Interferon-gamma–mediated tissue factor expression contributes to T-cell-mediated hepatitis through induction of hypercoagulation in mice§


  • Potential conflict of interest: Nothing to report.

  • This work was supported, in part, by JSPS KAKENHI (grant nos.: 22659328, 24659806, and 23659660) and a Hitec Research Center Grant from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.


Concanavalin A (Con A) treatment induces severe hepatitis in mice in a manner dependent on T cells, interferon (IFN)-gamma, and tumor necrosis factor (TNF). Treatment with the anticoagulant heparin protects against hepatitis, despite healthy production of IFN-γ and TNF. Here, we investigated molecular and cellular mechanisms for hypercoagulation-mediated hepatitis. After Con A challenge, liver of wild-type (WT) mice showed prompt induction of Ifnγ and Tnf, followed by messenger RNA expression of tissue factor (TF) and plasminogen activator inhibitor-1 (PAI-1), which initiate blood coagulation and inhibit clot lysis, respectively. Mice developed dense intrahepatic fibrin deposition and massive liver necrosis. In contrast, Ifnγ−/− mice and Ifnγ−/−Tnf−/− mice neither induced Pai1 or Tf nor developed hepatitis. In WT mice TF blockade with an anti-TF monoclonal antibody protected against Con A–induced hepatitis, whereas Pai1−/− mice were not protected. Both hepatic macrophages and sinusoidal endothelial cells (ECs) expressed Tf after Con A challenge. Macrophage-depleted WT mice reconstituted with hematopoietic cells, including macrophages deficient in signal transducer and activator of transcription-1 (STAT1) essential for IFN-γ signaling, exhibited substantial reduction of hepatic Tf and of liver injuries. This was also true for macrophage-depleted Stat1−/− mice reconstituted with WT macrophages. Exogenous IFN-γ and TNF rendered T-cell-null, Con A–resistant mice deficient in recombination-activating gene 2, highly susceptible to Con A–induced liver injury involving TF. Conclusions: Collectively, these results strongly suggest that proinflammatory signals elicited by IFN-γ, TNF, and Con A in both hepatic macrophages and sinusoidal ECs are necessary and sufficient for the development of hypercoagulation-mediated hepatitis. (HEPATOLOGY 2013)

Concanavalin A (Con A)-induced hepatitis is a well-characterized, representative mouse model of T-cell-mediated acute liver failure.1 After Con A challenge, mice show elevation of circulating proinflammatory cytokine levels, subsequently resulting in massive liver necrosis with dense infiltration of leukocytes. Because interferon (IFN)-γ or tumor necrosis factor (TNF) blockade and gene depletion of Ifnγ or Tnf rescues mice from Con A–induced hepatitis,2, 3 IFN-γ and TNF are convincingly regarded as the cytokines necessary for the development of this type of liver injury. Thus, one may assume that endogenous IFN-γ and TNF initiate both the hepatic inflammatory responses and the liver parenchymal cell death.4 However, as previously reported, massive liver necrosis is accompanied by severe thrombocytopenia and intrahepatic hemostasis, and heparin pretreatment substantially protects against liver injury without down-regulating the production of IFN-γ and TNF.2 This may imply that microcirculatory disturbances resulting from hepatic thrombosis contribute to liver injury independent of IFN-γ/TNF-mediated hepatic inflammation and hepatocytotoxicity. Alternatively, IFN-γ and/or TNF might be causative for hepatic thrombosis, perhaps by inducing procoagulant activity within the liver. Thus, it is important to elucidate whether and how IFN-γ and/or TNF contribute to the hepatic hypercoagulation and whether IFN-γ and/or TNF are sufficient to trigger these pathological changes.

Tissue factor (TF) is a transmembrane cofactor for the coagulation factor, VIIa, and is constitutively expressed in the blood vessel wall and its expression is induced by various mediators in several cell types including macrophage (Mø) and endothelial cells (ECs).5-7 Endothelial damage or TF expression on circulating monocytes/macrophages brings TF in contact with circulating factor VIIa to initiate the blood coagulation cascade, which eventually results in the activation of prothrombin, leading to fibrin formation and platelet activation. The coagulation system is tightly regulated by the fibrinolytic system, which comprises plasminogen, the tissue-type plasminogen activator (tPA), and its inhibitor, plasminogen activator inhibitor-1 (PAI-1).8, 9 Pai1−/− mice have been reported to be resistant to alcohol-induced or cholestatic liver injuries.10-12 Therefore, it is possible that PAI-1 as well as TF may play a role in coagulation-mediated liver injuries.

In this study, we investigated the mechanisms by which Con A treatment induces the prothrombotic state. We found strong induction of hepatic Tf and Pai1 expressions, dense hepatic fibrin deposits, and massive liver necrosis in Con A–treated wild-type (WT) mice, but not in Ifnγ−/−Tnf−/− mice. TF blockade protected WT mice from the intrahepatic fibrin deposition and resultant hepatitis. Both hepatic macrophages (Mø) and sinusoidal ECs (SECs) expressed Tf in Con A–challenged WT mice. IFN-γ signaling was crucial for Tf induction in both these cell types. Con A–resistant mice that have Mø and SECs, but not T cells, became highly susceptible to Con A when treated simultaneously with IFN-γ and TNF. Collectively, these results indicate that IFN-γ-, TNF-, and Con A–activated signaling pathways in hepatic Mø and SECs are necessary and sufficient for the development of intrahepatic hemostasis-mediated massive liver injuries.


Abs, antibodies; ALI, acute liver injury; ALT, alanine aminotransferase; B6, C57BL/6; BM, bone marrow; Ccl2, CC chemokine ligand 2 gene clodronate liposome, liposome-encapsulated dichloromethylene bis-phosphonate; Con A, concanavalin A; ECs, endothelial cells; H&E, hematoxylin and eosin; IFN, interferon; IFNAR1, IFN-α receptor 1; IgG, immunoglobulin G; IHC, immunohistochemistry; Il6, interleukin-6 gene; Il1β, interleukin-1β gene; IP, intraperitoneal; IV, intravenously; KO, knockout; Mø, macrophages; mAb, monoclonal antibody; mRNA, messenger RNA; PAI-1, plasminogen activator inhibitor-1; PBS, phosphate-buffered saline; qRT-PCR, quantitative real-time reverse-transcriptase polymerase chain reaction; RAG2, recombination-activating gene 2; rIFN-γ, recombinant IFN-γ; rRNA, ribosomal RNA; rTNF, recombinant TNF; SC, subcutaneously; SEC, sinusoidal endothelial cells; STAT1, signal transducer and activator of transcription 1; TAT, thrombin antithrombin III complex; TF, tissue factor; TNF, tumor necrosis factor; tPA, tissue plasminogen activator; TUNEL, terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphage nick-end labeling; WT, wild type.

Materials and Methods


Con A was purchased from J-Oil Mills (Tokyo, Japan). Neutralizing rat antimouse TF monoclonal antibody (mAb) (1H1) was described elsewhere.13 Purified rat immunoglobulin G (IgG) was purchased from Beckman Coulter (Fullerton, CA). Recombinant murine IFN-γ and TNF were from PeproTech (Rocky Hill, NJ). Liposome-encapsulated dichloromethylene bis-phosphonate (clodronate liposome) and phosphate-buffered saline (PBS) liposomes were prepared as described previously.14, 15

Induction of Acute Hepatitis.

Con A was administered to mice (20 mg/kg) through a tail vein.2 In some experiments, mice received Con A intravenously (IV), promptly followed by intraperitoneal (IP) treatment with recombinant IFN-γ (rIFN-γ; 500 ng) and recombinant TNF (rTNF; 500 ng). In some experiments, mice were treated IP with neutralizing anti-TF mAb, 1H1, or subcutaneously (SC) with heparin (5,000 U/kg) 30 minutes before Con A challenge. At various time points after challenge, plasma and liver specimens were sampled.16 Plasma alanine aminotransferase (ALT) and aspartate aminotransferase levels were measured (SRL, Osaka, Japan).

Quantitative Real-Time Reverse-Transcriptase Polymerase Chain Reaction.

We performed quantitative real-time reverse-transcriptase polymerase chain reaction (qRT-PCR), as shown in the Supporting Materials. RNA content was normalized based on amplification of 18S ribosomal RNA (rRNA) (18S).17 Change folds = normalized data of experimental sample/normalized data of control.

Assay for Thrombin Antithrombin III Complex.

Plasma levels of thrombin antithrombin III complex (TAT) were measured by commercially available kits for TAT (Enzyme Research Laboratories, South Bend, IN).16

Preparation of Liver Cells.

Hepatic nonparenchymal cells from 3 mice were pooled.15 CD11b+ hepatic Mø and CD146+ SECs were then enriched by magnetic-activated cell sorting (Miltenyi Biotec GmbH, Cologne, Germany) using anti-CD11b and anti-CD146 microbeads (Miltenyi Biotec), according to the manufacture's instruction, respectively. Stellate cells and liver parenchymal cells were prepared as described.18, 19

Histological and Immunochemical Analyses.

Formalin-fixed tissue sections were stained with hematoxylin and eosin (H&E).2

For detection of fibrin deposition, livers were perfused through a portal vein with PBS,2 and liver specimens were rapidly sampled, fixed in 10% zinc fixative (Becton Dickinson, San Diego CA), and embedded in paraffin. Tissue sections were incubated overnight with rabbit antimouse fibrinogen antiserum (1:5,000) (Molecular Innovations, Inc., Novi, MI), followed by treatment with the rabbit Vectastatin Elite ABC kit (Vector Laboratories, Burlingame, CA). Antigen-antibody (Ab) complexes were detected by using a DAB Substrate Kit (Vector Laboratories). Formalin-fixed liver sections were analyzed for apoptosis by terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphage nick-end labeling (TUNEL) assay.20

In Vivo Depletion of Mø.

Mø were depleted by the IV injection of clodronate liposome, as described previously.14

Mouse Reconstitution.

To abolish irradiation-resistant Mø, we injected IV clodronate liposome into host mice and, 2 days later, irradiated them, followed by transfer of donor bone marrow (BM) cells.21, 22 CD45.1 WT mice were transferred with CD45.2 WT or CD45.2 Stat1−/− BM cells, and CD45.2 Stat1−/− mice were transferred with CD45.1 WT BM cells.22 Two months later, the reconstituted mice were used.

Statistical Analyses.

All data are shown as the mean ± standard deviation of samples in each experimental group. Five to seven mice were used for each experimental group. Significance between the experimental and control groups was examined by the unpaired Student t test. P values less than 0.05 were considered significant. Two to three experiments were separately performed, and representative data were shown.


IFN-γ- and TNF-Dependent Hepatic Hypercoagulation Underlies Con A–Induced Hepatitis.

We previously reported that by use of electron microscopy, many microthrombi, consisting of platelets, red blood cells, and fibrin deposits were observed in hepatic sinusoids of Con A–treated mice.2 Immunohistochemistry (IHC) with antifibrinogen Abs further substantiated the dense fibrin deposition in the hepatic sinusoids (Fig. 1A, right middle lower panel). Pretreatment with the anticoagulant, heparin, protected against Con A–induced liver injuries with abundant TUNEL-positive hepatocytes and resulted in greatly reduced fibrin deposition (Fig. 1A). Consistent with our previous report,2 heparin pretreatment did not down-regulate hepatic Ifnγ or Tnf (Fig. 1B). This is also true for interleukin-1β (Il1β), interleukin-6 (Il6), and CC chemokine ligand 2 (Ccl2) genes (Supporting Fig. 1). These results clearly indicated the importance of intrahepatic fibrin deposition for liver injury, and suggested that induction of these proinflammatory cytokines/chemokine was insufficient for the development of liver injuries in the absence of hepatic thrombosis.

Figure 1.

Both IFN-γ and TNF are necessary for the development of thrombus-associated ALI. (A and B) WT mice were treated SC with heparin (red columns) or vehicle (Veh, closed columns), then with Con A. At the indicated time point, plasma and liver specimens were sampled for measurement of ALT (A) and hepatic Ifnγ (B) and Tnf (B), respectively. Fold increase of mRNA expression was calculated after normalization to 18S (B). Histological (H&E) and immunohistological study for TUNEL and fibrin deposition (A) were also performed. (C and D) WT (closed bars), Tnf−/− (hatched bars), Ifnγ−/− (gray bars), and Ifnγ−/−Tnf−/− mice (open bars) were treated with Con A. At the indicated time point, plasma and liver specimens were sampled for measurement of TAT (C) and histological/immunohistological study, as shown in (B), respectively. Representative data are shown (A and D left panels). Original magnification, ×10 (A and D, left upper panels) and ×50 (A and D, left lower panels). Arrowheads indicated necrotic area.

Because plasma TAT is an excellent indicator of thrombin formation in the circulation,16 we measured plasma TAT levels of Con A–challenged mice. Concomitant with dense fibrin deposition in the liver (Fig. 1A), plasma TAT levels were strongly elevated after challenge of WT mice with Con A, indicating that Con A treatment induced a systemic coagulation response along with hepatic hypercoagulation. Our previous report revealed that blockade of IFN-γ and TNF reduced hepatic coagulation response.2 The involvement of IFN-γ and TNF was examined in more detail by using knockout (KO) mice, including single- and double-KO mice. Expectedly, Ifnγ−/−, Tnf−/−, and Ifnγ−/−Tnf−/− mice had lower TAT elevation than WT mice (Fig. 1C). This clearly indicated the requirement of IFN-γ and TNF for the Con A–induced hypercoagulation response. In agreement, Ifnγ−/−Tnf−/− mice lacked fibrin deposition and were protected from Con A–induced hepatitis (Fig. 1D). Ifnγ−/− mice, like Ifnγ−/−Tnf−/− mice, were free from liver injury, whereas Tnf−/− mice showed only partial reduction of liver injury (Fig. 1D, right panel), suggesting that endogenous IFN-γ is more important than TNF for promoting liver injury. In contrast, Ifnγ−/−Tnf−/− mice showed significantly reduced, but still substantial induction of, Il1β, Il6, and Ccl2 (Supporting Fig. 2). Taken together, these results demonstrated that both IFN-γ and TNF are important initiators in the development of massive liver necrosis, which is mediated by the induction of intrahepatic hypercoagulation.

Requirement of IFN-γ and TNF for Hepatic Induction of Tf and Pai1.

Next, we investigated how IFN-γ and/or TNF contributed to hepatic thrombosis. Because TF and PAI-1 were reported to induce the prothrombotic state,5, 9 we measured both Tf and Pai1 levels in livers of Con A–challenged WT mice. Hepatic Tf levels started to increase at 2 hours, with a peak at 3 hours after Con A challenge (Fig. 2A,B). Hepatic Pai1 levels began to increase at 1 hour and peaked at approximately 3-6 hours (Fig. 2A,B). Intriguingly, both Ifnγ and Tnf levels increased immediately after Con A challenge, and the peaks of Ifnγ and Tnf preceded those of Tf and Pai1 (Fig. 2A). In sharp contrast to WT mice, Ifnγ−/− mice showed no increase in Tf and only little increase in Pai1 levels (Fig. 2B), indicating the importance of IFN-γ for the induction of both Tf and Pai1. This was also the case for Ifnγ−/−Tnf−/− mice (Fig. 2B). Tnf−/− mice showed poor induction of Tf and Pai1 as well, but their levels were significantly higher than those of Ifnγ−/− mice and Ifnγ−/−Tnf−/− mice (Fig. 2B). Compared to Tf and Pai1, the Con A–mediated increase of messenger RNA (mRNA) levels of tPA, a target protease of PAI-1, were much less pronounced and peaked at a much later time point in WT mice (Fig. 2B). In addition, tpa levels were only slightly reduced in Ifnγ−/−, Tnf−/−, and Ifnγ−/−Tnf−/− mice (Fig. 2B). These results strongly suggested that Con A stimulates hepatic T cells to produce both IFN-γ and TNF, which then induce the expression of hepatic Tf and Pai1.

Figure 2.

Requirement of IFN-γ and TNF for the induction of hepatic Tf and Pai1. WT (closed bars), Tnf−/− (hatched bars), Ifnγ−/− (gray bars), or Ifnγ−/−Tnf−/− mice (open bars) were challenged IV with Con A, and their liver specimens were sampled for measurement of mRNA expression levels of IFN-γ (A), TNF (B), TF (A and B), PAI-1 (A and B), and tPA (B) by real-time qRT-PCR. Fold increase of mRNA expression was calculated after normalization to 18S.

Importance of TF, but Not PAI-1, for Liver Injuries.

To examine the respective roles of TF and PAI-1 for the hypercoagulation response, we determined the effects of Con A treatment in WT mice pretreated with an anti-TF mAb and in Pai1−/− mice. Compared to mice receiving control rat IgG, treatment with the neutralizing anti-TF mAb, 1H1, just before Con A challenge reduced ALT plasma levels and fibrin deposition in a concentration-dependent manner (Fig. 3A). This indicated the importance of TF in mediating liver injury. Notably, TF blockade protected against plasma elevation of TAT without affecting hepatic Ifnγ, Tnf, Il1β, Il6, and Ccl2 inductions (Fig. 3B and Supporting Fig. 3). In contrast, Pai1−/− mice underwent massive liver injuries similar to WT mice in respect to hepatic fibrin deposition, plasma TAT elevation and induction of hepatic Ifnγ, Tnf, and Tf (Fig. 4). These results demonstrated a pivotal role for TF, but not PAI-1, in hypercoagulation response and the development of liver injuries.

Figure 3.

TF is necessary for the development of Con A hepatitis. We administered various doses of neutralizing anti-TF mAb (open bars) or control rat IgG (Ctrl IgG) (closed bars) into WT mice 30 minutes before Con A challenge. At the indicated time points after Con A challenge, plasma and liver specimens were sampled for measurement of ALT (A) and TAT (B) as well as measurement of Ifnγ and Tnf expressions (B) and histological/immunohistological studies (A), respectively. Arrowheads indicated necrotic area.

Figure 4.

Dispensability of PAI-1. WT mice (closed bars) or Pai1−/− mice (open bars) were challenged with Con A. At the indicated time points after Con A challenge, plasma and liver specimens were sampled, followed by the experiments, as shown in the legend to Fig. 3. Arrowheads indicated necrotic area.

Liver Cells Both Inside and Outside of the Sinusoid Expressed Tf.

Various cell types are localized within the hepatic sinusoid, such as SECs and liver Mø, including Kupffer cells. To identify the cell types that expressed TF mRNA upon Con A challenge, we isolated hepatic CD11b+ Mø and CD146+ SECs from Con A–treated mice and measured Tf expression. Both Mø and SECs prepared from livers of mice at 3 hours after Con A challenge showed a remarkable increase in Tf expression levels, as compared to naïve mice (Fig. 5A). Furthermore, cells outside of the sinusoid, such as hepatocytes and stellate cells, also increased the expression of Tf after Con A challenge (Fig. 5A). These results suggested that TF on Mø and SECs directly triggered the coagulation cascade within the hepatic sinusoid.

Figure 5.

Importance of macrophages for hepatic fibrin deposition. (A) Hepatic Mø (closed bars), SECs (open bars), hepatocytes (hatched bars), and stellate cells (gray bars) were isolated from WT mice at the indicated time points after Con A challenge. Tf was measured by qRT-PCR (A). (B-D) WT mice, having received clodronate liposome (Cld-lip) or control PBS liposome (PBS-lip), were challenged with Con A. At the indicated time points after Con A challenge, plasma and liver specimens were sampled, followed by the method shown in the legend to Fig. 1. Arrowheads indicated necrotic area.

To analyze the roles of Mø in liver thrombosis, we generated Mø-depleted mice by injection of clodronate liposome.15 Upon Con A challenge, Mø-depleted WT mice displayed significant diminution in hepatic Tf induction without reduction in hepatic Ifnγ and Tnf induction, as compared to PBS liposome-pretreated control mice (Fig. 5B). This suggested that the impaired TF induction was not attributed to the impaired induction of the upstream cytokines, but was rather the result of a decrease in the number of TF-expressing cells. Furthermore, Mø-depleted mice showed impairment in plasma TAT increase, hepatic fibrin deposition, and liver injuries (Fig. 5C,D). The findings suggested that Mø were an important cellular source of functional TF.

Requirement of IFN-γ/STAT1 Signaling in Both Mø and SECs for the Hepatic Tf Induction.

Because endogenous IFN-γ appeared more important than TNF for hepatic Tf induction (Fig. 2B), we further investigated the IFN-γ signaling pathway in liver cells (Fig. 6A). Stat1−/− mice, like Ifnγ−/− mice (Fig. 1D), showed a strongly impaired hepatic Tf induction and completely evaded Con A hepatitis (Fig. 6A), indicating the importance of the IFN-γ/STAT1-signaling pathway for these events. To exclude the possible involvement of type I IFN-mediated STAT1 signaling, we carried out experiments with mice deficient in the receptor for type I IFN, IFNAR. Ifnar−/− mice displayed healthy hepatic induction of Tf and Tnf (Fig. 6B), indicating that STAT1-mediated Tf up-regulation is not dependent on type I IFN. Next, we examined whether hepatic Mø or nonhematopoietic liver cells, including SECs, hepatocytes, and stellate cells, were responsible for STAT1-dependent Tf expression. We generated reciprocal BM chimeric mice by using WT and Stat1−/− mice. Mø are somewhat irradiation resistant. To improve depletion of host Mø, we pretreated host mice with clodronate liposome before reconstitution.22 WT mice reconstituted with WT hematopoietic cells (control mice) showed Tf induction in their livers after Con A challenge (Fig. 6A). WT mice transferred with Stat1−/− BM cells exhibited partly impaired induction of Tf, as compared to the control mice (Fig. 6A). Stat1−/− mice reconstituted with WT hematopoietic cells showed further reduction in Tf induction, as compared to Stat1−/− mice receiving WT BM cells (Fig. 6A). IHC with antiphosphorylated STAT1 mAb revealed its nuclear localization in the corresponding WT Mø and nonhematopoietic liver cells of the chimeric mice (Supporting Fig. 4). Thus, the Tf inductions in Mø and nonhematopoietic liver cells were largely dependent on STAT1. WT mice transferred with Stat1−/− hematopoietic cells and Stat1−/− mice with WT BM cells both developed significantly mild liver injuries, compared to control mice (Fig. 6A). Intriguingly, severities of the liver injuries were comparable between these two types of chimeric mice (Fig. 6A). Stat1−/− mice reconstituted with Stat1−/− BM cells exhibited the phenotypes equivalent to Stat1−/− mice (data not shown). Collectively, these results strongly indicated that the IFN-γ/STAT1 signalings in both hematopoietic Mø and nonhematopoietic liver cells are equally important for the development of Con A hepatitis.

Figure 6.

A pivotal role of IFN-γ/STAT1 signaling in both hematopoietic and nonhematopoietic cells in Con A hepatitis. (A) Host mice were treated with clodronate liposome 2 days before reconstitution. After irradiation, host CD45.1 B6 WT received BM cells from congenic B6 CD45.2 WT (WT>WT) or Stat1−/− CD45.2 B6 mice (Stat1−/−>WT). Irradiated Stat1−/− B6 CD45.2 mice received BM cells from CD45.1 B6 WT (WT>Stat1−/−). Stat1−/− mice were used as well. The reconstituted mice were challenged with Con A. At the indicated time points, plasma and liver specimens were sampled for measurement of hepatic Tf, TUNEL assay, and plasma ALT levels. (B) WT mice (closed bars) or Ifnar1−/− mice (open bars) were challenged with Con A. ns, not significant.

Figure 7.

IFNγ, TNF, and Con A signalings, likely in hepatic Mø and SECs, are necessary and sufficient for hepatitis involving hypercoagulation. (A) Rag2−/− mice were treated with rIFN-γ plus rTNF, Con A or Con A plus rIFNγ plus rTNF, or, additionally, with control rat IgG (Ctrl IgG) or neutralizing anti-TF mAb (αTF). At 3 hours, plasma and liver specimens were sampled for measurement of TAT and Tf expression, respectively. At 9 hours, plasma and liver specimens were sampled for measurement of ALT and histological study, respectively. ns, not significant. Arrowheads indicated necrotic area. (B) A proposal model for Con A–induced liver injury. Upon Con A challenge, T cells produce IFN-γ and TNF, which, together with Con A, cooperatively stimulate sinusoidal Mø and SECs to produce TF in a STAT1-dependent manner. TF then starts to rapidly and aberrantly activate the coagulation cascade to generate hepatic sinusoidal thrombus, eventually leading to massive liver injuries.

Con A Signaling in Non-T Non-B Cells Collaborates With IFN-γ and TNF Signaling in Thrombosis-Mediated Liver Injury.

T cells have been documented to be essential for Con A hepatitis.1 In agreement, Rag2−/− mice lacking T and B cells did not show hepatic Tf induction, elevation of plasma TAT concentrations, or liver damage after Con A challenge (Fig. 7A). T cells, including natural killer T cells, are necessary for the production of IFN-γ and TNF.2, 23 Both Ifnγ and Tnf inductions were absent in the liver of Con A–challenged Rag2−/− mice (Supporting Fig. 5). Because both IFN-γ and TNF mediate hypercoagulation and liver injury (Fig. 1C,D), we hypothesized that T cells may contribute to liver damage by producing IFN-γ and TNF. To test this possibility, we administered rIFN-γ plus rTNF into Rag2−/− mice and challenged them with Con A. This treatment resulted in hepatitis accompanied by considerable hepatic Tf induction, along with Il1β, Il6, and Ccl2 inductions (Supporting Fig. 6) and a strong increase in plasma TAT levels (Fig. 7A). Notably, TF blockade protected against the elevation of plasma TAT levels and the liver injuries in Con A plus rIFN-γ/rTNF-treated Rag2−/− mice (Fig. 7A and Supporting Fig. 7). However, in the absence of Con A, treatment with IFN-γ and TNF alone could not induce any of those alterations (Fig. 7A). Thus, in addition to signals elicited by IFN-γ and TNF, Con A signaling in the cells of Rag2−/− mice, likely mediated by hepatic Mø and SECs, was required for the development of thrombosis-mediated liver injuries.


Results presented here demonstrate that both endogenous IFN-γ and TNF are essential for the development of Con A–induced liver injuries through the induction of TF-dependent coagulation. In particular, the IFN-γ/STAT1-signaling pathway, in both hepatic Mø and SECs, was directly critical for the development of hypercoagulation and resultant acute liver injuries (ALIs). However, exogenous or endogenous IFN-γ and TNF were not sufficient to induce the hypercoagulation response or liver injuries. However, exogenous IFN-γ and TNF rendered Rag2−/− mice highly susceptible to Con A treatment, suggesting that Con A, IFN-γ, and TNF act in concert on hepatic Mø and SECs to elicit a procoagulant response. Based on these results, we propose a model of Con A–induced acute liver damage (illustrated in Fig. 7B). After stimulation with Con A, T cells produce IFN-γ and TNF. In hepatic Mø and SECs within the sinusoid, the cellular signaling pathways initiated by Con A, TNF, and IFN-γ through STAT1 activation synergize to elicit a robust expression of TF. TF then activates the coagulation system, leading to hepatic fibrin deposition and liver injury.

The IFN-γ/STAT1-mediated signaling in SECs is important for Con A–induced liver injury. Mø-depleted Stat1−/− mice reconstituted with WT Mø showed reduction in hepatic Tf induction and evaded Con A–induced liver injury. This suggested that IFN-γ/STAT1 induction of TF in SECs might evoke the hypercoagulation response relevant to the liver injury. A recent report verified a crucial role of endogenous IFN-γ in SEC damage of Con A–treated mice.24 SEC damage has been believed to be a potent inducer of intrahepatic coagulation.25 Therefore, IFN-γ/STAT1-mediated induction of TF in SECs may contribute to intrahepatic coagulation within the context of IFN-γ/STAT1-mediated cellular damage.

Under normal conditions, hepatocytes and stellate cells are anatomically segregated from the sinusoid. However, IFN-γ induction of SEC damage allows them to be exposed to the sinusoidal circulation, which might facilitate thrombosis. Thus, IFN-γ/STAT1-mediated induction of TF in hepatocytes and stellate cells might amplify procoagulant response.

Con A is a well-known T-cell mitogen, suggesting an important role of T cells in Con A–induced hypercoagulation. However, our present results verified the replacement of T cells by IFN-γ/TNF and the importance of Con A signaling in non-T cells, presumably exemplified by hepatic Mø and SECs. We are currently investigating the signaling pathway of Con A in Mø and SECs.

Thrombin-cleaved osteopontin was shown to be involved in this type of hepatitis.23, 26 These reports are consistent with the view that hepatic thrombosis is an essential contributor to Con A–induced liver injuries, at least through the induction of the thrombin-cleaved form of osteopontin and, perhaps, hepatic microcirculatory disturbance.

Mouse hepatitis virus infection is associated with intrahepatic thrombosis.27, 28 Patients with chronic hepatitis C show increase in plasma TF levels, whereas those with viral clearance by IFN-α therapy, such as healthy controls, do not.29 Furthermore, there is a growing recognition of the role of hypercoagulation in chronic liver injury and fibrosis.30 Although our current study was focused on ALI, the similar mechanism likely underlies acute and chronic viral hepatitis and fibrosis.

In summary, our present study demonstrates that IFN-γ/STAT1-mediated signaling in hepatic various cells, including Mø and SECs, is the underlying mechanism of Con A–induced aberrant activation of coagulation resulting in massive liver necrosis. In hepatic Mø and liver sinusoid, signaling through the IFN-γ/STAT1 pathway induced expression of Tf, which, in conjunction with IFN-γ/STAT1-mediated damage to the endothelium, triggered the coagulation reactions. The resulting formation of extensive microthrombi induced microcirculatory disturbances and hepatic inflammation involving thrombin-mediated conversion of precursor proteins, eventually leading to massive liver necrosis. It is conceivable that similar mechanisms are driving the progression of lethal fulminant hepatitis. Although our study did not formally address this question, the findings presented here may incite future studies to investigate this possibility and, perhaps, lead to novel therapeutic approaches.


The authors thank Drs. Mutoh and Nakanishi, and Ms. Iwami, Ms. Mitani, and Ms. Mizobuchi in our college for their enthusiastic discussion and excellent technical assistance, respectively.