Interferon regulatory factor 3 deficiency leads to interleukin-17-mediated liver ischemia-reperfusion injury


  • Potential conflict of interest: Nothing to report.

  • The Institute for Medical Immunology is sponsored by the government of the Walloon Region and GlaxoSmithKline Biologicals. This study was also supported by the Fonds National de la Recherche Scientifique (FNRS, Belgium) and an Interuniversity Attraction Pole of the Belgian Federal Science Policy. P.L. is supported by Roche grants.


Interferon regulatory factor 3 (IRF3) is an important transcription factor in Toll-like receptor 4 (TLR4) signaling, a pathway that is known to play a critical role in liver ischemia-reperfusion injury. In order to decipher the involvement of IRF3 in this setting, we first compared the intensity of hepatic lesions in IRF3-deficient versus wildtype mice. We found increased levels of blood transaminases, enhanced liver necrosis, and more pronounced neutrophil infiltrates in IRF3-deficient mice. Neutrophil depletion by administration of anti-Ly6G monoclonal antibody indicated that neutrophils play a dominant role in the development of severe liver necrosis in IRF3-deficient mice. Quantification of cytokine genes expression revealed increased liver expression of interleukin (IL)-12/IL-23p40, IL-23p19 messenger RNA (mRNA), and IL-17A mRNA in IRF3-deficient versus wildtype (WT) mice, whereas IL-27p28 mRNA expression was diminished in the absence of IRF3. The increased IL-17 production in IRF3-deficient mice was functionally relevant, as IL-17 neutralization prevented the enhanced hepatocellular damages and liver inflammation in these animals. Evidence for enhanced production of IL-23 and decreased accumulation of IL-27 cytokine in M1 type macrophage from IRF3-deficient mice was also observed after treatment with lipopolysaccharide, a setting in which liver gamma-delta T cells and invariant natural killer T cells were found to be involved in IL-17A hyperproduction. Conclusion: IRF3-dependent events downstream of TLR4 control the IL-23/IL-17 axis in the liver and this regulatory role of IRF3 is relevant to liver ischemia-reperfusion injury. (HEPATOLOGY 2013)

Interruption of liver blood supply during resection surgery, trauma, or transplantation rapidly results in tissue damage upon reperfusion. It has been clearly recognized as a cause of up to 10% early graft dysfunction but also as a cause of acute and chronic rejection.1 During this phenomenon of ischemia-reperfusion (IR) injury, resident macrophages (Kupffer cells), sinusoidal endothelial cells, and hepatocytes are important players in the sequence of events which lead to adenosine triphosphate (ATP) depletion and formation of reactive oxygen species in the ischemic liver lobe.2 Oxidant stress can directly damage endothelial cells and hepatocytes and the initial wave of cell death favors the release of damage-associated molecular patterns (DAMPs) molecules, including endogenous Toll-like receptor (TLR) ligands that provide initiating signals for the innate immune system.3, 4 A number of DAMPs released by necrotic cells can engage TLR, including heat shock proteins, fibrinogen, extra domain A of fibronectin, heparan sulfate, soluble hyaluronan, surfactant protein-A, β-defensin, high-mobility group box 1 protein (HMGB1), and heme.5-7 Several of these molecules are indeed ligands for TLR4 that has been shown to be critically involved in liver IR injury.4, 8, 9

Signal transduction through TLR4 involves two major pathways: (1) the MyD88-dependent pathway leading to activation of nuclear factor kappa B (NF-κB)-dependent genes, including interleukin (IL)-1α, IL-6, tumor necrosis factor alpha (TNF-α), IL-12/IL-23p40, and IL23p19, and (2) the TRIF-dependent pathway which is critical for the activation of IRF-3 dependent genes, including IFN-β, CXCL-10, IL-12p35, and IL-27p28.10

Here we investigated the role of IRF3 and TRIF during liver IRI by comparing the intensity of liver lesions and the expression of cytokines genes in IRF3-deficient versus WT mice. The observation of enhanced tissue damage, decreased accumulation of IL-27, and increased accumulation of IL-23 and IL-17A led us to further define the role of IL-17A during IRI in IRF3-deficient mice and to study the liver sources of IL-23 and IL-17 upon TLR4 ligation in vivo.


ALT, alanine aminotransferase; AST, aspartate aminotransferase; ELISA, enzyme-linked immunosorbent assay; IL, interleukin; I/R, ischemia/reperfusion; IRF3, interferon regulatory factor 3; TLR, Toll-like receptor.

Materials and Methods


Eight- to 12-week-old female C57BL/6 (B6, H-2b) mice were used (Harlan, Zeist, Netherlands). Age and sex-matched B6 IRF3-deficient (IRF3−/−) mice were obtained from the Riken BioResource Center (Ibaraki, Japan) with the approval of T. Taniguchi (University of Tokyo, Tokyo, Japan). TRIF-deficient mice were obtained from François Trottein (Institut Pasteur, Lille, France). Mice were housed and bred in our specific pathogen-free animal facility. All animal studies were approved by the Institutional Animal Care and Local Committee for Animal Welfare.

Liver Ischemia/Reperfusion.

Under anesthesia (xylazine 50 mg/kg and ketamine 100 mg/kg), a midline laparotomy was performed on prehydrated (0.9% NaCl, 300 μL) mice. The portal triad (hepatic artery, portal vein, bile duct) to the left and median cephalic liver lobes was occluded with an atraumatic microvascular clamp for 45 minutes (two-lobe IR model); or only the portal triad to the left cephalic lobe of the liver was occluded for 90 minutes (one-lobe IR model). We applied these two techniques since each of them were previously reported to mimic the pathological conditions of liver IR.4, 11-13 The two-lobe IR model of ∼70% segmental hepatic ischemia prevents mesenteric venous congestion by permitting portal decompression through the right and caudate lobes. The one-lobe IR model was designed for the same advantage with only 40% segmental hepatic ischemia. The 45-minute or 90-minute ischemia were chosen because it induced a reproducible sublethal liver injury in all the experiment groups. Reperfusion was initiated by removal of the clamp. Evidence of ischemia was confirmed by visualizing the pale blanching of the ischemic lobes. The absence of ischemic color changes or the lack of response to reperfusion was a criterion for immediate sacrifice. Mice body temperature was monitored by a rectal probe and maintained at 36.5-37.5°C during ischemia through the use of Homeothermic Blanket Control Unit (Harvard Apparatus, Natick, MA). Using a bleeding scale,14 we only kept mice with no to minimal bleeding during the surgical operation (bleeding <1+ on a scale of 0-5+ considered minimal), and euthanized mice with a small (2+) to moderate (3+) amount of bleeding. The abdomen was closed temporarily by two interrupted 5-0 sutures and covered with a sterile plastic wrap during ischemia to minimize evaporative loss, and closed by continuous 5-0 sutures in two layers after reperfusion. Blood samples were taken through the tail vein at several timepoints postreperfusion (<50 μL/timepoint). At 48 hours postreperfusion, the mice were sacrificed and blood samples were taken from the inferior hepatic vena cava before the ischemic lobes were harvested. Sham-operated animals underwent the same surgical procedure without portal triad occlusion. Serum aminotransferase (sALT and sAST) levels were measured using an autoanalyzer (Modular P800, Hitachi).

In Vivo Treatment.

Neutrophil depletion was obtained with 500 μg intraperitoneally on day 0 (2 hours before IR) and day 1 of anti-Ly-6G mAb (1A8, BioXCell, West Lebanon, NH). IL-17A neutralization was obtained with anti-IL-17A mAb (R&D Systems, 100 μg intravenously day 0 and 50 μg intraperitoneally day 1). The same protocols were used with control IgG (Jackson Immunoresearch Laboratories). Ten mg/kg ultrapure lipopolysaccharide (LPS) from Escherichia coli (0111:B4) (Cayla) was intravenously administered or used for in vitro experiments.


Formalin-fixed tissues samples were embedded in paraffin. Liver sections (4 μm) were stained with hematoxylin/eosin. Immunohistology was performed using the monoclonal antibodies: 50104 for IL-17A staining (R&D systems) and 1A8 for Ly-6G staining (BD Biosciences). Biotin-conjugated goat antirat IgG (Jackson Immunoresearch Laboratories) was used as secondary antibody.

Measurement of Myeloperoxidase Activity.

Liver myeloperoxidase (MPO) activity was determined as described37 with slight modifications. Briefly, 100 mg liver tissue was homogenized in 50 mmol/L potassium phosphate buffer (pH 6), centrifuged at 10,000 rpm and resuspended in 1,000 μL 0.5% hexadecyltrimethylammonium bromide in 50 mmol/L potassium phosphate buffer (pH 6). Tissue homogenates were sonicated, freeze/thawed, and centrifuged at 20,000 rpm. Fifty μL of supernatants were added to 200 μL 50 mmol/L phosphate buffer (pH 6) containing 0.167 mg/mL odianisidine hydrochloride and 0.0005% hydrogen peroxide, and the kinetics of absorbance at 460 nm was measured. Protein concentration of the supernatant was determined using a Micro BCA Protein Assay kit (Thermo Scientific) for calibration and values were standardized using MPO purified from human leukocytes (Sigma-Aldrich, St. Louis, MO).

RNA Purification and Real-Time Reverse-Transcription Polymerase Chain Reaction (RT-PCR).

RNA was extracted from the liver, gradient enriched liver and spleen cells or from purified liver and spleen CD11c+ or F4/80+ cells using MagnaPure LC RNA Isolation Kit (Roche Diagnostics, Brussels, Belgium). Reverse transcription and real-time PCR reactions were then carried out using LightCycler-RNA Master Hyprobes (one-step procedure) on a LightCycler apparatus (Roche Diagnostics) to measure IL-27p28, EBI3, IL-23p19, IL-12/23p40, IL-17A, IL-6, IL-10, IL-1ra, TNF-α, CXCL10, and β-actin. The sequences of primers and probes are available on request (

Cell Purification.

Livers were digested for 20 minutes at 37°C with 400 U/mL collagenase type III (Worthington Biochemicals), 10 mM HEPES in Ca2+-free Hank's buffered salt solution (HBSS) (Lonza). Digested suspensions were passed through a nylon mesh and centrifuged, and the cell pellet was resuspended in 5 mL of 45.5% Nycodenz solution (Nycomed). A centrifugation gradient was performed (1,700g for 15 minutes at 4°C). The cells at the interface were recovered as a low-density fraction, and CD11c+ or F4/80+ cells were isolated by magnetic cell sorting with anti-CD11c-coupled Microbeads on an MS column (Miltenyi Biotec) or with FITC-conjugated anti-F4/80 mAb (Serotec, Oxford, UK) and anti-FITC-coupled Microbeads (Miltenyi Biotec).

Intracellular Staining.

Liver cells were isolated by lymphoprep gradient before immunostaining. For IL-17A and IFN-γ intracellular staining, cells were stimulated with 50 ng/mL PMA (Sigma-Aldrich), 500 ng/mL ionomycin (Sigma-Aldrich), and GolgiPlug (BD Biosciences) for 6 hours. Cells were then incubated with Fc blocking antibodies (2.4G2), stained with FITC-anti-γδTCR or PE-Cy7-anti-NK1.1 and Pacific blue-anti-CD3, fixed and permeabilized with CytoFix/CytoPerm solution (BD Biosciences), and labeled with APC-conjugated anti-IFN-γ and PE-conjugated anti-IL-17A monoclonal antibodies (mAbs), or isotype controls (BD Biosciences). For M1 type macrophage analysis, 2 × 106/mL low-density fraction collected from Nycodenz gradient were incubated for 16 hours with 1 μg/mL LPS and GolgiPlug (BD Biosciences). Cells were stained with Pacific Blue anti-F4/80, PerCP anti-CD11c, Phycoerythrin-Cy7 anti-TNF-α, FITC anti-IL-10 (BD Bioscience), Alexa Fluor 647 anti-IL-27p28 (BioLegend), and eFluor 660 anti-IL-23p19 mAbs (eBioscience). Cells were analyzed on a Cyan ADP flow cytometer (DakoCytomation).

Quantification of Cytokine Production From the Liver Ischemic Lobe.

Liver ischemic lobe was incubated at 4°C in 500 μL 0.9% NaCl buffer for 24 hours. Supernatants were centrifuged at 300g for 10 minutes and then analyzed for the presence of cytokines. Quantification of TNF-α and IL-6 using commercially available enzyme-linked immunosorbent assay (ELISA) Duoset (R&D Systems).


Data are expressed as mean ± standard error of the mean (SEM). Statistical analysis of data was realized using a two-tailed no parametric Mann-Whitney test unless specified in the figure legends. P ≤ 0.05 was considered significant.


Liver IRI Is Enhanced in IRF3−/− and Not in TRIF−/− Mice.

We investigated the role of IRF3 and TRIF-mediated inflammatory response in TLR-4 dependent liver IR injuries. We first induced warm hepatic IR in IRF3−/− and TRIF−/− mice using a well-established two-lobe IR model. Compared with sham-operated controls (Supporting Fig. 1S), 45 minutes of 70% hepatic warm ischemia induce hepatic necrosis, 48-hour postreperfusion, as shown by the pale areas, most prominent in the posterior part of the left lobe in WT mice (Supporting Fig. 1S). Microscopically, the pale areas appeared as typical lesions of hepatocellular necrosis (Fig. 1, and Supporting Fig. 1S). Other common features of IR injury, including cellular edema, centrilobular vacuolization, endothelial cell disruption, and infiltration, were also observed at 48 hours postreperfusion in the ischemic lobe of WT mice (Fig. 1, Supporting Fig. 1S). IRF3−/−, but not TRIF−/− mice displayed severe IR injury as observed on liver lobes (Supporting Fig. 1S) and sections (Fig. 1, Supporting Fig. 1S) and evaluated by Suzuki's criteria (score, 8.6 ± 1.4 compared to 5.6 ± 1.6 in WT mice, P < 0.01) (Fig. 1). Liver injury was also assessed by increased sALT and sAST levels in WT mice, which peaked at 4 hours postreperfusion (1,971 ± 397 versus 27 ± 3 U/L [P < 0.001] for sALT, and 2,043 ± 556 versus 60 ± 9 U/L [(P < 0.001] for sAST) (Fig. 1). The serum transaminases levels remained elevated at 24 hours postreperfusion and recovered to normal level at 48 hours. Consistently, IRF3−/−, but not TRIF−/− mice displayed an increased acute injury (4 hours) (3.9-fold for sALT [P < 0.01], 3.7 folds for sAST [P < 0.05]), as compared with WT mice. The subacute injury (24 hours) in IRF3−/− mice was enhanced even further (6.4-fold for sALT [P < 0.05], 6.0-folds for sAST [P < 0.05]), implying a delayed recovery process (Fig. 1). No difference was found in the sALT or sAST levels among the sham-operated control groups of WT, IRF3−/− and TRIF−/− mice at 4 hours postsurgery (Fig. 1). We further confirmed the enhanced IR damage in IRF3−/− mice through the use of a milder, one-lobe IR model. In agreement with the previous observations, IRF3−/− mice displayed increased ALT serum levels at 4 and 20 hours postreperfusion compared with WT mice (P < 0.05 at 4 hours, P < 0.01 at 20 hours) (Supporting Fig. 2SA). Cellular damages were clearly visible in the ischemic liver lobe of IRF3−/− mice 20 hours postreperfusion, when 35 ± 10% hepatocellular necrosis areas could be detected, compared to 10 ± 2% in WT mice (Supporting Fig. 2SB). Forty-eight hours postreperfusion, hepatic lobe damages were maximal in IRF3−/− mice with more than 60 ± 10% of necrosis areas and were decreased to 5% in WT liver (Supporting Fig. 2SB). Anti-Ly-6G immunostaining revealed neutrophil infiltrates in the liver necrosis areas of WT and IRF3−/− mice 20 hours postreperfusion (Supporting Fig. 2SC). The neutrophils remain dominant in the liver lesions of IRF3−/− mice at 48 hours, whereas few neutrophil infiltrates were detected in WT mice. Ischemic liver lobe of IRF3−/− mice recovered completely 6 days postreperfusion (not shown). Overall, these clinical features revealed a more pronounced hepatic IRI in IRF3−/− but not TRIF−/− mice compared to their WT counterparts.

Figure 1.

Liver pathology after two-lobe liver IR. (A) Representative sample of at least 10 mice of each group (wildtype, IRF3−/−, and TRIF−/− mice), showing the H&E-stained liver sections (40×, 100× original magnification) from the same mouse. (B) Serum ALT and AST levels in wildtype, IRF3−/−, and TRIF−/− mice (n = 7-10) after partial ischemia (two-lobe 45-minute) and 4, 24, or 48 hours reperfusion (serially in each mice). Sham controls (n = 3 from each strain of mice) displayed blood sample at 4 hours after reperfusion. Results are mean values ± SEM. (B) Scoring of the ischemic liver sections according to the Suzuki's criteria at 48 hours. (*P < 0.05, **P < 0.01, ***P < 0.001).

Neutrophil Depletion Reduces Liver IRI in IRF3−/− Mice.

To evaluate the possible role of neutrophils in mediating tissue damage in this model, we treated WT and IRF3−/− mice with anti-Ly-6G mAb before IR and monitored hepatocellular injury 48 hours later. Neutrophil depletion was effective in both strains of mice, as no neutrophil infiltrate could be detected by immunostaining in the liver 2 days after the anti-Ly-6G treatment (Fig. 2A). Moreover, neutrophil depletion was associated with an almost complete resolution of liver parenchyma mortality within the ischemic lobe of IRF3−/− mice compared with WT mice (Fig. 2B) and a significant reduction in sALT levels in both strains (Fig. 2C).

Figure 2.

Liver histology after neutrophil depletion. (A) Liver sections were stained with anti-Ly-6G mAb 48 hours postliver IR on control Ig (upper panels) or anti-Ly-6G mAb (lower panels) treated wildtype or IRF3−/− mice. Sections are representative of 5 to 10 independent mice per experimental group (40× original magnification). (B) Scoring of the ischemic liver sections according to the Suzuki's criteria, showing the score of necrosis (*P < 0.05, **P < 0.01, ***P < 0.001 with the Wilcoxon statistical test). (C) sALT levels from 5 to 8 mice per experimental group. (*P < 0.05, **P < 0.01, ***P < 0.001).

IL-12/IL-23p40, IL-23p19, and IL-17A Messenger RNA (mRNA) Expression Is Increased While IL-27p28 Expression Is Decreased During IRI in IRF3−/− Mice.

We then wanted to characterize the inflammatory response that was triggered by IR in the liver of IRF3−/− mice. Ex vivo microdiffusion of the ischemic lobe revealed a reduced amount of TNF-α and an early enhancement of the IL-6 level in the IRF3−/− liver lobe compared with WT counterpart liver lobe 4 hours and 2 hours post-IR, respectively (Fig. 3A). IRF3−/− mice also displayed lower levels of liver CXCL10 and IL-27p28 mRNA, while a striking up-regulation of cytokines involved in Th17 differentiation was observed in IRF3−/− mice with a delayed kinetics. In particular, expression of IL23-related p19 subunit was elevated in IRF3-deficient mice 20 hours post-IR, while IL-17 production peaked at 48 hours (Fig. 3B).

Figure 3.

Liver cytokine production after IR. (A) Ischemic liver lobe from wildtype or IRF3−/− mice were collected at 2, 4, and 6 hours and incubated for 24 hours at 37°C for cytokine diffusion. TNF-α and IL-6 were measured in the supernatant by ELISA. Results are mean values ± SEM (n = 5-20/experimental group). (B) RNA were isolated from ischemic liver lobe from wildtype or IRF3−/− mice and subjected to quantitative real-time PCR analysis. CXCL-10, IL-27p28, IL-23p19, IL-12p40, and IL-17A mRNA contents were normalized using β-actin mRNA as reference. Results are mean values ± SEM and were collected from four independent experiments (*P < 0.05, **P < 0.01, ***P < 0.001).

Taken together, these data strongly suggest a role for IRF3 in controlling the coordinate expression of cytokines (IL-6, IL-23, and IL-17) of the Th17 family in response to hepatic IR.

IL-17A Is Critically Involved in Liver IRI in IRF3−/− Mice.

To evaluate the putative functional role of IL-17A following liver IR, we first identified IL-17A-producing cells by immunohistochemistry in ischemic livers. IL-17A-producing cells were only detected in the centrolobular space of IRF3−/− hepatic ischemic lobes 48 hours postreperfusion (Fig. 4A, Supporting Fig. 4SA). IL-17A neutralization using anti-IL-17A mAb during liver IR had no effect on hepatocellular damage in WT mice, but completely abrogated the liver toxicity in IRF3−/− mice, as witnessed by the reduction in liver necrosis areas neutrophil infiltration 48 hours postreperfusion (Fig. 4B,C, Supporting Fig. 4SA) and the reduction of liver MPO activity (Supporting Fig. 4SB), further confirming the role of IRF3 as a negative regulator of IL-17A-mediated liver pathology during IR.

Figure 4.

Liver histology after IL-17A neutralization. (A) IL-17A immunostaining on liver section from wildtype and IRF3−/− mice 48 hours post-IR. Sections are representative of five individual mice (100× original magnification). (B) Liver ischemic lobe sections were stained using anti-Ly-6G mAb on control Ig (upper panels) or anti-IL-17A mAb (lower panels) treated wildtype or IRF3−/− mice 48 hours postliver IR. Sections are representative of eight independent mice per experimental group (40× original magnification). (C) Scoring of the ischemic liver sections according to the Suzuki's criteria, showing the score of necrosis. (*P < 0.05, **P < 0.01, ***P < 0.001).

Liver Sources of IL-23 and IL-17 in IRF3−/− Mice upon TLR4 Engagement In Vivo.

As largely evidenced in the literature, ischemia-reperfusion injury is linked to LPS-induced liver injury.15 We then wanted to address the question of liver cellular sources of IL-23 and IL-17A in this more sensitive model of TLR4 ligation in vivo. We intravenously administered LPS in IRF3−/− and WT mice and compared 2 hours later the mRNA contents of the different IL-12 family members in the liver compared with the spleen. Low-density cellular fraction, purified CD11c+ dendritic cells, and CD11c F4/80+ macrophages were collected from the liver and the spleen. Under these settings, IL-12p35 and EBI3 mRNA synthesis were poorly induced in response to LPS in either type of mice, organ, or cellular fractions (not shown). IL-27p28 mRNA were significantly induced in all cellular preparations from the spleen and the liver of B6 WT mice after LPS exposure (Fig. 5A), but the LPS-induced IL-27p28 mRNA contents were strongly reduced in IRF3−/− low density, macrophages, and dendritic cells in both organs. In contrast, IL-23p19 mRNA contents were strongly enhanced only in the liver low density and macrophages compartments but not in the spleen of IRF3−/− mice as compared with WT mice (Fig. 5B). The IL-12p40 mRNA synthesis was equally induced in all fractions from the spleen of IRF3−/− and WT mice after LPS administration (Supporting Fig. 5SA). In the liver of IRF3−/− mice, a higher induction of IL-12p40 mRNA was only observed in the CD11cF4/80+ cell fraction compared with WT mice (Supporting Fig. 5SA). Collectively, these data show that the IL-27p28 gene activation is down-regulated in splenic and hepatic dendritic cells and macrophage populations in IRF3−/− mice upon TLR4 triggering. In contrast, IL-23p19 and p40 mRNAs are better induced only in hepatic F4/80+ Kupffer cells of IRF3−/− mice. We further demonstrated that M1-type macrophages (defined as TNF-αhigh IL-10low producing F4/80+CD11c cells) were the cellular source of this higher production of IL-23p19 and lower production of IL-27p28 in liver IRF3−/− mice after LPS activation compared with WT mice (Supporting Fig. 6S).

Figure 5.

Hepatic IL-23p19 and IL-27p28 mRNA contents after TLR4 engagement. Low density fraction, F4/80+ purified or CD11c+ purified cells were collected from the liver or the spleen of C57BL/6 wildtype or IRF3−/− mice that were intravenously injected with vehicle (sham) or LPS 2 hours before. RNA were extracted from a pool of four mice per experimental group and analyzed by real-time RT-PCR. IL-27p28 (A) and IL-23p19 mRNA (B) contents were normalized using β-actin mRNA as reference. Results are mean values ± SEM and were collected from four independent experiments (*P < 0.05, **P < 0.01).

We then addressed the question of whether IL-17A-producing immune cells would be preferentially expanded in the spleen or the liver of IRF3−/− mice upon TLR4 activation. Intracellular staining for IFN-γ and IL-17A were performed on CD4+, CD8+, γδ T cells or invariant natural killer T (iNKT) cells from the spleen and the liver of B6 WT and IRF3−/− mice 18 hours after LPS injection. As seen in Fig. 7S, in the liver (Supporting Fig. 7SA,B) but not in the spleen (Supporting Fig. 7SC), the percentage of both IL-17A-producing iNKT cells and γδ T cells was increased by LPS in IRF3−/− mice compared with WT mice. At the same time, the percentage of IFN-γ producing iNKT cells and γδ T cells was decreased by LPS in IRF3−/− mice. In this setting, the administration of LPS did not modify the percentage of IFN-γ and IL-17A-producing iNKT cells and γδ T cells in the spleen (Supporting Fig. 7SC). The percentage of IFN-γ and IL-17A-producing cells from CD4+ or CD8+ T cells in the liver (Supporting Fig. 7SB) or the spleen (not shown) was very low and was not affected by LPS administration in either strain of mice. Taken together, these results suggest that an IL-23/IL-17A axis involving innate immune cells such as macrophages, iNKT cells, and γδ T cells is particularly active in the liver of IRF3−/− mice upon TLR4 engagement.


As a potential receptor of DAMPs released by necrotic cells, TLR4 was demonstrated to be essential in the development of hepatic IR injury.8, 16, 17 TLR4 engagement on actively phagocytic nonparenchymal cells such as Kupffer cells is required for IR-induced liver inflammation and injury16 and TLR4 engagement on parenchymal cells such as hepatocytes and sinusoidal endothelial cells exacerbates IR injury by regulating the expression of ICAM-1.18 Two major downstream pathways are involved in TLR4 signaling: the MyD88-dependent pathway that causes early production of proinflammatory cytokines is more involved in the induction of host defense and liver regeneration; and the TRIF-dependent pathway that activates IRF3 and causes the late-phase activation, leading to the production of type one interferon (IFN-I) and IFN-stimulated genes (ISGs), is more involved in inflammation.10 Although studies demonstrated critical roles of the MyD88-dependent TLR4 signaling in the IR injury of heart19 and kidney,20 TLR4-mediated hepatocellular damage appears to be MyD88-independent and IRF3 dependent.3 Until now, it was reported that TRIF-IRF3 pathway contributes to liver injury either through repression of retinoid X receptor α (RXR) implicated in liver detoxification21 or through IFN-I and ISGs such as CXCL10 that targeted activated T cells and natural killer cells.3, 22, 23 However, an opposite role of IRF3 pathway has also been reported in the IR initiated TLR4 signaling in other organs. IRF3 was reported to be either not required for brain IR injury,19, 24 or neuroprotective against subsequent cerebral IR injury in an LPS preconditioning model.25 A protective effect of IFNβ administered either systemically or directly to the organ26, 27 was indeed reported. Furthermore, some ISGs, like Trim30 and Ifit1, were identified as antiinflammatory by inhibiting TLR4-induced NF-κB activation.28, 29

In the present study, we demonstrated for the first time a protective role of IRF3-dependent pathway in the late development of liver IR injury. Indeed, in the absence of IRF3, liver IR leads to an amplified neutrophil-associated inflammatory response, as evidenced by more deteriorated liver function, more severe histological damage, and more neutrophil infiltrates in the liver necrosis areas. We found that the intensified damage was associated with suppressed induction of IL-27 as well as enhanced induction of IL-6/IL-23 and IL-17A and with an IL-17A-dependent neutrophil recruitment that is dominant in the late phase (48 hours) of the injury. When IRF3 was deficient, TLR4 activation induced IL-27 expression was repressed in the spleen and liver nonparenchymal cells such as macrophages (Kupffer cells in liver) and dendritic cells, associated with strongly enhanced expression of IL-23p19/IL-12p40 in Kupffer cells, together probably resulting in expanded IL-17A-producing iNKT cells and γδ T cells in the liver. Furthermore, we observed that the increased IL-17A-associated inflammation that we observed in IRF-3 knockout mice is not related to impaired induction of interferon-stimulated genes encoding for antiinflammatory cytokines such as IL-10 and IL-1ra as reported30, 31 (Supporting Fig. 6S).

Interestingly, the IRF3-dependent protection we observed was TRIF-independent. TRIF deficit blocked not only the IRF3 pathway but also the late phase activation of NF-κB, which may enhance liver damage by producing proinflammatory cytokines including TNF-α, IL-6, and IL-1β. Therefore, the blockade of NF-κB activation caused protection may counterbalance the damage derived from the IRF3/IL-27/IL-23/IL-17 axis.

Resident lymphocytes found within the liver include conventional αβ T cells but also unconventional iNKT cells and γδ T cells. All of them can potentially regulate liver IR injury through IFN-γ, TNF-α, IL-17A, IL-4, IL-10 production and contribute to mediate additional death, although the mechanism by which they interact remains to be further studied.32, 33 The iNKT cells, which account for as much as 30% of the total lymphocyte population and as much as 50% of total αβ TCR+ T cells, were reported to be predominant mediators of liver IR injury.34 The γδ T-cells are preferentially localized in nonlymphoid tissues and appeared to be critical in the inflammatory response after renal IR.35 IL-17A has been found to be responsible for CD4+ T lymphocytes-induced liver IR damage by mediating neutrophil recruitment through production of chemokines like MIP-2.36 More recently, the regulatory role of IL-17A in initiating neutrophil-induced inflammatory response and hepatic injury, potentially through the induction of TNF-α production by Kupffer cells, was demonstrated.37 We observed a TLR4-induced expansion of IL-17-producing iNKT cells and IL-17-producing γδ T cells and a decrease in IFN-γ-producing iNKT cells and γδ T cells in the liver of IRF3−/− mice, while the IFN-γ and IL-17A-producing cells from CD4+ or CD8+ T cells accounts for the low percentage, suggesting an inhibitory role of IRF3 pathway in the “Th17-type” but not the “Th1-type” response dominated in unconventional T cells.

IL-23 critically influences the IL-17 type response. Its role in hepatic inflammation was suggested but the cellular source is not yet described.38, 39 We here demonstrated that M1 type Kupffer cells may express IL-23 as long as IRF3 is inactive, leading to IL-27 repression. IL-27 exerts both pro- and antiinflammatory functions.40 It can indeed promote T-bet-associated IFN-γ production in T cells but also efficiently suppresses pathogenic Th17 cell activities.41, 42 The balance between IL-23 and IL-27 is therefore critical for the induction of Th17-mediated inflammatory responses.40 Our data demonstrated that IRF3 deficit tipped the balance of IL-23 and IL-27 production in the liver nonparenchymal cell upon TLR4 engagement, and favored the IL-17-mediated liver inflammation.

Globally, we may consider that the immune response initiated upon liver IR consisted of two IRF3-dependent phases: an early reported phase involving a type I IFN-dependent neutrophil inflammation3 and a late protective phase controlling IL-17-type innate immune response that may act in an IL-27-dependent manner, as summarized in Fig. 6.

Figure 6.

Hepatocellular damage caused upon liver IR is controlled by two IRF3-dependent phases: an early cytotoxic phase involving a type I IFN-dependent neutrophil inflammation and a late protective phase controlling IL-17-type innate immune response that may act in an IL-27-dependent manner.


The authors thank Oberdan Leo for critically reading the article and Frédéric Paulart, Frédéric Lhommé, and Nicolas Passon for technical expertise and Philippe Horlait, Laurent Depret, Christophe Notte, Grégory Waterlot, and Samuel Vanderbiest for animal care.