A motive for killing: effector functions of regulated lytic cell death

Abstract

Immunological responses activated by pathogen recognition come in many guises. The proliferation, differentiation and recruitment of immune cells, and the production of inflammatory cytokines and chemokines are central to lifelong immunity. Cell death serves as a key function in the resolution of innate and adaptive immune responses. It also coordinates cell-intrinsic effector functions to restrict infection. Necrosis was formally considered a passive form of cell death or a consequence of pathogen virulence factor expression, and necrotic tissue is frequently associated with infection. However, there is now emerging evidence that points to a role for regulated forms of necrosis, such as pyroptosis and necroptosis, driving inflammation and shaping the immune response.

INTRODUCTION

There are several types of regulated necrosis, including caspase-1-mediated pyroptosis, caspase-4/11-driven pyroptosis and mixed lineage kinase domain-like/receptor interacting serine/threonine kinase 3 (MLKL/RIPK3)-dependent necroptosis. Each pathway can be distinguished at the genetic, biochemical, and cellular level. However, extricating the functional consequences of controlled cell death from the effects of immunomodulatory cytokines, such as IL-1α, IL-1β, IL-18, and IL-33 that can be released as part of these biochemical pathways, continues to be a major challenge in vivo. The development of new methods to detect pathway activation will help define the importance of their contribution to disease in vivo. In this review, we discuss recent literature demonstrating that regulated forms of apoptotic and nonapoptotic cell death serve as cell-intrinsic effector functions to kill intracellular and extracellular microbes, while also stimulating inflammation via the release of endogenous host molecules called DAMPs (damage-associated molecular patterns or danger-associated molecular patterns).

CASPASE-1 AND INFLAMMASOME-DEPENDENT PYROPTOSIS

Pyroptosis is a nonapoptotic form of cell death originally defined by the presence of caspase-1 activation.¹ Activation of caspase-4 in humans, or caspase-11 in mice, also leads to death associated with the release of highly inflammatory mediators, consistent with ‘pyro’-ptosis (Figure 1).^2–6 However, the precise cellular, biochemical, and structural features of this process appears to differ from caspase-1-dependent pyroptotic death. Additionally, caspase-4/11 cell death may not rely on a multiprotein cytoplasmic inflammasome complex as originally defined by Martinon and colleagues.^2–8 Therefore, the relationship between caspase-1-dependent pyroptosis and caspase-4/11-mediated cell death needs further examination.

Figure 1.

Immunomodulatory pathways of apoptotic and pyroptotic cell death. A full color version of this figure is available online at the Immunology and Cell Biology website.

Pyroptosis elicits a strong inflammatory response, in part due to the processing of pro-IL-1β and pro-IL-18 by caspase-1 to their bioactive forms.⁹ Defining cell-intrinsic effector functions that promote inflammation, beyond the release of these inflammatory cytokines, has been challenging. Many studies aimed at elucidating the function of caspase-1 have investigated the physiology and immunology of caspase-1-deficient mice. Reports have also examined biochemical readouts of caspase-1 and IL-1β processing; however, there have been few experiments outlining the effects on cell viability. As a result, we have an incomplete understanding of the biochemistry and cell biology of caspase-1-dependent cell death. From a limited number of studies, we know that caspase-1-dependent cell death can be a distinct process from IL-1β processing,¹⁰ which may reduce pathogen load,¹¹ and when triggered by non-infectious stimuli such as chemotherapy, has the capacity to kill hematopoietic stem and progenitor cells.¹² Additionally, when activated systemically, the NLRP1a inflammasome can deplete hematopoietic cells, resulting in cytopenia and immunosuppression.¹²

Early studies demonstrate that active caspase-1 can localize on the external surface of the plasma membrane, often in clusters on microvilli, but without signs of membrane disruption or loss of cell viability.¹³ Processed IL-1β can also be detected on the cell surface of monocytes with intact lipid bilayers and normal nuclear architecture. As illustrated in Singer et al.,¹³ the processing of IL-1β at the plasma membrane occurred in viable monocytes, suggesting that subsequent caspase-1-dependent loss of cell viability may be a late event in the activation of the inflammasome. Salmonella typimurium-induced macrophage pyroptosis involves cell swelling, poly ADP ribose polymerase (PARP) activation, and inhibitor of caspase-activated DNase (ICAD)-independent DNA fragmentation.^11,14 However, PARP cleavage and DNA cleavage are not required for pyroptosis. Pore size in pyroptotic cells is estimated at 1.1–2.4 nm, but these data are based on an indirect assay using osmoprotectants; of note, this method has the potential to interfere with other processes leading to inflammasome assembly and caspase-1 activation.¹⁴ More work is needed to determine the relationship between caspase-1-dependent death and cytokine processing and the extent to which these events are influenced by the type of inflammasome engaged and the cell type involved.

RECOGNITION OF INTRACELLULAR BACTERIAL INFECTION

The induction of cell death upon intracellular infection is a fundamental component of the innate immune response that is conserved across kingdoms.^15–17 In plants, the induction of cell death forms a dominant part of the innate immune response. In mammals, we now recognize and will summarize, a number of distinct biochemical pathways that can be activated during infection, leading to the loss of membrane integrity and cell viability.

When Gram-negative bacteria escape the phagosome and enter the cytoplasm, lipopolysaccharides (LPS) trigger activation of caspase-4 in humans and caspase-11 in mice.² Activation of caspase-4/11 can engage the NLRP3 inflammasome to drive caspase-1-dependent processing of IL-1β and IL-18.^4,18 Binding of caspase-4 to phosphate groups in the lipid A moiety of LPS occurs via basic residues in the caspase activation and recruitment domain (CARD domain). Introduction of K19E, K52E/R53E/W54A or K62E/K63E/K64E mutations in the CARD domain of caspase-4 abrogated binding of LPS.⁶ A caspase-1 chimeric protein containing the caspase-11 CARD domain sensitized cells to pyroptosis induced by cytoplasmic LPS.⁶ However, a crystal structure has yet to be reported demonstrating this structural interaction. Forms of LPS that fail to induce macrophage pyroptosis also fail to induce oligomerization of caspase-4/11. It remains to be determined if other LPS-binding cofactors participate in presentation of LPS to the inflammatory caspases. While caspase-5 is reported to be a redundant regulator of responses to cytoplasmic LPS,⁶ two subsequent studies using caspase-5-deficient myeloid cell lines failed to confirm a role for caspase-5 in responses to LPS.^18,19 Thus, while the activation of caspase-4/11 by LPS is unanimously supported by the literature, a role for caspase-5 still remains unclear.

CASPASE-4 AND CASPASE-11 ACTIVATE GASDERMINS TO KILL

Gram-negative bacteria in the cytoplasm cannot be sensed by Toll-like receptors (TLRs), which are restricted to the plasma membrane and endosomes. Outer membrane vesicles (OMV) are actively secreted from Gram-negative bacteria, such as Shigella and Burkholderia, or by vesiculating bacteria including: Escherichia coli, Vibrio cholera, Pseudomonas aeruginosa, and Neisseria meningitides as part of symbiotic and pathogenic interactions with mammalian hosts and other microorganisms.^20–22 LPS within OMV can be internalized by host cells via clathrin-mediated endocytosis and then released from early endosomes to engage caspase-4/11 in the cytoplasm.²³ Detection of cytoplasmic LPS by caspase-4/11 enables mammalian cells to mount an inflammatory response to this intracellular threat.

Caspase-4/11 activation triggers the oligomerization of gasdermin D (GSDMD) by cleaving human and mouse GSDMD at Asp275 and Asp276, respectively, to release an autoinhibitory C-terminal fragment. The resulting cleavage product (GSDMD-NT) oligomerizes to form 10–16 nm pores containing 16–24 protomers of the p30 fragment.^24–26 Reducing the concentration of GSDMD reduces the number of pores, but not the pore size. A GSDMD mutation has been described that impairs oligomerization but not recruitment of the protomers to the membrane, thus resulting in a reduction in pore number.²⁵ Caspase-4/11 triggers oligomerization of GSDMD and kills cells without involvement of nucleotide-binding domain and leucine-rich repeat (NLR) receptors or the adaptor protein apoptosis-associated speck-like protein containing a CARD (ASC),² thus it is questionable whether this complex constitutes any form of canonical or non-canonical inflammasome.

Other members of the gasdermin family include GSDMA, GSDMB, GSDMC, DFNA5 and DFNB59 in humans, and GSDMA1–3 and GSDMC1–4 in mice. The N-terminal fragments of GSDMD, GSDMA, and GSDMA3 exhibit pore-forming capacity but only GSDMD appears to be a substrate of caspase-4/11.^24,25 The stimuli driving activation of other gasdermin family members are not known. Mutation of four basic residues predicted to be in amphipathic α-helices of GSDMD, Arg138, Lys146, Arg152, and Arg154 prevented oligomerization, membrane binding, pore formation and pyroptosis.²⁴

GSDMD-NT binds mono-, bis- and tris-phosphorylated phosphatidylinositols (PtdIns(3)P, PtdIns(4)P, PtdIns(5)P, PtdIns(3,4)P₂, PtdIns(3,5)P₂, PtdIns(4,5)P₂, and PtdIns(3,4,5)P₃), phosphatidic acid, and phosphatidylserine, all of which are usually localized to the inner membrane of mammalian cells.^24–28 GSDMD-NT binding specificity has also been described for cardiolipin, which is found in bacterial cell membranes.^24,25,29 In contrast to GSDMD-NT, perforin kills cells from the exterior, consistent with its lipid-binding preference to phosphatidylethanolamine (PE), found on the outer leaflet of the membrane.^30,31 The inner membrane lipid-binding preference of GSDMD-NT enables specificity of killing. Consequently, surrounding mammalian cells would remain unaffected but the capacity to lyse extracellular bacteria would persist. The lipid composition of the phagosome and endosome membranes makes them a possible target of GSDMD-NT, in addition to the bacteria confined to these vacuoles. Whether extracellular GSDMD-NT can target apoptotic cells with externalized phosphatidylserine has not been addressed. These elegant studies provide evidence for cell-intrinsic effector functions of GSDMD-mediated pyroptosis in the face of intracellular and extracellular bacterial infection. Further work is needed to define the cell types that activate GSDMD, the role of extracellular inhibitors, and the mechanisms controlling its half-life.

CASPASE-1 ACTIVATES GSDMD

Caspase-1 can cleave GSDMD and is hypothesized to form GSDMD pores that control the release of IL-1β and the induction of pyroptotic death.³² Primary Gsdmd^−/− or Gsdmd^I05N/I105N bone marrow-derived macrophages (BMDM) are resistant to pyroptosis and produce less IL-1β when treated with ATP (NLRP3), poly(dA:dT) (AIM2), or TcdB (pyrin).³² GSDMD deficiency also protects macrophages from the cytopathic effects of nigericin stimulation.³³ However, loss of GSDMD does not fully protect BMDM from cell death. Therefore, it is possible that an alternative pathway such as NLRP3-dependent, caspase-8-mediated signaling may kill GSDMD-deficient BMDM.^34,35 More studies are required to investigate if GSDMD forms equal numbers and sizes of pores in response to canonical inflammasome stimuli.

It is not yet clear whether NLRP3-dependent and caspase-1-dependent GSDMD cleavage can trigger IL-1β release without pyroptotic death. However, our current understanding of GSDMD pore formation provides a clear mechanism for IL-1β and IL-18 secretion from viable cells and a framework to determine if IL-1β secretion can be completely uncoupled from subsequent loss of cell viability. For example, neutrophils produce IL-1β in response to caspase-1 activation but are resistant to caspase-1-induced cell death.³⁶ It remains unknown if neutrophils can also cleave GSDMD or if they respond to cytoplasmic LPS to induce GSDMD pores. Experiments in GSDMD- or caspase-11-deficient neutrophils will be needed to clarify the role of these proteins in neutrophil proinflammatory death.

MLKL: MEMBRANE-DISRUPTING OR PORE-FORMING PROTEIN?

Mixed lineage kinase domain-like (MLKL) mediates nonapoptotic cell death, specifically necroptosis, by membrane disruption. Phosphorylation of MLKL by RIPK3 at S345, S347, and T349 is followed by the translocation of MLKL to the membrane.³⁷ Translocation of MLKL to the membrane itself is not sufficient to cause cell death, therefore suggesting spontaneous assembly of an MLKL-dependent membrane-disrupting complex does not occur. The data suggest that additional events coordinated by MLKL or partner molecules are likely essential for MLKL-dependent cytotoxicity.³⁷ The proposed trimeric or hexameric pore in the membrane is yet to be visualized; however, structural information on the four helical bundle is available.^38–43

MLKL binds a similar lipid profile as GSDMD, interacting with PI (3)P, PI(4)P, PI(5)P, PI(4,5)P₂, PI(3,4,5)P₃, PI(3,4)P₂, PE(3,5)P₂ and cardiolipin, but only weakly associates with phosphatidylserine, phosphatidylglycerol, sulfatide, and phosphatidic acid.^38,40,44 Competitive PIP binding assays using overexpression of the PIP-binding pleckstrin homology (PH) domain-containing proteins phospholipase C δ (PLCδ), which binds PI(4,5)P₂, or Bruton’s Tyrosine Kinase (BTK), which binds PI(3,4,5)P₃, abrogated induction of necroptosis by MLKL.³⁸ This overlapping lipid-binding profile of GSDMD and MLKL suggests that MLKL may also have the capacity to lyse intracellular and extracellular bacteria via interactions with cardiolipin in bacterial membranes. Thus, in addition to the ability of necroptotic cells to release DAMPs to induce IL-1α and IL-33, this mode of cell death may also directly contribute to the killing of pathogens.^45,46

NEUTROPHIL EXTRACELLULAR TRAPS: AN EFFECTOR FUNCTION OF MLKL ACTIVATION?

Neutrophil extracellular traps (NETs) are networks of double stranded DNA, histones, and microbicidal proteins released from neutrophils to trap and kill pathogens.^47,48 NETs can restrict the replication of diverse pathogens including Staphylococcus aureus, Salmonella typhimurium, Streptococcus pneumonia, Shigella flexneri, and Candida albicans.^48–52 Systemic release of NETs, or the chronic release and prolonged presence of NETs, has been implicated in the pathogenesis of sepsis, systemic lupus erythematosus,⁵³ rheumatoid arthritis,⁵⁴ diabetes,⁵⁵ acute lung injury,⁵⁶ and thrombosis.^57–59 Neutrophil necroptosis may contribute to the pathogenesis of some of these infectious and inflammatory diseases. Both necroptosis and NETosis require the inhibition of apoptotic caspases to proceed, and they both result in a loss of plasma membrane integrity and the release of DAMPs. Using uric acid, LPS, and phorbol 12-myristate 13-acetate (PMA) to trigger NETs, two groups provide opposing data on the role of RIPK3 and MLKL in the generation of NETs.^60,61 The use of necrostatin-1 in the context of NET generation is also problematic because of off-target effects on indoleamine-2,3-dioxygenase (IDO).⁶² Therefore, additional experiments using alternate stimuli known to specifically engage RIPK3 and MLKL will be needed to determine if NET formation is an effector function of the necroptosis machinery.

CELL DEATH TO ELIMINATE INTRACELLULAR PATHOGENS

In mammals, apoptotic and nonapoptotic cell death pathways have fundamental roles in host-pathogen interactions. To counter the immunoregulatory roles of cell death and the cytokines elicited by regulated cell death, pathogens have evolved numerous mechanisms to suppress these pathways. The Myxoma virus pyrin domain-containing M13L protein inhibits caspase-1 processing and IL-1β production through an interaction with ASC.⁶³ It is unclear if pyroptosis has an effector role in restricting viral replication given the dual role of M13 in regulating NFκB and caspase-1-dependent processing of IL-1β.⁶⁴ However, in cell cultures of rabbit fibroblasts and lymphocytes, loss of M13L suppresses viral titers.⁶⁴ These data suggest that caspase-1-dependent pyroptosis may also have an effector function in the pathway to prevent viral replication and viral dissemination.

Death receptor-induced apoptosis of infected cells is a key antimicrobial defense strategy for enteropathogenic bacteria.^65,66 Mice deficient in Fas are highly sensitive to Citrobacter rodentium, an enteropathogenic-like mouse pathogen. The bacterial type III secretion system of enteropathogenic bacteria can deliver virulence factors to the host cytosol, including proteins that interfere with the formation of the death-inducing signaling complex (DISC). Modification of FADD by the N-acetylglucosamine transferase activity of NleB1 prevented apoptosis of cells, suggesting that bacteria target apoptosis to interfere with this antimicrobial host pathway.^65,66

Pseudomonas strains lacking the effector protein ExoU activate the NLRC4 inflammasome and trigger caspase-1-dependent death of BMDM.⁶⁷ Pseudomonas strains expressing ExoU can inhibit caspase-1 and drive caspase-1-independent cell death. In these strains, neutrophil elastase, but not NLRC4 or caspase-1, is responsible for IL-1β production in a mouse corneal model of Pseudomonas infection.⁶⁸ Like Pseudomonas, Legionella is an intracellular pathogen that activates the NLRC4 inflammasome, but when pyroptosis is not activated, it relies on expression of the pro-survival host protein BCL-XL to maintain host cell viability and enable bacterial replication.^69–71 Antagonists of BCL-XL, known as BH3 mimetics, trigger BAK/BAX-dependent apoptotic cell death. This induction of cell death eliminates intracellular Legionella and prevents fatal lung infection in mice.⁷¹ The extent to which apoptosis of Legionella-infected cells can amplify the inflammatory response, particularly in comparison with inflammatory forms of cell death such as NLRC4-dependent pyroptosis and necroptosis, remains to be determined.⁷⁰

Yersinia promotes pathogenesis by delivering virulence factors to the cytoplasm via a type III secretion system. The activity of these effector proteins can be recognized by the NLRP3, NLRC4, pyrin and NLRP12 inflammasomes, triggering cytokine processing and cell death. Yersinia outer proteins (Yop) YopE and YopT inactivate Rho GTPases, which trigger the formation of the pyrin inflammasome.⁷² YopM recruits ribosomal S6 kinase (RSK) and the serine/threonine protein kinase C-related kinases (PRK) PRK1 and PRK2 to phosphorylate 14-3-3 and prevent activation of the pyrin inflammasome.^72,73 Inflammasome activation and pyroptosis are also antagonized by YopK and JopJ.^74–78 It is therefore possible that gain-of-function mutations in PYRIN in patients with Familial Mediterranean Fever (FMF) or PYRIN-associated autoinflammation with neutrophilic dermatoses (PAAND) have been maintained in the population by conferring protection to Yersinia and other pathogens that inactivate the PYRIN inflammasome.^79–81

MLKL: A KILLER OR AN ACCOMPLICE?

There are many pathogens known to trigger alternate forms of cell death by impairing proapoptotic caspase function. For example, the cowpox serpin protein cytokine response modifier A (CrmA) inhibits caspase-1 and caspase-8.⁸² Mouse cytomegalovirus (MCMV) can inhibit both apoptosis and necroptosis by expressing viral inhibitor of caspase-8-induced apoptosis (vICA), an inhibitor of Caspase-8, and M45 (viral inhibitor of RIP activation; vIRA), a RHIM domain containing protein that binds to RIPK3.^83–85 MCMV mutants that are deficient in M45 (ΔM45), or with mutations in the RHIM domain, cannot inhibit RIPK3-dependent necroptosis and are avirulent.⁸⁵ Human cytomegalovirus is even more intriguing because it inhibits necroptosis downstream of RIPK3 activation and MLKL phosphorylation.⁸⁶ This therefore indicates the existence of a signaling network downstream of the MLKL phosphorylation event. This downstream signaling network may contribute to the membrane-disrupting capacity of MLKL or it may alter other cellular responses that contribute to the loss of cell viability.

The vaccinia virus F1L protein shares homology with Bcl-2 and can inhibit apoptosis.^87,88 F1L also suppresses NLRP1-induced IL-1β production.⁸⁹ Despite the ability of vaccinia virus to evade several cell death pathways, it fails to avoid the MLKL-dependent necroptosis pathway. This illustrates the benefits associated with the evolution of diverse biochemical pathways, all leading to cell death (Figure 1). Host-regulated cell death pathways therefore contribute to hostpathogen interactions, and they are targets of a diverse range of molecules deployed by pathogens. Pharmacological manipulation of these host cell death pathways may provide new opportunities to treat infection without exerting selective pressure on the pathogen.

SUMMARY

The activation of caspase-1-dependent, caspase-4/11-dependent and RIPK3/MLKL-dependent cell death pathways is inflammatory in vivo and thereby acts as an immunomodulatory event. Why these pathways have evolved to counter infection, how they are utilized in different innate immune cells, and whether they function beyond the indirect elicitation of high levels of inflammatory cytokines, has remained largely obscure. The study of how specific cell types have co-opted these pathways to serve key effector functions in resistance to infection, or aberrantly activate the pathway in other disease settings, will reveal new approaches to treat infection, kill cancer cells and provide novel targets for drug discovery.