Caspase-8 regulates TNF-alpha induced epithelial necroptosis and terminal ileitis

Caspase-8 is a cystein protease critically involved in regulating cellular apoptosis. Upon activation of death receptors, including TNF-receptor and Fas, caspase-8 is activated by limited autoproteolysis and the processed caspase-8 subsequently triggers the caspase cascade which finally leads to apoptotic cell death. Caspase-mediated apoptosis is important for the turnover of IEC and for shaping the morphology of the gastrointestinal tract⁴. Furthermore, recent data have indicated a role of caspase-mediated apoptosis of IEC in the pathogenesis of inflammatory bowel diseases (IBD) like Crohn's disease and ulcerative colitis^5,6.

To study the function of caspase-8 in the gut, we generated mice with an intestinal epithelial cell specific deletion of caspase-8 (Casp8^ΔIEC). Accordingly, mice with floxed caspase-8 alleles were bred with mice expressing the Cre-recombinase under the control of the IEC-specific villin promoter. Specific deletion of caspase-8 in IEC was confirmed by PCR and Western blotting (Supplementary Fig. 1a, b). Casp8^ΔIEC mice were born at the expected Mendelian ratios and developed normally, although weighing on average slightly less than control littermates at 8 weeks of age (data not shown). Despite the paradigm that apoptosis is important for regulating epithelial cell numbers⁴, histological and morphometrical analysis of the jejunum and colon of Casp8^ΔIEC mice showed no overt changes of tissue architecture or apoptosis dysregulation (Supplementary Figs. 1d-f, 2). While this suggested that caspase-8 is not essential for the structural development of the gut, high resolution endoscopy showed erosions in the terminal ileum but not in the colon of Casp8^ΔIEC mice (Fig. 1a, Supplementary Fig. 1c). Histological analysis demonstrated marked destruction of the architecture and signs of inflammation including bowel wall thickening, crypt loss and increased cellularity in the lamina propria (Fig. 1b) in more than 80% of all ileal specimens. This finding of spontaneous ileitis in the absence of caspase-8 in IEC was further supported by increased expression of the inflammation markers S100A9 and TNF-α and by elevated infiltration of the lamina propria with CD4⁺ T cells and granulocytes (Fig. 1c, Supplementary Fig. 3).

To investigate whether caspase-8 deficiency sensitizes mice to experimental intestinal inflammation in the large bowel, we subjected Casp8^ΔIEC and control mice to dextran sodium sulfate (DSS), a well established model of experimental colitis. Strikingly, Casp8^ΔIEC mice but not control mice showed high lethality and lost significantly more weight than control mice (Supplementary Fig. 4). All Casp8^ΔIEC but none of the control mice developed rectal bleeding, endoscopic and histological signs of very severe colitis with epithelial erosions, a finding which was confirmed by quantitative PCR for the IEC marker villin. Together, our data indicate that a lack of caspase-8 in IEC renders mice highly susceptible to spontaneous ileitis and experimentally induced colitis.

To screen for molecular mechanisms that sensitize Casp8^ΔIEC mice for intestinal inflammation, we next performed whole genome gene chip analysis of IEC from unchallenged control and Casp8^ΔIEC mice. Of the 45,000 expression tags analyzed, 197 were significantly downregulated and 136 upregulated in Casp8^ΔIEC mice (fold change > 2.0, p < 0.05) (Supplementary Table 1). Gene ontology (GO) analysis of downregulated genes showed that several biological pathways were impaired in Casp8^ΔIEC mice as compared to littermate controls, with the GO terms “defence response” and “MHC class II antigen presentation” reaching very high levels of significance (Fig. 1d). Within the group of genes upregulated in Casp8^ΔIEC mice, no GO term reached significance levels. Strikingly, within the defence response gene set, genes belonging to the family of antimicrobial peptides including several defensins, lysozyme and phospholipases were amongst the most strongly downregulated genes (Supplementary Fig. 5a), suggesting defects in the antimicrobial defence of the intestinal epithelium in the absence of epithelial caspase-8. Expression of antimicrobial peptides is a hallmark of Paneth cells, epithelial derived cells which are located at the base of crypts of Lieberkuehn in the small intestine⁷. Within Paneth cells, antimicrobial peptides are stored in cytoplasmatic granules, from which they can be released into the gut lumen, thereby contributing to intestinal host defences. Strikingly, Casp8^ΔIEC mice showed a complete absence of cells with secretory granules and lysozyme expression at the crypt base of the small intestine, suggesting that Paneth cells are lacking in the gut of these animals (Fig. 1e). Furthermore, the number of mucus secreting goblet cells was reduced, as indicated by staining with Ulex Europaeus Agglutinin 1 (UEA-1), a lectin binding to glycoproteins characteristic for these cells (Supplementary Fig. 5b). In contrast, we observed no changes in the appearance of enteroendocrine cells or absorptive enterocytes, as indicated by staining for chromogranin-A or alkaline phosphatase, respectively (Supplementary Fig. 5b). The lack of Paneth cells and partial lack of goblet cells was confirmed by quantitative gene expression analysis showing diminished expression of Paneth cell and goblet cell specific genes, while expression of genes specific for enteroendocrine cells, enterocytes and progenitor cells was unchanged (Supplementary Fig. 5c). Thus, collectively, our data implicate that deficient expression of caspase-8 in the intestinal epithelium results in diminished Paneth and goblet cell numbers and hence may lead to defects in antimicrobial host defence and terminal ileitis.

In addition to controlling apoptosis, there is growing evidence that caspase-8 regulates several non-apoptotic cellular mechanisms including proliferation, migration and differentiation^8,9. Accordingly, caspase-8 has been shown to promote the terminal differentiation of macrophages and keratinocytes¹⁰. Thus, we reasoned that caspase-8 might support intestinal immune homeostasis by promoting the terminal differentiation of Paneth and goblet cells. To verify this hypothesis, we performed long term organoid cultures of small intestinal crypts in vitro as previously introduced by Sato et al.¹¹. Isolated crypts from the small intestine underwent multiple crypt fissions forming large organoids in both control and Casp8^ΔIEC mice (Supplementary Fig. 6a). In striking contrast to the absence of Paneth cells in crypts from Casp8^ΔIEC mice in vivo, organoids grown from these crypts in vitro over a period of 1 to 4 weeks showed Paneth cells indistinguishable in localization and number from organoids cultured from control littermate mice (Fig. 2a). PCR-analysis confirmed deletion of the caspase-8 allele in Paneth cell positive organoids derived from Casp8^ΔIEC mice (Supplementary Fig. 6b). Thus, our data suggest that caspase-8 is not required for the differentiation of IEC into Paneth cells, but that a factor present in vivo either inhibited the development of or ablated Paneth cells in the absence of caspase-8 expression.

Indeed, Casp8^ΔIEC mice showed a large number of dying epithelial cells at the crypt base with pyknotic nuclei and a shrunken eosinophilic cytoplasm (Fig. 2b), implicating that caspase-8 deficient Paneth and goblet cells might be sensitive to cell death. Dying crypt cells usually lacked typical apoptotic body formation, suggesting necrotic rather than apoptotic cell death. This conclusion was supported by the observation that dying cells were TUNEL positive, but showed no activation of caspase-3 (Fig. 2b). The number of necrotic cells at the crypt base was significantly higher in Casp8^ΔIEC mice as compared to control mice (Fig. 2c). Finally, electron microscopy of the crypt area demonstrated cells with typical features of necrosis including mitochondrial swelling and extensive vacuole formation while typically lacking blebbing usually associated with apoptosis (Fig. 2d). Importantly, many cells with features of necrosis also showed electron dense granules indicating necrotic Paneth cells. This conclusion was supported by electron microscopy of mice in which the caspase-8 deletion was induced in adult mice by injection with Tamoxifen (inducible Casp8^ΔIEC) to detect early effects of caspase-8 deletion. Taken together, our data suggest that the lack of caspase-8 sensitizes Paneth cells in the crypts of the small intestine to necrotic cell death.

TNF-α stimulated death receptor signaling has been described to promote necrosis in a number of different target cell types, especially when apoptosis was blocked using caspase-inhibitors^12,13. We therefore reasoned that in the absence of caspase-8, TNF-α signaling might lead to excessive crypt cell death. To test this hypothesis, we intravenously (i.v.) administered mice with TNF-α using a dose which is not lethal to normal mice. While all control mice were still alive after 5 hours, Casp8^ΔIEC mice showed significantly more pronounced hypothermia and very high lethality (Supplementary Fig. 7a,b). Histological analysis demonstrated villous atrophy and severe destruction of the small bowel of Casp8^ΔIEC mice as compared to control littermates and an increased number of dying epithelial cells, as indicated by the pyknotic nuclei seen in the H&E stain of crypts and cells in the crypt lumen (Supplementary Fig. 7c-e). Similar to unchallenged mice, dying crypt cells were negative for active caspase-3 but positive for TUNEL staining (Supplementary Fig. 8a-c), suggesting that in the absence of caspase-8, TNF drives excessive necrosis of epithelial cells.

Recent data from other experimental systems have shown that inhibition of caspase activity in genetic models or by using specific caspase inhibitors can result in an apoptosis-independent type of programmed necrosis called necroptosis¹⁴. Necroptosis has been shown to be mediated by the kinases RIP1 and RIP3^15,16. Upon induction of necrosis, RIP3 is recruited to RIP1 to establish a necroptosis inducing protein complex. RIP3 seems to be essential for the molecular mechanisms driving necroptosis and expression of RIP3 has been demonstrated to correlate with the sensitivity of cells towards necroptosis^17,18. Moreover, deletion of RIP3 has recently been shown to rescue the lethal phenotype of general caspase-8 deficient mice by blocking cell death^19,20. Strikingly, expression of RIP3 mRNA, but not RIP1 mRNA was significantly increased in IEC isolated from unchallenged Casp8^ΔIEC mice as compared to controls (Supplementary Fig. 9). Condensed nuclei as observed in the crypts of Casp8^ΔIEC mice stained for RIP3 using immunohistochemistry (Fig. 3a, Supplementary Fig. 8e). Moreover RIP3 was overexpressed in the small intestine of TNF-α treated Casp8^ΔIEC mice when compared to control littermate mice (Fig. 3b, Supplementary Fig. 8d), suggesting that the lack of caspase-8 in the intestinal epithelium might sensitize IEC to RIP-mediated necroptotic cell death. In line with this hypothesis, RIP3 staining was detected especially in cells at the crypt base (Supplementary Fig. 10). Moreover, significant levels of TNF-α were detected in lamina propria cells adjacent to crypt IEC and both TNF-α and RIP3 expression were highest in the terminal ileum.

RIP-mediated necroptosis can be blocked in vitro and in vivo by using necrostatin-1 (nec-1), an allosteric small-molecule inhibitor of the RIP1 kinase²¹. Thus we reasoned that nec-1 might prevent TNF-α induced epithelial necroptosis and lethality in Casp8^ΔIEC mice. Indeed, in vitro, small intestinal organoid cultures from Casp8^ΔIEC mice but not from control mice exhibited necrosis within 24 hours after addition of TNF-α to the tissue culture. However, when cell cultures were pre-treated with nec-1, organoid necrosis was blocked (Fig. 3c, d). Moreover, pre-treatment with nec-1 significantly reduced TNF-α induced lethality and small intestinal tissue destruction in Casp8^ΔIEC mice (Fig. 3e, f). Collectively, our data suggest that deficient caspase-8 expression renders Paneth cells susceptible to TNF-α induced necroptosis highlighting a regulatory role of caspase-8 in antimicrobial defence and in maintaining immune homeostasis in the gut.

Interestingly, defects in Paneth cell function and in the expression of antimicrobial peptides have been described in Crohn's disease patients and accumulating evidence support a role for Paneth cells in the pathogenesis of this disease^1,7,22. As anti-TNF-α treatment is successfully used in the therapy of patients with Crohn's disease, we hypothesized that similar to Casp8^ΔIEC mice, human Paneth cells might be susceptible to TNF-α induced necroptosis. Paneth cell dysfunction has been reported in mice deficient of autophagy genes such as Atg16L1, a gene associated with Crohn's disease susceptibility³. Moreover, depletion of Paneth cells in the small intestine has been reported in patients with ileal Crohn's disease and patients with ulcerative colitis involving the terminal ileum²³. Strikingly, immunohistochemistry of samples derived from the terminal ileum of human patients undergoing endoscopic examination revealed high expression of RIP3 exclusively in Paneth cells (Fig. 4a), but not in other intestinal epithelial cell types. Importantly, analysis of histological samples from the terminal ileum of control patients and patients with active Crohn's disease showed a significant decrease in the number of Paneth cells and high numbers of dying cells with shrunken eosinophilic cytoplasm (Fig. 4b) at the crypt base, similarly to Casp8^ΔIEC mice. Electron microscopy of the Paneth cell area in the terminal ileum of patients with CD, showed increased necrotic cell death, as indicated by abundant organelle swelling, vacuole formation and the lack of blebbing (Fig. 4c+d). Moreover, crypt epithelial cells in areas of acute inflammation usually were TUNEL positive but lacked staining for active caspase-3 (Fig. 4e). Finally, ileal biopsies from control patients showed Paneth cell loss in the presence of high levels of exogenous TNF-α, an effect that was reversible by co-incubation with necrostatin-1 (Fig. 4f, Supplementary Fig. 11a). Thus, our data suggest that necroptosis of Paneth cells is a feature of Crohn's disease. Since it has recently been shown that anti-TNF treatment partially restores the deficient expression of antimicrobial peptides in Crohn's disease patients²⁴, our data suggest that the high levels of TNF-α present in the lamina propria of the inflamed ileum induce Paneth cell necroptosis and may provide a molecular explanation for the defects in antimicrobial defence observed in these patients.

In summary our data uncover an unexpected function of caspase-8 in regulating necroptosis of intestinal epithelial cells and in maintaining immune homeostasis in the gut. Caspase-8 deficient mice had no defect in overall gut morphology, demonstrating that cell death independent from the extrinsic apoptosis pathway can regulate intestinal homeostasis. Indeed, studies using electron microscopy have shown various different cell death morphologies in the small intestine including morphological changes usually seen in necrosis such as cell swelling and a degradation of organelles and membranes²⁵. Caspase-8 deficient mice completely lacked Paneth cells, suggesting that these cell types are highly susceptible to necroptosis. Crohn's disease patients frequently show reduced Paneth and goblet cell numbers and reduced expression of Paneth cell derived defensins in areas of acute inflammation, suggesting that necroptosis might be involved in the pathogenesis of human IBD²². Indeed, we were able to demonstrate constitutive expression of RIP3, a kinase sensitizing cells to necroptosis^17,18, in human Paneth cells. Moreover cells undergoing necroptosis were found at the crypt base in patients with Crohn's disease and Paneth cell death could be inhibited by blocking necroptosis.

Caspase-8 has recently been shown to suppress RIP3-RIP1-kinase-dependent necroptosis following death receptor activation^14,17,18. This has been highlighted by genetic studies demonstrating that deletion of RIP3 can rescue the embryonic lethality observed in mice with a general deletion of caspase-8^19,20. Thus it is getting increasingly clear, that caspase-8 has an essential function in controlling RIP-3 mediated necroptosis. On the molecular level, caspase-8 has been demonstrated to proteolytically cleave and inactivate RIP1 and RIP3 thereby regulating the initiation of necroptosis²⁰. Currently, to our knowledge, no study has demonstrated caspase-8 as an IBD-linked gene using genetic studies. However, caspase-8 was expressed at relatively low levels in the crypt area of the human terminal ileum and activation of caspase-8 was only seen sporadically along the crypt-villous axis (Supplementary Fig. 11b). Given the low expression of caspase-8 and high expression of RIP3 at the base of crypts, cells residing in this area may be susceptible to necroptosis upon stimulation with death receptor ligands such as TNF-α.

TNF-α is an important contributor to the pathogenesis of IBD and treatment with biological drugs targeting TNF-α is effective in patients with Crohn's disease. Interestingly, mice overproducing TNF-α develop transmural intestinal inflammation with granulomas primarily in the terminal ileum, similar to Crohn's disease²⁶. TNF-α was a strong promoter of intestinal epithelial necroptosis in our experiments. Since anti-TNF treatment has been shown to restore expression of certain antimicrobial peptides²⁴, it is tempting to speculate, that TNF-α induced necroptosis of intestinal epithelial cells contributes to the pathogenesis of IBD and that Paneth and goblet cell sensitivity towards TNF induced necroptosis may be an early event in IBD development. However, it remains to be determined, whether Paneth cell necroptosis in Crohn's disease patients is quantitatively sufficient to affect the disease process. Further studies will be needed to decipher the precise regulatory network of death receptors, RIP kinases and caspase-8 at the crypt base and it will be important to elucidate whether genetic, epigenetic or post-translational mechanisms restrict expression or activation of caspase-8 in IBD patients. Ultimately, our data for the first time demonstrate necroptosis in the terminal ileum of patients with Crohn's disease and suggest that regulating necroptosis in the intestinal epithelium is critical for the maintenance of intestinal immune homeostasis. Targeting necroptotic cellular mechanisms emerges as a promising option in treating patients with IBD.