True Grit: Programmed Necrosis in Antiviral Host Defense, Inflammation, and Immunogenicity

Regulated cell death is a potent arm of host defense (1–4), involving alternate strategies that evolved with animals to counteract pathogen-encoded cell death suppressors (3, 5). Intrinsic (mitochondrial) apoptosis is necessary for development (6), whereas extrinsic apoptosis and programmed necrosis play out as alternate innate immune countermeasures to control infection (3, 5, 7). Although mechanistically distinct from Casp8-mediated extrinsic apoptosis, receptor interacting protein kinase (RIP)3 necrosis similarly eliminates infected cells prior to release of viral progeny, halting infection and triggering an inflammatory response (7). Importantly, extrinsic apoptosis and necrotic cell death machinery is distributed in all somatic cells. These pathways reduce the burden of infection while also producing cell debris to promote Ag cross-presentation by dendritic cells (DCs), thereby supporting a robust adaptive immune response that ultimately controls infection. The study of virus-encoded cell death suppressor mutants brought RIP3 necrosis to light, revealing interdependencies fostered by a pathogen–host arms race centered on cell death measures and countermeasures (5). Based on the variety of strategies that have been observed, cell death suppressors are crucial to the pathogenesis of all large DNA viruses (2, 3, 5, 8, 9). Because cell death is triggered by pre-existing cellular machinery, dysregulation can inadvertently kill cells and cause inflammatory disease, even in the absence of infection (7). The distribution of these pathways in all somatic cells opens possible routes to improve host resistance to natural pathogens, as well as to prevent infection of novel biothreat agents. This review provides a perspective on recent advances in RIP3 necrosis. The intention is to highlight triggers and alternate pathways of extrinsic cell death where therapeutic intervention might improve innate resistance to infection or drive better cross-presentation during vaccination, without risking increased inflammatory disease (10). The derivation of viable, fertile, and immunocompetent mice with combined deficiency in Casp8 and RIP3 (11) dismisses any key role for Casp8-regulated pathways in development, but it certainly raises important questions as to how apoptosis, as well as necrosis, contributes to the function of the immune system.

To set the stage for discussing the current understanding of RIP3 kinase in host defense, it is important to consider the crucial role that TNF-mediated signal transduction has played in the elaboration of alternate cytoprotective and cytotoxic pathways (1, 12). Three distinct outcomes of signal transduction via the TNF death receptor, TNFR1, are recognized: cytokine activation, extrinsic apoptosis, and programmed necrosis. These converge on death domain (DD) signaling that is orchestrated via the adaptor FADD in complex with Casp8 and specific inhibitor, FLIP (12, 13) (Fig. 1). Pathogen sensors, IFNs, TCRs, and genotoxic stress all trigger analogous outcomes. Insights from TNFR DD signaling and identification of virus-encoded cell death suppressors using TNF-based assays (14–16) has brought an appreciation of core cell death machinery operating as an integrated pathogen sensor system (5). In line with the view that extrinsic death came into existence to support host defense, both TNF antagonist immunotherapy (17) and genetic linkage studies (18) show that TNF signaling contributes as a redundant factor in host defense, like many other innate immune mechanisms. A goal of this review is to highlight the growing evidence that TNF opened the awareness to a broadly distributed innate cell death system able to prevent infection.

TNFR1, as well as the Fas/CD95 and TRAIL death receptors, control NF-κB activation, extrinsic apoptosis, and programmed necrosis by DD signal transduction, functioning in collaboration with death effector domain (DED) interactions (6) mediated via a critical complex consisting of Casp8, FLIP, FADD, and RIP1 (denoted “Complex IIB” downstream of TNFR1 signal transduction) (19) and known as the “Necrosome” or “Ripoptosome” complex (3, 12, 13, 20, 21) (Fig. 1). This cytosolic complex maintains control over alternate death outcomes downstream of TNF family death receptors (22), while also metering RIP1-enhanced induction of NF-κB (19) and RIP1 kinase–dependent programmed necrosis, also called necroptosis (3, 5, 12, 22, 23). RIP1 kinase–dependent necroptosis is blocked by small molecule drugs, the necrostatins (24). Cell death triggered by death receptors, pathogen sensors (25–28), IFNs (29–31), Ag-specific TCR engagement (32, 33), or genotoxic stress (34) is regulated by heterodimeric Casp8-FLIP within this core complex, preventing Casp8 self-activation and extrinsic apoptosis (12, 13), while allowing sufficient basal protease activity to suppress necrosis (3, 12, 13, 20, 21). The ability of this core Casp8 complex to prevent extrinsic apoptosis, as well as necroptosis, first emerged in studies of TNFR1 DD signaling (22, 35, 36). Necroptosis is triggered when Casp8 becomes compromised during death signal transduction. In these settings, RIP1 functions as both a RIP homotypic interaction motif (RHIM)-dependent adaptor and a protein kinase to phosphorylate RIP3 (37–39), a partnership that results in formation of an amyloid-like complex (40). RIP3 kinase undergoes autophosphorylation and subsequently activates a target protein, mixed lineage kinase domain-like (MLKL), by phosphorylating key amino acids (7, 41, 42). The final steps in this pathway involve the formation of an MLKL homotrimer that translocates to the plasma membrane to mediate Ca²⁺ influx via a transient receptor potential melastatin related 7 channel (43). A similar RIP3–MLKL axis (28) is apparently shared by the three pathways leading to RIP3 necrosis, whether RIP1 dependent or RIP1 independent (Fig. 2).

Pathogen recognition receptors trigger NF-κB and IRF3/IRF7 (44–46), activating production of IFN and other cytokines (47, 48). These sensors regulate alternate activation of cytokines or cell death in a manner analogous to TNF family death receptor signaling (3, 49), subject to modulation by virus-encoded cell death suppressors (3, 23). RIP1–RIP3 necroptosis (5, 7, 50) occurs downstream of TLR signaling (27, 28, 51), as well as via retinoic acid–inducible gene 1 or melanoma differentiation-associated protein 5 dsRNA helicase enzymes (52, 53). Similar signaling also lies downstream of seemingly distinct pathogen response categories, including genotoxic stress (34), IFN activation (31), Ag-dependent activation of T cells (32, 33), or infection with viruses, such as vaccinia (37, 54), murine CMV (MCMV) (3, 25, 26), and reovirus (55). In addition to RIP1, two RHIM-containing adaptors are involved in activating RIP3: DNA-dependent activator of IRFs (DAI; also called ZBP1 or DLM1) (25) and TRIF (28), the key TLR3- and TLR4-signaling adaptor (Fig. 2). RIP3 directly engages the pathogen sensor DAI independent of RIP1 when cells or mice are infected with a mutant MCMV strain that lacks the M45-encoded viral inhibitor of RIP activation (vIRA) (25). Furthermore, RIP3 engages TRIF downstream of TLR3 in both RIP1-dependent and RIP1-independent pathways (27, 28, 51). DAI and TRIF engage in RHIM-dependent interactions that converge on RIP3 kinase (5). The necrotic death mediated by DAI-RIP3 (25, 26) and TRIF-RIP3 (28) may proceed independent of RIP1, but nevertheless follow similar parameters (37–39) and converge on MLKL (28, 41, 42, 56, 57) (Fig. 1). Thus, a RIP3 necrosis “trap door” lies downstream of innate immune signaling, and cell death may be triggered either independent of or dependent on death receptors and RIP1 kinase activity (3, 5, 7), through the various pathogen-related signaling events depicted in Fig. 2.

Type I (IFN-α and IFN-β) or type II (IFN-γ) IFNs act on receptors (INFαβR or IFNγR, respectively) to trigger JAK-STAT signal transduction and mediate antiviral and immunomodulatory outcomes. IFNs also induce cell death (58) analogous to death receptors and pathogen sensors (3, 49), but it is carried out by distinct JAK-STAT–signaling cascades. Recent studies implicate both IFN-β (29) and IFN-γ (30) in RIP1-dependent cell death with characteristics of RIP3 necrosis. Activation of cell death by IFN-γ requires JAK-STAT function, as well as RIP1 and RIP3 (31). The impact of either FADD or caspase compromise produces a picture that points to the same core cytosolic Casp8–FADD–FLIP–RIP1 “Ripoptosome” complex (Fig. 1). This complex forms in response to intracellular pathogens as well as death receptor signaling. It forms as a result of pathogen alarm and control mechanisms as diverse as IFNs, genotoxic stress, and Ag activation of lymphocytes. This places the core Casp8–FADD–FLIP complex (27, 32, 34, 37–39, 51, 52) in the role of a mammalian pathogen supersensor (28).

DNA virus–encoded cell death suppressors are crucial to pathogenesis of viral infection and disease progression (1–4). These functions have contributed to dissection of extrinsic death pathways (3, 5, 7). Initially, apoptosis-prone L929 (59) and necrosis-prone L-M variant cell line (60) led to assays (61) that allowed for the identification of adenovirus-encoded cell death suppressors (14). Poxviruses and herpesviruses provided the first examples of DED-containing Casp8 suppressors, so-called “viral FLIPS” [(v)FLIPs] (62, 63), opening the way toward understanding the prosurvival role of NF-κB (64), as well as the consequences of cellular FLIP-Casp8 heterodimerization (12). Cowpox caspase and serine protease inhibitor, CrmA, was crucial in characterizing necrosis as a caspase-independent pathway triggered by TNF under conditions that prevent Casp8-dependent extrinsic apoptosis (65). The concept that RIP3 necrosis may be a host countermeasure against viruses encoding caspase inhibitors (3, 9, 33) has been refined with the demonstration that the highly specific CMV-encoded viral inhibitor of Casp8 activation (vICA) predisposes to RIP3 necrosis (11). Consistent with this understanding, TNF-induced necroptosis makes a striking contribution to host defense against the poxvirus and vaccinia in mice (37, 54) where a virus-encoded inhibitor related to CrmA likely unleashes the pathway.

Wild-type (WT) MCMV is insensitive to RIP3 necrosis; however, MCMV-encoded M45 mutant viruses that are deficient in vIRA induce necrosis within a few hours of invading cells (3, 25, 26) via a DAI–RIP3 complex. Curiously, mathematical models of MCMV vIRA and vICA function (66) have completely missed the mark (11, 20). vIRA acts as a virion protein (67) to block RHIM-dependent signaling (3, 25, 26) upon delivery to cells during initial penetration (67–70). vIRA-deficient virus fails to gain a foothold in the host because infection halts as a result of the elimination of virus-exposed cells prior to the production of progeny virus. Two key issues remain to be fully addressed: whether MCMV-encoded vICA suppression of apoptosis is responsible for unleashing DAI–RIP3 necrosis under natural infection conditions in the host animal (3, 28) and how the core Casp8 complex communicates with RIP3 kinase without the benefit of the adaptor RIP1 (Fig. 1).

Human CMV (HCMV) and MCMV both encode vICA (71) and block Casp8 apoptosis, a viral strategy that is particularly important during infection of macrophages (72). HCMV has a homolog of M45 (73), called UL45, but this fails to suppress cell death (74, 75). The parallels of vICA function aside, most of the immunomodulators encoded by MCMV and HCMV act independently on conserved host defense pathways. HCMV infection blocks necrosis (S. Omoto and E.S. Mocarski, manuscript in preparation); however, the nature of the HCMV-encoded necrosis inhibitor remains to be determined. Based on early experimental data (3, 9), some vFLIPs may act as suppressors of necrotic death. RIP3 necrosis plays out in humans, as well as mice, although humans encode two self-processing caspases (Casp8 and Casp10), whereas rodents have only Casp8. Nevertheless, primary human peripheral blood cells retain the capacity for necroptosis under experimental conditions that parallel what is known in mice (76).

The increased susceptibility of RIP3-deficient mice to vaccinia infection (37) stands in striking contrast to natural infection with MCMV, in which vIRA sustains infection by preventing DAI-RIP3 necrosis (25) (Fig. 2) but where elimination of the RIP3 pathways does not alter WT virus pathogenesis or control (26). Thus, RIP3 necrosis is a mechanism of host defense that threatens virus-infected cells, making the dialogue between vIRA and RIP3 crucial in both immunocompetent (26, 69), as well as in immunodeficient, mice lacking NK, T, and B cell functions (25, 26, 67). The need for vIRA is reversed in RIP3- and DAI-deficient mice (25) where the mutant virus replicates and disseminates. Thus, RIP3 necrosis operates within infected cells. RIP3 does not make an apparent contribution to the function of immune cells that respond to and control infection. RIP3 partner DAI has the capacity to trigger RHIM- and RIP3-dependent IFN activation in mouse and human cells (77, 78), although neither NF-κB nor IFN contributes to virus-induced necrosis (25). DAI-dependent IFN activation can be suppressed by MCMV-encoded vIRA (77, 79); however, the contribution of this one pathogen sensor in dictating levels of NF-κB and IRF3 activation during natural infection in the host has not been gauged with any accuracy. Parenthetically, DAI certainly contributes to HCMV virion–induced IFN response in cell culture (80). With MCMV infection, RIP3 necrosis becomes subdued, enabling virus infection and dissemination to proceed (25, 26, 68). In aggregate, the elaboration of a potent suppressor by MCMV reveals that RIP3 necrosis can completely arrest viral infection by killing off infected cells independent of other immune mechanisms, as well as the particular inoculation route. It would be a remarkable feat to harness RIP3 by therapeutic intervention to confer pathogen-independent resistance to a wide range of infectious agents, such as those that pose a potential biothreat.

Although potentiators of RIP3 necrosis have not been investigated, necrotic death is experimentally blocked by recently described small molecule RIP3 kinase inhibitors (28) that act regardless of whether triggered by RIP1–RIP3, DAI–RIP3, or TRIF–RIP3 complex formation. Such inhibitors promise to expand our understanding of RIP3 kinase in necrotic death, similar to the powerful impact that necrostatins have had on defining the specific role of RIP1 kinase activity in necroptosis (22).

Host defense mechanisms involving immune cells that protect from infection through innate and adaptive mechanisms have the potential to trigger immunopathology (10). Death pathways emanating from the core Casp8 complex are known to undermine development and tissue homeostasis by unleashing RIP3 necrosis and inflammation (3, 5, 12, 49). When germline Casp8 or FADD deficiency is rescued by elimination of RIP3 (11, 20) or RIP1 (81), respectively, RIP1–RIP3 necroptosis emerges as a specific risk when the Casp8 complex becomes compromised during development (3, 5, 12). Although RIP3 engages DAI (25) and TRIF (28) as alternatives to RIP3, neither of these RHIM adaptors contributes to midgestational death in mice. Casp8-FLIP association within the core complex (Fig. 1) blocks RIP3 necrosis (20), potentially targeting RIP1, RIP3, or some component of polyubiquitylation/deubiquitylation machinery (Fig. 1); however, the precise target(s) of basal caspase activity that prevents necrosis remains to be clarified. Casp8-deficient humans survive development but exhibit immunodeficiency (82), a phenotype that is remarkably similar to T cell–specific disruption of either Casp8 (32, 83) or FADD (84) in mice where T cells die by necroptosis upon TCR activation (32, 84). The ability of TCR to trigger RIP3 necrosis indicates that the CARMA1–BCL10–MALT1 complex that normally activates NF-κB also influences the core “Ripoptosome” complex or, alternatively, contributes to increased production of TNF, followed by TNFR1-induced necroptosis (Fig. 2).

All settings in mice in which deficiency of Casp8 (11, 20), FADD, or FLIP (21) has been rescued by eliminating RIP3 produce viable and fertile mice that exhibit lymphoid hyperplasia accompanied by the accumulation of an abnormal B220⁺ T cell population as they age (85). This phenotype aligns with the importance of Fas-dependent extrinsic apoptosis in the homeostatic turnover of T cells. tCasp8^−/−Rip3^−/− or tFadd^ddRip3^−/− (84) mice phenocopy this defect, revealing a requirement for Casp8 function to eliminate excess T cells that is independent of RIP3. Curiously, humans with Casp10 deficiency exhibit an autoimmune lymphoproliferative syndrome characteristic of Fas signaling deficiency (82) that also matches the phenotype of Fas signaling deficiency in mice.

Other than lymphoid hyperplasia that develops with age, Casp8^−/−Rip3^−/− mice exhibit none of the severe developmental defects, homeostatic collapse, or increased inflammation that result from the disruption of either Casp8 or FADD in specific tissues (49, 52, 86–100). Thus, RIP3 necrotic death and unleashed inflammation are both consequences of compromised Casp8 function. Compromise in E3 ubiquitin ligases cIAP1 and cIAP2 or in the SHARPIN component of the linear ubiquitination complex results in similar inflammatory outcomes (34, 101, 102), and it was the topic of a recent review (103). Thus, the interdependency of Casp8 and RIP3 pathways, which evolved for host defense, leads to serious developmental, homeostatic, and inflammatory complications. Disparate observations in the fields of immunology and cell and development biology, as well as in signal transduction, center on dysregulation of a core “Ripoptosome” complex that can sidetrack cell cycle progression, NF-κB activation, autophagy, cell adhesion and migration, and inflammation (12, 32, 33, 49). The picture reveals a striking system-wide role for Casp8 in silencing RIP3-dependent pathways to prevent inflammatory damage and disease throughout development and during life (7).

Despite deficiency in extrinsic apoptosis and RIP3 necrosis, Casp8^−/−Rip3^−/− mice can control viral infection like WT or RIP3-deficient mice (11), mounting CD8 T cell responses to control acute MCMV infection (D. Livingston-Rosanoff and E.S. Mocarski, manuscript in preparation) that compare with matched C57BL/6 mice (104, 105). In addition, tCasp8^−/−Rip3^−/− (32, 83) and tFadd^ddRip3^−/− (84) mice retain full immune control over the RNA viruses lymphocytic choriomeningitis virus and mouse hepatitis virus. Casp8^−/−Rip3^−/− mice support MCMV-specific CD8 T cell expansion, contraction, and recall like WT controls, including characteristic memory inflation and complete protection from secondary challenge. Thus, extrinsic death pathways are redundant, in a pattern that also characterizes other immune mechanisms (10). Experiments with Casp8^−/−Rip3^−/− mice showed that cytotoxicity, turnover of responding cells, memory T cell maintenance and recall, as well as aspects of the cellular immune response to virus infection that interface with other immune and nonimmune cell types, can proceed completely independent of extrinsic apoptosis. This comes as a surprise given the range of lymphocytes, macrophages, and DCs that is known to collude in control of viral infection and the repeated implication of death receptors, as well as pathogen sensors and other signaling pathways, that trigger extrinsic cell death in the overall immune response to infection. Most surprising of all is that elimination of extrinsic cell death does not impact the intensity of the Ag-specific CD8 T cell response, which is dependent on APCs that present viral peptides by cross-presentation (106). In line with other infections (50, 107, 108), the immune response to MCMV within WT mice is influenced by levels of cross-presentation that depend on dying virus–infected cells for a protective CD8 T cell response (109). This occurs via immunoproteasome-dependent APC function (110, 111) that counters virus-mediated MHC class I downregulation in infected APCs (112). Casp8^−/−Rip3^−/− mice probably depend on intrinsic, Bcl2 family member Bim-dependent apoptosis (6) for purposes of lymphocyte contraction, as well as for turnover in the antiviral CD8 T cell response, because there is no other known pathway to support cross-presentation (108). Casp8^−/−Rip3^−/− mice are able to support memory inflation that accompanies latent infection, a pathway that depends on direct Ag presentation (113), as well as CD4 T cell function (114). The ability of Casp8^−/−Rip3^−/− mice to mount diverse innate and adaptive immune responses in the absence of extrinsic death machinery indicates that Fas, other death receptors, or any innate signaling via the “Ripoptosome” complex is dispensable for a cellular immune response that controls infection (84).

The remarkable ability of mice lacking Casp8 and RIP3 pathways to mount a protective CD8 T cell immune response raises a very important question about the contribution of extrinsic cell death to the immune response in the WT host. This question is important on several levels: the basis of vaccine immunogenicity rests on empirical comparisons to natural virus infection (115), little is known about the independent contribution of innate immune cell death independent of innate immune induction of cytokine production, and experimental studies have long suggested a correlation between induction of cell death and immunogenicity (50, 107, 108), although this has not been addressed in hosts that are deficient in major cell death pathways or with pathogens that are specifically susceptible to apoptotic or necrotic death. Extrinsic apoptosis and programmed necrosis certainly influence immune response parameters toward virus-infected cells in infection models (50, 107, 108). In the setting of systemic MCMV infection, where cross-presentation dominates CD8 T cell priming (109) and relies on the CD8α subset of DCs (116), the impact of particular cell death pathways on CD8 T cell immunity remains to be established. This area has relevance because CMVs have potential as vaccine vectors to protect against pathogens, such as HIV (117–120). In natural infection, viral load is a major driver of adaptive immunity. Attenuated or replication-defective viral vectors typically drive a weaker T cell response that may exhibit different qualitative parameters than the original viral pathogen (115). This has triggered a growing literature on the topic of rational vaccine vector design (121), as well as the search for vectors that have the potential to deliver supernatural immunogenicity, such as observed with rhesus macaque CMV (120). Contributions of cross-presentation to CD8 T cell priming can be addressed by disrupting either MCMV (109) or mouse (116) genetic determinants, such as virus-encoded cell death suppressors, or the extrinsic apoptosis and programmed necrosis pathways that they target (16).

Proapoptotic vICA (121) and pronecrotic vIRA (26) mutant MCMV induce premature Casp8 apoptosis (121, 122) and RIP3-dependent necrosis (25, 26), cutting short infection (16). Replication levels of vICA- and vIRA-deficient viruses become normalized in Casp8^−/−Rip3^−/− (double knockout) mice due to the absence of the pathways that these cell death suppressors target. Systemic inoculation (123) delivers sufficient virus to trigger a cross-presentation–mediated response, even when a replication-defective virus is used, although peak antiviral responses are lower (109). WT MCMV produces sizeable virus loads in spleens, livers, and lungs of WT and double-knockout mice, and virus disseminates to salivary glands, where persistent infection occurs for at least a month. In contrast, virus strain–matched proapoptotic vICA (∆M36) (121, 122, 124) and pronecrotic vIRA (M45mutRHIM) (25, 26) mutant viruses are attenuated and fail to produce significant virus loads in any tissue. The differences between replication-competent and replication-deficient MCMV significantly impact the size of virus-specific CD8 T cell responses (109). Despite replication compromise, proapoptotic and pronecrotic mutants induce a MCMV-specific CD8 T cell response that is as robust as WT virus when assessed by cytoplasmic IFN-γ in virus-specific T cells. In this setting, the increased cross-presentation from either enhanced apoptosis (121, 122) or enhanced necrosis (25, 26) may compensate for reduced Ag load (Fig. 3), a relationship that contrasts observations with the replication-deficient viral mutant that does not enhance cross-presentation, leaving peak MCMV-specific IFNγ⁺ CD8 T cell levels that are many fold lower than WT (109). It appears that pronecrotic or proapoptotic viruses promote Ag cross-presentation (D. Livingston-Rosanoff and E.S. Mocarski, manuscript in preparation), leading to a model that relates Ag load and CD8 T cell response (Fig. 3). The recent success of rhesus macaque CMV as a model SIV vaccine vector (120, 125) may certainly stem from the predicted proapoptotic nature of this vector due to the disruption of vICA activity (71). It will be of interest to determine whether matched proapoptotic or pronecrotic rhesus macaque CMV mutants show enhanced cross-presentation properties comparable to murine CMV mutants in mice.

The capacity of the host to switch-hit between apoptosis and necrosis pathways very likely facilitates innate clearance of many intracellular pathogens, despite the fact that adenoviruses, poxviruses, and herpesviruses all encode potent cell death suppressors that have limited the impact of cell death in host defense (3, 14, 15, 25, 26, 37, 68). The balance of cell death contributes to inflammation, cross-presentation and control of viral infection, as well as life-long adaptive immune memory that prevents reinfection. Cell death pathways that evolved to support host defense are completely dispensable for development. It is attractive to consider how these pathways may be harnessed to enhance innate host resistance to infection and improve immunogenicity of vaccines.

We thank colleagues at the Emory Vaccine Center for the environment where the concepts presented in this article were developed and discussed.

The authors have no financial conflicts of interest.