The Antisocial Network: Cross Talk Between Cell Death Programs in Host Defense

Abstract

Nearly all animal cells contain proteins evolved to trigger the destruction of the cell in which they reside. The activation of these proteins occurs via sequential programs, and much effort has been expended in delineating the molecular mechanisms underlying the resulting processes of programmed cell death (PCD). These efforts have led to the definition of apoptosis as a form of nonimmunogenic PCD that is required for normal development and tissue homeostasis, and of pyroptosis and necroptosis as forms of PCD initiated by pathogen infection that are associated with inflammation and immune activation. While this paradigm has served the field well, numerous recent studies have highlighted cross talk between these programs, challenging the idea that apoptosis, pyroptosis, and necroptosis are linear pathways with defined immunological outputs. Here, we discuss the emerging idea of cell death as a signaling network, considering connections between cell death pathways both as we observe them now and in their evolutionary origins. We also discuss the engagement and subversion of cell death pathways by pathogens, as well as the key immunological outcomes of these processes.

1. INTRODUCTION

Programmed cell death (PCD) constitutes an essential biological process in diverse physiological settings, including embryonic development, maintenance of tissue homeostasis, and host defense against pathogens. In contrast to unprogrammed cellular rupture and lysis following injury, PCD is governed by distinct molecular signaling pathways whose engagement leads to the deletion of either unneeded cells or those that have been compromised by infection or transformation. As numerous groups have provided comprehensive summaries of the signaling mechanisms that define distinct PCD modalities (1–8), this review provides an abbreviated summary of PCD signaling pathways. The bulk of our discussion is then primarily focused on two topics: (a) emerging evidence of functional and evolutionary cross talk between PCD pathways, and (b) the immunological effects of PCD engagement in the context of pathogenic infection. In each section, we compare and contrast three forms of PCD: apoptosis, necroptosis, and pyroptosis. Other forms of PCD have been identified, including ferroptosis (9), neutrophil NETosis (10), and engulfment or entosis of whole cells (11); however, this review is focused on the aforementioned programs to constrain our discussion to a smaller subset of PCD pathways that display clear mechanistic interconnectivity.

Recent dogma within the cell death field has coalesced around the idea that cells dying by apoptosis are nonimmunogenic while necroptosis and pyroptosis drive immune responses (12–14). While this distinction is broadly useful, we suggest that it does not account for the complexity of either the immune system or the cell death pathways themselves. We now recognize that the immune system carries out numerous functions beyond the elimination of pathogens, including roles in development, tissue homeostasis, and wound healing; consideration of how dying cells may drive such immune functions does not fit neatly within the immunogenic/nonimmunogenic dichotomy. Furthermore, as discussed in Section 3 we now recognize that extensive cross talk exists between cell death pathways and that effectors of multiple pathways are likely activated simultaneously within tissues or even within individual cells to help coordinate host defense (Figure 1). We therefore hope that this review will highlight that the ability of PCD to instruct varying qualities of downstream immune responses will be, like many topics in biology, highly dependent on the responding cell types, microenvironmental cues, and specific disease settings involved.

Figure 1.

Programmed cell death pathways function as interconnected signaling networks that coordinate host defense. Stress- or infection-associated stimuli are detected through a repertoire of innate immune receptors, resulting in the concomitant or sequential activation of cell death signaling programs. Apoptotic, necroptotic, and pyroptotic signaling contribute various immunological outputs, including entry of cell- or pathogen-associated antigens into presentation pathways, the removal of a replicative niche for intracellular pathogens, de novo production of inflammatory cytokines and chemokines, and release of DAMPs. The composition of these outputs tailors the generation of an immune response mounted in response to programmed cell death. Abbreviations: DAMP, damage-associated molecular pattern; PAMP, pathogen-associated molecular pattern.

2. SIGNALING OVERVIEW

The engagement of PCD is tightly controlled by a complex network of signaling mechanisms that often exhibit cross talk between receptors, enzymes, and downstream signaling products. Here, we provide a simplified overview of apoptotic, necroptotic, and pyroptotic pathways. Additionally, the molecular circuitry associated with each form of PCD is summarized in Figure 2.

Figure 2.

Overview of programmed cell death signaling pathways. (a) Apoptosis occurs following extrinsic activation of death receptor family members or intrinsic perturbation of mitochondrial membrane dynamics. Extrinsic apoptosis requires the formation of proapoptotic CASP8 homodimers paired with suppression of the inhibitory FADD/CASP8/cFLIP complex, while intrinsic apoptosis results in CASP9 activation downstream of MOMP triggered by cellular stresses that skew the ratio of proapoptotic Bcl-2 family members. Both CASP8 and CASP9 activate the executioner caspases CASP3 and CASP7, which cleave cellular substrates to induce apoptotic cell death characterized by membrane blebbing and the expression of eat-me signals on apoptotic bodies including phosphatidylserine and calreticulin. (b) Necroptosis occurs upon activation of death receptors, certain TLRs, or ZBP1. When paired with suppression of the aforementioned inhibitory FADD/CASP8/cFLIP complex, these signals trigger either RIPK1-dependent or RIPK1-independent assembly and phosphorylation of RIPK3 oligomers. Activated RIPK3 then mediates two separate signaling outcomes: phosphorylation and activation of the necroptosis executioner MLKL, which forms pores in cellular membranes, and activation of transcription factors such as NF-κB and IRF1 that mediate inflammatory gene expression. This leads to lytic cell death that releases cell-associated antigens and DAMPs, paired with de novo cytokine and chemokine production. (c) Pyroptosis is engaged following the detection of PAMPs or pathogen-associated cellular stressors by cytosolic sensors such as NLRP3, NLRC4, or AIM2, which recruit pro-CASP1 and form a multimeric signaling complex termed the inflammasome. Inflammasome assembly results in activation of CASP1, which carries out two effector functions: the cleavage and activation of the pyroptosis executioner GSDM-D, which forms pores in the plasma membrane, and the processing and activation of IL-1 family cytokines. CASP1 activation therefore results in the secretion of IL-1 cytokines through GSDM pores and eventual cell lysis due to imbalanced osmolarity. Abbreviations: cFLIP, cellular FLICE-inhibitory protein; CRT, calreticulin; cytoC, cytochrome c; FADD, Fas-associated death domain; DAMP, damage-associated molecular pattern; dsDNA, double-stranded DNA; GSDM, gasdermin; LPS, lipopolysaccharide; MOMP, mitochondrial outer membrane permeabilization; PAMP, pathogen-associated molecular pattern; PS, phosphatidylserine; ROS, reactive oxygen species; SMAC, second mitochondria-derived activator of caspases; tBID, truncated BID; TLR, Toll-like receptor; XIAP, X-linked inhibitor of apoptosis.

2.1. Apoptosis

Apoptosis can be engaged through two pathways: Intrinsic apoptosis is triggered by cell-intrinsic stressors such as DNA damage or endoplasmic reticulum stress, and extrinsic apoptosis is initiated through activation of death receptor (DR) family members located at the plasma membrane. Although intrinsic and extrinsic apoptosis differ with respect to their upstream stimuli and apical signaling components, both forms of apoptosis require the function of a family of cysteine-aspartate proteases termed caspases, whose activation mediates apoptotic cellular suicide.

Intrinsic apoptosis is triggered when cell-intrinsic stresses impinge upon the balance of the Bcl-2 family of proteins (4, 15). The Bcl-2 family consists of both proapoptotic members (Bax, Bak, Bok, Bid, and Bim, among many others) and antiapoptotic members (Bcl-2, Bcl-XL, and Mcl-1), and complex interactions between these proteins integrate pro- and antiapoptotic signals. In situations where the former predominate, effector members of the Bcl-2 family can trigger mitochondrial outer membrane permeabilization (MOMP) (16) by forming pores in the outer mitochondrial membrane. MOMP leads to the release of several key proapoptotic molecules into the cytosol, including the second mitochondria-derived activator of caspases (SMAC) and the electron transport chain component cytochrome c. Once released, SMAC antagonizes X-linked inhibitor of apoptosis (XIAP), a suppressive protein that binds and inhibits apoptotic caspase-9, −3, and −7 (17). Meanwhile, released cytochrome c is bound by the cytosolic sensor APAF-1, which in turn recruits and activates caspase-9; the resulting multiprotein signaling complex is termed the apoptosome (18). Activated caspase-9 can then activate the executioner caspase-3 and −7, which cleave aspartate residues found in a wide array of intracellular proteins. Cleavage of these substrates by executioner caspases leads to an irreversible loss of cellular function and subsequent cellular dismantling, leading to characteristic morphological changes discussed below.

In contrast to intrinsic apoptosis, extrinsic apoptosis is engaged through plasma membrane DR signaling (19). This includes binding of TNFR1/2 to TNF (20), Fas/CD95 to FasL (21), or TRAILR to TRAIL/Apo2 (22). Each of these receptor-ligand pairs differs slightly with respect to the composition of signaling proteins that are recruited to cytosolic receptor domains, but in all cases induction of cell death by these receptors requires the recruitment of an adaptor protein termed Fas-associated death domain (FADD), which then recruits and activates caspase-8, the apical caspase family member required for extrinsic apoptosis (23). In some cases, caspase-8 can directly activate caspase-3 and −7, but in most cells effective apoptosis requires that caspase-8 cleave and activate the proapoptotic Bcl-2 family member Bid (17, 24). This triggers MOMP, which amplifies the proapoptotic signal.

Notably, DR signaling alone is often insufficient to engage apoptosis, as binding between DR family members and their conjugate ligands typically activates prosurvival signaling. Signaling via TNFR1 is a well-studied example: Receptor ligation leads to the recruitment of numerous signaling proteins and complex ubiquitination and phosphorylation events carried out by the IKK, TAK1, cIAP, and LUBAC enzymes (reviewed in 3). The output of this signaling is generally transcriptional activation via the MAPK and NF-κB pathways, leading to upregulation of prosurvival factors. These include cellular FLICE-inhibitory protein (cFLIP), a caspase paralog that forms a suppressive complex with caspase-8 and FADD, sequestering caspase-8 to prevent homodimer formation and therefore inhibiting extrinsic apoptosis (25). Therefore, TNF engagement alone triggers prosurvival signaling, and extrinsic apoptosis can only occur upon concurrent suppression of this transcriptional signaling. As discussed below, this may represent a checkpoint by which this cytokine can trigger cell death if pathogen-encoded effectors that block inflammatory transcriptional programs are present.

As both intrinsic and extrinsic apoptosis converge upon the activation of the executioner caspase-3/−7, which subsequently cleaves intracellular substrates, it follows that the morphology of cells dying via either form of apoptosis is the same (26). Apoptotic cells exhibit chromatin condensation, DNA cleavage, and cellular shrinkage, and the cell eventually dissociates into numerous plasma membrane–bound vesicles termed apoptotic bodies. These vesicles serve an important function by tidily packaging the intracellular contents of a dying cell, including various potentially immunostimulatory molecules such as self–nucleic acids and other damage-associated molecular patterns (DAMPs), shielding these molecules from the extracellular space, where they may initiate inflammatory signaling (27). As apoptotic bodies are coated in various phagocytic eat-me signals including phosphatidylserine (PS) and calreticulin, under normal conditions these vesicles are rapidly taken up by phagocytes through a process referred to as efferocytosis.

Efferocytosis plays a critical role in the maintenance of tissue homeostasis, as billions of cells undergo apoptosis within the body every day, the vast majority of which must be cleared without inducing inflammation. Therefore, homeostatic efferocytosis is typically viewed as an immunologically silent or even immunosuppressive phenomenon that has important implications in the maintenance of tissue homeostasis and the resolution of inflammation (27, 28). This is partially attributed to the rapid clearance of apoptotic bodies by phagocytes before they can release potentially immunostimulatory intracellular contents via secondary necrosis. Importantly, apoptosis can also actively suppress the induction of inflammatory responses by apoptotic cell-intrinsic neutralization of immunogenic molecules, including caspase-directed oxidation of high-mobility group box-1 protein (HMGB1) (29) and caspase-mediated blockade of type I IFN signaling (30). Additionally, the uptake of apoptotic cells often relays immunomodulatory signals to recipient phagocytes, as exemplified through the binding of PS on apoptotic cells to the regulatory receptors AXL and MERTK on macrophages, which upon coincidental sensing of the type 2 cytokine IL-4 skew macrophages toward an anti-inflammatory wound healing phenotype (31).

The importance of apoptosis in suppressing inflammation is highlighted by the fact that mutations in genes associated with the packaging, recognition, and degradation of apoptotic bodies by phagocytes are associated with a break in immunological tolerance characteristic of autoimmune diseases such as systemic lupus erythematosus (32–34). However, under particular cellular contexts such as infection, apoptosis can also be associated with inflammatory gene expression that supports the generation of type 1 immunity (35, 36). As with many cell death pathways, the immunogenicity of apoptosis can vary widely according to the composition of inflammatory cues present in the local tissue microenvironment as well as the specific cellular context in which apoptosis occurs.

2.2. Necroptosis

Necroptosis is a lytic form of cell death, defined by activation of the kinases RIPK1 and RIPK3, which can form a cytosolic complex termed the necrosome via amyloid-like interactions between the RIP homotypic interaction motif (RHIM) domains present in both kinases (37). This can lead to activation of the downstream effector MLKL (38) following its RIPK3-mediated phosphorylation. Phosphorylated MLKL assembles into a multimeric complex that disrupts the plasma membrane, leading to cell lysis (39). However, under most physiological settings this signaling is antagonized by the caspase-8/cFLIP complex (40, 41), which is recruited to the necrosome via interactions between the adapter FADD and RIPK1. The proteolytic activity of caspase-8/cFLIP can be directed against RIPK1, whose cleavage suppresses necroptosis (42). For this reason, experimental induction of necroptosis generally involves inhibition or ablation of caspase-8. This has led to the hypothesis that necroptosis is a trapdoor to cell death that can occur in the presence of pathogen-mediated inhibition of caspase-8 or cFLIP, supported by the finding that several viral and bacterial pathogens encode such inhibitors (43, 44).

Although necroptosis activation downstream of the death receptors has been confirmed extensively using in vitro models of caspase deficiency or pharmacological inhibition, a clearly beneficial physiological role for DR-dependent necroptosis in immune defense remains elusive. Clearer evidence for necroptosis as a component of host defense comes from studies of virus-induced necroptosis downstream of the nucleotide sensor ZBP1. ZBP1 contains multiple RHIM domains by which it can recruit and activate RIPK1 and RIPK3 (45), and this signaling pathway is engaged by multiple viruses, including cytomegalovirus (CMV) (46), herpes simplex virus (HSV) (47, 48), vaccinia virus (VACV) (49), influenza A virus (IAV) (50, 51), West Nile virus (WNV) (52), and Zika virus (53). RIPK1/3 signaling can also occur downstream of the Toll-like receptors TLR3 and TLR4, via the adapter TRIF, which also contains a RHIM domain (54). This pathway has been implicated in defensive necroptosis following bacterial infection, as discussed below.

Due to the formation of MLKL pores following RIPK3 activation, necroptosis is a rapid and lytic form of cell death that lacks the tidy packaging of intracellular molecules observed in apoptotic cells. Necroptosis therefore releases cell-associated antigens accompanied by an array of DAMPs (55), constitutively expressed molecules that can trigger innate immune signaling through pattern recognition receptors (PRRs) to alert the immune system at sites of tissue damage and infection (56–58). Potential DAMP molecules shown to be released from necroptotic cells include self–nucleic acids; polymerized actin that binds to CLEC9A/DNGR1 (59); HMGB1, which triggers RAGE (60); ATP that binds to P2X7 (61); and IL-33, which is recognized by ST2 (62). Notably, immunogenic signaling through DAMP and/or extracellular enzymatic activity is not restricted solely to MLKL-driven cellular lysis, as these molecules can be released following unprogrammed necrosis induced by physical injury. In many cases our understanding of the roles of these DAMPs in specific pathologies has remained incomplete, as several proposed DAMP molecules are essential for cell function and cannot be studied via knockout approaches.

Several groups have reported that activation of the necroptotic program leads to release of immunogenic signals that are produced independently of MLKL-mediated pore formation and cell death, implying that additional functions of RIPK1/RIPK3 activation beyond DAMP release drive the immunostimulatory effects of these cells (63–65). Transcriptional programs activated downstream of RIPK3 may be primarily responsible for driving the immunogenic effects of necroptosis in disease settings ranging from infection to oncogenic transformation and autoimmunity. It is indeed possible that RIPK1- and/or RIPK3-dependent transcriptional targets, when presented in conjunction with potentially immunogenic DAMPs and antigens released from dying cells, serve as a more potent source of inflammatory immune stimulation when compared to some of the immunoregulatory properties associated with apoptotic cells (63, 66). It is also notable that genetic studies have identified several MLKL-independent roles for RIPK1/3 signaling, including antiviral responses within the central nervous system (CNS) and pathogenic responses during kidney ischemia-reperfusion injury (52, 53, 67). Thus, while RIPK1/3 are considered key components of the necroptotic pathway, their activation may also lead to important death-independent signaling outcomes.

2.3. Pyroptosis

Pyroptosis is also a form of lytic cell death, but it differs from necroptosis in that it is engaged independently of DR or TLR activation (68, 69). Instead, pyroptosis occurs following the assembly and activation of multiprotein signaling complexes termed inflammasomes, which are activated in response to an array of signals ranging from molecules associated with cellular stress or damage to pathogen-associated molecular patterns (PAMPs) that are typically encountered during infection (70, 71). Inflammasomes are classified as either canonical or noncanonical subtypes, which differ with respect to their upstream activating ligands and induce the activation of a set of pyroptotic caspases that are entirely distinct from those associated with apoptosis: caspase-1 and −11 (in rodents; caspase-11 is orthologous to caspase-4 and −5 in humans) (72).

Noncanonical inflammasome activation is defined by direct binding of caspase-4, −5, and −11 to intracellular lipopolysaccharide (LPS) derived from gram-negative bacterial membranes (73–76). LPS binding is sufficient for the dimerization and activation of these caspases, which can then act on downstream substrates as discussed below. In contrast, canonical inflammasomes require the assembly of cytosolic multiprotein signaling complexes that recruit and activate caspase-1, which can occur in response to either infection-associated PAMPs or cellular stress and injury signals (70). Canonical inflammasomes include absent in melanoma 2 (AIM2) and nucleotide-binding domain-like receptor (NLR) family members including NLRP3 and NLRC4. Some of these function as sensors that directly bind to PAMPs, such as AIM2, which binds to virus- or bacterium-derived cytosolic double-stranded DNA via interactions with its HIN200 domain (77). NLR proteins can also be activated indirectly, as is the case with NLRC4: Cytosolic flagellin monomers or bacterial secretion system components are bound by a class of sensors termed NLR family apoptosis inhibitory proteins (NAIPs), which in turn activate the NLRC4 inflammasome (78, 79). NLRP3 is likely activated in an indirect fashion, as it can be triggered by a myriad of diverse signals including microbial cell wall components, pore-forming toxins, crystalline aggregates such as silica and alum, and DAMPs such as ATP. It has therefore been proposed that these stimuli converge upon a shared secondary signal that reflects a state of cellular stress, such as reactive oxygen species (ROS) or potassium ion efflux, which in turn stimulates NLRP3 oligomerization and activation (80). Regardless of the initiating signal, oligomerization of the canonical AIM2, NLRC4, or NLRP3 inflammasomes leads to the recruitment of caspase-1; in the case of AIM2 and NLRP3, caspase-1 recruitment requires binding of the adaptor protein ASC (81). Upon recruitment, caspase-1 undergoes autocatalytic cleavage into its activated subunits, p46 and p33/p10, which enable the execution of downstream effector functions.

Both canonical inflammasome activation and noncanonical inflammasome activation converge on a common downstream substrate: gasdermin D (GSDM-D) (73, 82, 83). GSDM-D is viewed as the executioner of pyroptosis, analogous to MLKL in necroptosis. Once cleaved by pyroptosis-associated caspases, GSDM-D oligomerizes and forms large pores in the plasma membrane (84, 85). Like MLKL-mediated pore formation in necroptosis, the formation of GSDM-D pores disrupts ion homeostasis, leading to cell swelling as extracellular fluid flows into the cell and eventual plasma membrane rupture. Therefore, induction of pyroptosis prompts the lysis and release of intracellular DAMPs, similarly to necroptosis. Notably, activated caspase-1 (but not caspase-4/−5/−11) also catalyzes the processing of the cytokines IL-1β and IL-18 from their inactive zymogen forms into mature cytokines, which can be released either through GSDM-D pores or following pyroptotic cell lysis. The activation of these IL-1 family cytokines holds important implications for immune stimulation downstream of pyroptosis, as IL-18 can stimulate IFN-γ production by CD4⁺ type 1 T helper (Th1) cells and prime natural killer (NK) cell responses, while IL-1β can support T cell expansion, prolong T cell help to B cells to stimulate antibody production, and support the differentiation of Th17 cells (reviewed in 86–88).

In summary, both canonical pyroptosis and noncanonical pyroptosis lead to DAMP release, but only canonical inflammasome activation via caspase-1 produces active IL-1β and IL-18. Pyroptosis therefore resembles necroptosis in that both forms of PCD lead to rapid osmolysis and the release of potentially immunostimulatory DAMPs, coupled to the release of cytokine signals—IL-1β and IL-18 in the case of pyroptosis, and NF-κB-dependent cytokines produced downstream of RIPK1/3 signaling in the case of necroptosis. Thus, while both necroptosis and pyroptosis are pathogen-induced forms of lytic cell death, important differences are apparent in both the signals that initiate them and the immunological outputs of these two cell death programs.

3. CROSS TALK AND EVOLUTIONARY INTERPLAY BETWEEN CELL DEATH PROGRAMS

The description above presents the canonical programs of apoptosis, necroptosis, and pyroptosis as generally distinct signaling cascades. However, numerous recent studies have indicated functional cross talk between these programs; some examples of the mechanisms by which these programs interface are discussed below. In considering these examples, it is noteworthy that in many cases the diversion from one cell death pathway to another—a necroptotic stimulus that engages pyroptotic effectors, for example—is obvious only when components of the canonical pathway are absent, inhibited, or mutated. This may reflect a networking of cell death pathways that ensures that a cell can divert to an alternative pathway even in the presence of pathogen-encoded inhibitors of the primary pathway. The resulting hybrid forms of PCD are discussed below and summarized in Figure 3.

Figure 3.

Network interactions between cell death pathways lead to hybrid forms of programmed cell death. (a) Apoptotic and necroptotic signaling machinery are tightly connected, as extrinsic death receptor signaling through the RIPK1/FADD/CASP8 complex can lead to pleiotropic signaling outcomes. In the absence of RIPK or caspase inhibition, prosurvival programs through NF-κB dominate; if RIP kinases are inhibited, extrinsic apoptosis via CASP8 homodimerization occurs; if caspases are inhibited; necroptosis is enabled through RIPK1-RIPK3 oligomerization. (b) Necroptotic signaling interfaces with pyroptosis through the executioner MLKL. RIPK3 oligomerization downstream of extrinsic DR ligation or cytosolic ZBP1 activation leads to the phosphorylation of the executioner molecule MLKL, which forms pores in cellular membranes including the plasma membrane. MLKL-mediated K⁺ efflux can subsequently trigger NLRP3 activation, leading to inflammasome assembly and CASP1 cleavage that mediates GSDM-D pore formation and the processing of IL-1 family cytokines. (c) Expanded substrate targeting by caspases connects apoptotic and pyroptotic signaling components. Inflammasome engagement leads to CASP1 activation, which in addition to cleaving its canonical substrates of GSDM-D and IL-1 family cytokine precursors can mediate the activation proteolysis of BID into proapoptotic tBID, triggering MOMP and apoptosome formation. CASP1 can also directly cleave CASP11 to promote inflammatory pyroptotic signaling, through poorly understood mechanisms. Additionally, the apoptotic executioner CASP3 can target the pyroptotic executioner GSDM-E for proteolysis, leading to gasdermin pore formation and morphological aspects of pyroptosis. Abbreviations: cFLIP, cellular FLICE-inhibitory protein; DAMP, damage-associated molecular pattern; FADD, Fas-associated death domain; GSDM, gasdermin; MOMP, mitochondrial outer membrane permeabilization; tBID, truncated BID.

3.1. Apoptosis-Necroptosis Cross Talk

As summarized in Section 2, the apoptotic and necroptotic pathways are tightly linked through the activity of caspase-8 (40, 41); caspase-8 is a canonical activator of extrinsic apoptosis, but it also inhibits necroptotic signaling through the cleavage of RIPK1 (42) and possibly RIPK3 (89). This occurs in the necrosome, a complex comprising minimally caspase-8, FADD, RIPK1, and RIPK3. However, in bringing together these key components of apoptotic and necroptotic signaling, the necrosome has the potential to drive both pathways. For example, while addition of the caspase-8 inhibitor zVAD-fmk promotes necroptotic signaling via RIPK3-dependent MLKL phosphorylation, the converse is also true: Small-molecule inhibitors of RIPK3 stabilize the necrosome, allowing it to act as a platform for caspase-8 activation and apoptosis (90). This finding is consistent with the idea that pathogen-mediated inhibition of one pathway may potentiate the other; indeed, influenza infection activates the nucleic acid sensor ZBP1, which recruits the same necrosome components and can trigger either apoptosis or necroptosis (50, 91) (and perhaps even pyroptosis; see below). Also consistently, CMV infection also activates ZBP1, but CMV encodes inhibitors of both apoptosis and necroptosis (92–94).

3.2. Apoptosis-Pyroptosis Cross Talk

Apoptosis and pyroptosis both involve activation of members of the caspase family of proteases, and as discussed below they likely share a common evolutionary origin. It is therefore not surprising that cross talk exists between these pathways on multiple levels. One example is the finding that the pyroptosis-inducing caspase-1 protease can cleave the Bcl-2 family member Bid, leading to MOMP and activation of downstream apoptotic signaling (95). Another example is the finding that the apoptotic initiator caspase-8 can interact with the pyroptotic adapter protein ASC (96, 97). ASC forms supramolecular oligomers during inflammasome activation, and caspase-8 can be activated at these sites. Furthermore, consistent with the concept of bidirectional signaling mentioned above in the context of the necrosome, the ASC–caspase-8 interaction can apparently also lead to caspase-8-dependent activation of pyroptosis in some settings. For example, mice homozygous for a form of caspase-8 that lacks catalytic activity display embryonic lethality, and in contrast to the case of Casp8^−/− animals, this developmental defect is not fully rescued by ablation of necroptotic effectors. This difference is apparently due to a scaffolding function of catalytically inactive caspase-8, which is able to recruit ASC and caspase-1, leading to pyroptosis and lethal inflammation during development (98, 99). Further complicating matters, knockout experiments implicate noncanonical inflammasome signaling via caspase-11 in aspects of the resulting pathology, as well as nonnecroptotic inflammation driven by RIPK3. Together, the picture that emerges is of a signaling complex containing initiators of apoptosis, pyroptosis, and necroptosis (caspase-8, ASC/caspase-1, and RIPK1/3, respectively) that is capable of promoting all three forms of cell death. Importantly, however, these effects become obvious only upon mutation of caspase-8, along with ablation of combinations of downstream cell death effectors; how an analogous complex might influence the response to infection—upon sensing of pathogen-encoded cell death inhibitors, for example—remains an area of active investigation.

A second and more fundamental link between apoptosis and pyroptosis has recently been proposed. This connection is based on the observation that gasdermin E (GSDM-E), which is homologous in structure and pore-forming function to the pyroptotic effector GSDM-D, can be cleaved and activated by caspase-3, the executioner of apoptosis (100). The resulting pore-forming activity of GSDM-E can thereby regulate the membrane permeabilization that occurs late in apoptosis, an event termed secondary necrosis that has long been considered unprogrammed. This finding has led to the suggestion that apoptotic events that terminate in GSDM-E-mediated secondary necrosis are a form of pyroptosis (101). While this is mainly a semantic argument, the question of how GSDM-E engagement during apoptosis influences the release of immunostimulatory molecules from apoptotic cells during normal or pathogenic processes remains to be fully addressed. As we discuss below, recent reports implicate this process in promoting immune responses to dying tumor cells.

3.3. Necroptosis-Pyroptosis Cross Talk

Necroptosis and pyroptosis share a lytic and potentially inflammatory morphology, and this feature has led to the identification of mechanistic cross talk between these pathways. Pyroptosis can be initiated by the NLRP3 inflammasome, which can be activated in response to changes in cellular ion homeostasis; NLRP3 has been proposed to thereby sense cellular integrity and respond to membrane or lysosomal damage that may accompany infection (80). However, this property also allows its activation in response to membrane disruption caused by MLKL, the terminal effector of necroptosis. Through this mechanism, cells undergoing necroptosis can activate the pyroptotic machinery, such that their death is accompanied by the activating cleavage of IL-1β and IL-18 (102, 103). As mentioned above, IAV infection has been reported to trigger simultaneous hallmarks of apoptosis, pyroptosis, and necroptosis, at least in cultured cells (104). All these effects are reportedly dependent on the nucleotide sensor ZBP1, canonically considered a driver of necroptosis. However, while pyroptotic responses clearly contribute to both immunity and immunopathology to IAV in vivo (105), the phenotypes of mice entirely lacking pyroptotic responses are distinct from those lacking ZBP1 or MLKL (50, 91); thus, while necroptosis may drive pyroptosis in some settings, redundancies allow for pyroptotic engagement in the absence of necroptosis.

3.4. Evolutionary Development and PCD Interactions

The cell death pathways discussed here interact with one another not only functionally but evolutionarily. These connections are particularly interesting in considering the origin of the apoptotic and pyroptotic pathways. Pyroptosis is initiated by sensors such as the NLRs, whose activation by pathogen-encoded molecules leads to formation of multimeric inflammasomes that recruit caspase-1 via its CARD domain. Strikingly, the activation of APAF-1 by cytochrome c following MOMP—central to the apoptotic pathway—closely mirrors inflammasome activation (106). APAF-1 shares sequence homology with the NLRs, the structure of the apoptosome is similar to that of an inflammasome, and the end result is the CARD-dependent recruitment and activation of caspase-9. Notably, binding of mitochondrial cytochrome c by APAF-1 parallels binding of pathogen-derived molecules by the NLRs. This is particularly intriguing given the prokaryotic origin of the mitochondria, as it suggests that apoptosis may have evolved from a form of pyroptosis-like cell death that predates the symbiotic relationship with a bacterium that led to formation of the mitochondria (107). In this scenario, a proto-eukaryote established an ancient form of cell death in which an APAF-1-like molecule recognized cytochrome c–like molecules in invading bacteria, triggering defensive cell suicide. Later in evolution this proto-eukaryote established a symbiotic relationship with a similar bacterium, based in part on the useful redox chemistry made possible by the latter’s cytochrome c. Still later in evolution, perhaps coinciding with the emergence of multicellular life, it became necessary to delete some cells in a developmentally programmed manner. To achieve this, evolution repurposed components of the cytochrome c–mediated pyroptotic pathway: As the key PAMP driving this pathway (cytochrome c) was now present in all cells but compartmentalized in the mitochondria, achieving its controlled release could allow for PCD to occur. While speculative, this scenario has at least two striking implications. First, it implies that pyroptosis-like defensive cell death was among the most ancient forms of immune defense, predating even mitochondrial symbiosis and the emergence of eukaryotes. Second, it implies that while apoptosis was the first form of PCD discovered and remains the best understood, apoptosis originated as a repurposing and taming of pathogen-triggered pyroptosis.

The evolutionary origins of necroptosis are less clear. Unlike the case of apoptosis, which is present in some form in all animals, many components of the necroptotic pathway are apparently poorly conserved even among vertebrates (108). For example, marsupials lack obvious homologs of MLKL and RIPK3, while birds lack RIPK3 and mammalian carnivores lack MLKL. Intriguingly, however, amyloid-forming RHIM domains—central and unique to necroptotic signaling in mice and humans—were recently identified as a key component of the immune deficiency (IMD) pathway in Drosophila, which regulates the activation of the NF-κB homolog Relish (109). To draw additional parallels, Drosophila IMD is activated by a PAMP-sensing adapter, forms RHIM-dependent amyloid structures, and interacts with the fly homologs of FADD and caspase-8; these properties are shared by mammalian RIPK1. Interestingly, however, Drosophila IMD is activated in response to peptidoglycan, and this activation drives NF-κB responses; these properties are shared by the mammalian NOD2-RIPK2 signaling axis, which involves a homolog or RIPK1 (RIPK2) but does not depend on RHIM domain signaling.

Taken together, these observations imply that, like many immune pathways, cell death signaling has evolved and diversified under the selective pressure of pathogen infection. Across evolutionary time, genomic changes such as gene duplication and domain swapping have likely provided the evolutionary plasticity needed to ensure that protective cell death can occur in the face of pathogen-encoded PCD inhibitors. This plasticity has likely also driven much of the cross talk between cell death pathways described above.

4. CELLULAR SUICIDE IN HOST DEFENSE AGAINST PATHOGENS

A primary purpose of the immune system is to defend organisms from infectious agents that threaten host health and function. This is accomplished through a wide variety of mechanisms that range from the generalized responses engaged by innate immune modules that prime responses against classes of pathogens to the specificity of adaptive lymphocyte receptors that enable long-term recognition of specific molecular epitopes associated with prior episodes of infection. PCD is an important arm of innate immunity, as it can eliminate replicative niches for intracellular pathogens, engage both rapid innate immune signaling pathways through DAMP-PRR ligation, and influence the quality of adaptive immune memory generated against cross presented pathogen antigens from dying cells through the release of death-associated chemokines and cytokines. Notably, many pathogens express virulence factors that target multiple host PCD pathways to maximize their likelihood of survival and dissemination; given this broad targeting, it follows that multiple aspects of apoptosis, necroptosis, and/or pyroptosis can be observed upon infection with a single microorganism. Here, we discuss the contributions of apoptosis, necroptosis, and pyroptosis in tailoring host immune responses to subsets of pathogens. Our examination should not be considered exhaustive, as more comprehensive discussion of specific pathogens has been provided elsewhere (110–112).

4.1. Viruses

Viruses are obligate intracellular pathogens that hijack host cell machinery to replicate and propagate infection. Due to this requirement for maintaining physiological host cell processes that support viral replication, the deletion of infected cells through any form of PCD is an effective strategy of host defense by limiting infection while exposing the virus to immune recognition. Beyond this important consequence of PCD, other molecular signals that are produced by specific PCD modalities further contribute to the degree of inflammation and quality of an antiviral immune response.

Although apoptosis is typically viewed as less immunogenic due to the expression of numerous molecules that dampen inflammation, it still plays an important role in eliminating virally infected cells (113). Importantly, apoptosis is the main mechanism of PCD engaged in infected cells targeted by cytotoxic immune cells, which are critical for controlling infection with intracellular pathogens such as viruses. These include NK cells and cytotoxic T lymphocytes (CTL), which use perforin-granzyme delivery and FasL expression to target infected cells through a variety of apoptosis-inducing mechanisms (114). (As discussed below, very recent studies indicate that granzymes may activate gasdermin-dependent pyroptosis in some situations, though the immunological outcome of this cross talk is not completely understood.) The importance of apoptosis in host defense is clearly illustrated by the fact that numerous viruses encode virulence factors that prevent killing either through cell-autonomous apoptosis or by cytotoxic lymphocytes. These include caspase inhibitors such as vICA from murine cytomegalovirus (MCMV) (93), the cowpox virus effector CrmA (115, 116), and B13R/Spi2 expressed by VACV (117, 118). In addition to caspase inhibitors, many viruses express Bcl-2 homologs that can directly inhibit Bax and Bak, such as gammaherpesvirus M11 (119, 120). However, in many of these cases, the successful evasion of apoptosis by these pathogens can render them vulnerable to alternative forms of PCD that may have counterevolved to detect and eliminate infected cells that cannot undergo apoptosis.

This case is best exemplified by necroptosis, which serves as an alternative form of PCD that can be engaged in cells where caspase function has been compromised. Beyond the elimination of caspase-deficient cells, necroptosis can also function more broadly by eliminating cells whose transcriptional and/or translational outputs are altered by viral infection, as RIPK1/NF-κB-mediated expression of prosurvival cIAPs and cFLIP is also required for the suppression of necroptosis. Indeed, IAV does not specifically encode a dedicated caspase-8 inhibitor, but it can trigger RIPK3-dependent necroptosis (91) that may be caused by IAV-mediated blockade of mRNA translation (121), including prosurvival proteins like cFLIP. As with apoptosis, the importance of necroptosis as a mechanism of host immunity to infection is highlighted by the expression of RIPK inhibitors by viruses. In some cases these proteins contain a RHIM domain that effectively blocks RHIMRHIM interactions to prevent RIPK1/RIPK3 oligomerization, and such inhibitors generally occur in combination with inhibitors of apoptosis (44, 46–48, 92). Interestingly, while MCMV encodes a RHIM-containing necroptosis inhibitor, human CMV also blocks necroptosis but appears to do so downstream of RIPK1/3 oligomerization, by targeting MLKL-dependent membrane destruction (122).

Beyond blunt restriction of viral replication through the deletion of their replicative niche, necroptotic signaling also contributes to antiviral immunity through death-independent functions of RIPK3. Indeed, the gene regulatory functions of RIPK3 are required for mounting successful immune responses to viral infection in certain cases where MLKL activation is dispensable for host immunity. This includes neuronal RIPK3-mediated expression of inflammatory chemokines including CCL2 and CXCL10 following WNV infection, which act in a cell-extrinsic fashion to recruit inflammatory monocytes and WNV-specific CTLs to the CNS that are critical for viral clearance (52). RIPK3 can also function in a cell-intrinsic manner by driving expression of the enzyme IRG1 in neurons during ZIKV infection, which promotes a protective antiviral metabolic state (53). It should be noted that both of these cases examine RIPK3 signaling in CNS neurons, a nonrenewing cell type located in tissue that has strict requirements for restraining excessive immunopathology; it is therefore likely that this cell type is particularly reliant on death-independent mechanisms of host defense. However, it appears that RIPK1- and RIPK3-mediated gene expression is primarily responsible for driving the immunogenicity of necroptotic signaling pathways in numerous cell types across a variety of disease contexts (63, 66, 123), highlighting the fact that de novo transcription and translation associated with this PCD pathway can promote the immunogenic effects of necroptosis independently of MLKL-mediated DAMP release.

Considering that several upstream PRRs capable of initiating pyroptosis can sense viral PAMPs or pathogenic activities, inflammasome activation is a well-established response to viral infection (124). A diverse array of both DNA and RNA viruses can induce inflammasome activation, including IAV and respiratory syncytial virus activation of NLRP3 (125–127), and HSV-1 and MCMV activation of AIM2 (128, 129). Not surprisingly, the extent to which pyroptotic signaling promotes overall host health depends on the viral infection model. Indeed, inflammasome activation can be protective during IAV or WNV infection, where Casp1^−/−Casp11^−/− mice exhibit increased mortality (130–132). However, during HIV-1 infection, the DNA sensor IFI16 detects viral DNA intermediates and can trigger pyroptosis in CD4⁺ T cells (133), contributing to the lymphocyte depletion that drives patient progression to AIDS. Beyond the elimination of necessary cell types, pyroptosis can also contribute to pathological outcomes, as Nlrp3^−/− or Gsdmd^−/− mice exhibit prolonged survival and decreased intestinal inflammation during the course of murine norovirus infection (134). These studies highlight the important balance that must be struck between producing sufficient inflammatory signals to stimulate immune responses and limiting tissue immunopathology and resultant lethality.

Similar to how necroptotic cell contributions to host immunity can be parsed into lytic cell death and cytokine or chemokine production, pyroptosis produces both DAMPs through cytolysis as well as IL-1 family cytokines that can instruct downstream immune responses. In some cases, IL-1 and IL-18 production alone does not fully account for host protection: For example, Casp1^−/−Casp11^−/− mice exhibited greater mortality and higher clinical disease scoring compared to Il1b^−/−Il18^−/− or wild-type controls following CNS infection with an attenuated strain of rabies virus (135), implying that IL-1 family–independent functions of caspase-1/−11 were required for host protection. The recent identification of GSDM-D as the executioner molecule of pyroptosis and the generation of Gsdmd^−/− mouse lines will help attribute various host immune responses to either IL-1 family cytokine production or cell death, analogous to the usage of Ripk3^−/− and Mlkl^−/− mice to separate the gene regulatory and cell death arms of necroptosis.

4.2. Bacteria

Unlike viruses, bacteria can exist as obligate or facultative intracellular pathogens or circumvent intracellular replication entirely and replicate as extracellular microorganisms. Cell death is therefore thought to promote host defense particularly during intracellular phases of infection, again illustrating the importance of PCD in eliminating the propagation compartment for intracellular bacteria. As with viral infection, PCD also plays an important role in influencing the uptake and cross presentation of bacterial antigens by APCs that, in conjunction with the specific cytokine milieu derived from dying cells, can orchestrate the generation of protective adaptive immune responses against bacteria.

Apoptotic signaling can promote host defense against bacterial infection through a number of mechanisms. For intracellular bacteria such as Mycobacterium tuberculosis, apoptosis of infected alveolar macrophages within granulomas is thought to prevent bacterial replication within its niche of immature phagosomes, thereby suppressing M. tuberculosis proliferation (136). The dissociation of infected cells such as macrophages into apoptotic bodies can sequester intracellular pathogens such as Salmonella enterica serovar Typhimurium, Listeria monocytogenes, or M. tuberculosis within vesicles that can then be easily phagocytosed, digested, and processed for antigen cross presentation to better stimulate adaptive antibacterial immunity (137–139). Indeed, the importance of apoptosis in eliminating many bacterial infections is demonstrated through the bacterial evolution of diverse apoptosis evasion strategies. This includes secreted effector proteins whose sole purpose is to block apoptotic signaling, such as Legionella pneumophila SidF, which inhibits BNIP3 and Bcl-rambo to inhibit intrinsic apoptosis (140), and Chlamydia trachomatis chlamydial protease-like activity factor (CPAF) (141, 142), which cleaves and deactivates proapoptotic Bim, Puma, and Bad. Interestingly, cytosolic Shigella flexneri inhibits caspase-3 activity through direct binding to its LPS O antigen moiety to suppress caspase activity, an inhibitory mechanism that is essential for S. flexneri virulence in animal models (143). This case reveals a novel method of effector-independent apoptosis inhibition, where a bacterial PAMP that typically triggers inflammatory PRR signaling surprisingly mediates evasion of innate immune signaling and apoptosis. This potentially constitutes an underlying molecular mechanism of apoptosis inhibition that is shared by gram-negative bacteria including Rickettsia rickettsii (144) and Francisella tularensis (145), which share similar O antigen structure and evasion of apoptosis without the expression of clearly defined antiapoptotic effectors. Whether through the expression of virulence factors or caspase inhibition through O antigen binding, blockade of apoptosis preserves the bacterial ability to propagate intracellularly, highlighting a protective function of apoptosis in certain types of bacterial infections.

Inflammasome activation and subsequent pyroptosis can be triggered by a wide array of bacteria-associated PAMPs or virulence activities, making this a key defense pathway in the context of bacterial infection. This is demonstrated through the susceptibility of Casp1^−/−Casp11^−/− mice to a vast array of bacteria, including Citrobacter rodentium (146), Legionella spp. (147), S. flexneri (148), and S. Typhimurium (149, 150). The Yersinia genus of bacteria provides an interesting case study in PCD subversion by host-adapted pathogens, as Yersinia spp. express a sophisticated repertoire of secreted virulence effectors termed Yersinia outer proteins (Yops) that act as countermeasures to evade host detection (151, 152). Yops interfere with a wide variety of host innate immune signaling mechanisms, including evasion of inflammasome activation and canonical apoptosis through the effector YopK, which regulates the translocation of proteins through the type III secretion system (T3SS). This includes T3SS components YopB and YopD, which can activate NLRP3 in the absence of YopK (153, 154). YopM suppresses inflammasome activation through a more direct mechanism of caspase-1 inhibition, through either direct binding to caspase-1 or interactions with IQGAP1, a GTPase involved in inflammasome activation (155, 156). Additionally, the Y. pseudotuberculosis and Y. pestis effector YopJ (YopP in Y. enterocolitica) blocks NF-κB and MAPK signaling in infected host cells, which results in a blockade of proinflammatory cytokine production and induction of cell death (157–160). Interestingly, in response to YopJ-mediated blockade of innate immune activation, Yersinia spp.–infected cells undergo caspase-1 cleavage and cell death downstream of RIPK1/FADD/caspase-8 complex activation, which effectively overrides blockade of innate immune pathways by Yersinia spp. (161, 162). Yersinia spp. are therefore an interesting model to study interactions among host PCD pathways and bacterial pathogenesis mechanisms, as extensive coevolution with humans has presumably led to the induction of hybrid forms of PCD with properties of both apoptosis (caspase-8 activation) and pyroptosis (caspase-1 activation and IL-1 family cytokine secretion) as a result.

As Yersinia spp. are host-adapted human pathogens, it follows that this serves as a prime example of the evolutionary arms race between pyroptotic signaling and bacterial effector proteins to subvert these pathways. Conversely, the importance of inflammasome activation and pyroptosis as a first line of host defense against bacterial pathogens is demonstrated by opportunistic infection with the environmental bacterium Chromobacterium violaceum. C. violaceum is a ubiquitous environmental pathogen that is only known to infect humans with immunodeficiency disorders (163). Similarly, rodent models of C. violaceum infection are highly resistant to the pathogen (164). NLRC4 is required for this resistance, as Nlrc4^−/− mice are dramatically susceptible to C. violaceum–induced lethality following even very low-level infection. This is due to both cell-intrinsic deletion of infected splenic macrophages as well as IL-18-dependent priming of cytotoxic NK cells in the liver to kill infected targets through cell-extrinsic mechanisms. This finding implies that inflammasome activation and pyroptosis may be engaged in our bodies continually to defend us against bacterial infections that are acquired environmentally and that bacterial species that have not yet evolved strategies to subvert pyroptosis are rapidly cleared without any establishment of productive infection.

Unlike apoptosis and pyroptosis, the importance of necroptosis in host defense from bacterial infection is less clearly defined. Ripk3^−/− mice do not have any overt survival defects following infection with several extensively studied bacterial species, including S. Typhimurium, Y. pseudotuberculosis, C. rodentium, and Staphylococcus aureus (161, 165–167). As discussed above, numerous bacteria encode caspase inhibitor virulence factors, which should theoretically sensitize cells to necroptosis. Indeed, enteropathogenic Escherichia coli (EPEC) expresses a type III effector, NleB1, that efficiently antagonized DR-mediated apoptosis via FADD binding, an event predicted to sensitize cells to necroptosis (168). However, subsequent studies also identified EspL, another EPEC type III effector that specifically cleaves the RHIM domains of RIPK1, RIPK3, ZBP1, and TRIF (169). Notably, numerous other bacterial species encode orthologs of EspL, raising the possibility that inactivation of necroptotic signaling is a common feature of many pathogenic bacteria. Thus, the relative paucity of studies of necroptosis as an antibacterial event may be due to widespread subversion of this pathway by pathogenic bacteria.

In summary, pathogenic infection can elicit immunological outputs associated with multiple forms of PCD, each of which can have distinct effects on downstream immune responses. Any type of PCD succeeds in eliminating the replicative niche for intracellular pathogens, but these modalities can vary with respect to the inflammatory signals that accompany the death of infected cells and must be tightly regulated to avoid detrimental immunopathology. The important role of PCD in shaping host-pathogen interactions over the course of metazoan evolution is evidenced by the repertoire of both viral and bacterial virulence factors that inhibit cell death signaling components, indicating that modulation of PCD constitutes a key aspect of pathogenesis. Additionally, the fact that many pathogens can engage networks of multiple cell death pathways indicates how distinct forms of PCD signaling synergize to constrain pathogen replication while concomitantly stimulating pathogen-specific adaptive immune responses required for pathogen clearance.

5. FUTURE PERSPECTIVES

5.1. Therapeutic Applications of Programmed Cell Death

The examples provided above highlight the interconnected nature of PCD signaling pathways in the context of infection, as well as the significant understanding that has emerged regarding the immune responses elicited by dying cells in this context. These findings hold promising implications in the design of novel therapeutics for the treatment of numerous diseases, ranging from infectious disease and inflammatory autoimmune disorders to cancer. Notably, our understanding of the consequences of PCD has directly informed efforts to adapt cell death inducers for therapeutic use in cancer immunotherapy. The goal of cancer immunotherapies is to engage cytotoxic immune responses to kill cancer cells; as these responses evolved to fight intracellular pathogens—the same pathogens controlled by defensive cell suicide—therapeutic activation of these same cell death and innate immune pathways in cancer cells may provide immunotherapeutic benefit (66). Furthermore, dying tumor cells are the key source of tumor-associated antigens by which the immune system may recognize transformed cells. It therefore stands to reason that strategies to reconstitute PCD signaling to eliminate transformed cells while optimally alerting the immune system to the presence of a tumor are an important area of tumor immunotherapy research.

Efforts toward manipulating PCD to this effect include targeting of caspase-independent cell death (CICD) in tumor cells, which occurs upon inhibition of executioner caspases downstream of MOMP, leading to inflammatory cytokine secretion via NF-κB (170) and type I IFN production via cGAS-STING-mediated recognition of mitochondrial DNA (30, 171). CICD-induced inflammation enhanced tumor control, driven by inflammatory macrophage polarization and T lymphocyte activation. The successful manipulation of immunogenic cell death in tumors is not limited to apoptosis, as ectopic activation of RIPK3 or MLKL in the tumor microenvironment has been shown to stimulate RIPK1- and NF-κB-dependent tumor suppression accompanied by increased activation of tumor APCs and tumor-specific CTLs (123, 172). Additionally, expression of the pyroptosis executioner GSDM-E in tumor cells leads to increased tumor infiltration by leukocytes and improved CTL- and NK cell–mediated tumor suppression (173). These beneficial antitumor effects of pyroptotic signaling can be targeted therapeutically, as GSDM-A-conjugated nanoparticles demonstrated similar mechanisms of tumor control (174). In these cases of RIPK3 activation and GSDM-A delivery, both treatments synergized robustly with the checkpoint inhibitor α-PD-1 to promote tumor clearance, suggesting that manipulation of immunogenic cell death signals in the tumor microenvironment could prove to be an effective agent for coadministration with existing tumor immunotherapies.

While manipulation of the immunogenicity of cell death during cancer treatment holds promise, clinical deployment of such inflammatory treatments will need to be tightly regulated to avoid off-target effects such as cytokine release syndrome and autoimmunity that occur in some cancer patients following chimeric antigen receptor (CAR) T cell therapy. Interestingly, two recent reports indicate that CTL-dependent killing of target cells via granzyme A or B delivery can respectively engage GSDM-B or -E, rather than (or in addition to) canonical apoptosis (175, 176). Activation of the granzyme B–GSDM-E axis by CAR T cells was found to drive pathogenic cytokine release syndrome, while the granzyme A–GSDM-B pathway could promote immune-mediated tumor clearance. These findings, respectively demonstrating pathogenic and beneficial engagement of immune responses by immunogenic cell death, highlight the promise and challenge of developing analogous therapies for clinical use in the treatment of cancer and other diseases.

6. CONCLUSIONS

Over the last two decades, the study of PCD has been transformed by the description of non-apoptotic cell death programs such as pyroptosis and necroptosis and by the discovery that these cell death programs are mechanistically linked. As a result, it is now appropriate to consider PCD signaling as a network of connected programs, capable of compensating for inhibition of one program with the activation of another. These observations open new questions related to the immunological consequences of cell death, as we must now consider the effects not only of canonical apoptosis, pyroptosis, or necroptosis in vivo but also of hybrid cell death modalities. For example, how do immune cells perceive a cell undergoing apoptosis with or without activation of GSDME, or necroptosis that does or does not concurrently engage NLRP3-dependent inflammasome activation? As the answers to these questions are likely to be highly context dependent, they will need to be addressed in specific tissue settings and disease models. The task of understanding these questions, and of developing therapeutics that target the activation or selective inhibition of this signaling in the design of cutting-edge immunotherapies, is an exciting new frontier for the cell death field.