A RIPK3–Caspase 8 Complex Mediates Atypical Pro–IL-1β Processing

Interleukin-1β is an inflammatory cytokine that is closely associated with acute and chronic inflammation and has emerged as a therapeutic target for various systemic and local inflammatory diseases (1). In response to pattern recognition receptor stimulation, APCs such as dendritic cells (DCs) and macrophages produce pro–IL-1β through NF-κB–dependent transcription. Full maturation and secretion of IL-1β require caspase 1–mediated proteolytic processing. Caspase 1 is the enzymatic component of a macromolecular structure termed the inflammasome, which also contains the adaptor protein ASC and a sensor protein such as NLRP3 or AIM2.

Caspase 8 is the initiator caspase for death receptors in the TNFR family. It is recruited to the receptor through binding to its upstream adaptor Fas-associated death domain protein (FADD). Interestingly, recent studies show that caspase 8 can also regulate IL-1β expression by promoting de novo synthesis of pro–IL-1β and processing of pro–IL-1β into its cleaved mature form (2). For instance, in DCs, fungi and mycobacteria stimulate Dectin-1–mediated and caspase 8–dependent IL-1β secretion (3, 4). Infection of macrophages with Salmonella, Yersinia, Citrobacter rodentium, or Escherichia coli also induced caspase 8–dependent IL-1β secretion (5–7). In addition, stimulation with Fas ligand, chemotherapeutic agents, or endoplasmic reticulum stress elicited IL-1β secretion in a caspase 8–dependent manner in LPS-primed macrophages or DCs (8–11). However, the biochemical mechanism that stimulates caspase 8–mediated pro–IL-1β processing is undefined.

Receptor interacting protein kinase (RIPK)3 is a serine/threonine kinase that is crucial for necroptosis in response to ligand binding to TNFR-like death receptors, TLR3 and TLR4 (12). RIPK3 interacts with its upstream activator RIPK1 via the receptor-interacting protein homotypic interaction motif (RHIM) to form an amyloid-like complex termed the necrosome to signal for necroptosis downstream of TNF receptor-like death receptors (13). The kinase activities of RIPK3 and RIPK1 are critical for stabilizing the necrosome and to promote necroptosis. In contrast, RIPK3 interacts with another RHIM-containing adaptor Toll/IL-1R domain–containing adapter inducing IFN-β (TRIF) to mediate TLR3- and TLR4-induced necroptosis (14, 15). Interestingly, both RIPK1 and RIPK3 are cleaved and inactivated by caspase 8 (16, 17). Hence, necroptosis is optimally induced when the FADD–caspase 8 complex is inactivated (18).

Although necroptosis is the most prominent function of RIPK3, several studies showed that RIPK3 could also promote nonnecrotic signaling under certain conditions. For example, RIPK3 drives IL-1β secretion in LPS-primed macrophages or DCs when cellular inhibitor of apoptosis (cIAP) proteins or caspase 8 are depleted (19–21). In addition, in cells that lack TAK1, RIPK3 promotes TNF-induced apoptosis (22). Mice genetically engineered to express the kinase-dead RIPK3 mutant D161N (Ripk3^D161N-KD mice) died at midgestation because of hyperactivation of caspase 8 and excessive apoptosis (23). However, because the RIPK3-dependent effects on IL-1β and apoptosis were detected in highly manipulated experimental systems, it remains unclear whether RIPK3 has physiological functions beyond necroptosis. In this study, we show that RIPK3 promotes IL-1β secretion through assembly of an alternative FADD–caspase 8–activating complex. This necroptosis-independent effect of RIPK3 also requires TRIF and RIPK1. In contrast to necroptosis, the kinase activities of RIPK1 and RIPK3 are dispensable for assembly of this complex and IL-1β secretion. In fact, RIPK3 kinase inhibitors facilitate conformational change that is conducive for assembly of this alternative caspase 8–activating complex. These results reveal RIPK3 as a positive regulator of nonapoptotic caspase 8 activation and highlight the complex interplay between RIPK3 and caspase 8 in cell death-dependent and independent signaling.

Ripk3^−/− (V. Dixit, Genentech), Trif^−/−, Myd88^−/− (E. Lien, University of Massachusetts Medical School), Casp1^−/−Casp11^−/− (K. Rock, University of Massachusetts Medical School), RIPK1-K45A (Ripk1^K45A-KD), and RIPK3-K51A kinase-dead (Ripk3^K51A-KD) knock-in mice (GlaxoSmithKline [GSK]) were used (24, 25). Fadd^fl/fl:Cd11c-Cre (dcFadd^−/−) mice were provided by A. Winoto (University of California Berkeley) (26). Wild-type (WT) mice from Ripk3^+/− intercross were used. All mice were housed in specific pathogen-free facility at University of Massachusetts Medical School. All animal experiments were approved by the institutional animal care and use committee.

LPS (Invivogen), z-VAD-fmk, z-YVAD-fmk, z-IETD-fmk, Necrostain-1 (Nec-1), MG-132 (Enzo Life Sciences), cycloheximide (CHX; Sigma-Aldrich), and N-acetyl-l-cysteine (Calbiochem) were used. RIPK3 kinase inhibitors GSK’840, GSK’872, and GSK’843 were provided by GSK and have been described in previous publications (15, 25).

Bone marrow cells were cultured for 1 wk with 10 ng/ml GM-CSF and 5 ng/ml IL-4 to generate bone marrow–derived DCs (BMDCs). Fadd^fl/fl and Fadd^fl/fl:Cd11c-Cre (dcFadd^−/−) bone marrow cells were differentiated into BMDCs by GM-CSF alone. Ripk1^+/− and Ripk1^−/− newborn liver cells were differentiated to DCs with GM-CSF and IL-4. After stimulation, culture media and cells were used for ELISA, protein, and cell death analyses (CellTiter-Glo Luminescent Cell Viability Assay; Promega). The untreated sample is defined as having a value of 0% cell death. IL-1β and IL-18 were measured by ELISA sets from BD Biosciences and R&D Systems, respectively.

Whole-cell extracts were prepared in radioimmunoprecipitation assay buffer. Protease inhibitor mixture (Roche) and phosphatase inhibitor mixture (Sigma-Aldrich) were included in lysis buffer. Secreted proteins in culture media were precipitated by trichloroacetic acid. For immunoprecipitation, anti-RIPK3 (Prosci) or anti–caspase 8 (Enzo Life Sciences) Abs were used. Western blotting was performed with anti-RIPK1 (BD Biosciences), RIPK3 (Prosci), FADD (provided by A. Winoto), caspase 8 (Enzo Life Sciences), MyD88 (eBiosciene), IL-1β, cIAP1/2 (R&D Systems), GFP (Roche), caspase-1 (Santa Cruz), and IL-18 (BioVision) Abs. Anti-HSP90 Ab (BD Biosciences) was used as a loading control.

Various mutants of mouse RIPK3 were cloned into pTRIPZ vector (Open Biosystems). The 293T cells were transfected by LV-GFP pLKO.1 (addgene 25999), LV-Cre pLKO.1 (addgene 25997), or mRIPK3/pTRIPZ in combination with pMD2.G and psPAX2 to generate lentivirus. On day 4 during BMDC differentiation, floating cells were collected; plated in media containing lentivirus, 10 μg/ml polybrene, GM-CSF, and IL-4; and incubated for an additional three days prior to stimulation. Lentivirus carrying RIPK3 or its mutant genes was transduced to Ripk3^−/− 3T3 cells, and then cells were selected for puromycin resistance. To induce RIPK3 expression, cells were treated with 1 μg/ml doxycycline.

The p values were calculated using unpaired t test with Welch’s correction. The p values <0.05 were considered statistically significant.

IL-1β production generally requires two distinct signals, as follows: a first signal that elicits pro–IL-1β expression through NF-κB pathway, and a second signal that causes inflammasome activation and caspase 1–dependent processing of pro–IL-1β. However, LPS alone was sufficient to induce a low level of IL-1β secretion in DCs (27). Surprisingly, we found that LPS-induced IL-1β secretion was strongly suppressed in Ripk3^−/− BMDCs (Fig. 1A, Supplemental Fig. 1A, 1B) (28). LPS alone did not lead to secretion of the related cytokine IL-18 (Supplemental Fig. 1B, 1C), although pro–IL-18 was constitutively expressed in BMDCs (Supplemental Fig. 1D). Stimulation of the inflammasome with ATP led to greater level of secreted IL-1β and detectable IL-18 secretion that were also dependent on intact RIPK3 (Supplemental Fig. 1C). Genetic deletion of caspase 1, Nlrp3, Asc (27), or inhibition of caspase 1 with the inhibitor z-YVAD-fmk only partially suppressed LPS-induced IL-1β secretion (Fig. 1B). This suggests that RIPK3 is responsible for caspase 1–dependent and independent pathways of pro–IL-1β processing.

Besides caspase 1 activation, LPS also triggered caspase 8 activation in WT BMDCs (Fig. 1C, Supplemental Fig. 1D). Strikingly, caspase 8 activation was also abolished in Ripk3^−/− BMDCs (Fig. 1C, Supplemental Fig. 1D) (28). RIPK3 did not affect the priming of BMDCs by LPS, because pro–IL-1β induction 1 h after LPS treatment was normal in Ripk3^−/− BMDCs (Fig. 1C, Supplemental Fig. 1D). However, the level of intracellular pro–IL-1β was higher in Ripk3^−/− BMDCs than in WT BMDCs 6 h after LPS treatment (Supplemental Fig. 1E). These results are most consistent with impaired cleavage of pro–IL-1β in Ripk3^−/− BMDCs. Consistent with a deficiency in cleavage of pro–IL-1β, the caspase 8–specific inhibitor z-IETD-fmk partially suppressed LPS-induced IL-1β secretion, but not pro–IL-1β induction (Fig. 1D, Supplemental Fig. 1F). Coadministration of caspase 1 and caspase 8 inhibitors further abolished the residual IL-1β secretion (Fig. 1D), indicating that both caspases contributed to IL-1β secretion by LPS-treated BMDCs.

IAP proteins are natural inhibitors of RIPK3-dependent IL-1β secretion (19, 20). However, LPS treatment did not alter cIAP1/2 expression in BMDCs (Supplemental Fig. 1G), indicating that RIPK3-dependent IL-1β secretion was not due to loss of cIAP expression. Although it is an initiator caspase for apoptosis, caspase 8 activation did not cause cell death in LPS-treated WT BMDCs. However, inhibition of de novo synthesis of prosurvival molecules with CHX led to LPS-induced cell death that was dependent on RIPK3 (Fig. 1E). These results indicate that, besides caspase 1, RIPK3 also stimulates caspase 8–mediated pro–IL-1β processing in LPS-treated BMDCs. Moreover, in contrast to death receptor–induced apoptosis, LPS-induced caspase 8 activation did not result in cell death.

MyD88 is required for signaling by all TLR family receptors except TLR3, whereas TRIF is a unique signal adaptor for TLR3 and TLR4. LPS-induced caspase 8 activation was abolished in Trif^−/− BMDCs (Fig. 1F). In contrast, Myd88^−/− BMDCs exhibited normal caspase 8 activation. These results are consistent with a previous report that in macrophages, TRIF is required for caspase 8–dependent IL-1β secretion downstream of TLR3 or TLR4 and CHX stimulation (2). Moreover, these results also indicate that TRIF acts upstream of RIPK3 to promote caspase 8 activation.

Our recent work shows that RIPK3 promotes caspase 1 activation in BMDCs through reactive oxygen species production (28). However, unlike caspase 1, the reactive oxygen species scavenger N-acetyl-l-cysteine did not block LPS-induced caspase 8 activation (Fig. 2A). This indicates that RIPK3 controls caspase 1 and caspase 8 activity through different mechanisms. Binding to the adaptor FADD is a requisite step for autocleavage and activation of caspase 8 in response to death receptor stimulation (29). Interestingly, we detected constitutive interaction between caspase 8 and FADD in BMDCs (Fig. 2B, Supplemental Fig. 2A). The interaction between FADD and caspase 8 was not changed upon LPS treatment (Fig. 2B), suggesting that LPS causes a qualitative, but not quantitative change in FADD–caspase 8 interaction. This is in contrast to TNF- and CHX-stimulated 3T3 fibroblasts, which showed inducible interaction between FADD and caspase 8 (Supplemental Fig. 2A).

Because RIPK3 was required for LPS-induced caspase 8 activation (Fig. 1C), we examined RIPK3 recruitment to the caspase 8 complex. Indeed, LPS significantly enhanced the interaction of RIPK3 with caspase 8 and FADD (Fig. 2C). RIPK1 is a RHIM- and death domain–containing adaptor that facilitates death receptor–induced apoptosis and necroptosis. We reasoned that RIPK1 might interact with RIPK3 via the RHIM and FADD or caspase 8 via the death domain to mediate caspase 8 activation. To ascertain this possibility, we took advantage of Ripk1^K45A-KD mice, which express a knock-in version of kinase-inactive RIPK1 (24). Because exon 3 of the Ripk1^K45A-KD allele is flanked by loxP sites, we can specifically delete Ripk1 by Cre-mediated recombination in BMDC cultures. We transduced WT or Ripk1^K45A-KD BMDCs with lentivirus expressing Cre or GFP as control and observed significant reduction of RIPK1 expression in Ripk1^K45A-KD, but not WT BMDCs transduced with Cre-expressing lentivirus (Fig. 2D, compare lanes 3 and 7). In these cells, LPS-induced caspase 8 activation and IL-1β secretion were suppressed compared with the cells transduced with GFP vector (Fig. 2D [compare lanes 6 and 8], Supplemental Fig. 2B). In contrast, WT and Ripk1^K45A-KD BMDCs transduced with control GFP lentivirus exhibited equal caspase 8 activation and IL-1β secretion (Fig. 2D [compare lanes 2 and 6], Supplemental Fig. 2B). Consistent with results from Ripk1^K45A-KD BMDCs, the RIPK1 kinase inhibitor Nec-1 also did not affect TLR4-induced caspase 8 activation (Supplemental Fig. 2C). These results indicate that, whereas an intact RIPK1 is essential for TLR4-induced caspase 8 activation, its kinase activity is dispensable. Moreover, our results suggest that LPS triggers formation of a RIPK1–RIPK3–FADD–caspase 8 complex. This complex is reminiscent of the apoptosis complex driven by the kinase-inactive RIPK3 mutant D161N (23). However, recruitment of RIPK1 to the RIPK3–FADD–caspase 8 complex in LPS-treated BMDCs was weak (see below), in part because of a reduction in overall RIPK1 expression (Fig. 2D). The reduction in cellular RIPK1 was not due to caspase-mediated cleavage or proteasome-dependent degradation (Supplemental Fig. 2D), but might be caused by translocation to detergent-insoluble compartment (30).

The results shown above indicate that RIPK3 licenses activation of the FADD–caspase 8 complex. In necroptosis, RIPK3 kinase activity is essential for assembly of a signaling complex termed the necrosome (31). To address whether RIPK3 kinase activity is similarly required for LPS-induced caspase 8 activation, we used the recently described RIPK3-specific kinase inhibitors GSK’872 and GSK’843 (15, 25). Treatment with the second mitochondria–derived activator of caspase (SMAC)–mimetic LBW242 and the pan-caspase inhibitor z-VAD-fmk induced autocrine TNF production and RIPK3-dependent necroptosis in BMDCs (Fig. 3A) (32). This necrotic cell death was suppressed by GSK’872 and GSK’843, but not by the human RIPK3-specific inhibitor GSK’840 (25). Surprisingly, GSK’872 and GSK’843 significantly increased LPS-induced IL-1β secretion in a dose-dependent manner (Fig. 3B). The enhanced IL-1β secretion requires TLR4 signaling, because the inhibitors did not induce IL-1β secretion without LPS (Fig. 3C). The effect of the inhibitors was RIPK3 specific, because neither GSK’872 nor GSK’843 increased IL-1β secretion in Ripk3^−/− BMDCs (Fig. 3B). The RIPK3 kinase inhibitor-induced increase in IL-1β secretion was not driven by caspase 1, because enhanced LPS-induced IL-1β secretion was also detected in RIPK3 inhibitor-treated Casp1^−/−Casp11^−/− BMDCs (Fig. 3D, filled bars).

FADD is an essential adaptor of caspase 8 activation. DC-specific Fadd-deficient (dcFadd^−/−) mice developed systemic inflammation associated with elevated proinflammatory cytokines, including IL-1β (26). BMDCs generated from dcFadd^−/− mice showed significantly increased IL-1β secretion upon treatment with LPS (Fig. 4A) (26). The enhanced IL-1β secretion was due to increased pro–IL-1β induction and caspase 1, but not caspase 8, activation, because caspase 8 activation was abolished in dcFadd^−/− BMDCs (Fig. 4B) (21). Moreover, the RIPK3 inhibitor GSK’872 did not enhance IL-1β secretion or caspase 8 activation in dcFadd^−/− BMDCs (Fig. 4A, 4B). Because GSK’872 only enhanced caspase 8 activation, but not pro–IL-1β expression and caspase 1 activation (Fig. 4C), these results indicate that the RIPK3 inhibitor enhanced pro–IL-1β processing through caspase 8, but not caspase 1. In addition to FADD, RIPK1, and RIPK3, LPS-induced caspase 8 activation also requires an intact TRIF, but not MyD88 (Supplemental Fig. 2E).

Consistent with the specific stimulation of caspase 8–mediated pro–IL-1β processing, GSK’872 augmented LPS-induced interaction of RIPK3 with FADD and caspase 8 (Fig. 4D). Specifically, GSK’872 significantly enhanced LPS-induced RIPK1 recruitment to RIPK3 (Fig. 4D). Moreover, deletion of RIPK1 by Cre-mediated recombination in Ripk1^K45A-KD BMDCs repressed the GSK’872-mediated enhancement of caspase 8 activation (Fig. 4E). The residual caspase 8 activation by GSK’872 was most likely due to incomplete deletion of RIPK1 (Fig. 4E, lane 10, second panel). Indeed, GSK’872-mediated enhancement of caspase 8 activation was completely abrogated in Ripk1^−/− DCs generated from Ripk1^−/− newborn liver cells (Supplemental Fig. 2F). The RIPK1 kinase inhibitor Nec-1 failed to reverse the enhanced caspase 8 activation by the RIPK3 inhibitor (Supplemental Fig. 2G). Hence, the enhanced RIPK3-dependent caspase 8 activation and pro–IL-1β processing by RIPK3 kinase inhibitors require an intact RIPK1, but not its kinase activity.

The results with the RIPK3 inhibitors are surprising given the positive role of RIPK3 in pro–IL-1β processing. One possible explanation for these results is that RIPK3 phosphorylates and inhibits an unknown substrate that is required for switching on caspase 8 activity. In support of this model, mice expressing the kinase-inactive RIPK3 mutant D161N (Ripk3^D161N-KD) were recently reported to suffer from embryonic lethality due to massive caspase 8–mediated apoptosis in the yolk sac (23). However, mice expressing another kinase-inactive RIPK3 mutant K51A (Ripk3^K51A-KD) were viable (25), suggesting that inhibition of RIPK3 kinase activity per se does not necessarily induce caspase 8 activation. To further test the role of RIPK3 kinase activity in LPS-induced caspase 8 activation, we generated BMDCs from Ripk3^K51A-KD mice. Treatment of DCs or macrophages with SMAC mimetics in the presence of caspase inhibition causes autocrine TNF production and necroptosis (32). Consistent with the essential role of RIPK3 kinase activity in necroptosis, Ripk3^K51A-KD BMDCs were protected from SMAC-mimetic and zVAD-induced necroptosis (Fig. 4F). In contrast to the RIPK3 inhibitors, LPS-induced IL-1β secretion was not increased in Ripk3^K51A-KD BMDCs (Fig. 4G). In fact, Ripk3^K51A-KD BMDCs secreted reduced level of IL-1β (Fig. 4F). However, the reduced IL-1β secretion was most likely caused by reduced RIPK3 expression in Ripk3^K51A-KD BMDCs (Fig. 4F). Importantly, GSK’872 still enhanced IL-1β secretion by LPS-treated Ripk3^K51A-KD BMDCs (Fig. 4G). These results strongly suggest that RIPK3 kinase activity is not required for LPS-induced caspase 8 activation. Rather, the RIPK3 kinase inhibitors promote IL-1β secretion by inducing a conformational change that is amenable for assembly of the alternative caspase 8 complex.

Although the low doses of the RIPK3 kinase inhibitor we used in BMDCs did not cause any toxicity (Supplemental Fig. 3A), higher concentrations of GSK’872 enhanced caspase 8 activation and apoptosis in RIPK3-positive cells (25). Strikingly, the apoptosis induced by high doses of the RIPK3 kinase inhibitors requires a similar RIPK1–RIPK3–FADD–caspase 8 complex (25). Therefore, we used this system to further examine the requirement of the RHIM, which is essential for necroptosis (13), for RIPK3-dependent caspase 8 activation. We expressed WT or mutant RIPK3 in Ripk3^−/− 3T3 cells under the control of doxycycline. Expression of WT RIPK3 and RIPK3-D161N, but not RIPK3-K51A, led to caspase 8–dependent apoptosis (25) (Supplemental Fig. 3B). The cell death induced by RIPK3-D161N is apoptotic, because it was reversed by the pan-caspase inhibitor z-VAD-fmk (Supplemental Fig. 3C). However, z-VAD-fmk did not rescue cell death in WT RIPK3-expressing cells due to a switch from apoptosis to necroptosis (Supplemental Fig. 3C). Caspase 8–dependent apoptosis was enhanced by GSK’872 in cells expressing WT RIPK3, RIPK3-K51A, and RIPK3-D161N (Supplemental Fig. 3B), indicating that the RIPK3 kinase inhibitor enhanced caspase 8 activation independent of the inhibition of RIPK3 kinase activity. In contrast, tetra-alanine substitution in the RHIM abolished caspase 8 activation and apoptosis in WT RIPK3- and RIPK3-D161N–expressing cells (Supplemental Fig. 3B) (25). In addition, the effect of GSK’872 was no longer observed in the RHIM mutant (Supplemental Fig. 3B). These results indicate that RHIM-mediated interaction is crucial for necroptosis as well as RIPK3-dependent caspase 8 activation and pro–IL-1β processing.

RIPK3 is touted as a major driver of inflammation due to necroptosis-associated release of intracellular immunogenic contents. However, recent evidence shows that RIPK3 can promote inflammation independent of necroptosis (33). Specifically, RIPK3 positively promotes IL-1β secretion in macrophages or DCs when caspase 8 or IAP proteins are depleted (19–21). In contrast, several recent studies show that RIPK3 is dispensable for IL-1β secretion in macrophages when caspase 8 and IAP functions are preserved (5–9). In this study, we show that RIPK3 functions as an unexpected positive promoter of caspase 8 activation and IL-1β secretion in BMDCs. This is a surprising result because caspase 8 cleaves RIPK1 and RIPK3 and is widely recognized as a negative regulator of necroptosis (16–18). The RIPK3-dependent effect on IL-1β secretion indicates that, under certain conditions, RIPK3 is a positive activator of caspase 8. Our results therefore revealed a complex interplay between the RIPKs and FADD/caspase 8 in cell death and inflammatory signaling.

Mechanistically, we show that RIPK3 promotes assembly of an alternative caspase 8–activating complex that comprises RIPK1, RIPK3, FADD, and caspase 8. The TLR3/4 adaptor TRIF is also essential for caspase 8 activation. However, because Abs that recognize endogenous mouse TRIF are not available (data not shown), we cannot confirm whether TRIF is also a component of this complex. RIPK3 interacts with RIPK1 or TRIF via the RHIM (14, 15, 34, 35). Interestingly, the RHIM of TRIF was reported to be important for caspase 8–dependent IL-1β maturation (2). Therefore, we speculate that similar RHIM-mediated interaction among TRIF, RIPK1, and RIPK3 may be involved in the assembly of the caspase 8–activating complex downstream of TLR4. This notion is bolstered by the observation that RIPK3 is dispensable in situations in which caspase 8 activation occurs independent of TRIF, such as that in β-glucan or Fas-induced IL-1β secretion (4, 9). Hence, it appears that there are multiple distinct mechanisms by which caspase 8 facilitates pro–IL-1β processing.

The ripoptosome is a macromolecular complex that activates apoptosis and is made up of RIPK1, FADD, and caspase 8 (36, 37). Recruitment of RIPK3 switches the ripoptosome toward necroptosis. Thus, the ripoptosome resembles the LPS-induced caspase 8–activating complex in composition. Although its primary function is to promote IL-1β maturation, the LPS-induced caspase 8 complex has the capacity to induce apoptosis when de novo protein synthesis is blocked by CHX. Moreover, RIPK1, RIPK3, FADD, and caspase 8 are also components of the necrosome (31). That means that the same adaptors can initiate at least three distinct signaling outcomes, as follows: apoptosis, necroptosis, and IL-1β processing. How is this remarkable diversity in signaling outcome achieved? Although the details are yet to be determined, additional adaptors may be involved in this decision. For example, formation of the ripoptosome and the necrosome requires depletion of the cellular IAPs. Although depletion of cellular IAPs is not a prerequisite for the caspase 8 complex that cleaves pro–IL-1β, other adaptors may yet determine the signaling outcome of this alternative caspase 8–activating complex. It is also noteworthy that the kinase activities of RIPK1 and RIPK3, which are essential for necrosome and ripoptosome assembly, are dispensable for the LPS-induced caspase 8 activation. Hence, posttranslational modifications such as ubiquitination, protein phosphorylation, or binding by pathway-specific adaptors may determine the signaling outcome of the RIPK1–RIPK3–FADD–caspase 8 complex.

We thank D. Porter (Novartis) for LBW242 and V. Dixit, E. Mocarski, W. Kaiser, K. Rock, M. Kelliher, E. Lien, and A. Winoto for providing different knockout mice used in this study.

J.B. and P.J.G. are employees of GlaxoSmithKline.