NKT cells are required to induce high IL-33 expression in hepatocytes during ConA-induced acute hepatitis



Interleukin-33 (IL-33) is thought to be released during cellular death as an alarmin cytokine during the acute phase of disease, but its regulation in vivo is poorly understood. We investigated the expression of IL-33 in two mouse models of acute hepatitis by administering either carbon tetrachloride (CCl4) or concanavalin A (ConA). IL-33 was overexpressed in both models but with a stronger induction in ConA-induced hepatitis. IL-33 was weakly expressed in vascular and sinusoidal endothelial cells from normal liver and was clearly induced in CCl4-treated mice. Surprisingly, we found that hepatocytes strongly expressed IL-33 exclusively in the ConA model. CD1d knock-out mice, which are deficient in NKT cells and resistant to ConA-induced hepatitis, no longer expressed IL-33 in hepatocytes following ConA administration. Interestingly, invariant NKT (iNKT) cells adoptively transferred into ConA-treated CD1d KO mouse restored IL-33 expression in hepatocytes. This strongly suggests that NKT cells are responsible for the induction of IL-33 in hepatocytes.


Interleukin-33 (IL-33) is a newly identified member of the IL-1 cytokine family and mainly induces IL-4, IL-5 and IL-13 via its interaction with a heterodimer composed of ST2 (IL-1 receptor-like 1) and IL-1RAcP (IL-1 receptor accessory protein) receptors. IL-33 is involved in many immune pathologies [1–5], and cellular sources in most tissues are mainly endothelial and epithelial cells [6], as well as smooth muscle cells, fibroblasts, myofibroblats, keratinocytes, DCs and activated macrophages.

Among the cytokines which can induce IL-33 in various cell types, we and others have identified that a cocktail of IL-1β, IFN-γ and TNF-α, but not IL-6 or PDGF-BB, can increase IL-33 expression [2, 7–9]. Furthermore, bacterial LPS stimulates production of IL-33 in cultured human monocytes [10]. Pro-inflammatory cytokines such as TNF-α and IL-6 are potent inducers of soluble ST2 (sST2), which acts as a decoy receptor of IL-33 and abrogates the biological effects of IL-33 under different inflammatory conditions [11, 12].

Regarding chronic hepatitis, we found that overexpression of IL-33 positively correlated with liver fibrosis and necro-inflammatory conditions in mice and humans [9]. Furthermore, elevated and positively correlated serum IL-33 and sST2 levels were found in chronic liver failure and acute hepatitis patients [13]. Thus, sST2 is an early biomarker of acute hepatitis. Limited data are available concerning the regulation of increased IL-33 expression in vivo or in pathological conditions such as immune-mediated induction of IL-33. This information would help to further understand the proposed role of IL-33 as an ‘alarmin’ of the immune system [6, 14]. The present study aimed to identify and compare the cellular sources and in vivo regulation of IL-33 in two major models of acute hepatitis. One model was induced by a hepatotoxic molecule, carbon tetrachloride (CCl4), whereas the other was induced by the T-cell activator concanavalin A (ConA). Our findings provide evidence that (i) IL-33 expression is strongly induced in liver from ConA-treated mice compared with CCl4-treated mice, (ii) hepatocytes are a major and novel cellular source of IL-33 in vivo during ConA-induced acute hepatitis, but this is not the case for the CCl4 model and (iii) specific hepatocyte expression of IL-33 is NKT-cell dependent because mice deficient in NKT cells (CD1d KO) were resistant to ConA-induced hepatitis and displayed no IL-33 expression in hepatocytes; these phenomena were restored after iNKT-cell adoptive transfer in these mice.

Results and discussion

Hepatocytes are a major and novel source of IL-33 during acute hepatitis in mice

To investigate the in vivo regulation and expression of IL-33 during acute hepatitis, we studied both CCl4- and ConA-induced acute hepatitis. In CCl4-induced acute hepatic injury, the levels of aspartate aminotransferase (AST) significantly increased (59.2-fold) compared with oil control mice (Fig. 1A). The levels of IL-33 mRNA in livers from CCl4-treated mice were significantly higher (3.5-fold) than in controls (Fig. 1A). We observed a large hepatic injury in ConA-induced acute hepatitis indicated by the significant increase in serum AST levels after 8 h of ConA administration (Fig. 1B). This hepatic injury was associated with a robust upregulation (25-fold) of IL-33 mRNA during ConA-induced hepatitis (Fig. 1B), which was much stronger than during CCl4-induced hepatitis. Histological examination of liver tissues revealed a normal tissue structure of PBS/control-treated mice livers. As expected, a perivascular zone of necrosis was seen in CCl4-treated mice livers, and ConA-treated mice showed a diffuse perivascular and parenchymal zone of hepatic injury (Fig. 1C, dotted line area). We further identified the cellular sources of IL-33 in mouse livers following treatment with CCl4 and ConA. IL-33 was constitutively and weakly expressed in liver vascular endothelial and sinusoidal endothelial cells in PBS/control mice. The same cells strongly upregulated IL-33 expression following 48 h of CCl4 administration (Fig. 1D). In the livers of ConA-treated mice, a large number of cells presenting all the features of hepatocytes strongly expressed IL-33 along with vascular and sinusoidal endothelial cells (Fig. 1D). Indeed, the IL-33-positive parenchymal cells exhibited a hexagonal or polyhedral pattern, bile canaliculi, and few were bi-nucleated (Fig. 1E). All these features are typical of hepatocytes. Only the 33 kDa full-length form of IL-33 was detected by Western blot (see Supporting Information). This has been proposed to act via ST2/IL-1RAcP receptors [15].

Figure 1.

Liver injury and IL-33 expression in mice treated with CCl4 and concanavalin A. Levels of serum AST (IU/L) and relative mRNA expression of IL-33 in livers were measured in mice treated with (A) CCl4 (2.4 g/kg by gavage, n=4) for 48 h or (B) concanavalin A (ConA) (20 mg/kg by i.v., n=16 for AST and n=5 for transcripts) 8 h post-injection. The oil-control (n=2–3) or PBS-treated (n=2–3) mice served as a reference for mRNA expression (control mRNA level was arbitrarily taken as 1). Results are represented as mean+SD of each group and p-values as a measure of significance between continuous variables/or groups are denoted as *p<0.05 and **p<0.01 (Mann–Whitney U-test). (C) Sections of liver tissues taken after each sacrifice were stained with H&E and (D) immunolocalization of IL-33 was performed. PBS or oil-treated control mice livers were seen as normal; the zone of necrosis observed in both hepatic models was surrounded by dotted-line area. Dark brown IL-33-positive cells in the liver are indicated with black arrows (vascular endothelial cells) and pink-red arrows (sinusoidal endothelial cells, D); typical hepatocyte-positive cells observed in ConA-treated livers are further enlarged in the squared area (D). Scale bars: (C) 500 μm, (D) either 200 or 50 μm. (E) Detection of IL-33-positive cells by immunofluorescence was performed upon cryosections of liver using anti-mIL-33 goat IgG followed by Cy5-conjugated secondary antibody (green color for IL-33). Counterstaining with Hoechst and Texas red-phalloidin allowed visualization of nuclei in blue and F-actin filaments in orange. Typical hexagonal or polyhedral hepatocytes surrounded by orange F-actin filaments are indicated by white dotted line, and bile canaliculi between hepatocytes are indicated by white arrows. Fluorescent microscope magnification 40×. The data presented are representative of three independent experiments.

These findings show that acute hepatitis is associated with the induction of IL-33 and highlight major differences between CCl4- and ConA-induced hepatitis both in terms of the level of IL-33 induction and cellular sources, i.e. IL-33 is induced in endothelial/sinusoidal cells in each type of hepatitis but hepatocytes express IL-33 only in the ConA model. Such differences may be due to distinct mechanisms of triggering liver injury specific to each molecule. CCl4 is a hepatotoxic molecule: it activates cytochromes that form the trichloromethyl radical that reacts with oxygen to form a highly reactive species, which initiates a chain reaction of lipid peroxidation affecting the permeabilities of cell organelles [16]; this results in severe hepatic damage. In contrast, ConA is a lectin which acts as a T-cell mitogenic agent and induces hepatic injury mediated by NKT cells [17, 18]. Thus, induction of IL-33 in hepatocytes that is concomitant with cellular death or hepatic injury could be dependent on NKT cells.

Specific expression of IL-33 in hepatocytes during ConA-induced acute hepatitis depends on NKT cells

NKT cells are necessary for hepatocyte cell death [17, 18] in ConA-induced hepatitis, and IL-33 is produced by hepatocytes in this model. We thus studied the expression of IL-33 in CD1d KO mice, which are deficient in NKT cells during ConA-induced acute hepatitis. After 8 h of ConA administration, an elevated and significant increase in serum AST was observed in WT ConA-treated mice, whereas CD1d KO mice exhibited similar AST levels to PBS-injected mice (Fig. 2A). These findings were consistent with the histological observations where WT livers showed a diffuse perivascular and parenchymal zone of liver injury, whereas CD1d KO livers appeared normal (Fig. 2B). An initial investigation of IL-33 expression using RT-qPCR indicated elevated mRNA levels in livers from CD1d KO-treated mice (12.8-fold versus PBS), although the increase was lower than in WT mice (25-fold versus PBS). However, this difference was not statistically significant (Fig. 2C). Immunolocalization of IL-33 in ConA-treated liver revealed that the IL-33 expressing hepatocytes seen in WT mice (in perivascular and parenchymal zones) were completely absent from CD1d KO mice livers (Fig. 2D). Nevertheless, in WT and CD1d KO mice, IL-33 was markedly upregulated following ConA treatment in liver vascular and sinusoidal endothelial cells (Fig. 2D). NKT cells therefore appear necessary for the upregulation of IL-33 expression in hepatocytes. To further confirm a direct link between hepatocyte IL-33 expression and NKT-cell activity, we adoptively transferred NKT cells isolated from WT mice into CD1d KO mice with a simultaneous injection of ConA. After 8 h of ConA injection, CD1d KO mice adoptively transferred with NKT cells developed a moderate liver injury (AST level of 1300 IU/L) associated with nuclear IL-33 expression in hepatocytes (Fig. 2E). This restored phenomenon further demonstrates that IL-33 expression in hepatocytes is NKT-cell dependent.

Figure 2.

Liver injury and IL-33 expression in WT and CD1d KO mice treated with concanavalin A. (A) Levels of serum AST (IU /L) and (C) relative mRNA expression of IL-33 in livers were measured in WT mice versus CD1d KO mice treated with concanavalin A (20 mg/kg i.v., n=16 for AST and n=5 for transcripts in each WT and CD1d KO group) 8-h post injection. The PBS-treated mice (n=2–3 in each case) served as a reference for mRNA expression (control mRNA level was arbitrarily taken as 1). Results are represented as mean+SD of each group and p-values as a measure of significance between continuous variables/or groups are denoted as **p<0.01 and ***p<0.001 (one-way ANOVA; Kruskal–Wallis test with post Mann–Whitney U-test). (B) Sections of liver from WT and CD1d KO mice were stained with H&E. (D) Immunolocalization of IL-33 was also performed with hematoxylin counterstaining. Zone of necrosis observed only in WT mice is surrounded by the dotted line. Dark-brown IL-33-positive vascular endothelial cells in livers are indicated with black arrows; typical IL-33-positive hepatocytes observed in WT ConA-treated livers are further enlarged in squared area. Scale bars: (B) 500 μm, (D) either 200 or 50 μm. (E) NKT cells were isolated from WT mice by flow cytometry as CD1d tetramer-α-GalCer and CD3+ cells and were injected into CD1d KO mice (n=3) which also received ConA (20 mg/kg i.v.) injection. Immunolocalization of IL-33 was performed on sections of livers collected 8 h after treatment. Massive staining of parenchymal cells was observed in NKT-cells-adopted-ConA-treated livers with typical IL-33-positive hepatocytes (dark brown color) further enlarged in the squared area (E). The data presented are representative of three independent experiments.

Inflammation and pro-inflammatory cytokines such as IL-1β, TNF-α, IL-6 and IFN-γ play a crucial role in the progression of liver injury [19]. Most of these cytokines are important inducers of IL-33 expression in vitro [9]. We therefore measured the mRNA expression of IL-1β, TNF-α, IFN-γ and IL-6 in treated livers. There was a strong induction of IL-1β, TNF-α, IFN-γ and IL-6 in both WT and CD1d KO ConA-treated mice livers compared with PBS/control mice (Fig. 3A). The mRNA levels of TNF-α, IFN-γ and IL-1β were comparable in WT mice and CD1d KO mice livers following administration of ConA, but this was not true for IL-6 (Fig. 3A). Both WT and CD1d KO mice livers showed marked leukocyte infiltration (CD45+ cells, Fig. 3B) indicating an ongoing inflammatory process, where none of the CD45+ cells were producing IL-33. This latter observation ruled out the possibility that the increase in IL-33 in treated livers was due to infiltration of IL-33 producing leukocytes.

Figure 3.

mRNA expression of pro-inflammatory cytokines and immunolocalization of infiltrate cells in livers of WT and CD1d KO mice treated with concanavalin A. (A) Relative mRNA expression of TNF-α, IFN-γ, IL-6 and IL-1β were measured in livers of WT mice (n=5) versus CD1d KO mice (n=5) treated with ConA 8 h post-injection. The PBS-treated mice (n=2–3 for each group) served as a reference for mRNA expression (control mRNA level was arbitrarily taken as 1). The results are represented as mean+SD of each group and p-values as a measure of significance between continuous variables/or groups are denoted as *p<0.05 (one-way ANOVA; Kruskal–Wallis test with post Mann–Whitney U-test). (B) For co-immunolocalization of IL-33 and CD45+ cells, cryosections of liver (PBS control and ConA treated) were incubated with two distinct primary antibodies directed against IL-33 and CD45, respectively, stained with cyanine 5 (green color for mIL-33-positive cells) and cyanine 3 (pink color surrounding blue nuclei for CD45+ cells). Nuclei were stained with Hoechst (blue); fluorescent microscopic magnification was 40×. The data presented are representative of three independent experiments.

The cytokinic environment could explain the induction and strong expression of IL-33 in liver vascular and sinusoidal endothelial cells that was observed in WT and CD1d KO mice. However, IL-33 was absent in the hepatocytes of CD1d KO livers where leukocyte infiltration and pro-inflammatory cytokines were much elevated. This suggests that the inflammatory conditions may not be sufficient for the induction of IL-33 in hepatocytes, and that the presence of NKT cells in the liver is essential for the induction of IL-33 in hepatocytes.

Several papers suggest that NKT cells mediate liver injury via induction of cell-death signals (FasL, TRAIL or TNF-α) following ConA administration [20, 21]. The concomitance of IL-33 induction in hepatocytes during ConA-induced hepatic cell death suggests that, like HMGB1, IL-33 acts as an ‘alarmin mediator’. HMGB1 has recently been reported to be released by hepatocytes in concomitance with hepatocyte death during ConA-induced acute hepatitis [22]. It has also been suggested that IL-33 may be released passively from the nucleus or cytosol during cell lysis and could serve as a ‘necrocrine’ [23]. Indeed, IL-33 plays a crucial role as an amplifier of innate immunity in various physio-pathological conditions including colitis or airway inflammation [24]. Based on our findings, we propose that activated NKT cells are responsible for hepatocyte cell death and induction of IL-33 expression. In turn, this can modulate inflammation and consequently liver pathology. To date, NKT cells have been found to be involved in numerous liver pathologies where such mechanism could take place [25, 26]. This could also be true for other tissues or inflammatory diseases involving NKT cells, such as colitis and airway inflammation [24].

Concluding remarks

The present study uncovers a novel mechanism of IL-33 induction and provides evidence that in vivo IL-33 expression in hepatocytes during ConA-induced acute hepatitis is NKT-cell dependent. This mechanism could also apply to other physio-pathologies where NKT cells are involved.

Materials and methods


BALB/c WT and CD1d KO (NKT-cell deficient) mice were obtained respectively from Janvier (Le Genest-sur-isle, France) and Dr. Valerie Julia (Inserm, E0344, Nice, France) and were bred in specific pathogen-free (SPF) conditions in the local animal house facilities in accordance with the French laws and the institution's guidelines for animal health care (agreement of M. Samson ♯3596). Ten-wk-old mice were treated with CCl4 (Sigma-Aldrich, St. Louis, MO, USA) diluted in olive oil at a dose of 2.4 g/kg and administered by gavage. The mice were killed 48 h following CCl4 injection. Alternatively, mice were injected intravenously with ConA (Sigma-Aldrich) at a dose of 20 mg/kg body weight and were sacrificed at 8 h post-injection. Control mice were treated with vehicle only in both CCl4 and ConA-induced hepatic models. Fragments of mouse livers were fixed in 4% paraformaldehyde and embedded in paraffin or frozen in liquid nitrogen in the presence of isopentane. Serum AST was measured according to the IFCC primary reference procedures using Olympus AU2700 Autoanalyser® (Olympus Optical, Tokyo, Japan). For histopathology, hematoxylin and eosin (H&E) staining of liver tissues was carried out to access the liver injury.

RNA isolation and RT-qPCR

Total RNA was extracted from mice livers using TRIzol (Invitrogen) reagent. First-strand cDNA was synthesized using the SuperScriptTM II Reverse Transcriptase (Invitrogen). The cDNA or RT amplification was further verified by PCR amplification using the house-keeping gene, GAPDH. Real-time quantitative PCR was performed using the fluorescent dye SYBR Green with the double-strand specific SYBR® Green system (Applied Biosystems) and the ABI 7000 Prism sequence detector (Applied Biosystems). The protocol and conditions for qPCR were as reported earlier by our laboratory [9] using specific primers for 18S, IL-33, TNF-α, IL-6, IFN-γ and IL-1β (Table 1). The relative gene expression was normalized against 18S gene expression.

Table 1. Sequence of primers used for RT-qPCR

Immunolocalization of IL-33 in liver tissues

Paraformaldehyde-fixed and paraffin-embedded mouse liver sections (7 μm) followed by antigen retrieval were incubated with primary antibody (goat IgG anti-murine IL-33, R&D Systems) in a Ventana CT/09/021 automated machine (Ventana Medical Systems, USA). Revelation of primary antibody was carried out using HRP-conjugated rabbit polyclonal anti-goat (Dako, USA) secondary antibody followed by diamino-benzidine and hematoxylin coloration. Paraformaldehyde-fixed cryosections (8 μm) of mouse livers were permeabilized by 0.1% Triton X-100 and non-specific sites were blocked with 2% BSA. Liver sections were then incubated with primary antibodies (goat IgG anti-mouse IL-33, R&D Systems, and rat IgG anti-mouse CD45, BD Sciences) for 2 h at ambient temperature. For immunofluorescence detection, fluorochrome-conjugated secondary antibody (Cy5-conjugated bovine anti-goat IgG, Cy3-conjugated donkey anti-IgG rat, Jackson Immuno Research Laboratories) was used for 1 h at room temperature. Nuclei were counterstained with Hoechst (Molecular Probes) and F-actin filaments were stained by phalloidin Texas red (Sigma).

Liver NKT-cell isolation and adoptive transfer

Livers obtained from 10-wk-old BALB/c WT mice were perfused with PBS, and the purification of liver leukocytes was carried out as previously described [27, 28]. After a blocking step with anti-CD16/32 (BD Pharmingen), total infiltrating leukocytes were labeled with appropriate dilutions of CD1d tetramer-α-GalCer-PE (provided by Dr. M. Leite de Morales) and rat anti-mouse CD3 FITC (BD Biosciences). The CD3+CD1d tetramer+ NKT-cell population was sorted using a FACSAriaTM II flow cytometer together with BD FACSDiva software (BD Bioscience), and 40 000 cells were injected intravenously in each CD1d KO mice as previously described [27]. Simultaneously, ConA was injected into CD1d KO mice at a dose of 20 mg/kg, and mice were then sacrificed 8 h post-injection and their sera and livers were analyzed.

Statistical analysis

Data are expressed as the mean+SD for all mice treated similarly. Kruskal–Wallis one-way analysis of variance (ANOVA) was performed, and mean differences between experimental groups were assessed using the non-parametric Mann–Whitney U-test and GraphPad Prism5 software. For all statistical analyses, *p<0.05, **p<0.01 and ***p<0.001.


This work was supported by INSERM, the Ministère de l'Education Nationale de la Recherche et de la Technologie, the University of Rennes 1, the Région Bretagne, National French Society of Gastro-Enterology (SNFGE), the ‘Ligue contre le cancer’ and the ‘IFR 140’. Muhammad Imran Arshad was supported by a PhD fellowship from the Government of Pakistan (Higher Education Commission, University of Agriculture, Faisalabad). For immunohistochemistry and flow cytometry analyses, animal house facilities, we would like to thank the dedicated platforms (i.e. H2P2 platform, Impac cells platform, cytometry platform and animal house platform) of IFR140, University of Rennes 1, France.

Conflict of interest: The authors declare no financial or commercial conflict of interest.