Receptor-interacting protein (RIP) kinases, in particular RIPK1, RIPK2 and RIPK3, have emerged as pleiotropic modulators of inflammatory responses that act either by directly regulating intracellular inflammatory signaling pathways or by causing apoptotic or necrotic cell death. In this Review, we discuss the signaling pathways and immunological functions of these RIP kinases in the inflammatory response to microbial infection and tissue injury, as well as their potential roles in the pathogenesis of inflammatory disease and aging.
The immune system provides protection against infection and tissue injury by eliciting various defense responses, including the induction of inflammatory cytokines, chemokines and interferons, as well as the activation of cell-death pathways to eliminate pathogen-infected or damaged cells. The receptor-interacting protein kinases RIPK1, RIPK2 and RIPK3 have emerged as key molecules in the regulation of inflammatory signaling and cell-death pathways. The RIP kinase family has seven members that share a homologous serine-threonine kinase domain (Fig. 1). RIPK1, the ‘founding member’ of the family, was initially identified as a death domain (DD)-containing protein that interacts with the DD of the death receptor CD95 (Fas)1. RIPK1 bears an amino-terminal kinase domain, a carboxy-terminal DD and a bridging intermediate domain (ID) that contains a RIP homotypic interaction motif (RHIM). RIPK2 contains a carboxy-terminal caspase-activation-and-recruitment domain (CARD) and an ID that lacks a RHIM. RIPK3 has a carboxy-terminal region that harbors a RHIM but lacks a DD. RIPK4 and RIPK5 contain an ID and carboxy-terminal ankyrin domains, while RIPK6 (LRRK1 (‘leucine-rich-repeat kinase 1’)) and RIPK7 (LRRK2 (‘leucine-rich-repeat kinase 2’)) have leucine-rich repeats, ankyrin repeats, and ROC (‘Ras (GTPase) of complex proteins’) and COR (‘carboxyl terminus of ROC’) domains. Few studies have connected inflammation and immunity with the functions of RIPK4–RIPK7.
One of the most unique activities of RIPK1 and RIPK3 is the regulation of apoptosis and necrosis. Necrosis, a form of cell death that results in the compromise of cell-membrane integrity, has long been recognized as a major trigger of inflammation, because as cells die, their cellular contents ‘spill out’ and activate the host immune response. Despite the obvious importance of necrosis to health and disease, the study of necrosis came to focus only in recent years as research on the biochemical pathways of apoptosis, a form of cell death that retains cellular membrane integrity, started to reveal the ‘cross-talk’ between apoptotic cell death and necrotic cell death. Not only do some of the components of apoptosis signaling pathway also participate in necrosis and vice versa, but also some antagonize necrosis. Researchers thus started to realize that regulated cell death, defined here as cell death executed by intrinsic biochemical programs, can manifest via either apoptosis or necrosis, or even a mixture of both, with the outcome depending on the cell-death ‘executioner’ proteins involved. In this Review, we discuss the latest progress in this fast-moving field, with a focus on the signaling pathways and the physiological and pathological implications of RIP kinase–mediated necrosis in the context of the diverse functions of RIP kinases in regulating inflammatory and immune responses.
The apoptosis signaling pathway
Apoptosis was initially defined as a form of programmed cell death with distinct morphological features; dying cells display cell shrinkage, membrane blebbing, chromatin condensation and DNA fragmentation, which are ultimately followed by the formation of membrane-bound apoptotic bodies that are engulfed and digested by phagocytic cells2. A series of genetic and biochemical studies have identified a subfamily of cysteine proteases, known as ‘caspases’, as the executioners of apoptosis. It is now understood that apoptosis can be induced by a variety of signals via activating cell-surface death receptors in the extrinsic pathway or mitochondrial effectors in the intrinsic pathway.
The extrinsic apoptotic pathway is initiated by the binding of death ligands of the TNF (‘tumor-necrosis factor’) superfamily of cytokines, including TNF, FasL, and TRAIL (‘TNF-related apoptosis-inducing ligand’), to their respective death receptors: TNFR1, Fas, and DR4 (‘death receptor 4’) or DR53,4. For example, the binding of TNF to TNFR1 triggers the formation of a TNFR1-containing membrane-signaling complex (complex I) that consists of the adaptors TRADD (‘TNFR1-associated DD protein’), TRAF2 (‘TNFR-associated factor 2’) and cIAP (‘cellular inhibitor of apoptosis protein’), plus RIPK15. This complex promotes activation of the NF-κB (‘nuclear factor-κB’) signaling pathway, which negatively regulates apoptosis by inducing the transcription of genes encoding anti-apoptotic proteins, including members of the IAP (‘inhibitor of apoptosis protein’) family6, the pro-survival factor and caspase-8 homolog c-FLIP (‘cellular Fas-associated DD–like interleukin 1β–converting enzyme–inhibitory protein’)7 and members of the anti-apoptotic Bcl-2 family8,9. To initiate TNFR1-mediated apoptosis, the membrane-associated complex I must convert into a cytosolic signaling complex (complex II) within which the caspase-8 proenzyme, in association with FADD (‘Fas-associated DD’) and RIPK1, undergoes auto-activation to initiate extrinsic apoptosis5,10.
In mammals, intrinsic apoptosis is triggered by the mitochondrial release, into the cytoplasm, of several apoptogenic proteins, including cytochrome c, that normally reside in the intermembrane space of mitochondria. The released cytochrome c interacts with the central mediator of mitochondria-dependent apoptosis, Apaf-1 (‘apoptotic protease activating factor-1’), and pro-caspase-9 to form a protein complex (called an ‘apoptosome’), which leads to the activation of caspase-911. Another mitochondrial apoptogenic protein is Smac (Diablo), which interacts with IAPs to relieve their inhibition of caspases and induce auto-degradation of cIAP1 and cIAP2 by activating their own E3 ubiquitin-ligase activity12. In addition to regulating IAPs, c-FLIP regulates the extrinsic pathway by forming a catalytic inactive dimer with caspase-8, and members of the the Bcl-2 family regulate the intrinsic pathway by controlling the permeability of mitochondria to these apoptogenic proteins.
After stimulation with TNF, RIPK1 and either cIAP1 or cIAP2 are recruited to the TNFR receptor complex. Interestingly, a Smac mimetic (a small molecule that mimics Smac), sensitizes a variety of cancer cells to TNF-induced apoptosis by triggering the RIPK1-dependent activation of caspase-810,13,14. The binding of a Smac mimetic to cIAP1 or cIAP2 triggers its degradation, which results in the disassociation of RIPK1 from the receptor complex to form a cytosolic RIPK1–FADD–caspase-8 complex; this complex leads to the activation of caspase-8 and apoptosis10,12,13 (Fig. 2). In some cells, Smac mimetic compounds also trigger autocrine production of TNF by activating non-canonical NF-κB signaling, which in turn induces apoptosis mediated by the RIPK1–FADD–caspase-8 complex13,15,16. The kinase activity of RIPK1 is essential for Smac mimetic–induced TNF-mediated apoptosis10. In contrast, RIPK1 is not required for apoptosis induced by co-treatment with TNF and cycloheximide, a protein-synthesis inhibitor that blocks the synthesis of c-FLIP17. Therefore, RIPK1 promotes either cell survival or apoptosis after binding TNF, depending on the activities of cIAPs.
Necroptosis signaling pathway
Necroptosis, a form of regulated necrosis, is tightly regulated by the RIPK1 and the closely related kinase RIPK3 (RIP3). The earliest hint of necroptosis was recorded in 1988, when TNF was found to be able to trigger both apoptotic cell death and necrotic cell death in certain cell lines; it was later found that these cells could switch from apoptosis to necrosis when caspases were inhibited, either by viral proteins or by chemical inhibitors such as the pan-caspase inhibitor z-VAD-FMK18. It was thus clear that cross-talk occurs between the extrinsic apoptosis pathway and the necroptosis pathway. Furthermore, death ligands such as FasL and TRAIL have been confirmed to induce necrosis independently of caspase activation3.
In a seminal paper in 2000, RIPK1 was identified as an essential molecule required for necrosis (before the name ‘necroptosis’ was proposed) induced by death-receptor ligands19. Because of the requirement for specific molecules, including RIPK1, this form of death-receptor-induced necrosis has been referred to as ‘programmed necrosis’20. The importance of the kinase activity of RIPK1 was further demonstrated by the ability of necrostatin-1, an inhibitor of the kinase activity of RIPK1, to specifically block necrosis mediated by death receptors21,22. Given the programmed nature of this form of necrosis, and that this signaling pathway shares several common components with the extrinsic apoptotic pathway, such RIPK1-dependent necrosis has been called ‘necroptosis’21. It has become clear that RIPK1 is a truly pleiotropic protein capable of initiating several downstream signaling pathways, either by functioning as an adaptor through its DD domain for activating NF-κB in a kinase-independent manner, or through its RHIM domain, which can interact with other RHIM-containing proteins, including RIPK3, the adaptor TRIF (‘Toll–interleukin 1 receptor domain–containing adaptor inducing IFN-β’) and the cytosolic receptor DAI (‘DNA-dependent activator of interferon-regulatory factor’; also known as ZBP1 or DLM1)23,24,25. In addition, RIPK1 can function as a true kinase that triggers both apoptosis and necroptosis. These diverse functions of RIPK1 are controlled by different cellular contents, as described below.
Activation of RIPK3 is a key step in the initiation of necroptosis26,27,28. RIPK3 can interact with RIPK1 through their respective RHIM domains to form a protein complex (necrosome). This process activates RIPK3, which leads to the phosphorylation of RIPK3 at Ser227 (human RIPK3) or at Thr231 and Ser232 (mouse RIPK3). The RHIM-dependent interaction between RIPK3 and RIPK1 during necroptosis results in the formation of a large amyloid-like signaling complex that might serve as a unique RIPK3-activation platform for necroptosis29. RIPK1-mediated activation of RIPK3 also occurs in necroptosis initiated by type I or type II interferons30,31. In addition to being activated by RIPK1, RIPK3 can be activated by other cellular RHIM-containing proteins, including TRIF32 and DAI33. TRIF is an adaptor downstream of TLR3 and TLR4, which are pathogen-recognition receptors of the TLR (‘Toll-like receptor’) family that initiate activation of NF-κB and transcription factor IRF3–mediated production of type I interferons through the recognition of pathogen-associated microbial patterns34. TLR3 recognizes viral double-stranded RNA (dsRNA) or the synthetic dsRNA poly(I:C), while TLR4 is activated by the lipopolysaccharide (LPS) of Gram-negative bacteria. Activation of TLR3 or TLR4 causes RIPK3-mediated necroptosis in the presence of the pan-caspase chemical inhibitor z-VAD-FMK32. This process is dependent on TRIF, which is recruited to RIPK3 through their RHIM domains after activation of TLR3 and/or TLR432. Inhibition of the kinase activity of RIPK1 abolishes TLR4-induced necroptosis in macrophages32,35. However, RIPK1 is dispensable for TLR3-induced necroptosis in mouse fibroblasts, which suggests that RIPK3 can also be activated by the binding of TRIF, in addition to RIPK135. DAI was identified as a sensor of dsDNA involved in the activation of innate immune responses36. RIPK3 thus is a signal-integration molecule for necroptosis that can receive a necroptosis-inducing signal from various pathways via a homotypic interaction (Fig. 2). Ironically, RIPK3 was first identified as an apoptosis-inducing protein37,38, and indeed, when RIPK3 is over-expressed in various cell lines, robust apoptosis can be detected39. Interestingly, subsequent studies have found that several inhibitors of RIPK3’s kinase activity and the D161N kinase-inactive mutant of RIPK3 unexpectedly cause apoptosis, both in cell lines and in animals40,41. It is thus obvious that RIPK3, like RIPK1, can also function as an adaptor when its RHIM domain is exposed, presumably caused by specific mutations or the binding of small molecules to its kinase site, or simply by overexpression, which leads to the recruitment of RIPK1 with the attendant binding and activation of FADD–caspase-8 and causes apoptosis. Unlike RIPK1-mediated apoptosis induced by TNFR1, this form of RIPK3- and RIPK1-mediated apoptosis is not sensitive to inhibitors of RIPK1’s kinase activity41. These two pathways of apoptosis, in addition to the RIPK3-mediated necroptosis that critically requires its kinase activity, can happen simultaneously in fibroblasts and lung epithelial cells after infection with influenza virus42,43.
The pseudokinase MLKL (‘mixed lineage kinase domain-like protein’) has been identified as a downstream target of RIPK3, and its phosphorylation by RIPK3 is known to be a key step in the execution of necroptosis44,45. MLKL is a pseudokinase with a kinase-like domain located in the carboxy-terminal region that is normally folded back at its amino-terminal coiled-coil region to keep it in an inactive state46. Activated RIPK3 binds to and phosphorylates MLKL in its pseudokinase domain, at Thr357 and Ser358 (human MLKL) or Ser345 and Ser347 (mouse MLKL)44,47,48. The RIPK3-mediated phosphorylation of MLKL triggers a conformational change and thus allows exposure of the amino-terminal four-helix bundle of MLKL47. This event leads to the oligomerization of MLKL49,50,51. The oligomerized MLKL is able to bind to negatively charged phospholipids, including cardiolipin and the highly phosphorylated inositol phosphate molecule IP6, which displaces the auto-inhibitory region of MLKL and thus allows binding to the membrane and insertion of its amino terminus into various cellular organelles and plasma membranes49,52,53. The amino-terminal four-helix bundle of MLKL is sufficient to induce necroptosis when MLKL is ectopically expressed in cells54,51.
The precise mechanism that leads to necroptotic membrane disruption is yet to be defined. One hypothesis for the execution of necroptosis is that MLKL directly disrupts membrane integrity by forming cell-content-permeable pores49. However, that model is supported only by in vitro liposome studies. Moreover, strong RIPK3-specific phosphorylated-MLKL signals can be detected in cultured live cells and in tissues, which suggests that phosphorylation of MLKL and its translocation to membranes do not cause spontaneous cell death, as would occur with membrane-permeable pores49. An alternative hypothesis is that MLKL activates calcium, sodium or magnesium ion channels to mediate the influx of ions, which could cause cell death through osmotic pressure50,51,55. However, that hypothesis is inconsistent with the observation that necroptosis still happens when cells are cultured in medium lacking these ions. Another perplexing observation is that activated MLKL seems to target intracellular membranes as well49. If MLKL simply causes necroptotic cell death, disrupting the plasma membrane would seem to suffice. Thus, manipulating MLKL to specifically target different membrane compartments in cultured cells and in animals should be an interesting research avenue through which to further delineate the regulatory mechanisms and biological importance of necroptosis.
Regulation of necroptosis
Like apoptosis, a process that can be regulated at multiple points of the pathway, necroptosis is also known to be highly regulated. A key regulatory point of necroptosis is the activity of caspase-8. With the exception of the mouse sarcoma cell line L929, in which spontaneous necroptosis was first observed in the presence of TNF, robust necroptosis has been seen only when caspase-8 was either inhibited or genetically deleted. It is now apparent that both RIPK1 and RIPK3 are substrates for caspase-8, and the close proximity of these proteins in receptor-induced signaling complexes favors the cleavage of RIP kinases by caspase-8 and thus re-directs the cell fate toward apoptosis20,56. The strongest evidence for caspase-8-mediated restriction of necroptosis was provided by the finding that the embryonically lethal phenotype observed in caspase-8-deficient mice was completely ‘rescued’ by ablation of RIPK357,58. Notably, adult Casp8−/−Ripk3−/− mice are fertile and seem immunocompetent, and they display progressive lymphadenopathy and splenomegaly marked by the accumulation of lymphocytes57,58. Consistent with that, RIPK3 deficiency also ‘rescues’ the embryonic lethality caused by deletion of FADD59. c-FLIP is a caspase-8 homolog that can form heterodimers with pro-caspase-8 to prevent its activation60. Consistent with that, the embryonic death of c-FLIP-deficient mice is also ‘rescued’ by ablation of both FADD and RIPK3. Such mouse gene-knockout studies support the idea that caspase-8 is a gatekeeper for RIPK3-mediated necroptosis during embryonic development59. Notably, mice that lack RIPK3 or MLKL exhibit normal development and are immunocompetent and fertile47,61,62.
Another key regulatory step is mediated by cIAPs, which are subunits of the TNFR-activated intracellular signaling complexes. cIAPs, through their ubiquitin-ligase activity, form Lys63 (K63)-linked polyubiqutin chains and/or a linear ubiquitin-chain-assembly complex (LUBAC) on RIPK1 to signal activation of NF-κB63. Such modification precludes the binding of RIPK3 or FADD to RIPK1 and thereby blocks the induction of necroptosis and apoptosis. Smac, by binding to cIAPs, induces the self-destructing K43-linked ubiquitin chain on cIAPs, which leads to their degradation12. The RIPK1-containing complex then disassociates from the plasma membrane, and the ubiquitin chains on RIPK1 are subsequent removed by the de-ubiquitin enzymes cylindromatosis (CYLD) or A20 (TNFAIP3)64,65,66. Subsequently, the cytoplasmic RIPK1-containing complex (also called ‘complex II’5,10) is able to recruit FADD–caspase-8 through its DD and to recruit RIPK3 through its RHIM domain, both in a RIPK1-dependent manner10,26,67. These two events do not seem to be mutually exclusive. The complex usually signals apoptosis through caspase-8 activity; however, when caspase-8 is insufficient to cleave and inactive RIPK3, the complex causes necroptosis. Since Smac is a mitochondrial intermembrane-space protein that is released into the cytosol by the pro-apoptotic Bcl-2 family of proteins, it is conceivable that apoptosis and necroptosis engage in cross-talk again at the level of mitochondria.
As phosphorylation of RIPK3 is critical for its recruitment of MLKL, de-phosphorylation of RIPK3 is predicted to repress necroptosis. Indeed, the phosphatase Ppm1b was found to diminish the phosphorylation of RIPK3 and negatively regulate RIPK3-dependent necroptosis68. Moreover, Ppm1b deficiency sensitizes mice to TNF-induced cecal damage that is prevented by deletion of RIPK3. In addition, studies have shown that formation of the RIPK1–RIPK3 necrosome can be affected by the ubiquitination status of constituent proteins. One report has shown that the ubiquitin-editing enzyme A20 inhibits the ubiquitination of RIPK3 at Lys5 to prevent formation of the necrosome69. A20-deficient cells are thus susceptible to necroptosis, and deletion of RIPK3 prolongs the survival of A20-deficient mice. Moreover, the E3 ubiquitin ligase PELI1 has been shown to cause K63-linked ubiquitination of RIPK1 at Lys115, which promotes formation of the RIPK1–RIPK3 complex70. Deficiency in PELI1 blocks necroptosis and sensitizes cells to apoptosis. Although such studies suggest that ubiquitination is involved in the regulation of RIPK1 and RIPK3 in necroptosis, further investigation will be needed to delineate the precise mechanism(s) underlying these regulatory events.
Interestingly, before membrane disruption, activation of MLKL causes the formation of plasma-membrane bubbles characterized by exposed phosphatidylserine, a phagocytosis-signaling molecule known to be associated with apoptotic cell death71,72,73. ESCRT-III components are required for the formation of plasma-membrane bubbles and subsequent shedding of exosomes containing MLKL71,72. Suppression of ESCRT components sensitizes cells to necroptosis71,72. Such results support the proposal of a protective function for the ESCRT machinery directed against MLKL-mediated membrane damage via active removal of the damaged sites through ESCRT-dependent exocytosis. The observation of necroptosis-induced exosome formation also raises the interesting possibility that cells undergoing necroptosis actively signal to the environment through these exosomes, including the induction of cytokine secretion by surrounding macrophages72.
Although RIPK1 is required for necroptosis induced by the TNFR family, the evidence obtained from studies of RIPK1-deficient mice as well as mice expressing kinase-inactive RHIM or mutant RHIM (via knock-in mutation) also suggest a protective role for RIPK1 in the activation of RIPK3 as well74,75,76,77,78,79,80. In addition to NF-κB signaling mediated by RIPK1 that suppresses apoptosis and necroptosis by the transcriptional activation of genes encoding c-FLIP and cIAPs, RIPK1 also protects cells from RIPK3-mediated necroptosis by sequestering other RHIM-containing proteins, such as DAI from RIPK379,80.
Necroptosis and inflammation
The possibility of a role for necroptosis in triggering inflammation is supported by in vivo studies of mice that lack caspase-8 or FADD. Mice with intestinal epithelial cell–specific deficiency in FADD or caspase-8 show epithelial-cell necrosis and develop spontaneous intestinal inflammation with loss of Paneth cells81,82. RIPK3 deficiency provides protection against the necrosis of intestinal epithelial cells and prevents the intestinal inflammation induced by intestinal epithelial cell–specific deletion of FADD or inducible deletion of caspase-8 in adult mice81,83. Likewise, ablation of RIPK3 prevents necrosis of keratinocytes and skin inflammation in mice lacking FADD or caspase-8 in keratinocytes83,84. Further support for the idea that necroptosis drives inflammation has been provided by studies of RIPK1-deficient mice. Such mice die around birth and exhibit systemic inflammation, which is diminished in Ripk1−/−Ripk3−/− and Ripk1−/−Mlkl−/− mice but is not affected in Ripk1−/−Casp8−/−mice74. The keratinocyte hyperplasia in Ripk1−/− mice is prevented by deletion of either Ripk3 or Mlkl but not by deletion of Casp874. Consistent with that, mice with epidermis-specific RIPK1 deficiency develop keratinocyte death and severe skin inflammation, each of which is prevented by the loss of RIPK3 or MLKL but not by the loss of FADD85. Collectively, these studies suggest that the necroptosis of epithelial cells is a critical pathological mechanism for triggering epithelial barrier disruption and that necroptosis contributes to the pathogenesis of intestinal and skin inflammation. In these cases, RIPK1 functions mainly by holding necroptosis in check, rather than via the activation of necroptosis.
Pediatric patients with inflammatory bowel disease show increased levels of RIPK3 and MLKL and a reduced level of caspase-8 in their inflamed tissues, which provides correlative evidence of a link between necroptosis and inflammatory bowel disease86.
Emerging evidence suggests that in addition to its involvement in intestinal and skin inflammation, necroptosis seems to function in a wide variety of experimentally inducible inflammatory conditions that involve tissue injury. RIPK3 deficiency, MLKL deficiency or inactivation of RIPK1’s kinase activity ameliorates inflammation and tissue injury in various inflammatory disorders, including TNF-induced systemic inflammatory response syndrome78,87, cecal-ligation-and-puncture–induced sepsis87,88, systemic inflammatory responses in A20-deficient mice69,89, dermatitis in mice deficient in the LUBAC component SHARPIN78, atherosclerosis90,91, acute pancreatitis26,62, colitis92, chronic obstructive pulmonary disease93,94, rheumatoid arthritis95, liver injury96,97 and allograft rejection98,99, as well as in brain injury induced by ischemia–reperfusion21, kidney injury98,100 and cardiac injury101,102. In studies benefitting from the development of specific antibodies that can specifically recognize phosphorylated RIPK3 and phosphorylated MLKL, activation of necroptosis has been detected in damaged tissues, including mouse atherosclerotic plaques91, the liver of patients with drug-induced liver injury30, and renal biopsies from patients after renal transplantation71. In conclusion, such studies provide strong evidence suggesting that necroptosis triggers inflammation and participates in the pathogenesis of many inflammatory diseases. We note that discrepant results have been reported for the contribution of RIPK3 to cecal-ligation-and-puncture–induced septic shock, acute pancreatitis, major cerebral-artery-occlusion stroke and liver injury62,89,103. Hence, further functional validation will be needed to clarify the exact role of necroptosis in such conditions.
Given the essential role of necroptosis in driving inflammation, it is clearly important to understand the underlying mechanisms. DAMPs are molecules released or exposed by dying cells that act as endogenous danger signals to elicit inflammatory responses. Some of the most-studied DAMPs include HMGB1 (‘high mobility group box-1’), the cytokines IL-1α and IL-33 (of the IL-1 family), the S100 calcium-binding proteins S100A8 and S100A9, heat-shock proteins, uric acid, ATP and mitochondrial factors. DAMPs are generally recognized by pattern-recognition receptors, such as TLR and IL-1 receptors, and this recognition induces the production of cytokines and chemokines. Release of HMGB1 into the culture medium, from various cells undergoing necroptosis, has been detected26,104,105. RIPK3 deficiency results in suppression of the release of HMGB1 in a model of retinal degeneration induced by dsRNA106. In TNF-induced systemic inflammatory response syndrome, the abundance of DAMPs (including circulating mitochondrial DNA and IL-1α) was elevated in the plasma after challenge with TNF, while levels were attenuated in RIPK3-deficient mice87. Moreover, loss of RIPK3 diminishes the abundance of cytokines, including IL-1α, and causes inflammation in atherosclerotic plaques91. Collectively, these studies demonstrate that necroptosis-induced inflammation causes the release of DAMPs. Thus, it is conceivable that necroptosis, by inducing the release of DAMPs, is a more pro-inflammatory form of cell death than is apoptosis. In agreement with that, loss of RIPK3 or MLKL prevents the release of IL-33 and IL-1α into the plasma and causes systemic inflammation in RIPK1-deficient mice, but the loss of caspase-8 does not74. Consistent with that, keratinocyte necroptosis has been shown to drive skin inflammation in mice with epidermis-specific RIPK1 deficiency, but keratinocyte apoptosis has not, although both keratinocyte apoptosis and necroptosis occur in these mice85. Comparison of apoptosis and necroptosis in the context of immune responses has been undertaken in ligand-free systems that exclusively activate apoptosis or necroptosis. Fibroblasts or cancer cells undergoing necroptosis induced by RIPK3 release high concentrations of HMGB1 and ATP, whereas these DAMPs are not secreted by cells undergoing apoptosis induced by caspase-8104,107. Notably, injection of necroptotic cells into mice induces more recruitment of immune cells than of apoptotic cells. Such studies support the proposal that DAMPs released by necroptosis contribute to the induction of inflammation. Although HMGB1 seems to be the most common DAMP released from necroptotic cells, its abundance in the plasma of RIPK1-deficient mice is not different from that in the plasma of wild-type mice74. We are thus tempted to speculate that diverse DAMPs, including these newly discovered necroptosis-induced exosomes71,72,73, might be selectively released from necroptotic cells to trigger or amplify distinct inflammatory signals.
In addition to regulating innate immunity, necroptosis is involved in adaptive immunity. Cells undergoing RIPK3-mediated necroptosis induce the activation of dendritic cells (DCs) in vitro and cause efficient cross-priming of CD8+ T cells in vivo, but those undergoing caspase-8-mediated apoptosis do not104,107. Interestingly, necroptosis induced by dimerization of a variant form RIPK3 lacking its RHIM does not activate cross-priming, but it does induce the release of both HMGB1 and ATP by necroptosis104. This suggests that DAMPs such as HMGB1 and ATP that are released after necroptosis are insufficient to activate adaptive immunity, despite the ability of these DAMPs to promote innate immunity104. Instead, RIPK1-mediated NF-κB signaling is needed to achieve efficient cross-priming104. However, there is also evidence showing that the status of NF-κB’s activation does not correlate with the capacity of necroptotic cancer cells to activate DC cross-priming after vaccination with these necroptotic cancer cells107. In addition, poly(I:C)-induced necroptosis in cervical carcinoma cells induces the release of HMGB1 and IL-1α105. IL-1α promotes the activation of DCs and production of IL-12, but HMGB1 does not105. Although such studies do demonstrate that necroptotic cells can be inducers of DC activation, additional studies will be needed to elucidate the underlying mechanism(s) and to assess the functional effect of necroptosis on these adaptive immune responses.
Necroptosis-independent activities of RIP kinases in inflammation
In addition to its roles in regulating cell death, RIPK1 is a critical activator of NF-κB signaling downstream of death receptors, TLRs and the cytosolic receptor RIG-I (‘retinoic acid–inducible gene I’). K63-linked polyubiquitination of RIPK1 at Lys377 is required for the activation of NF-κB, but its kinase activity is not108. Such polyubiquitination of RIPK1 promotes its recruitment to a complex of the kinase TAK1 and the TAK1-binding proteins TAB2 and TAB3, and to the IKK (‘inhibitor of κB kinase’) complex, to activate NF-κB108,109. Additionally, RIPK1 mediates DNA damage–induced activation of NF-κB110,111. After genotoxic stress, RIPK1 is recruited to the apoptotic effector molecule PIDD (‘p53-induced protein with a DD’), followed by formation of the PIDD-RIPK1–sumoylated NF-κB modulator Nemo complex in the nucleus, which ultimately leads to the activation of NF-κB111. In contrast, RIPK3-deficient cells show normal NF-κB signaling in response to activation of TNFR1 and TLR2 or TLR461. Interestingly, RIPK1 and RIPK3 also have important roles in regulating the production of pro-inflammatory cytokine in immune cells95,112,113,114,115,116,117,118,119,120,121. For example, the NLRP3 (NALP3) inflammasome is a caspase-1-activating complex that consists of NLRP3 (‘Nod-like receptor family pyrin domain–containing 3’), caspase-1 and the adaptor ASC (‘apoptosis-associated speck-like protein’)122. Activated caspase-1 further cleaves pro-IL-1β and pro-IL-8 into mature IL-1β and IL-8, respectively. In addition to caspase-1, caspase-8 has been suggested to cleave these inflammatory cytokines123. Genetic deletion of IAPs or treatment with Smac mimetics promotes the LPS-induced cleavage of IL-1β in bone marrow–derived macrophages and DCs117. This effect is dependent on RIPK3, which is thought to induce not only maturation of IL-1β mediated by the NLRP3–caspase-1 inflammasome but also caspase-8-dependent processing of IL-1β. In the presence of IAPs, caspase-8 seems dispensable for RIPK3-mediated activation of IL-1β, as caspase-8 deficiency in DCs still allows RIPK3-mediated activation of the NLRP3 inflammasome and production of IL-1β in response to LPS118. Furthermore, treatment of DCs with LPS alone induces minimal secretion of IL-1β and activation of NF-κB, which are suppressed by deletion of RIPK3112,113. In macrophages, RIPK1 and RIPK3 promote the production of inflammatory cytokines induced by LPS and z-VAD-FMK, and this effect is attenuated by inhibition of signaling via the kinase Erk and the transcription factor c-Fos or via NF-κB114. In addition, there is evidence showing that RIPK1 and RIPK3 activate the NLRP3 inflammasome in macrophages in response to RNA viruses113,121. The precise mechanism by which NLRP3 is activated by RIPK1 and RIPK3 is not known, although induced production of reactive oxygen species has been proposed to have a role in this113,114. Such activity clearly does not occur via necroptosis, since the kinase activities of RIPK1 and RIPK3 as well as MLKL are dispensable for LPS-induced activation of the NLRP3 inflammasome, maturation of IL-1β and production of inflammatory cytokines95,112,113,115,117.
RIPK2 is another RIP kinase involved in the activation of NF-κB, mitogen-activated protein kinases (MAPKs) and apoptosis124,125,126. The kinase activity of RIPK2 is dispensable for signaling via NF-κB and the MAPK JNK (‘Jun N-terminal kinase’) but is required for activation of the MAPK ERK2126,127. The best-characterized function of RIPK2 is its mediation of signal transduction from the NOD (‘nucleotide-binding oligomerization domain’) proteins NOD1 and NOD2, which are cytosolic pathogen-recognition receptors that activate pro-inflammatory and antimicrobial responses in response to bacterial peptidoglycans in macrophages128. NOD1 and NOD2 are homologous proteins composed of CARDs, nucleotide-binding domains and leucine-rich repeats129,130. After recognition of its ligands, NOD1 or NOD2 recruits RIPK2 via CARD–CARD homotypic interactions130,131. This process promotes the ubiquitination of RIPK2 and activation of the TAK1 and IKK complexes, which leads to the activation of NF-κB and MAPKs, as well as the production of pro-inflammatory cytokines by macrophages. RIPK2 also recruits various ubiquitin E3 ligases to the NOD2 complex, including XIAP (‘X-chromosome-linked inhibitor of apoptosis’), cIAP1 and cIAP2, PELL3 and LUBAC132. It has been demonstrated that XIAP deficiency causes impaired NOD1- or NOD2-mediated ubiquitination of RIPK2 and inflammatory signaling133,134,135. XIAP is recruited to the NOD2–RIPK2 complex via binding to the kinase domain of RIPK2, which leads to K63-linked ubiquitination of RIPK2 and recruitment of LUBAC; this results in efficient activation of NF-κB and MAPKs, as well as cytokine production in macrophages134,135. The ubiquitination sites of RIPK2 (Lys209, Lys410 and Lys538) are essential for its function in mediating signaling via NOD1 and NOD2134,136. The D146N kinase-inactive mutant of RIPK2 retains the ability to bind XIAP and to activate NOD2 signaling, which suggests that the kinase activity of RIPK2 is not required for signaling via NOD1 and NOD2134. Furthermore, there is evidence showing that RIPK2 deficiency results in impaired T cell proliferation and differentiation into the TH1 subset of helper T cells, as well as reduced production of interferon-γ in both TH1 cells and natural killer cells137,138. Although such studies suggest that RIPK2 is an important mediator of both innate immune responses and adaptive immune responses, further studies will be needed to elucidate the molecular mechanisms and functions of its action in adaptive immunity.
Necroptosis and aging
A surprising recent finding is that mice without the key necroptosis signaling components RIPK3 and MLKL show a remarkable delay in the aging of the male reproductive system139. Normal signs of aging in male mice include decreased sperm counts, drops in testosterone levels and enlargement of seminal vesicles when the mouse reaches more than 1 year of age; these signs are not observed in age-matched RIPK3- or MLKL-deficient mice. Moreover, chronic treatment with an inhibitor of the kinase activity of RIPK1 for up to 5 months after wild-type male mice reach 1 year of age also prevents the appearance of these signs of aging.
The mechanism of the delayed aging phenotype noted above is believed to be the blocking of necroptosis in spermatogonium stem cells and in their supporting cells (Sertoli cells) in the testes. Both cell types show necroptosis-specific staining of phosphorylated MLKL when wild-type mice reach more than 1 year of age. The elimination of RIPK3 or MLKL, as well as feeding wild-type mice an inhibitor of RIPK1, prevents the appearance of these necroptosis-specific signals. A single injection of a necroptosis-inducing agent into the testes of young wild-type mice results in accelerated aging of the reproductive system but does not do so in age-matched RIPK3- or MLKL-deficient mice. However, although these genetically or pharmacologically manipulated mice seem to be young in their reproductive behavior, the offspring of these chronologically old mice are not healthy and show a variety of birth defects139. Necroptosis thus seems to be critical in eliminating old sperm (which can still accumulate DNA damage) from the reproductive pool, a function that might be important for the health of the offspring.
Necroptosis and host defense
Cell death is considered to be a part of the host defense against pathogen infection; for example, it facilitates the elimination of virus-infected cells before the production of progeny virions. In response to host defense, pathogens develop various strategies to evade cell-death-based defenses. Indeed, many viruses, including vaccinia virus (VV) and cowpox virus, express caspase inhibitors to subvert apoptosis. The potential link between necroptosis and host defense was first observed in cells infected with VV that expresses B13R (Spi2), a potent inhibitor of caspase-1 and caspase-820,27. Cells infected with VV are highly sensitive to TNF-induced necroptosis20,27. RIPK3-deficient mice show impaired control of VV replication in vivo27. The proposal of an anti-viral function for necroptosis is strongly supported by the discovery that the murine cytomegalovirus (MCMV) protein M45 (also called vIRA (‘viral inhibitor of RIP-activation protein’) acts as a viral inhibitor of RIPK3. M45 is a RHIM-containing protein that suppresses necroptosis140. A recombinant MCMV expressing a vIRA RHIM-domain mutant activates RIPK3-mediated necroptosis through DAI, not through RIPK133. Therefore, vIRA-mediated disruption of the DAI–RIPK3 interaction prevents necroptosis and benefits viral infection and pathogenesis. Another VV protein, E3L, has been suggested to block DAI-mediated activation of RIPK3 via its z-DNA-binding domain141. Mutant VV lacking the z-DNA-binding domain of E3L causes DAI–RIPK3–mediated necroptosis, again independently of RIPK1141 (Fig. 2). Similarly, both ICP6 and ICP10 of human herpesvirus have a RHIM-competitor function that counteracts TNF-induced necroptosis by inhibiting the activation of RIPK3 in their natural host cells142,143. More interestingly, the human CMV protein IE1 inhibits necroptosis by acting downstream of the phosphorylation of MLKL, a mechanism certainly worthy of further exploration144. In addition to being a DNA sensor, DAI has been shown to recognize the genomic RNA of influenza A viruses and thus to elicit RIPK3-mediated necroptosis and apoptosis145. Moreover, a contribution of necroptosis to bacterial clearance has been observed during infection with Staphylococcus aureus146,147. However, activation of type I interferon–induced macrophage necroptosis actually promotes bacterial survival during infection with Salmonella enterica serovar Typhimurium30. Given the fact that necroptosis is a strong trigger of inflammation, it is conceivable that necroptosis can act as a defense mechanism directed against pathogens by mounting inflammatory responses, in addition to its direct cell-death function, which could either eliminate pathogens or release pathogens. Interestingly, anti-viral activities of RIPK1 and RIPK3 beyond cell death have been demonstrated in infection with West Nile virus, during which they promote neuro-inflammation148.
Despite enormous progress in the field in recent years, key questions remain about the regulation and execution of necroptosis, as well as about the complex cross-talk among apoptosis, necrosis and the non–cell-death roles of RIP kinases in regulating inflammation. In particular, how MLKL destroys the plasma membrane and what, if any, role MLKL has in targeting the intracellular membrane during necroptosis need to be investigated. A related question is how phosphorylated MLKL, especially under non-lethal conditions, is subjected to additional regulation. For example, how is ESCRT-mediated exocytosis regulated and executed? What is (are) the function(s) of this process, in addition to attenuation of necroptosis? Another fruitful field of inquiry should be the roles of RIPK1 and RIPK3 in controlling cytokine production via activation of the inflammasome. Such activity connects to another form of regulated necrosis called ‘pyroptosis’, a form of cell death triggered by inflammatory caspases, including caspase-1, caspase-4 and caspase-5, and their substrate pyroptosis ‘executioner’, GSDMD (‘gasdermin D’)149 (Fig. 3).
Finally, RIP kinases are recognized as a potential therapeutic targets for many diseases, given their broad linkage to the pathogenesis of inflammatory diseases, degenerative diseases (Box 1) and cancer (Box 2), and even the normal aging process. Greater understanding of these biochemical functions of RIP kinases in relevant disease models will certainly be instrumental for the development of valuable life-saving therapeutic strategies.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supported by the National Basic Research (973) Program of China (2013CB910102), the National Natural Science Foundation of China (31471303 and 31671436), a Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions and Fok Ying Tung Education Foundation for Young Teachers (151020) and the Non-profit Central Research Institute Fund of Chinese Academy of Medical Sciences (2017NL31002,2017NL31004).