Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Potential conflict of interest: Nothing to report.
Interleukin (IL)-33, a member of the IL-1 cytokine family, positively correlates with acute hepatitis and chronic liver failure in mice and humans. IL-33 is expressed in hepatocytes and is regulated by natural killer T (NKT) cells during concanavalin A (ConA)-induced acute liver injury. Here, we investigated the molecular mechanisms underlying the expression of IL-33 during acute hepatitis. The expression of IL-33 and its regulation by death receptor pathways was investigated after the induction of ConA-acute hepatitis in wildtype (WT), perforin−/−, tumor necrosis factor related apoptosis inducing ligand (TRAIL)−/−, and NKT cell-deficient (CD1d−/−) mice. In addition, we used a model of acute liver injury by administering Jo2/Fas-antibody or D-galactosamine-tumor necrosis factor alpha (TNFα) in WT mice. Finally, the effect of TRAIL on IL-33 expression was assessed in primary cultured murine hepatocytes. We show that IL-33 expression in hepatocytes is partially controlled by perforin during acute liver injury, but not by TNFα or Fas ligand (FasL). Interestingly, the expression of IL-33 in hepatocytes is blocked during ConA-acute hepatitis in TRAIL-deficient mice compared to WT mice. In contrast, administration of recombinant murine TRAIL associated with ConA-priming in CD1d-deficient mice or in vitro stimulation of murine hepatocytes by TRAIL but not by TNFα or Jo2 induced IL-33 expression in hepatocytes. The IL-33-deficient mice exhibited more severe ConA liver injury than WT controls, suggesting a protective effect of IL-33 in ConA-hepatitis. Conclusion: The expression of IL-33 during acute hepatitis is dependent on TRAIL, but not on FasL or TNFα. (HEPATOLOGY 2012)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Interleukin-33 (IL-33) or IL-1F11 is a recently described member of the IL-1 cytokine family which includes IL-1α, IL-1β, and IL-18.1 IL-33 is widely expressed in various tissues and the cellular sources of IL-33 are mostly endothelial and epithelial cells, as well as smooth muscle cells, keratinocytes, astrocytes, adipocytes, fibroblasts, monocytes, macrophages, hepatic and pancreatic stellate cells, and hepatocytes.2-5 Once released from the cells, IL-33 mediates its cytokine functions by interacting with its specific heterodimeric receptor comprising ST2 (IL-1 receptor-like 1) and IL-1RAcP (IL-1 receptor accessory protein) in an autocrine or paracrine manner.6, 7 IL-33 targets cells of the immune system mainly by way of the ST2 receptor. The ST2 is expressed by T helper 2 (Th2)-cells, basophils, eosinophils, natural killer (NK) cells, natural killer T (NKT) cells, and dendritic cells. Functionally, IL-33 can act as a chemoattractant for Th2 lymphocytes and induce various proinflammatory cytokines or inflammatory mediators.5, 8 The expression of IL-33 has been clearly associated with many acute and chronic inflammatory diseases including arthritis, asthma, lung inflammation, allergy, Crohn's disease, ulcerative colitis, sepsis, anaphylactic shock, and hepatitis.2, 3, 9
Recently, we and others have shown that the IL-33/ST2 axis could play an important role in acute and chronic liver diseases.2, 3, 10 Surprisingly, we found that IL-33 is strongly expressed in hepatocytes and demonstrated that NKT cells were responsible for regulating IL-33 in hepatocytes during concanavalin A (ConA)-induced acute hepatitis.2 Pretreatment of WT mice with recombinant IL-33 prevents the severity of ConA-induced liver damage by recruiting regulatory T-cells and CD4+ T-cells into the liver.10 However, the molecular mechanisms that trigger nuclear IL-33 expression in hepatocytes remain unknown.
The administration of ConA in mice provides an experimental model of T-cell-mediated liver injury, which resembles viral or autoimmune hepatitis in humans.11, 12 In the ConA model, the cytotoxic effector molecules and their receptors like tumor necrosis factor alpha (TNFα), perforin-granzyme, FasL/Fas, and tumor necrosis factor related apoptosis inducing ligand (TRAIL)/DR5 play a crucial role for hepatitis development as well as hepatocyte cell death.13-25 Indeed, TRAIL and DR5 expression increase and liver injury is suppressed in TRAIL−/− mice or after blockage of the DR5 receptor, which suggests a critical role of TRAIL/DR5 in the pathophysiology of ConA-induced acute hepatitis.23, 24
In the present study, we aimed to better characterize the expression of IL-33 during acute hepatitis by using wildtype (WT), perforin−/−, TRAIL−/−, and CD1d−/− mice as well as in primary murine hepatocytes. We also demonstrate the functional impact of endogenous IL-33 in ConA-hepatitis by using IL-33 deficient mice. Our results suggest that IL-33 expression in hepatocytes is partially dependent on perforin, but not on FasL or TNFα in acute hepatitis. Furthermore, we show that TRAIL is essential for inducing IL-33 expression in hepatocytes during T-cell-mediated hepatitis in mice or in cultured murine hepatocytes.
The C57Bl/6 WT mice (8-10-week-old, Janvier, Le Genest-sur-isle, France) were injected intravenously with ConA (Sigma-Aldrich, St. Louis, MO) to induce acute hepatitis at a dose of 20 mg/kg body weight. Mice were sacrificed from 6 to 10 hours postinjection. Intraperitoneal injection of anti-Fas/Jo2 antibody (Purified Hamster Anti-Mouse CD95, BD Pharmingen) agonist antibody was administered at a dose of 0.15 μg/g of body weight to induce hepatic injury and mice were sacrificed at 2, 4, 6, 10, and 24 hours postinjection. Recombinant murine (rm)-TNFα (PeproTech, USA) was injected intravenously (10 μg/kg body weight) alone or in combination with D-galactosamine (D-GalN, Sigma) at a dose of 15 mg/mouse (intraperitoneal) in WT mice and sacrificed at 8 hours postinjection.
Mice C57Bl/6 perforin-KO, TRAIL-KO and IL-33-KO (provided by Dr. Jean-Philippe Girard26 and bred in our local animal facility) were injected intravenously with ConA (20 mg/kg body weight) and sacrificed at the designated timepoints. The C57Bl/6 CD1d-KO mice were primed with ConA for 2 hours followed by injection of rm-TRAIL (30 μg/mouse, intravenous, PeproTech, USA). Mice were sacrificed 8 hours after injection of ConA. In each experiment the control mice were treated with phosphate-buffered saline (PBS) or vehicle only. All the mice were bred in specific pathogen-free conditions in the local animal house facilities and all treatment protocols were in accordance with the French laws and the institution's guidelines for animal welfare (agreement of M. Samson #3596).
Histopathological and Biochemical Examination.
The histopathological and serum biochemical analysis was performed as reported.2
RNA Isolation and Quantitative Reverse-Transcription Polymerase Chain Reaction (RT-qPCR).
The protocol and conditions for RNA extraction, RT-PCR, and qPCR were the same as reported earlier by our laboratory2, 3 using specific primers for 18S, IL-33, FasL, Fas, TRAIL, DR5, TNFα, TNFR1, and TNFR2 (Table 1). The relative gene expression was normalized against 18S gene expression. The control mice in each treatment group served as a reference for messenger RNA (mRNA) expression (control mRNA level was arbitrarily taken as 1).
Table 1. Sequences of Primers Used for RT-qPCR
5′-CAG AAC ACC GTG TGT AAC TGC-3′
5′-GCA AGC GGA GGA GGT AGG-3′
5′-CGC TGG TCT TCG AAC TGC-3′
5′-CAG GAG GAC ACT TAG CAC AGC-3′
Immunolocalization of IL-33 and DR5 in Liver Tissues.
Cryosections or paraformaldehyde-fixed and paraffin-embedded mouse liver sections (7 μm) followed by antigen retrieval were incubated with primary antibody (goat IgG antimouse IL-33, R&D Systems) in a Ventana automated machine (Ventana Medical Systems, USA). Revelation of primary antibody was carried out using horseradish peroxidase (HRP)-conjugated rabbit antigoat (Dako, USA) and for immunofluorescence detection by Cy5-conjugated bovine antigoat IgG secondary antibody and nuclei were stained with Hoechst (Molecular Probes) as described.2 For immunolocalization of DR5 in liver tissues, rabbit anti-DR5 polyclonal antibody (ProScience) was used in a Ventana machine and detection was carried out by using antirabbit biotinylated antibody (Vector Laboratories, USA) followed by diamino-benzidine and hematoxylin coloration or Cy2-conjugated donkey antirabbit (Jackson ImmunoResearch Laboratories) for immunofluorescence. The counting of IL-33-positive hepatocytes was carried in at least 20 different microscopic fields corresponding to 2.67 mm2 surface area using image analysis software (Compix, Hamamatsu, Japan).
Isolation and Primary Culture of Murine Hepatocytes.
The primary murine hepatocytes were obtained after liver perfusion and librase (Sigma) treatment as described27, 28 and hepatocytes were cultured for 12 hours in modified Eagle's medium (MEM) medium supplemented with 10% fetal bovine serum, insulin (500 μg/mL), hydrocortisol (20 ng/mL), and antibiotics. The hepatocytes were stimulated by rm-TRAIL (100 ng/mL) for 0, 4, 6, 7, and 8.5 hours or with rm-TNFα (10 ng/mL) and with FasL/Jo2-antibody (1 or 10 ng/mL) for 8.5 hours as described.29 The immunolocalization of IL-33 or DR5 were carried out using primary antibodies as described above, and detection with secondary antibodies, i.e., Cy5-conjugated bovine antigoat IgG or DyLight-549-conjugated donkey antirabbit IgG for IL-33 and DR5, respectively (Jackson ImmunoResearch Laboratories). Nuclei were counterstained with Hoechst (Molecular Probes).
The results in each model are represented as means ± standard error of the mean (SEM) of each group. A Mann-Whitney U test was used for comparison of control group parameters with treatment group and multiple group analysis was carried out by one-way analysis of variance (ANOVA) with post Mann-Whitney U test as reported.2 The correlation between continuous variables was analyzed using GraphPad Prism5 software. For all statistical analyses, *P < 0.05, **P < 0.01, and ***P < 0.001.
Expression of IL-33 in Hepatocytes Is Partially Dependent on Perforin During ConA-Induced Hepatitis.
Perforin is involved in contributing to ConA-induced liver injury,14 thus we tested whether perforin is involved in regulating IL-33 expression in hepatocytes. We administered ConA into WT or perforin-deficient mice and observed a moderate protection of liver injury in perforin−/− mice as evidenced by serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels compared with WT mice (Fig. 1A; Supporting Fig. S1). The liver IL-33 mRNA expression was increased in WT and perforin−/− mice following ConA-hepatitis compared to unstimulated mice (Fig. 1B). However, we observed a significant difference in liver IL-33 transcript levels between WT and perforin−/− mice 6 hours after ConA administration, which was not evident at the 10-hour timepoint (Fig. 1B). Immunohistochemical analysis showed IL-33 expression in hepatocytes of WT mice 6 and 10 hours after ConA injection. In contrast, in hepatocytes of perforin−/− mice IL-33 expression was only evident 10 hours, but not 6 hours after ConA administration (Fig. 1C). No significant difference in IL-33 expression between WT and perforin−/− mice was evident in liver sinusoidal and vascular endothelial cells after ConA injection (Fig. 1C).
Fas-Induced Liver Injury Does Not Increase IL-33 Expression in Hepatocytes.
Fas stimulation of hepatocytes by way of the agonistic Fas antibody Jo2 triggers hepatocyte apoptosis and severe acute hepatitis.21, 22 We hypothesized that Fas-induced liver injury might directly increase IL-33 expression in hepatocytes. Jo2 stimulation of WT mice triggered severe liver injury as evidenced by hematoxylin and eosin (H&E) staining of liver sections (Fig. 2A). However, Jo2 stimulation had no impact on IL-33 expression in hepatocytes, but only in vascular and sinusoidal epithelial cells (Fig. 2A). A dramatic increase in transaminases was observed following Jo2 administration (Fig. 2B; Fig. S2A) but mRNA expression of IL-33 was not much augmented (Fig. 2C). Jo2 administration resulted in a minor up-regulation of FasL, Fas (Fig. S2B,C) and TRAIL liver mRNA expression (Fig. 2D), whereas a strong increase in DR5 transcript levels with a peak 10 hours after stimulation was evident (Fig. 2E). These findings suggest that the FasL/Fas axis and the increased DR5 expression have no impact on the regulation of IL-33 in hepatocytes during acute liver injury.
TNFα Induces Liver Injury but Does Not Induce IL-33 Expression in Hepatocytes.
ConA stimulation triggers higher TNFα expression (Fig. 3A) and earlier reports demonstrated that this cytokine is essential to trigger liver injury in this model.17–20 Therefore, we tested whether TNFα is involved in triggering higher IL-33 expression in hepatocytes. We thus stimulated WT mice with doses of TNFα (10 μg/kg) previously reported to induce cell adhesion molecules on endothelial cells in mice30 or in combination with D-GalN to induce acute liver injury.31, 32 Eight hours after TNFα stimulation the mice experienced signs of fever and a mild increase in serum transaminases levels (Fig. S3A,B). However, no change in liver IL-33 mRNA expression was evident after TNFα stimulation (Fig. S3C). Histopathological analysis of liver tissues showed no major signs of hepatic injury following TNFα administration and immunolocalization studies revealed any IL-33 expression in hepatocytes (Fig. S3D). The D-GalN-TNFα administration induced severe liver injury in mice as evident from serum ALT (Fig. 2B) or AST (Fig. S3E) levels or liver histology (Fig. 3C). However, IL-33 was not expressed in hepatocytes, whereas the induction of IL-33 expression in vascular and sinusoidal endothelial cells was observed in these livers (Fig. 3C). These findings demonstrate that TNFα does not directly control IL-33 expression in hepatocytes.
TRAIL Controls IL-33 Expression in Hepatocytes.
TRAIL is involved in triggering ConA-induced liver injury.12, 23, 24 We thus aimed to determine a possible contribution of TRAIL in regulating IL-33 expression in hepatocytes. ConA was injected into WT and TRAIL−/− mice. TRAIL−/− mice were significantly protected against ConA-hepatitis compared to WT mice as depicted by serum transaminases levels (Fig. 4A; Fig. S4A). A significant difference in liver IL-33 mRNA expression was observed between WT and TRAIL−/− mice. The up-regulation IL-33 expression was significantly reduced in TRAIL−/− livers (Fig. 4B). Histopathology of liver tissues revealed perivascular and parenchymal zones of liver injury in TRAIL−/− and WT mice (Fig. 4C). Interestingly, only few hepatocytes were positive for IL-33 in TRAIL−/− livers compared with WT mice (Fig. 4D).
Following ConA-hepatitis, we observed increased liver mRNA expression for TRAIL, DR5, FasL, and Fas but not for TNFR1 or TNFR2 in WT mice at different time intervals (Fig. S4B). Localization of the DR5 receptor was further addressed in WT mice by immunohistochemistry showing DR5 receptor expression dominantly in noninjured liver areas, whereas low DR5 expression was evident in necrotic areas (Fig. S4C). Moreover, the expression of hepatocyte-specific nuclear IL-33 and membrane DR5 expression was selectively colocalized in the noninjured area of liver (Fig. S4C). WT and TRAIL−/− mice showed no significant difference in the regulation of liver DR5, TNFR1, and TNFR2 mRNA expression before and 10 hours after ConA injection (Fig. S4D). However, a significant increase in liver FasL mRNA expression was observed in TRAIL−/− mice compared to WT mice 10 hours after ConA injection, whereas liver Fas and TNFα mRNA levels were significantly down-regulated in TRAIL−/− compared to WT mice at this timepoint (Fig. S4D).
To address the functional role of IL-33 in ConA liver injury, we compared hepatic injury in WT and IL-33-deficient mice. We demonstrated significantly increased levels of serum ALT in IL-33−/− mice than WT controls at 24 hours of ConA liver injury, suggesting a protective effect of IL-33 during ConA-hepatitis (Fig. 4E).
Recombinant TRAIL Induces IL-33 Expression in Hepatocytes.
To test an essential role of TRAIL for inducing IL-33 expression in hepatocytes, we reconstituted CD1d−/− mice (NKT cells deficient) with rm-TRAIL following ConA-priming. There was no significant difference in basal TRAIL mRNA expression between WT and CD1d−/− livers (Fig. 5A). However, after ConA injection liver mRNA TRAIL expression was significantly lowered in CD1d−/− compared to WT mice (Fig. 5A). Additionally, the kinetic of higher liver DR5 mRNA expression after ConA administration was also significantly less evident in CD1d−/− compared to WT livers (Fig. 5B).
We next tested the possibility of whether TRAIL administration after ConA injection was able to induce liver injury in CD1d−/− mice. The simultaneous injection of TRAIL and ConA in CD1d−/− mice significantly induced stronger liver injury as evidenced by increased serum ALT and AST levels in the ConA/TRAIL- compared to the ConA/PBS-treated group (Fig. 5C; Fig. S5). Additionally, TRAIL/ConA stimulation in CD1d−/− mice also resulted in a significant increase in liver IL-33 mRNA expression compared to the ConA/PBS-treated mice (Fig. 5D). Immunohistochemistry revealed that higher IL-33 expression in ConA/TRAIL treated CD1d−/− mice could be localized mainly in hepatocytes (Fig. 5E). There was a significant increase (3.6-fold) in IL-33 positive hepatocytes in ConA/TRAIL- compared to ConA/PBS-treated CD1d−/− mice (Fig. 5F). In summary, these results indicate that TRAIL is essentially involved in inducing IL-33 expression in hepatocytes during ConA-induced liver injury.
Recombinant TRAIL Induces IL-33 Expression in Primary Cultured Murine Hepatocytes.
Finally, we were interested to investigate the direct link between TRAIL stimulation and IL-33 expression in primary murine hepatocytes. We first tested whether primary murine hepatocytes express the corresponding TRAIL death receptor. As shown by immunostaining, DR5 (TRAIL-R2) could be detected in murine hepatocytes in culture (Fig. 6A). We next stimulated hepatocytes with (100 ng/mL) rm-TRAIL, (10 ng/mL) rm-TNFα, or (10 ng/mL) Jo2 antibody (FAS agonist). Interestingly, while rm-TNFα or FasL/Jo2 antibody stimulation (8.5 hours) did not induce IL-33 expression in murine hepatocytes (Fig. 6A), TRAIL significantly induced IL-33 expression in hepatocytes (Fig. 6A) with a progressive relative increase in IL-33-positive hepatocytes at 4, 6, 7, and 8.5 hours following TRAIL stimulation (Fig. 6B). These data clearly demonstrate that TRAIL can induce IL-33 expression in hepatocytes.
IL-33 and its receptor ST2 have been linked to the progression of liver diseases, as recent findings demonstrated overexpression of IL-33 and ST2 in liver fibrosis3 as well as in acute, acute-on-chronic, and chronic hepatic failure.33 Moreover, an immunomodulatory role of IL-33 mediated by regulatory T-cells during ConA-induced acute hepatitis has been shown. These results suggested that the IL-33/ST2 axis has a protective role during liver injury.10 IL-33 is known to be expressed by several cell types in many tissues, especially by endothelial and epithelial cells where it can act as an “alarmin mediator” of the immune system.4, 34–36 Up to now, especially hepatic stellate cells, sinusoidal epithelial, and vascular endothelial cells have been shown to be cellular sources of IL-33 expression in the liver.3 However, we recently found increased IL-33 expression during ConA-induced liver injury and we demonstrated NKT cells-dependent regulation of IL-33 in hepatocytes.2 In the present study, we aimed to better characterize the molecular regulation of IL-33 expression in vivo and in vitro in hepatocytes. We investigated the contribution of different effector molecules like perforin, FasL/Fas, TNFα, and TRAIL/DR5 for controlling IL-33 expression in hepatocytes.
Our first results demonstrated that perforin contributes to IL-33 expression in the liver, as we found a delayed IL-33 expression in perforin−/− hepatocytes compared to WT livers. The perforin-granzyme system is known to be involved in mediating ConA-hepatitis.14 Our findings of ConA-induced liver injury in perforin−/− mice indicated partial contribution of perforin in hepatocyte-specific expression of IL-33 in vivo.
Previous data demonstrated that stimulation of the Fas receptor on hepatocytes by the agonist anti-Fas-antibody Jo-2 causes hepatocyte apoptosis and severe acute hepatitis in mice.21, 22, 37, 38 In our present study, Jo2 (FAS agonist) stimulation (in vivo and in vitro) did not have an impact on the IL-33 expression level in the liver; even a similar kinetic of liver injury was evident as found after ConA stimulation. These results suggested that higher IL-33 expression in hepatocytes is not relevant for Fas-induced liver injury. ConA administration induces higher TNFα expression and earlier work demonstrated that this cytokine is crucially involved in the pathogenesis of this model.17, 19, 25 Here, we found a significant increase in TNFα mRNA expression after ConA-administration, which was associated with higher IL-33 expression. In contrast, administration of recombinant TNFα alone or in combination with D-GalN had no impact on hepatocyte-specific IL-33 expression, suggesting that this cytokine is not involved in triggering its higher expression in liver cells. Further, the rm-TNFα stimulation could not induce IL-33 expression in cultured murine hepatocytes.
The crucial role of TRAIL/DR5 axis during ConA-induced liver injury has been demonstrated earlier showing higher TRAIL expression and its receptor after ConA administration.23, 24 In our present study, TRAIL−/− mice were significantly protected from ConA-hepatitis, as shown by reduced serum AST and ALT levels compared to WT mice, which is in accordance with previous findings.24 This observation was associated with a reduced induction of IL-33 mRNA levels in TRAIL−/− mice compared to WT mice. Immunohistochemical analysis also confirmed a lower number of IL-33-positive hepatocytes in TRAIL−/− mice after ConA stimulation. These data led to our working hypothesis that TRAIL might be involved in triggering IL-33 expression in hepatocytes in vivo. Additionally, comparable DR5 mRNA expression was found in unstimulated WT and TRAIL−/− mice, in agreement with previous studies,23-25, 39, 40 whereas Fas mRNA expression was significantly higher. ConA injection induced higher DR5 expression, which may sensitize hepatocytes for TRAIL-mediated cell death-like in bile duct ligated liver injury in mice.41
Our findings suggest that in the absence of TRAIL, even when DR5 is present, IL-33 expression was not increased in hepatocytes. However, FasL expression was significantly increased in TRAIL−/− mice following ConA injection (Fig. S4D), but obviously this increase had no impact on hepatocyte-specific IL-33 expression in TRAIL−/− mice. The expression of TNFR1/TNFR2 was comparable in WT and TRAIL−/− mice but TNFα expression decreased in TRAIL−/− mice, suggesting a limited role of TNFα and its receptors in hepatocyte-specific IL-33 expression. These results indicate that even some of the intracellular pathways between TRAIL, TNFα, and Fas are redundant and IL-33 seems to be specifically triggered by a distinct cascade by way of TRAIL. Additionally, we demonstrated that IL-33−/− mice were more sensitized to ConA hepatic injury than WT controls, in agreement with a protective effect of IL-33/ST2 axis in ConA-hepatitis.10 Our findings are closer to earlier data describing an increased tendency of liver injury in IL-33−/− mice42; however, we speculate that IL-33 may not be implicated in death cascade; rather, it is expressed/released by dying cells as a readout marker to justify its proposed “alarmin” functions during necrosis.43
Finally, we primed the CD1d−/− mice that are deficient in NKT cells by ConA injection along with a simultaneous injection of rm-TRAIL. Previously, it has been reported that an abundant amount (i.e., nearly 500 μg/mouse) of rm-TRAIL injected into mice is necessary to induce moderate liver injury.24 In our present work, the injection of only 30 μg/mouse of rm-TRAIL (i.e., 10 times less than used earlier) was sufficient to trigger severe ConA-induced hepatitis in CD1d−/− primed mice. Interestingly, the reconstitution of TRAIL in CD1d−/− mice induced liver IL-33 expression, which was localized in hepatocytes. We stimulated primary hepatocytes in vitro with rm-TRAIL in order to exclude an indirect effect of TRAIL on IL-33 expression during ConA-induced liver injury. Interestingly, TRAIL readily induced IL-33 expression in cultured murine hepatocytes.
In conclusion, our work demonstrates that the molecular regulation of IL-33 in hepatocytes during acute hepatitis is not dependent on FasL or TNFα, but on TRAIL.
For immunohistochemistry analysis and animal house facilities, the authors thank the dedicated platforms (i.e., H2P2, ImPACell, and animal house platforms) of SFR BIOSIT, University of Rennes 1, France.