavalin A (Con A), which rapidly induces clinical and histological evidence of hepatitis, including elevation of transaminase activities, within 8–24 hours (11). Liver histology shows massive granulocyte accumulation, T cell infiltration, necrosis, and apoptosis. The molecular and cellular mechanisms underlying Con A–induced hepatitis have been extensively investigated, but it is still not fully understood. Multiple cells and cytokines have been implicated in Con A–induced hepatitis (11–15), suggesting that T cell–mediated hepatitis results from the interactions of multiple cells and cytokines. However, the nature of such interactions and their roles in T cell–mediated hepatitis remain poorly understood.
The Janus kinase-signal transducer and activator of transcription factor (JAK-STAT) signaling pathway, activated by more than 40 cytokines and growth factors, has been implicated in a variety of cellular functions (16–19). In general, binding of these cytokines to their receptors induces receptor dimerization, tyrosine phosphorylation of the receptor-associated JAKs, followed by activation of the STATs, including STAT1, 2, 3, 4, 5, and 6. Phosphorylated STATs form dimers and translocate to the nucleus to activate the transcription of many target genes, including a family of inhibitory proteins, suppressor of cytokine signaling (SOCS) (20–22), which then turn off the JAK-STAT signaling pathway. The SOCS family consists of eight proteins: SOCS1–SOCS7 and cytokine-inducible SH2-containing protein (CIS). SOCS1–SOCS3 and CIS are rapidly induced in response to stimulation by a wide variety of cytokines and negatively regulate cytokine signaling by inhibiting the JAKs. There is little evidence that SOCS4–SOCS7 are induced by cytokines and inhibit cytokine signaling. In the liver, the JAK-STAT signaling pathway plays an important role in liver development (23), acute-phase response (24), regeneration (25), protection against liver injury induced by carbon tetrachloride (CCl4) (26), Fas (27), and alcohol (28), as well as antiviral defense in the liver (29–31).
In this study, we demonstrate that injection of Con A rapidly induces activation of multiple JAK-STATs and their inhibitory proteins, SOCSs (20–22). Our findings suggest that Con A–mediated hepatitis is tightly regulated by STAT1 and STAT3, which modulate CD4+ and NKT cell activation, and expression of apoptosis-associated genes. Activated STAT1 and STAT3 in hepatocytes also negatively regulate one another through the induction of SOCS.
Anti-STAT1, anti–phospho-STAT1 (Tyr701), anti–phospho-STAT3 (Tyr705), anti-STAT3, anti–phospho-STAT5 (Tyr694), anti–phospho-STAT6 (Tyr641), and anti–phospho-Tyk2 (Tyr1054/1055) were obtained from Cell Signaling (Beverly, Massachusetts, USA). Anti–phospho-STAT4 (Tyr693) was obtained from Zymed Laboratories Inc. (South San Francisco, California, USA). Anti-STAT2, anti-p21cip1/waf1, anti-p53, anti–caspase-3, anti–phospho-JAK2 (Thr1007/Tyr1008), and anti–phospho-tyrosine Ab’s were purchased from Upstate Biotechnology Inc. (Lake Placid, New York, USA). Anti–Bcl-2 and anti–Bcl-XL Ab’s were obtained from BD PharMingen (San Diego, California, USA). Anti-CIS and anti–interferon regulatory factor-1 (anti–IRF-1) were purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, California, USA). Anti–phospho-JAK1 (Thr1022/Tyr1023), murine IFN-γ and IL-6 were purchased from BioSource International (Camarillo, California, USA).
Male C57BL/6J background IL-6–/– mice and control mice, male BALB/cJ background IFN-γ–/– and control mice weighing 20–25g were purchased from The Jackson Laboratory (Bar Harbor, Maine, USA). Male 129/SvEv background STAT1–/– mice and control mice were purchased from Taconic Farms (Germantown, New York, USA). Preliminary data showed that C57BL/6J mice exhibit more susceptibility to Con A–induced liver injury than BALB/cJ and 129/SVEv mice. Thus, in this article, 10 μg/g of Con A was injected into C57BL/6J background wild-type and knockout mice, 15 μg/g of Con A was injected into BALB/cJ background wild-type and knockout mice, and 22 μg/g of Con A was injected into 129/SVEv background wild-type and knockout mice.
Primary mouse hepatocyte isolation and culture were performed as described previously (31).
Tissue extraction, SDS-PAGE, and Western blot analysis were performed as described previously (31). All wild-type and knockout mouse samples for each individual protein or RNA (14 samples, see Figures Figures2d,2d, d,4a,4a, and and5a;5a; 12 samples, see Figure Figure3,3, a and b) were done in the same blot to accurately quantify and compare them. Protein bands were visualized by an enhanced chemifluorescence reaction (ECF) (Amersham Pharmacia Biotech, Piscataway, New Jersey, USA) and scanned with a Storm PhosphoImager (Molecular Dynamics, Sunnyvale, California, USA). The densities of bands were analyzed with ImageQuant software (Molecular Dynamics). The pSTAT1, pSTAT3, IRF-1, and Bcl-XL protein levels were quantified using ImageQuant software and normalized to β-actin protein levels at each time point. Fold induction is the relative induction compared with untreated wild-type control mice.
RT-PCR was performed as described previously (31), except 20–25 cycles were used in this study. The sequences of the primers used in the study for murine SOCSs are as follows: SOCS1 (210 bp), 5′ GAG GTC TCC AGC CAG AAG TG 3′ (forward), 5′ CTT AAC CCG GTA CTC CGT GA 3′ (reverse); SOCS2 (300 bp), 5′ AAG ACA TCA GCC GGG CCG ACT A 3′ (forward), 5′ GTC TTG TTG GTA AAG GTA GTC 3′ (reverse); SOCS3 (450 bp), 5′ GGA CCA GCG CCA CTT CTT CAC 3′ (forward), 5′ TAC TGG TCC AGG AAC TCC CGA 3′ (reverse). The β-actin gene was also amplified as an internal control. β-actin primers were 5′ AAG GTG ACA GCA TTG CTT CT 3′ (forward) and 5′ GGC TGC CTC AAC ACC TCA 3′ (reverse). RNA without reverse transcription did not yield any amplicons; therefore, no genomic DNA contamination occurred. The PCR bands were scanned with a Storm PhosphoImager, and inverted ethidium bromide-stained bands are shown. The densities of bands were analyzed with ImageQuant software. SOCS1 and SOCS3 mRNA levels were quantified using ImageQuant software and normalized to β-actin mRNA levels at each time point. Fold induction is the relative induction compared with untreated wild-type control mice.
Following fixation of the livers with 4% formalin/PBS, livers were sliced and stained with hematoxylin-eosin (H&E) for histological examination.
Liver injury was quantified by measurement of plasma enzyme activities of alanine transaminase (ALT) using a kit from Sigma-Aldrich (St. Louis, Missouri, USA).
Plasma levels of cytokines were measured by using standard ELISA sandwich kits as specified by the manufacturer (BioSource International).
Wild-type and knock-out mouse livers were removed and pressed through a 200-gauge stainless steel mesh. The liver cell suspension was collected and suspended in RPMI-1640 medium (Life Technologies Inc., Gaithersburg, Maryland, USA). Parenchymal cells (pellet) were separated from mo-nonuclear cells (MNCs) (supernatant) by centrifugation at 50 g for 5 minutes. Supernatant containing MNCs was collected, washed in PBS, and resuspended in 40% Percoll (Sigma-Aldrich) in RPMI-1640 medium. The cell suspension was gently overlaid onto 70% Percoll and centrifuged for 20 minutes at 750 g. MNCs were collected from the interface, washed twice in PBS, and resuspended in RPMI-1640 medium.
Activation of CD4+ T cells was determined by anti-CD4 plus anti-CD69 (early activation marker) (BD PharMingen) using a FACScalibur (Becton Dickinson Immunocytometry Systems, Mountain View, California, USA). NK T cell number was determined by anti-NK1.1 plus anti-CD3 Ab’s (BD PharMingen).
For comparing values obtained in three or more groups, one-factor ANOVA was used, followed by Tukey’s post hoc test, and P values less than 0.05 were taken to imply statistical significance.
After injection of Con A, plasma levels of a wide variety of cytokines are dramatically elevated (13). To determine the interaction of these cytokines in Con A–induced hepatitis, activation of the JAK-STAT signaling pathway in the liver was examined. As shown in Figure Figure1a,1a, Con A injection rapidly induced activation of JAK1 and JAK2, but not Tyk2. Consistent with activation of JAK1 and JAK2, multiple STATs were also activated. As shown in Figure Figure1a,1a, STAT1, STAT3, STAT5, and STAT6 were activated in the liver with the peak effect occurring at 1–3 hours, whereas peak STAT4 activation was detected at 9 hours. Activation of STAT2 was not detected at any time point tested. Furthermore, injection of Con A significantly induced expression of both SOCS1 mRNA and SOCS3 mRNA, whereas it only slightly induced SOCS2 and CIS expression (Figure (Figure11a).
The robust activation of STAT1 and STAT3 suggested that the downstream effectors of STAT1 and STAT3 may be important mediators of T cell–mediated hepatitis. STAT1 and STAT3 play complimentary, counterregulatory roles in apoptosis during Con A–induced hepatitis, because STAT1 activates the proapoptotic proteins IRF-1 and p21cip1/waf1 (30, 32), while STAT3 activates the antiapoptotic proteins Bcl-2 and Bcl-XL (33). As shown in Figure Figure1b,1b, IRF-1 was activated in a biphasic response, with significant peak induction at 3 hours and again at 9 hours. Expression of Bcl-XL proteins was elevated with peak effect between 9 hours and 48 hours. Additionally, Bax, a proapoptotic protein, was significantly induced with peak effect at 9 hours. Caspase-3, an indicator of apoptosis, is generated as a 32-kDa precursor, which is enzymatically cleaved into p17/p12 mature subunits (34). Caspase-3 was activated 3–9 hours after Con A injection, as assessed by generation of the mature p17 subunit. On the contrary, expression of p21cip1/waf1, p53, and Bcl-2 proteins was not altered after injection of Con A.
To examine the potential role of STAT1 activation in T cell–mediated hepatitis, we compared Con A–induced hepatitis in STAT1–/– and STAT1+/+ mice. As shown in Figure Figure2,2, a and b, after injection of Con A, ALT activities were dramatically elevated in wild-type mice, but were not increased in STAT1–/– mice. Examination of liver pathology showed massive necrosis in STAT1+/+ mice, but not in STAT1–/– mice. A representative example of liver histology from STAT1+/+ and STAT1–/– mice obtained 9 hours after Con A injection is shown in Figure Figure22b.
Since IFN-γ has been shown to play an important role in Con A–induced hepatitis (14, 35), we compared the serum IFN-γ levels in both wild-type and STAT1–/– mice after injection of Con A. As shown in Figure Figure2c,2c, Con A injection–induced elevation of IFN-γ was significantly attenuated in STAT1–/– mice compared with wild-type mice. Next, we compared activation of the JAK-STAT signaling pathway in the liver of Con A–induced hepatitis in STAT1+/+ and STAT1–/– mice. As shown in Figure Figure2d,2d, JAK1 activation remained unchanged and JAK2 activation was decreased in STAT1–/– mice compared with STAT1+/+ mice. Activation and induction of STAT1, as expected, were absent in STAT1–/– mice, whereas activation of STAT3 was significantly enhanced and prolonged in these mice. Consistent with absent STAT1 activation and strong STAT3 activation in STAT1–/– mice, induction of IRF-1 protein was not detected, whereas induction of Bcl-XL protein was significantly enhanced in these mice (Figure (Figure2d).2d). Expression of SOCS in the liver of Con A–induced hepatitis was also examined. A 20- to 25-fold induction of SOCS1 mRNA expression was observed between 3 hours and 9 hours in STAT1+/+ mice, whereas five- to eightfold induction was detected in STAT1–/– mice (Figure (Figure2d).2d). In contrast with decreased SOCS1 induction, Con A injection–induced SOCS3 mRNA expression was significantly enhanced at 1 hour and 48 hours in STAT1–/– mice compared with STAT1+/+ mice. Similar weak activation of SOCS2 and CIS was detected in both STAT1+/+ and STAT1–/– mice.
The above data indicate that Con A injection-induced STAT3 activation in the liver is prolonged in STAT1–/– mice, which could be due to either increased levels of STAT3-activating cytokines or increased response of STAT1–/– hepatocytes to STAT3 activation. Our observation that Con A injection–induced activation of JAKs was not enhanced in STAT1–/– mice (Figure (Figure2d)2d) suggested that Con A–induced prolonged activation of STAT3 in STAT1–/– mice is not due to increased levels of STAT3 activating cytokines such as IL-6. Indeed, Con A injection–induced elevation of serum IL-6 levels was decreased in STAT1–/– mice compared with wild-type mice (data not shown). To examine whether STAT1–/– hepatocytes were more susceptible to STAT3 activation, mouse hepatocytes from STAT1+/+ and STAT1–/– mice were stimulated with IFN-γ and IL-6, which are the two major cytokines responsible for STAT1 and STAT3 activation, respectively, in the livers of mice with Con A–induced hepatitis (see Figures Figures44 and and5).5). As shown in Figure Figure3a,3a, IFN-γ treatment induced STAT1 activation and IRF-1 expression in STAT1+/+ mouse hepatocytes, but not in STAT1–/– mouse hepatocytes. On the contrary, the same IFN-γ treatment caused prolonged STAT3 activation and enhanced Bcl-XL expression in STAT1–/– mouse hepatocytes (Figure (Figure3a).3a). IFN-γ–induced SOCS1 mRNA expression was completely abolished in STAT1–/– mouse hepatocytes, whereas induction of SOCS3 mRNA expression was enhanced and prolonged in these cells, compared with wild-type cells (Figure (Figure33a).
The effects of IL-6 on STAT activation and SOCS induction in STAT1+/+ and STAT1–/– mouse hepatocytes are shown in Figure Figure3b.3b. IL-6 treatment caused weak STAT1 activation and IRF-1 induction in STAT1+/+, but not in STAT1–/– mouse hepatocytes. On the contrary, the same IL-6 treatment slightly enhanced STAT3 activation and Bcl-XL protein expression in STAT1–/– mouse hepatocytes compared with STAT1+/+ mouse hepatocytes. IL-6 treatment induced enhanced and prolonged SOCS3 mRNA expression in STAT1–/– mouse compared with STAT1+/+ mouse hepatocytes, but did not affect SOCS1 mRNA expression in either cell type. Taken together, these findings suggest that Con A–induced prolonged activation of STAT3 in STAT1–/– mice is caused by increased response of STAT1–/– hepatocytes to STAT3 activation.
IFN-γ, which is elevated in the serum after administration of Con A (Figure (Figure2c),2c), is an important activator of STAT1 signaling. To explore the role of IFN-γ/STAT1 signaling in T cell–mediated hepatitis, we injected IFN-γ–/– mice with Con A and assessed the subsequent molecular and pathological changes in the liver. Con A injection–induced STAT1 activation and induction were almost completely abolished in the liver of IFN-γ–/– mice, whereas activation of STAT3 was slightly but not significantly enhanced in IFN-γ–/– mice compared with IFN-γ+/+ mice; JAK1 and JAK2 activation were decreased in IFN-γ–/– mice (Figure (Figure4a).4a). Consistent with weak STAT1 activation, IRF-1 gene expression was not upregulated in IFN-γ–/– mice. Bcl-XL protein expression was slightly reduced in IFN-γ–/– mice compared with IFN-γ+/+ mice. Additionally, a 13- to 24-fold induction of SOCS1 mRNA was observed between 3 hours and 9 hours after injection of Con A in IFN-γ+/+ mice, whereas only a eight- to tenfold increase was detected in IFN-γ–/– mice (Figure (Figure4a).4a). Furthermore, induction of SOCS3 expression was slightly but not significantly enhanced in IFN-γ–/– mice compared with IFN-γ+/+ mice. Weak activation of SOCS2 and CIS was seen in both IFN-γ+/+ and IFN-γ–/– mice.
Next we compared Con A–induced liver injury in IFN-γ+/+ and IFN-γ–/– mice. As shown in Figure Figure4b,4b, injection of Con A significantly induced serum ALT levels in IFN-γ+/+ mice, with a peak effect at 9 hours; in comparison, serum ALT levels were not significantly increased in IFN-γ–/– mice. Examination of liver histology showed massive necrosis in IFN-γ+/+ mice, whereas IFN-γ–/– mice appeared to be largely protected from hepatic injury. A representative example of liver histology from 9-hour Con A–injected IFN-γ+/+ mice and IFN-γ–/– mice is shown in Figure Figure44c.
STAT3 and its downstream antiapoptotic factors such as Bcl-2 and Bcl-XL are largely activated by the cytokine IL-6. To determine the role of IL-6/STAT3 signaling in hepatitis, IL-6–/– mice were injected with Con A, and the subsequent molecular and pathological changes in the liver were assessed. As shown in Figure Figure5a,5a, Con A injection–mediated activation of STAT3 was markedly attenuated at 1 and 3 hours, whereas STAT1 activation was significantly potentiated at 3, 6, and 9 hours in IL-6–/– mice compared with IL-6+/+ mice; JAK1 and JAK2 activation were downregulated in IL-6–/– mice. Consistent with upregulation of STAT1 activation, induction of IRF-1 gene expression was enhanced in IL-6–/– mice. On the contrary, induction of Bcl-XL protein was diminished in IL-6–/– mice, which correlated with downregulation of STAT3 activation in these mice. Con A injection–induced SOCS mRNA expression was also examined. As shown in Figure Figure5a,5a, Con A injection–induced SOCS1 mRNA expression was enhanced and prolonged, whereas SOCS3 mRNA induction was attenuated in IL-6–/– mice compared with IL-6+/+ mice. Furthermore, weak induction of CIS observed in IL-6+/+ mice was barely detected in IL-6–/– mice. Similar activation of SOCS2 was seen in both IL-6+/+ and IL-6–/– mice.
Liver injury was also compared in IL-6+/+ and IL-6–/– mice. As shown in Figure Figure5b,5b, Con A injection–induced elevation of ALT levels was markedly increased and prolonged in IL-6–/– mice compared with IL-6+/+ control mice. Likewise, histological examination of liver revealed much more sever necrosis in IL-6–/– mice compared with IL-6+/+ control mice (Figure (Figure5c).5c). Finally, the levels of serum IFN-γ in wild-type and IL-6–/– mice were determined. As shown in Figure Figure5d,5d, elevation of serum IFN-γ levels at 3 and 9 hours after administration of Con A is significantly enhanced in IL-6–/– mice compared with wild-type mice.
The above data indicate that Con A–induced liver injury is exacerbated in IL-6–/– mice, suggesting that IL-6 imparts protective effects against hepatic injury. To explore this hypothesis, we examined whether administration of IL-6 protects against Con A–induced liver injury. As shown in Figure Figure5e,5e, injection of IL-6 markedly suppressed Con A–induced liver injury, consistent with a previous report (36).
Activation of CD4+ and NK T cells, which have been shown to play an important role in Con A–induced hepatitis (11, 37), was examined in the IFN-γ, STAT1, and IL-6 knockout mice. Liver lymphocyte activation was determined by FACS analysis of early activation marker CD69+. As shown in Figures Figures6,6, a, b, and c, administration of Con A caused a significant increase in CD69+ positive cells (lower- and upper-right quadrant in CD4+/CD69+ panel) in three strains of wild-type mice and IL-6–/– mice, but such activation was almost completely abolished in STAT1–/– mice and IFN-γ–/– mice. These findings suggest that activation of total liver lymphocytes including CD4+ and NK T cells is suppressed in STAT1–/– mice and IFN-γ–/– mice. Furthermore, Con A injection caused a significant increase in CD4+CD69+ double-positive cells (upper-right quadrant in CD4+/CD69+ panel) in wild-type mice and IL-6–/– mice, but not in STAT1–/– mice and IFN-γ–/– mice (Figure (Figure6,6, a, b, and c), suggesting that activation of CD4+ cells is inhibited in STAT1–/– mice and IFN-γ–/– mice.
It has been shown that NK T cells are rapidly activated and consequently depleted after administration of Con A (11, 37), suggesting that NK T cell activation results in depletion. As shown in Figure Figure6,6, injection of Con A led to significant depletion of NK T cells (NK1.1+CD3+ double-positive cells in upper-right quadrant) in wild-type mice, but not in STAT1–/– and IFN-γ–/– mice. On the contrary, similar NK T cell depletion was observed in IL-6–/– mice compared with wild-type mice (Figure (Figure6,6, c and d). These findings suggest that NK T cell activation is impaired in STAT1–/– and IFN-γ–/– mice, but not in IL-6–/– mice in Con A–mediated hepatitis.
Here, data from studies using an established model of Con A–induced hepatitis suggest that T cell–mediated liver injury is tightly controlled by STAT1 and STAT3, which is summarized in a proposed model in Figure Figure7.7. In this model, IFN-γ activation of STAT1 plays a harmful role in Con A–induced liver injury by activation of CD4+ and NK T cells and directly inducing hepatocyte death. IL-6 activation of STAT3 protects against Con A–induced liver injury by suppression of IFN-γ signaling and induction of Bcl-XL. STAT1 and STAT3 in hepatocytes also negatively regulate one another through the induction of SOCSs.
STAT1 was activated and induced after injection of Con A (Figures (Figures1,1, ,2,2, ,4,4, and and5).5). IFN-γ is the major cytokine responsible for STAT1 activation and induction because such activation was almost completely suppressed in IFN-γ–/– mice (Figure (Figure4).4). Disruption of the IFN-γ gene or the STAT1 gene abolished elevated ALT activities and necrosis in Con A–induced hepatitis (Figures (Figures22 and and4),4), suggesting that IFN-γ/STAT1 play an essential role in Con A–induced hepatitis. In addition to this model, IFN-γ/STAT1 have also been implicated in LPS/D-galactosamine–induced liver injury (our unpublished data) and SOCS1 deficiency–induced hepatitis (38, 39). Activation of CD4+ and NK T cells has been shown to play a critical role in Con A–mediated hepatitis (11, 37); however, such activation was markedly attenuated in STAT1–/– or IFN-γ–/– mice (Figure (Figure6),6), indicating that IFN-γ/STAT1 is essential for CD4+ and NKT cell activation in Con A–mediated hepatitis. The mechanism underlying IFN-γ/STAT1-mediated activation of CD4 and NK T cells is not clear. Ivashkiv et al. (40) has reported that treatment of T cells with phytohemagglutinin resulted in a delayed activation of STAT1, possibly through induction of IFN-γ. Therefore, it is plausible that Con A activates CD4+ and NK T cells by an IFN-γ/STAT1-dependent pathway, leading to liver damage. Additionally, IFN-γ/STAT induction of the proapoptotic protein IRF-1 in hepatocytes and subsequent induction of hepatocyte apoptosis may also contribute to liver damage in Con A–mediated hepatitis, because significant apoptosis has been reported in Con A–induced hepatitis (13, 14, 41–45) and IFN-γ is able to directly cause or synergistically potentiate TNF-α–induced hepatocyte apoptosis by a IRF-1–dependent mechanism (ref. 46, our unpublished data).
IL-6 activation of STAT3 plays an important role in protection against liver injury induced by various toxins (25–28). Here we demonstrate that attenuated STAT3 signaling in IL-6–/– mice correlates with exacerbated liver injury (Figure (Figure5b)5b) and that administration of exogenous IL-6 prevents Con A–induced liver injury (Figure (Figure5e),5e), suggesting that IL-6/STAT3 protects against liver damage in Con A–mediated hepatitis. The protective effects of IL-6/STAT3 is unlikely to be mediated by suppression of inflammation because similar CD4+ and NK T cell activation were observed in IL-6–/– and wild-type mice in Con A–mediated hepatitis (Figure (Figure6).6). Con A injection–induced elevation of serum IFN-γ levels, STAT1 activation, and STAT1-controlled proapoptotic IRF-1 gene are enhanced in IL-6–/– mice (Figure (Figure5)5) but suppressed by administration of recombinant IL-6 (our unpublished data and ref. 47), suggesting that suppression of IFN-γ production and signaling is probably one of the major mechanisms contributing to the protective effect of IL-6/STAT3 in Con A–induced hepatitis. Additionally, induction of the antiapoptotic Bcl-XL protein expression in hepatocytes may be another mechanism contributing to IL-6/STAT3-mediated protection against Con A–induced liver injury because Bcl-XL, which is elevated and positively regulated by IL-6/STAT3 in Con A–induced hepatitis (Figure (Figure2d2d and Figure Figure5a),5a), has been implicated in protection against hepatocyte death induced by a variety of hepatoxins, including CCl4 (26), Fas ligand (27, 48), alcohol (28), and TNF-α (48).
IL-6 is the major cytokine responsible for STAT3 activation in liver regeneration (25) and CCl4-induced liver injury (26), since such activation is not detectable in IL-6–/– mice under these conditions. However, STAT3 activation in Con A–induced hepatitis was reduced only about 50% in IL-6–/– mice (Figure (Figure5),5), suggesting that other cytokines are also involved in STAT3 activation in this model. Three lines of evidence suggest that IFN-γ partially contributes to STAT3 activation in Con A–induced hepatitis. First, IFN-γ activated STAT3 in primary mouse hepatocytes in vitro (Figure (Figure3)3) and in primary human hepatocytes (31). Second, injection of IFN-γ induced significant activation of STAT3 in the liver (data not shown). Finally, elimination of STAT1 activation in STAT1–/– mice markedly enhanced STAT3 activation after injection of Con A, while disruption of STAT1 activation in IFN-γ–/– mice only slightly enhanced STAT3 activation. This is probably because IFN-γ–mediated activation of STAT3 signal is also disrupted in IFN-γ–/– mice.
In the present paper, we clearly demonstrate that activation of STAT1 promotes, whereas activation of STAT3 protects against, Con A–induced liver injury. In addition to such functional antagonism, STAT1 and STAT3 also mutually inhibit one another, at least in hepatocytes, by the induction of SOCS. To date, eight SOCS proteins (SOCS1–SOCS7 and CIS) have been identified (20–22). Among them, SOCS1 and SOCS3 appear to be the most potent inhibitors of cytokine signaling (49). Here, we demonstrate that both SOCS1 and SOCS3 are markedly induced, whereas SOCS2 and CIS are only slightly enhanced in the liver of Con A–induced hepatitis, suggesting that SOCS1 and SOCS3 may be the most important suppressors of STAT1 and STAT3 signaling in this model. Several lines of evidence indi-cate that induction of SOCS1 in the liver is mainly controlled by STAT1 activation. First, disruption of the STAT1 gene in STAT1–/– mice almost completely abolished Con A–mediated induction of SOCS1, although STAT3 activation was markedly enhanced and prolonged in these mice (Figure (Figure2).2). Second, IFN-γ–mediated induction of SOCS1 was completely abolished in primary STAT1–/– mouse hepatocytes, whereas induction of SOCS3 was enhanced in these cells. Third, enhanced STAT1 activation correlated with prolonged induction of SOCS1 mRNA expression in IL-6–/– mice (Figure (Figure5).5). Finally, STAT1 plays an essential role in induction of IRF-1 (Figures (Figures2,2, ,3,3, and and4),4), a key transcription factor controlling SOCS1 gene transcription (50). SOCS1 was originally identified in a functional screen for factors that inhibit IL-6 signaling (51) and later identified as a general negative feedback inhibitor for cytokine signaling (21, 22). Thus, STAT1-mediated induction of SOCS1 not only is a negative feedback loop for STAT1 activation, but also attenuates activation of STAT3 in Con A–induced hepatitis.
SOCS3 was induced in Con A–mediated hepatitis, and such induction positively correlated with STAT3 activation (Figure (Figure2d2d and Figure Figure5a).5a). Elimination of the STAT1 gene did not reduce, but rather enhanced, SOCS3 induction (Figure (Figure2d2d and Figure Figure3).3). Collectively, these findings suggest that STAT3 plays a more important role than STAT1 in inducing SOCS3 in this model of hepatitis. Induction of SOCS3 controlled predominantly by STAT3 in the liver is also suggested in another model of liver regeneration (52). In this model, SOCS3 but not SOCS1 is markedly induced, which correlates with activation of STAT3 but absent STAT1 activation (52). SOCS3 has been shown to negatively regulate STAT1 and STAT3 activation by inhibition of JAKs (21, 22). Thus, it is very likely that induction of SOCS3 attenuates both STAT1 and STAT3 in Con A–induced hepatitis.
In summary, our findings suggest that T cell–induced hepatitis is tightly controlled by mutual functional antagonism of IFN-γ/STAT1 and IL-6/STAT3, which negatively regulate one another through the induction of SOCS. Elevation of IFN-γ and IL-6 is observed not only in this Con A–induced murine model of hepatitis, but also in a variety of human liver disorders, including viral hepatitis (7, 53–56), autoimmune hepatitis (6, 56, 57), primary biliary cirrhosis (58), and alcoholic liver disease (59), suggesting that human liver disorders may also be controlled by mutual antagonism of IFN-γ/STAT1 and IL-6/STAT3. Modulation of such mutual antagonism between STAT1 and STAT3 could offer a novel approach in the treatment of T cell–mediated liver damage in human liver disease.
Conflict of interest: The authors have declared that no conflict of interest exists.
Nonstandard abbreviations used: concanavalin A (Con A); Janus kinase-signal transducer and activator of transcription factor (JAK-STAT); suppressor of cytokine signaling (SOCS); cytokine-inducible SH2-containing protein (CIS); carbon tetrachloride (CCl4); interferon regulatory factor-1 (IRF-1); enhanced chemifluorescence reaction (ECF); hematoxylin-eosin (H&E); alanine transaminase (ALT); mononuclear cells (MNCs).