Inflammasomes are potent regulators of innate immunity that act as the first line of defense in response to numerous damage-associated molecular patterns (DAMPs) and pathogen-associated molecular patterns (PAMPs), thereby shaping adaptive immunity (1). Canonical inflammasomes are supramolecular organizing centers (SMOCs) that comprise a hierarchical architecture, with the sensor, the apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC), and the executor (mainly caspase-1), leading to Gasdermin D (GSDMD)-mediated pyroptotic cell death and IL-1β and IL-18 secretion (2–4). The structures and functions of nucleotide-binding oligomerization domain (NOD)-like receptors (NLRs) such as NLRP3, NLRC4, NLRP1, and NLRP6, absent in melanoma 2 (AIM2)-like receptors (ALRs) and pyrin, have been reviewed elsewhere (5–10); they are common canonical platforms for the activation of caspase-1 and the maturation of its substrates. The non-canonical caspase-11/-4/-5 play pivotal roles in sensing intracellular LPS, inducing pyroptosis and non-canonical cytokine secretion in a caspase-1-dependent manner (11). Two downstream events, pyroptosis and cytokine secretion, serve as key weapons in maintaining homeostasis and protecting against infection, stress, and damage under various pathological conditions such as COVID-19, cancers, and Alzheimer’s disease (12–17). While pyroptosis and cytokine secretion may be usually observed as coupled events, evidence has shown that this scale can tip under the influence of inflammasome signaling with unclear mechanisms (18). This review aims to discuss recent advancements and provide insights into the pyroptosis-predominant and cytokine-predominant uncoupling propensities that lie downstream of inflammasome signaling. Overall, pyroptotic death and cytokine secretion are introduced, the association between ASC-mediated inflammasome assembly and caspase activation is traced, and the significance of these uncoupling preferences in physiological and pathological conditions is discussed. Our review will help future research in better understanding the inflammasome tilting toward pyroptosis or cytokine secretion.

Pyroptosis, a type of regulated cell death, is mainly mediated by gasdermin family members and is induced by transmembrane pore formation via the active N-terminal fragments, leading to cell swelling, plasma membrane rupture, and intracellular content release (19). Pyroptosis eliminates pathogen niches, exposes microbes, or maintains pathogens within pyroptotic corpses to propagate local inflammatory response (20–22). Pyroptotic pores and cell lysis also enable the release of certain intracellular cytokines and contents into the extracellular environment (23), facilitating the transferal of other immune sentinels to the point of injury or infection (24). Pyroptosis, therefore, plays a key role in homeostasis and host defense.

GSDMD, encoded by the gene GSDMD on chromosome 8q24.3, is the main executor of inflammasome-driven pyroptosis (25); it is widely expressed in various tissues (e.g., the colon, liver, and brain) and immune cells (26–28). GSDMD is exclusive to the mammalian genome (26). GSDMD-mediated pyroptosis has been observed in immune cells (29–31) such as macrophages and monocytes and in nonimmune cells (32–34) such as epithelial cells and osteoblasts. Upon inflammasome activation, canonical caspase-1 and noncanonical caspase-4/-5 (in humans)/-11 (in mice) are activated, cleaving GSDMD at D275 (FLTD|GVP in humans) or D276 (LLSD|GID in mice) and separating its active 31kDa N-terminal (GSDMD-NT) from its 22kDa autoinhibitory C-terminal fragments (35–37). GSDMD-NT interacts preferentially with negatively-charged membrane lipids, such as phosphatidylserine in the inner leaflet of the cell membrane and cardiolipin in the inner and outer leaflets of bacterial membranes; here, it oligomerizes into a ring-like structure that targets cell plasma, mitochondria, nucleus, and bacterial membranes, increasing the permeability and leading to intracellular architecture loss (38–40). The events between GSDMD cleavage and membrane pore formation could be controversial, because biophysical studies may have difficulties in determining whether GSDMD-NT monomers assemble into pore-forming oligomers before or after inserting into membranes. Recent evidence may suggest the latter, because membrane-associated GSDMD-NT monomers form oligomers when treated with ROS-inducing agents, while GTPases RagA and RagC are required in GSDMD-NT oligomerization but not in GSDMD-NT plasma membrane localization (41). The phospholipid phosphatase PtpB secreted from Mycobacterium tuberculosis (Mtb) can dephosphorylate cell membrane lipids to inhibit the membrane trafficking of GSDMD-NT (42). Therefore, membrane interaction and oligomerization are critical steps that are regulated by both the host and pathogens. GSDMD can also be cleaved by neutrophil elastase and cathepsin G to promote its pore-forming ability; it can be cleaved by caspase-3/-7 to generate an inactive p43 fragment and by caspase-8, which is driven by TAK1 inhibition in Yersinia infection at the same site as the inflammatory caspases in mouse macrophages (43–47). Other gasdermin family members such as GSDMA3, GSDMB, and GSDME can also form pores in cell membranes in an inflammasome-independent manner (48–50). Herein, we focus on the role that GSDMD plays in pyroptosis downstream of inflammasome signaling.

The exact structure and functionality of GSDMD pores remain elusive. GSDMA3, which is highly similar to GSDMD, organizes pore structures comprising 26–28 protomers that are 70 Å in height, with inner and outer diameters of 180 Å and 280 Å, respectively. These pores are likely to allow the cytokines IL-1β and IL-18, with diameters of around 4–8 nm (51, 52) to be released. GSDMA3 and GSDMD are thought to assemble prepores or prepore-like structures (arcs, slits, and rings) that form the functional larger pores (49, 53, 54). The 33 (varying between 31–34)-subunit GSDMD pore is characterized by a height of 80 Å and inner and outer diameters of 215 Å and 310 Å, respectively, and features a globular domain (more membrane-distal than GSDMA3 by approximately 16°) on the cytosolic side with a large transmembrane β-barrel (53). More precisely, four solvent-exposed acidic patches (AP1 and 4 in the globular domain, and AP2 and 3 in the β-barrel) are found near the conduit, contributing to negative potentials in the pore passage (53). GSDMD pore formation is regulated at different levels: autoinhibition, proteolytic cleavage, membrane lipid selectivity, and oligomerization (55–57). The presence of this conduit facilitates IL-1β and IL-18 secretion (discussed later) and ion passage, and the flux of K⁺ ions through these pores as a result of caspase-11 or AIM2 activation may act as a secondary signal for NLRP3 inflammasome activation (58–60). The flux of K⁺ ions through GSDMD pores in AIM2 inflammasome signaling may also restrict the response of the cGAS-dependent type I interferon (IFN) or promote the nonclassical release of IFN-β, indicating an additional role for GSDMD (61, 62). In contrast, the increased influx of Ca⁺ ions through the GSDMD pores may aid in membrane repair by recruiting endosomal sorting complexes required for transport (ESCRT) and removing damaged membranes that contain GSDMD pores from macrophages and HeLa cells (63). These results indicate a pro-inflammatory and pro-death role as well as the association of positive and negative feedback loops with GSDMD pores in inflammasome regulation.

Under the increased intracellular pressure that follows ion and water flux, the pyroptotic cell may undergo swelling, rupture, and lysis, allowing for the extracellular release of relatively larger intracellular components that can act as DAMPs. Lactate dehydrogenase (LDH, diameter 8–10 nm) release is therefore regarded as a classic readout of lytic cell death (64). However, pyroptotic cell death can be separated from cell lysis ( Figure 1 ). LDH release is inhibited by an anti-lytic agent glycine in bone marrow-derived macrophages (BMDMs), challenged by RodTox, an NLRC4 activator, and detected once the glycine is removed (65). In provoked BMDMs that express fluorophores, the loss of cytosolic tdTomato and GFP occurs 4–5 min after the first Sytox internalization, which can enter cells without intact plasma membranes and is interpreted as indicating the increased permeability that is initiated by GSDMD pore formation in this study. Glycine inhibits LDH leakage but cannot delay the loss of tdTomato or GFP, although their kinetics are slowed (65). More importantly, when Sytox enters the cell, cell movement stops, mitochondrial activity slows, and the cell starts to swell; these effects cannot be abolished by glycine (65). These results indicate that the GSDMD pore-mediated membrane permeability starts before cell lysis. Another independent study showed that mitochondrial membrane depolarization, Ca²⁺ internalization, phosphatidylserine exposure, cell swelling, and lysosome decay occur 18–21 min, 12–15 min, 9–12 min, 13 min, and 6–9 min, respectively, before PI or Sytox enter the cell (which is interpreted as cell membrane damage in this study) in NLRP1b signaling, and approximately 20 min, 6–9 min, 3 min, 13 min, and 6–9 min before PI or Sytox enter the cell in NLRC4 signaling (66). In contrast, nuclear rounding and condensation occur concurrently with the loss of cell membrane integrity (66). Despite the different interpretations, these results show that pyroptotic cell death closely coincides with GSDMD-mediated cell permeability, which occurs earlier than cell lysis, indicating that the inhibition of cell lysis may not block pyroptotic cell death. In line with these data, a recent study reported that deficiency of the cell-surface protein NINJ1 in BMDMs under LPS electroporation or nigericin treatment results in impaired pyroptotic cell membrane rupture and LDH release, while this deficiency does not significantly influence IL-1β secretion, suggesting that NINJ1 functions downstream of the GSDMD pores as LDH is only released during lysis (67). Ninj1^–/– macrophages die with ballooned morphology rather than a lytic appearance, indicating that pyroptotic cell death differs considerably from cell lysis (67). The separate events of GSDMD pore formation and cell membrane rupture are also evidenced by the differential release styles of intracellular contents; IL-1β, IL-18, and rhoGTPase Rac1 (5 nm) are secreted through GSDMD pores (often accompanied by detectable PI intake) and ruptured cell membranes, and HMGB1 (7.9 nm), caspase-1 heterotetramer (6.8 nm), and LDH leakage occur only as a result of cell lysis (52, 68). In contrast, one study found that glycine cannot inhibit the release of LDH or HMGB1 from pyroptotic THP-1 cells and that the swollen and unruptured cells release some pyroptotic contents such as IL-1β, LDH, and HMGB1, rendering the use of LDH and HMGB1 release as readouts for lysis suspicious (20). In addition, cell swelling could be an active process that is regulated by Ca²⁺ (from GSDMD pores and/or intracellular lysosomal leakage)-dependent calpain-driven vimentin cleavage and the consequent loss of intermediate filament (20). The cell rupture that occurs under the associated shear stress and compressive force enables the release of ASC specks and other macro-DAMPs (20). The authors further suggest that intracellular soluble microDAMPs (as large as 200 kDa) could escape through the GSDMD pores, while larger or nonsoluble macroDAMPs require cell rupture for release (20). Together, GSDMD pores could thus act as a filter that determines which “small” molecules can pass before cell lysis.

However, not all small molecules can be easily secreted through GSDMD pores. Pro-IL-1β, which is similar in size to IL-1β, is not secreted via GSDMD pores, indicating that GSDMD pores may act as filters that repel the precursor (53). This observation leads to the hypothesis that the negatively charged GSDMD pores preferentially allow the release of positively (e.g., IL-1β, which basifies during maturation) and neutral charged, as compared to negatively (e.g., pro-IL-1β, which has an acidic domain) charged, molecules. To support this assumption, mutations that diminish the negative potential of GSDMD pores markedly increase the release of negatively charged small dextrans (40 kDa) (53). Extrinsic factors (e.g., lipid and salt) may also affect the electrostatic environment of GSDMD pores and IL-1 transport (69). These data indicate that along with size, the charge of a cargo molecule is also important for passing through GSDMD pores. Notably, both anionic and cationic ultrasmall (<10 nm) nanoparticles could enter the pyroptotic macrophages through the GSDMD pores via passive diffusion and in a microtubule-independent way (70). GSDMD-dependent membrane perforation may also allow access for extracellular nanobodies targeting ASC to inhibit further inflammasome activity without affecting initial and pre-pyroptotic IL-1β secretion (71). These evidence may lead to the hypothesis that GSDMD pores may act as useful windows or passages for diagnostic and therapeutic strategies associated with extracellular drugs targeting pyroptotic cells. Collectively, these studies exhibit the discrepancies between active and passive swelling and rupture, and GSDMD pore-mediated and membrane rupture-mediated extracellular release of cell contents, exhibit the significance of different mechanisms of intracellular content release. Notably, the formation of GSDMD pores could occur on a sublytic level, avoiding cell rupture and LDH release while preserving IL-1β secretion (72).

Interesting questions, therefore, arise: what are the exact criteria by which “microDAMPs” and “macroDAMPs” are distinguished, and is the difference dependent on cell type (e.g., BMDMs versus THP-1 cells)? Is there a structural mode [e.g., the formation of large pores from GSDMD-NT prepores or prepore-like structures as mentioned previously, resembling the Bax pore formation on the mitochondrial outer membrane (73, 74)] resulting in early-stage ion-selectivity and late-stage non-selectivity in GSDMD pores, which may be independent of any subsequent cell membrane rupture? Would charge-based modification act as an effective strategy for developing new drugs? Further studies are required to explain the underlying mechanisms of GSDMD-mediated pyroptosis downstream of inflammasome signaling.

IL-1β, a potent pro-inflammatory cytokine in the IL-1 family, is induced mainly in myeloid cells and some non-myeloid cells (e.g., epithelial cells), but it is not always constitutively expressed (12, 75). Its precursor is inactive and needs processing to generate a bioactive mature form (76). Hence, the production of IL-1β requires two steps in most cells: a priming signal (e.g., TLR ligation) for precursor expression, and a processing procedure for maturation. Caspase-1, the main effector caspase in canonical inflammasomes, cleaves pro-IL-1β, removing the pro-peptides and creating the mature form, while caspase-4/-11 are ineffective and caspase-5 has a weak ability in processing the precursor (12, 77, 78). IL-1β may also be processed by non-caspase-1 proteases such as caspase-8, proteinase-3, elastase, granzyme A, matrix metalloprotease 9, or chymase (79). However, IL-1β lacks a signal peptide for its release into extracellular space via the classical ER/Golgi pathway (80–82). Such a poorly understood release mechanism is denoted “unconventional protein secretion (83)”, for which several differential mechanisms and multiple models have been proposed. Upon inflammasome activation, mature IL-1β is released, both from GSDMD pores and passively via pyroptotic lysis, which are regarded as non-vesicular pathways. Intracellular mature IL-1β is detected in murine macrophages that have been primed with LPS (1 μg/ml, 2 h) and treated with ATP for 5 min, along with a detectable level of extracellular IL-1β in the supernatant (84). The amount of mature IL-1β within the cell lysates subsequently decreases from this point, correlating with an increase in the supernatant (84). LDH release is significantly enhanced after 10 min treatment with ATP (approximately 5 min after extracellular IL-1β detection), indicating that IL-1β release is not just a non-specific process in cell lysis, but also occurs in dying cells before rupture, most likely through GSDMD pores as mentioned previously (84). Mature IL-1β can accumulate within the cytosol of GSDMD^-/- macrophages, further indicating that the IL-1β maturation and GSDMD pore formation could act as parallel events during inflammasome activation (25). Besides the non-vesicular model, IL-1β may also be secreted in a vesicular manner that is dependent on GSDMD but not on the formation of pores, although other vesicular secretion models such as secretory autophagy, microvesicle shedding, multivesicular bodies and exosomes, and secretory lysosomes have also been proposed (52, 75, 85). The different unconventional protein secretion pathways of IL-1β via plasma membrane pores or vesicular carriers are closely dependent on cell types and the type and strength of stimulus (76, 86). Metabolic condition in host cells may also affect the secretion route of IL-1β, with mechanisms of pore-formation and cell lysis in nutrient repletion versus autophagic capture and vesicle intermediate translocation during starvation and ER stress (87–89). Notably, the polybasic motif may help direct mature IL-1β to the inner face of the plasma membrane for colocalization with PIP2 (perhaps indirectly), and this IL-1β enrichment may act as a prerequisite for its subsequent release via both GSDMD pore-dependent and independent mechanisms (90). Interestingly, the discrepancies between GSDMD-associated pyroptosis and unconventional IL-1β secretion may contribute to their uncoupling downstream of inflammasome signaling.

IL-1α, another form of IL-1, may also be released during inflammasome activation. Unlike IL-1β, active IL-1α precursor is constitutively expressed in numerous cells of different organs such as the kidney, liver, and lung, and is processed by elastase, calpain, granzyme B, and thrombin into a cleaved form with increased activity and receptor affinity (91–94). Human recombinant pro-IL-1α is cleaved by caspase-5, but not by caspase-1/4 (95), and murine IL‐1α is not processed by murine caspase-1 (95). The release of cleaved IL-1α from human and murine macrophages primed with LPS and transfected with intracellular LPS requires the presence of caspase-5 and caspase-11, respectively (95). Caspase-1 may indirectly promote IL-1α release by cleaving cytosolic IL-1R2, which binds to pro-IL-1α to prevent cleavage by calpain, thus dislocating pro-IL-1α for further processing in macrophages (96). IL-1α may shuttle to the nucleus to act as a transcription factor regulating gene expression (e.g., IL-8), or bind to the cell membrane receptor complexes IL-1R1/IL-1R3, sharing a proinflammatory function with IL-1β (12, 97). Compared to IL-1β, IL-1α is more likely secreted as a key alarm in cell lysis during pyroptosis, necrosis, and necroptosis (85). IL-1α has also been detected on cell membranes, indicating a possible association with the inadvertent permeabilization and leakage of IL-1α within the cell, and cell surface IL-1α could be further cleaved and released (98). IL-1β and IL-1α, together termed IL-1, could promote innate immunity by inducing CXC- and CCL- chemokines (e.g., IL-8) for neutrophil recruitment (99–101), directly [e.g., in myocardial infarction (102)] or indirectly [e.g., by enhancing TH17 cell differentiation (103–105)] upregulating granulopoiesis and mature neutrophil release from the bone marrow, increasing the formation of neutrophil extracellular traps (NETs) for the trapping and killing of bacteria (106–108), and driving M1/M2 polarization and macrophage activation (109–111). Some of these mechanisms [e.g., NET formation (112) and M1/Th1 activation (113)] could further amplify IL-1 production. IL-1 signaling also plays a vital role in shaping adaptive immunity by increasing dendritic cell (DC) maturation and chemokine secretion (114–116), directly or indirectly promoting T cell expansion, differentiation, and survival (117–119), and enhancing Tfh-mediated B cell proliferation and antibody production (120–122). In addition, IL-1 is critical in maintaining the epithelial barrier as epithelial cells encounter various PAMPs and DAMPs under both physical and pathological conditions (123–125). These results render IL-1 an important pro-inflammatory regulator in infectious, autoinflammatory, and autoimmune diseases such as HIV (126), COVID-19 (127–129), cancers (130–132), diabetes (133–135), and rheumatoid arthritis (136).

Another IL-1 family member, pro-IL-18, is processed by caspase-1/-4 into its mature form during inflammasome activation and released via an unconventional secretion pathway (137). Like IL-1β, the IL-18 precursor could be cleaved by caspase-8 (138); however, unlike IL-1β, IL-18 is constitutively expressed in myeloid cells and epithelial cells, functions via IL-1R5 (IL-18 receptor alpha chain)/IL-1R7 (IL-18 receptor beta chain) ligation, and is antagonized by the IL-18–binding protein (139–141). The predominant cytokine downstream of inflammasome signaling could be either IL-1β (142), IL-18 (143), or a combination of the two (144) in a cell type- and context-dependent manner. IL-18 upregulates the expression of cell adhesion molecules (145), nitric oxide synthesis (146), cytokine and chemokine production (147–149), and IFNγ production in CD4+/CD8+ T cells and NK cells via combined action with IL-12 (150). These pro-inflammatory IL-18 effects contribute to the pathogenesis and development of diseases such as COVID-19 (151), inflammatory bowel disease (152), diabetes (153), and cancer (154–156). However, IL-18 also has a protective effect against the progression of age-related macular degeneration (157) and Alzheimer’s disease (158). Therefore, IL-18 downstream inflammasome signaling plays an important role in maintaining homeostasis and inducing inflammation.

Human IL-37, an anti-inflammatory IL-1 family member that has no homolog in mice or chimpanzees, has isoforms (b-e) that contain caspase-1 cleavage sites and enable maturation for increased activity compared to the active precursor without a classical signal peptide (79, 159). Active caspase-1 is required for the secretion of mature IL-37 and the nuclear translocation of intracellular IL-37 in NLRP3 signaling but is not required for the release of its precursors (160). Peritoneal macrophages from transgenic mice expressing native human IL-37 decrease the LPS-induced production of IL-6, IFNγ, TNFα, and IL-1β, along with suppressing the activation of NFκB and MAP kinase as compared to mice harboring caspase-1-uncleavable D20A mutant IL-37, further indicating an unexpected role of caspase-1 in limiting inflammation (161). IL-37 could ameliorate the inflammation in insulin resistance (162), allergic rhinitis (163), and asthma (164–166). However, its expression is upregulated in rheumatoid arthritis (167–169), ankylosing spondylitis (170), Grave’s disease (171), and systemic lupus erythematosus (172–174), with a relatively lower IL-37 level correlating with higher disease activity (175); this indicates the potential anti-inflammatory effect of IL-37 against inflammatory situations.

The maturation and secretion of cytokines downstream of inflammasome signaling, therefore, play an important role in regulating innate and adaptive immunity ( Figure 2 ). When considering the extracellular function, cytokine maturation and secretion could be uncoupled from pyroptosis in live and lytic cells. Since IL-1α and IL-37 precursors are active, the role of caspase-1 in pro-IL-1α cleavage could be indirect, and IL-18 may be preferentially produced in non-myeloid cells [e.g., bronchial epithelial cells (143), but PBMCs may act as an exception with large amount of IL-1β production (176)] following inflammasome activation,, this review, therefore, uses IL-1β secretion as a main readout of cytokine maturation and release (if not always), downstream of inflammasome signaling.

The two major functional outcomes in inflammasome signaling are pyroptosis and cytokine secretion, which may be usually observed as coupled events ( Figure 5 ). However, discrepancies among the diverse key steps in the signaling transduction include (1) caspase-1 processing, activation, and deactivation dynamics mediated by ASC specks (if not always) (2), GSDMD pore formation, pyroptotic cell death, and cell membrane rupture, and (3) IL-1β maturation and secretion in a GSDMD-dependent or -independent manner, or via cell lysis. These discrepancies may affect our understanding of the uncoupling events that occur downstream of inflammasome signaling. Besides possible misunderstandings, the underlying mechanisms of uncoupling remain elusive. Besides, the inflammasome sensor/ASC/caspase-1/GSDMD/IL-1β functions, the type and intensity of stimuli, the expression, and activation of SARM, other gatekeepers, participants, or executors of different levels, and the cell type and its microenvironment, all determine the occurrence of coupling or uncoupling and affect the direction in which specific uncoupling falls. Thus, subtle interrelations within inflammasome machinery and other non-inflammasome components exert significant effects on cell fate decisions. This raises an interesting question into why our immune system is so costly in terms of multilevel regulation of the coupling and uncoupling of pyroptosis and cytokine secretion, or why, how, and when a certain uncoupling appears to be preferred in some scenarios, since coupling is more commonly observed, at least in vitro, and is generally more rapid and robust when pyroptotic cells release DAMPs and IL-1β together as a massive inflammatory response.

There appear to be at least three reasons associated with uncoupling preferences. Firstly, hyperactivation and pyroptosis (even along with cytokine secretion) differ considerably in terms of host defense. Hyperactivation allows the living immune cells to add IL-1β and/or IL-18 into the cytokine repertoire, improving upon the repertoire of inflammatory mediators such as TNF-α and IL-6 in response to TLR agonists and resulting in prolonged pro-inflammatory effects on neighboring cells in situ, and even at distance via circulation. A well-characterized example is the salmonella infection in neutrophils. During acute salmonella-induced mouse peritonitis, resident macrophages secrete the first wave of IL-1β within 1 h post-infection, followed by rapid pyroptosis, and pyroptosis-resistant neutrophils are subsequently recruited as the main source of IL-1β 1–12 h post-infection (142). Neutrophil depletion increases the bacterial burden in mouse liver and spleen (142). These data suggest that the capacity of live neutrophils to fight pathogens is cytokine dependent. Additionally, hyperactive cells also contribute to shaping adaptive immunity by promoting T-cell differentiation and effector function (338). Hence, hyperactivation may initiate a modest and prolonged cytokine-mediated immunomodulatory effect, compared to the robust but relatively short (at least at a single-cell level) effects of pyroptosis coupled with cytokine secretion, and excessive hyperactivation may also be detrimental to the host defense (75, 350). In contrast, pyroptotic cell death, which can lead to massive inflammatory responses by exposing intracellular PAMPs and DAMPs, is so costly that dead cells cannot produce cytokine (359). The cost of inflammatory death at the single cell level may be compensated for by warning other cells to provide further protection against danger. However, the cost could be fatal at the host level; in some scenarios, cell pyroptosis (along with cytokine secretion) is a dangerous “signal” that is associated with in vivo lethality as compared to hyperactivation. LPS-primed mice all die from a second dose of LPS within 30 h, but sustain viability when challenged by oxPAPC or PGPC, indicating that stimuli (e.g., double LPS treatment) that trigger pyroptosis in vitro result in lethal sepsis in vivo, whereas the stimuli that induce hyperactivation in cells may promote inflammation without increased lethality in animal models (337). Sarm^-/- mice that are doubly challenged with LPS also exhibit an improved clinical score and survival compared to WT mice (356). These data help in further understanding two options: first, rapid death in a single cell or host to protect other cells and hosts and second, hyperactivation, wherein hyperactive cells secrete IL-1β and call for help from other cells to fight against DAMPs and PAMPs. Hence, the balance between benefit and cost needs to be carefully considered for the most appropriate decisions to be made in terms of cell/host fate.

Secondly, on a single cell level, hyperactivation enables delicate reactions to different stimuli at certain intensities in specific conditions. In terms of stimulus intensity, even the classic pyroptosis-inducer nigericin has been shown to promote hyperactivation in iBMDMs at relatively low doses (0.5 μM) as compared to higher doses (20 μM), affecting both IL-1β secretion and pyroptosis (53). In terms of stimuli type and microenvironmental context, oxPAPC alone (without priming) induces CD14 deficiency in cell membranes and inhibits its binding to LPS and the TLR4 signaling, whereas LPS-primed and oxPAPC-treated DCs become hyperactive (337). The former may imply sterile inflammation in which DAMPs alone are detected, and a less sensitive response to PAMPs is therefore subsequently required to avoid excessive autoinflammation; however, if the immune system detects PAMPs followed by DAMPs, a dangerous infection is implied, which results in inflammatory hyperactivation to cope with virulent pathogens and tissue damage. Notably, these differential reactions are also organized in a cell type- and species-dependent manner. oxPAPC induces robust IL-1β secretion from GMCSF-DCs, and to a lesser extent from FL3TL-DCs (338), and can also exert a cytotoxic effect on cDC1 but not cDC2, whereas PGPC does not induce the release of LDH in either population (338). These data suggest the differential effects of hyperactivating stimuli on DCs in different subgroups. In addition, as the proteins CD14 and TLR4 are important in LPS internalization, LPS alone activates the alternative inflammasome signaling in human monocytes that express abundant CD14, whereas IL-1β secretion is not detected in LPS-treated murine PBMCs or human macrophages and DCs with lower CD14 expression unless a second signal occurs to induce the simultaneous release of LDH and IL-1β (343, 346). It is conceivable for monocytes to be more sensitive than macrophages or DCs under hyperactivation as a result of the recruitment and differentiation of monocytes in the blood into macrophages and DCs in LPS concentration-rich bacterial infection sites; the rapid activation of live monocytes with IL-1β secretion could act as an acute response in the frontline against pathogens. Another cell type-dependent example is pyroptosis-resistant neutrophils. Human neutrophils, which have a relatively short life span (< 1 day), require timely replacement and efficient anti-bacterial actions, rendering anti-pyroptosis a considerate strategy (360). Neutrophils have lower GSDMD mRNA and p30 fragment expression, with less caspase-1 and ASC levels and smaller specks than macrophages and DCs, facilitating longer caspase-1 activity IL-1β secretion (12, 39, 52, 305, 352, 355). While macrophages undergo caspase-1-dependent pyroptosis to prevent intracellular S. Typhimurium replication and raise further anti-bacterial responses from neighboring cells (e.g., more effective killing by recruited neutrophils), caspase-1 deficiency has no effects on the intracellular bacterial burden of neutrophils, further suggesting that these cells may adopt a strategy other than pyroptotic lysis in canonical inflammasome signaling (142, 359). In contrast, one may expect that neutrophils committing inflammatory suicide utilize a rapid way to kill intracellular pathogens over the short-term instead of pyroptosis resistance. Indeed, pyroptosis-resistant neutrophils do not promote macrophage efferocytosis, which enables macrophages to engulf dying neutrophils, rendering pyroptosis resistance an unsatisfactory means of building an immune response (355). However, pyroptosis exemption could extend the lifespan of neutrophils, allowing for degranulation, reactive oxygen species (ROS) production, chasing and killing bacteria such as those released by pyroptotic macrophages, and recruiting more neutrophils to the infection site due to the consistent direct and indirect GSDMD-dependent IL-1β secretion (352, 361). While pyroptosis-resistant neutrophils secrete comparable levels of IL-1β when primed with LPS and treated with ATP as compared to macrophages, IL-1β release is further increased in neutrophils but significantly decreased in macrophages when cells are pre-treated with injured cell-derived supernatants or extracellular ATP, further suggesting the significance of neutrophils as an important IL-1β source in the danger signal-enriched milieu (355). Therefore, by regulating coupling and uncoupling within the cell population, a highly effective cooperation and amplification system can be organized with each component involved in both anti-infection and anti-danger functions. Notably, the exact role of GSDMD-NT, in classic NETOsis (which is supposed to trap extracellular pathogens) is contradictory; however, under cytosolic Gram-negative bacteria infection, caspase-11/GSDMD-mediated noncanonical NETosis would protect neutrophils from bacterial invasion and decrease the intracellular bacterial burden under GSDMD cleavage and IL-1β secretion (39, 43, 44). Furthermore, whether and how neutrophiles balance or “unbalance” pyroptosis and IL-1β release may be influenced by host species, readout timing, priming requirement, and notably the level of neutrophiles granules that may act as “sinks” to capture GSDMD-NT and avoid pyroptosis (349). Hence, these multi-layered anti-bacterial strategies allow neutrophils to behave versatilely in a context-dependent manner. Hyperactivation may also occur in nonimmune cells. Caspase-1 dimerization induces IL-1β release in viable immortalized mouse embryonic fibroblasts without rupturing the cell membrane (312). A possible explanation is that some nonimmune cells do not endogenously express inflammasome components for massive GSDMD pore formation (312). Likewise, pyroptosis-predominant uncoupling is also tailored to specific conditions where an inflammasome without an ASC focus is achieved under experimental conditions (e.g., NLRC4-caspase-1 complex in ASC^-/- macrophages), mutations (e.g., ASC^Y59A and ASC^E80R in NLRP3/AIM2/PYRIN signaling), and infections, and is also dependent on cell type (neutrophils are an exception with smaller ASC specks for p46 and p33/p10 species accumulation but with poor pyroptotic modalities and dynamics, as mentioned previously). Since different immune/nonimmune cells and their differential subgroups are responsible for multiple distinct and overlapping (if any) roles in building innate and shaping adaptive immunity, the alternative strategies of coupling or uncoupling are supposed to function in a specific manner. The fate decision is made not only on the single cell level but also in the cell population within the microenvironment, where different and associated cells cooperate to cope with intricate challenges, foreign or endogenous. Therefore, further in vivo studies are required to clarify how the cell players work together as a whole if different strategies are preferentially chosen by different subgroups.

Finally, uncoupling may be adopted by bacteria or host to fight against each other in different situations. From the bacterial perspective, as some Gasdermin family members are also found in fungi and bacteria, their pore structures may function as transport systems that allow microbes to release proteins into the periplasm or extracellular space (55). This hypothesized mechanism is designed so that unreversible cell death is not triggered, even in host cells. In this way, bacteria (e.g., S. Typhimurium) may evade neutrophil-induced death by avoiding pyroptotic-mediated disruption of the intracellular niche, at least in the short term. Therefore, from the host’s perspective, the prolonged IL-1β secretion is induced by caspase-1 activation following ASC speck formation to fight bacteria. However, bacterial infection may also regulate the host’s defense by utilizing pyroptosis. L. pneumophila infection inhibits NLRC4 and ASC expression in human monocytes to avoid robust IL-1β secretion, and the innate immune system has the other means of fighting back; the relatively inactive NLRC4 inflammasome with less ASC participants is still able to induce cell death while caspase-1 remains unprocessed, however, the intracellular bacteria count may still increase in the cytotoxic background (180). In this situation, caspase-7 (in a GSDMD-independent manner) and the GSDMD are responsible for pore formation-induced cell death and the restriction of bacterial replication in response to L. pneumophila infection, both in vivo and in vitro, indicating the significance of cell death in the anti-bacterial response (204, 362, 363). However, as mentioned previously, excessive pyroptosis is dangerous to hosts. In some scenarios, pyroptosis may lead to bacteria (e.g., Mtb) spreading to adjacent healthy cells as new hosts, facilitating replication (364). Pyroptosis (and the backup pathways of apoptosis, if any) may even promote mouse death as a result of NLRC4 overactivation in response to non-propagative S. Typhimurium, indicating that inflammasome-induced damage is lethal, whereas bacterial burden is not, although the propagation of WT strains is hampered by inflammasome signaling (e.g., IL-1β secretion in hyperactive neutrophils/monocytes and pyroptotic macrophages) (142, 344, 365). The balance between pathogen and host renders our understanding of inflammasome signaling and downstream events complex, where the very same strategy (e.g., pro-pyroptosis or anti-pyroptosis) may dynamically benefit one side in some situations, and the other in others.

Despite these possible reasons, more detailed answers are required to clarify the exact mechanisms, timing, and regulation of coupling and uncoupling, especially the poorly understood roles of the uncommonly localized inflammasome components (e.g., extracellular ASC specks), non-inflammasome participants, the connection between and pyroptosis and other forms of regulated cell death, and other inflammatory pathways and inflammasome activation mechanisms. Uncoupling downstream of inflammasome signaling has an outstanding role in homeostasis and host defense and is thus attracting increasing interest, rendering this field worthy of further exploration. However, more attention should be paid to the design of lab investigations. For example, because GSDMD-mediated pyprotosis and cytokine secretion are observed using numerous readouts, the accurate interpretation of the meaning of such readouts is important. Although the mechanisms surrounding glycine and punicalagin are unclear, the effects of the two cytoprotectants may differ. Punicalagin inhibits the release of IL-1β and LDH at similar half-maximal inhibitory concentrations (IC50) of 3.91 and 3.67 μM in ATP-treated macrophages, whereas glycine prevents the release of LDH (not in THP-1 cells) but not IL-1β secretion or PI intake, suggesting different roles of blocking plasma membrane permeabilization and rupture, respectively (20, 52, 68). PI uptake may always be detected alongside IL-1β secretion, although in some scenarios the latter is observed without the former, perhaps because of the subthreshold GSDMD pore formation in hyperactive cells or the partially indirect role of GSDMD-NT in neutrophil IL-1β release (52, 344, 352, 353). In contrast, LDH release is more likely considered a result of cell membrane rupture, although controversial data shows that glycine cannot inhibit the release of LDH from pyroptotic THP-1 cells (20). The evaluation of LDH release may be used in combination with other means that assess cell viability to further conclude hyperactivation in IL-1β secreting cells. Moreover, since LDH release could indicate other forms of cell membrane rupture, attention should be paid when using it as a sole marker for pyroptosis (39). Additionally, the possibility of detecting extracellular pro-IL-1β using ELISA should be considered to avoid the incorrect detection of IL-1β (68). More importantly, as the relative responses are unique in each species, the interpretation and extrapolation of data from animal models should be carefully considered when searching for potential targets or developing novel therapies for human diseases. Collectively, studies investigating the downstream events of inflammasome signaling should be carefully designed to consider species, cell type, and readouts to avoid misunderstanding.

Furthermore, when considering pathological conditions, uncoupling can be protective or detrimental to the host, and excessive or inadequate activation may backfire in the initial attempt to maintain homeostasis. Hyperactive-like neutrophils are involved in the pathogenesis and development of autoinflammatory disorders (e.g., cryopyrin-associated periodic syndromes [CAPS]) in gain-of function models (366, 367), and hyperactivating stimuli may also lead to inflammation but not death in mouse sepsis (337). Pyroptotic death and its backup apoptosis may help to limit bacterial reproduction in cells and tissues but may be lethal to mice under overactive inflammasome signaling (365). “Incompetent” uncoupling could be too weak to control bacterial replication or too strong to avoid massive and widespread damage that may threaten host survival (368). Hence, when developing potential therapies for uncoupling-associated pathological conditions, the use of pro-uncoupling, coupling, or inhibition of inflammasome activation should be carefully balanced to consider the benefits and costs. As multicellular hosts possess a sophisticated network of differential and cooperative cellular and molecular players, therapies need to be balanced so that any deleterious effects can be limited on all components. Further studies that shed light on the roles of uncoupling events downstream of inflammasome signaling may thus create promising opportunities for novel drug development.

YL wrote the manuscript and created the figures. QJ reviewed the manuscript and provided guidance. Both authors contributed to the article and approved the submitted version.