Gasdermin pore forming activities that promote inflammation from living and dead cells

Abstract

Gasdermins are proteins that can self-assemble into membrane channels (also known as pores). These pores can serve as conduits for the secretion of cytosolic molecules, with the most commonly studied being members of the interleukin-1 family of cytokines. However, gasdermin pore forming activities must be tightly regulated, as the channels that they form can lead to a lytic proinflammatory form of cell death known as pyroptosis. Recent studies have revealed multiple mechanisms that control gasdermin activities within cells and identified gasdermin proteins in organisms as diverse as bacteria, humans and yeast. In this Review, we discuss the diverse molecular and cellular mechanisms that regulate gasdermin pore formation. These mechanisms of gasdermin regulation likely explain the flexibility of these proteins to display cell type specific (and potentially organism specific) functions.

Overview

The gasdermin family was first used to refer to a group of proteins homologous to gasdermin A (GSDMA), which is selectively expressed in the mouse gastrointestinal tract and skin and was associated with alopecia-like skin disease in mice [1, 2]. These proteins share a conserved N-terminal domain that has the intrinsic ability to oligomerize to form membrane channels that are commonly referred to as pores. This intrinsic pore forming activity is restrained in resting cells by an auto-inhibitory C-terminal domain. Select proteolytic cleavage events that liberate the C-terminal domain enable pore forming activity to occur. Gasdermin pores serve as conduits that link the cytosol with the extracellular space, thereby enabling the secretion of intracellular molecules [3, 4]. Gasdermin activities were initially described as executioners of an inflammatory form of lytic cell death called pyroptosis [5], as gasdermin pore forming activities can promote membrane rupture. Pyroptosis is distinct from other programmed cell death pathways, including apoptosis and necroptosis. While pyroptotic cells release inflammatory molecules into the extracellular space to enhance immune responses, apoptotic cells are immunosuppressive [6]. At a molecular level, pyroptosis is defined as a gasdermin-dependent form of lytic cell death [7], which is most commonly associated with the actions of caspases that can cleave these pore forming proteins.

Recent studies have demonstrated that gasdermin activities are regulated at multiple levels to prevent spontaneous pore formation and to control the abundance of pores at the cell surface. These mechanisms include autoinhibitory structures, proteolytic cleavage events that promote (or prevent) pore assembly, posttranslational modifications to regulate lipid binding and oligomerization, and pore removal by membrane repair mechanisms (Figure 1). In this Review, we discuss mechanisms that regulate gasdermin pore formation and how these regulations diversify cellular responses.

Figure 1.

Intermediate steps regulating gasdermins activation of pore formation. 1) Autoinhibition - most gasdermins contain a pore-forming fragment and an inhibitory fragment, which through intramolecular interactions mask the surfaces required for lipid-binding and oligomerization. 2) Proteolytic cleavage - proteases are required to cleave the linker between the pore-forming and inhibitory fragments. These proteases include proinflammatory caspases like caspase-1/4/5/11, proapoptotic caspases like caspase-8/3, granules-associated proteases like elastase and cathepsin G. Cells can regulate the rate of gasdermin pore formation by controlling the degree of proteolytic activity. 3. Lipid selectivity - the pore-forming fragment of most gasdermins shows selectivity toward negatively charged phospholipids, which associate with the inner leaflet of the plasma membrane and the outer leaflet of intracellular vesicles. The pore-forming fragments of human and murine gasdermins also bind to cardiolipin, a phospholipid found in mitochondria and the plasma membrane of bacteria. The lipid preference ensures that gasdermin activation only kills the cells in which they are activated. 4. Oligomerizations - the pore-forming fragments oligomerize to form pore-like structures on cellular membranes. The oligomerization step requires covalent modifications on conserved cysteines and reactive oxygen species (ROS). However, the exact mechanism and the nature of the covalent modifications remain largely unknown. 5. Amplifications - the formation of gasdermins pores usually leads to further cleavages of gasdermins through feed-forward loops by releasing proteases into the cytosol or activating pathways responsible for gasdermin cleavage. 6. Membrane repair - there are mechanisms to remove perforated membranes and organelles, such as membrane remodeling by the ESCRT-III complex and autophagy, which may establish an equilibrium with the pore-forming activities to prevent cell death to preserve other cellular functions.

Autoinhibition prevents spontaneous gasdermin pore forming activity

Gasdermin-like proteins are best-studied in mammals but are also present in bacteria, fungi, early metazoa and invertebrates [8-10]. Despite their highly diverse amino acid sequences, gasdermin-like proteins share conserved structures. Most gasdermins consist of a pore-forming N-terminal domain (NT-GSDM) and an autoinhibitory C-terminal domain (CT-GSDM), which are connected by a disordered linker [11].

The intramolecular interactions between NT-GSDM and CT-GSDM mask the surfaces on NT-GSDM required for lipid binding and oligomerization, thus preventing spontaneous pore formation in membranes. This autoinhibitory structure was first revealed by the crystal structure of gasdermin A3 (GSDMA3) [11]. The pore-forming domain NT-GSDMA3 contains two subdomains—a globular domain and an extended β-barrel that can insert into the lipid bilayer. The globular domain contains a positively charged α-helix (α1) followed by β1-β2 hairpin, which is important for interaction with acidic phospholipids [11, 12]. CT-GSDMA3 is a compact globular fold composed primarily of α-helices. This domain interacts with NT-GSDMA3 through the α1 helix and the β1-β2 hairpin. Mutations of residues in CT-GSDMA3 that mediate interdomain interactions abolish autoinhibitory activity, resulting in spontaneous pore formation and cell death [11]. Conversely, mutations of residues in NT-GSDMA3 that interact with the C-terminal domain decrease lipid binding and pore formation. These findings suggested a molecular model of gasdermin autoinhibition, whereby the amino acids that enable lipid binding and pore formation of NT-GSDMA3 can be masked by interactions between these same residues and the C-terminal domain [11, 12]. Subsequent structural analysis demonstrated similar autoinhibitory mechanisms associated with human GSDMA, GSDMC, GSDMD [11]. Furthermore, the crystal structure of full-length murine and human GSDMD also revealed similar intramolecular interactions between NT-GSDMD and CT-GSDMD, whose interface overlaps with the lipid-binding surface on NT-GSDMD [13]. Although bacterial gasdermins (bGSDMs) contain a shorter C terminal domain than mammalian counterparts, the intramolecular inhibition is conserved. The structures of Runella, Bradyrhizobium, and Vitiosangium gasdermins revealed a C-terminal peptide that binds the pore forming domain of these bGSDMs. Removal of the C-terminal peptide enables the N-terminal fragment of these bGSDMs to insert into the plasma membrane, form pore-like structures, and arrest cell growth [8]. These observations suggest that autoinhibition is a conserved mechanism that controls gasdermin pore forming activity.

Cleavage of gasdermins to release autoinhibition and promote pore formation

Due to the autoinhibitory mechanism of regulation, gasdermin pore formation requires a proteolytic cleavage event to separate the inhibitory domain from the pore-forming domain. This principle was established though the study of gasdermin D (GSDMD) [3, 4]. GSDMD can be cleaved by several inflammatory caspases, such as caspase-1, −4, −5, and −11 [3, 4]. GSDMD can also be cleaved by caspase-8, neutrophil elastase, cathepsin G, and granzyme A [14-17].

Caspase-1, human caspase-4 and −5 and murine caspase-11 cleave GSDMD within the central linker region, starting at amino acid 272 in human (272-FLTD-275) and amino acid 273 in mice (273-LLSD-276). This cleavage event releases a 31-kDa pore-forming NT-GSDMD and a 22-kDa CT-GSDMD. Contrary to the initial belief that caspases specifically recognize the tetrapeptide at the cleavage site, the tetrapeptide sequence contributes little to the recognition of GSDMD by caspase-4 or −11 [18]. Mutation of 272-FLTD-275 in hGSDMD into AAAD does not prevent cleavage of GSDMD by caspase-4 [18]. To explain this observation, the cleavage of GSDMD was proposed to proceed in several steps. First, activated caspases bind to GSDMD through a conserved hydrophobic exosite present in the C-terminal domain. The interaction between the caspase and CT-GSDMD guides the tetrapeptide into the catalytic pocket, resulting in gasdermin cleavage. This exosite-guided mechanism of substrate specificity explains the minimal contribution of the tetrapeptide sequence in GSDMD for cleavage by inflammatory caspases. Multiple caspases bind to GSDMD using the same exosite, suggesting that the recognition of the exosite in CT-GSDMD is a conserved mechanism that allows GSDMD to be recognized by inflammatory caspases [18].

Caspase-8 can contribute to GSDMD cleavage and induce pyroptosis [14, 19, 20]. The cleavage of GSDMD by caspase-8 has been studied in macrophages that have been infected with pathogenic Yersinia [14, 20]. The Yersinia effector protein YopJ interferes with mitogen activated protein kinase (MAPK) activities downstream of the innate immune kinase TAK1, which are critical for cell survival [21, 22]. When these activities are blocked by YopJ (or by chemical inhibitors of TAK1), caspase-8 activity is stimulated. Activated caspase-8 cleaves GSDMD at the same site as the aforementioned caspases and induces pore formation and cell death [14]. This pathway may play an important role in the activation of GSDMD within cells that do not express inflammatory caspases or when factors released by pathogens block these caspases.

Proteolytic cleavage of GSDMD is not limited to the caspase family. In neutrophils, GSDMD is cleaved by elastase to form a pore-forming fragment [17, 23]. In phorbol 12-myristate 13-acetate (PMA)-treated neutrophils, elastase that is released from granules can cleave GSDMD, which subsequently permeabilizes primary azurophilic granules to release more neutrophil elastase as a positive feedback mechanism [17]. Similarly, cathepsin G, another granule-associated protease, cleaves murine GSDMD at a site two amino acids upstream of the caspase-cleavage site. This cleavage event generates a pore-forming fragment [15].

Similar to GSDMD, other gasdermins require proteases to generate pore-forming fragments. Gasdermin E (GSDME) can be cleaved by caspase-3 to produce a pore-forming fragment that permeabilizes the plasma membrane [24, 25]. The involvement of caspase-3, which is present in most cell types, suggests that the activation of GSDME may be a common and ancient pathway to trigger inflammatory cell death. Indeed, GSDME homologs were found in several invertebrate species, including corals. Coral GSDME is cleaved by coral caspase-3 to generate a pore-forming fragment to induce pyroptosis-like cell death. This process was proposed as a defense mechanism in corals infected with the Vibrio coralliilyticus [10].

Similar to GSDMD, the cleavage of GSDME is not only restricted to the caspase family. Granzyme B, secreted by cytotoxic lymphocytes, can cleave GSDME at the same site as caspase-3 to induce pyroptosis in GSDME-expressing cells [26]. The cleavage of GSDME within tumor cells by granzyme B, in addition to cleavage by caspase-3, seems to be an essential immune-mediated mechanism to prevent tumor cell growth [26, 27]. Similarly, GSDMB can be cleaved by granzyme A, and GSDMC can be cleaved by caspase-8 following TNF-mediated death receptor signaling to trigger pyroptosis [16, 28].

Even within single celled organisms, gasdermin-like proteins require proteases for their activation. Podospora anserina, a filamentous fungus, has a gasdermin homolog that triggers cell death [9]. P. anserina has a pair of idiomorphic genes named het-Q1 and het-Q2, which are alleles of the same locus that contain different sequences. het-Q1 encodes a gasdermin, while het-Q2 encodes a subtilisin-like serine protease. Somatic fusion between different fungal strains allows HET-Q1 and Q2 to be co-present, resulting in the cleavage of a 5kDa C-terminal fragment of HET-Q1. This cleavage event liberates HET-Q1 from its autoinhibited state, resulting in cell death. In silico analyses of het-Q1 homologs in other fungi revealed that most gasdermin-like genes are clustered with protease-encoding genes [29].

Genes that encode bGSDMs are often associated with those that are predicted to encode caspase-like proteases. Some proteases in the bGSDM loci are fused with domains known to recognize pathogens in the mammalian innate immune system, such as leucine-rich repeats, tetratricopeptide repeats, WD40 repeats, and NACHT domains [8]. Proof of principle analysis revealed that coexpression of Runella bGSDM and its associated proteases induced cytotoxicity in E. coli [8]. In the Runella system, the associated protease cleaves the gasdermin at a specific site before the C-terminal peptide and induces cytotoxicity. Cleaved Rubella gasdermin also localizes to the plasma membrane and forms pore-like structures on artificial liposomes [8].

The almost perfect parallel between the activation of bacterial, fungal, and mammalian gasdermins suggests that gasdermin-mediated cell death may have originated from an ancient defense mechanism that senses proteases. Many viruses, including bacteriophages, carry proteases that assist their entry and replication in the host cells. Gasdermins may act as protease-sensors that generate pore-forming fragments upon detection of compatible protease activities. Thus, a gasdermin-based protease-sensing system may be a primordial defense mechanism to guard against intracellular pathogens.

Lipid binding determines the subcellular sites of gasdermin pore activities

Gasdermin cleavage may not be sufficient to separate the pore-forming and inhibitory domains. In the case of GSDMD, these domains remain bound together after cleavage in vitro [11, 30]. The noncovalent complex between NT-GSDMD and CT-GSDMD only dissociates in the presence of liposomes containing negatively charged phospholipids [11, 30]. NT-GSDMD binds preferentially to negatively charged phospholipids, including phosphatidylinositol (PI) phosphates (PIPs), PI(4)P and PI(4,5)P2 and cardiolipin. NT-GSDMD also binds to phosphatidic acid (PA) and phosphatidylserine [11, 30]. Similarly, human NT-GSDMA, NT-GSDME, and murine NT-GSDMA3 bind to negatively charged phospholipids, suggesting a conserved mechanism that enables the pore-forming fragment to bind and insert into membranes [11, 25].

Crystal structures of the GSDMA3 and GSDMD pores demonstrated that the α1 helix, which contains a positive patch (basic patch 1, BP1), promotes lipid binding and pore formation [12, 13, 31]. Consistent with this idea, alanine mutations at three conserved basic residues in α1 helix abolished the abilities of GSDMA3 and GSDMD to permeabilize liposomal membranes [12, 31]. In addition, the β1-β2 loop, which is membrane-proximal and contains a hydrophobic tip flanked by basic residues (basic patch 2, BP2), is important for interaction with acidic lipids [31]. These findings suggested a lipid-binding model in which the hydrophobic tip of the β1-β2 loop anchors the pore-forming fragment into the lipid bilayer, followed by the interaction between the basic residues of BP2 and negatively charged acidic lipids [31]. Finally, in GSDMD, a basic patch on the β7-β8 hairpin, which is adjacent to the globular domain, is also potentially involved in the membrane interaction [31].

Within cells, the lipids that are bound by NT-GSDMD are located on different organelles. Despite this diversity, the lipid preference of NT-GSDMD and other gasdermin homologs suggests that the pore-forming fragment selectively binds to the inner leaflet of the plasma membrane [30]. No lipids found in the outer leaflet of the plasma membrane have been identified as high affinity ligands for NT-GSDMD. This selectivity for intracellular lipids may be important to ensure that the pore-forming fragment only perforates the cell that expressed the gasdermin and does not kill bystander cells when released into the extracellular space. The preference of the gasdermin pore-forming domain for cardiolipin, a negatively charged phospholipid found in bacteria, suggests that the pore-forming fragment may target and control bacterial growth. Supporting this prediction, the pore-forming fragments of GSDMD, GSDMA, and GSDMA3 can lyse protoplasts of Bacillus megaterium. Concentrated pyroptotic cell supernatants kill E. coli in a dose-dependent manner, further suggesting a function of GSDMD in direct anti-bacterial activities [30].

Cardiolipin is also present in the inner membrane of mitochondria, which has been confirmed as another intracellular target of gasdermins [11, 30]. NT-GSDMD and NT-GSDME can permeabilize mitochondria to release cytochrome c, which activates caspase-3. In the case of cells that dually express GSDMD-and GSDME, activated caspase-3 may reinforce the commitment of individual cells to pyroptosis by cleaving GSDME [32]. Mutants of GSDMA3 that lose autoinhibition can target mitochondria, resulting in an increase in reactive oxygen species (ROS) production and the disruption of mitochondrial membrane potential [33]. However, the mechanism of how the pore-forming fragment gains access to the inner membrane where cardiolipin localizes remains unknown.

Although structural studies on gasdermin pores have elucidated the chemical nature of the interaction between the pore-forming fragments and lipids, these studies relied on recombinant proteins at high concentrations and liposomes. In the cellular context, the pore-forming fragments may be exposed to lower concentrations of lipids that are distributed across a range or organelles. Moreover, additional molecules may impact the access of cleaved gasdermins to lipids that promote pore formation. It is possible that chaperones exist to ensure gasdermins are positioned in regions of the cells that are conducive for rapid interactions with membranes. Upon cleavage, the pore-forming fragment would then be exposed to relatively high local concentration of target lipids, thereby ensuring pore formation. Interestingly, a genome-wide CRISPR screen using Cas9-expressing immortalized macrophages treated with LPS and a TAK1 inhibitor identified the Ragulator-Rag complex as a lysosomal platform to activate caspase-8, which cleaves GSDMD and triggers pyroptosis [34]. The membrane of lysosomes is an interesting target for gasdermins, as its convex curvature (curve towards the cytosol) may impact GSDMD pore formatting efficiency [31]. In addition, the surface area of lysosomes is smaller than the plasma membrane, which may allow the pore-forming fragments to oligomerize in these organelles efficiently. Finally, the perforation of lysosomes can release cathepsins into the cytosol that can cleave GSDMD, resulting in a positive feedback mechanism to enhance the commitment to pyroptosis [15]. Future work is needed to test the validity of this proposed model.

Oligomerization of gasdermins into pore forming membrane channels

The GSDMA3 pore is predominantly composed of 27-subunits, whose transmembrane β-barrels together form a conduit with a diameter of 180 Å [12]. The NT-GSDMA3 pore-forming fragment, when inserted into the lipid bilayer, undergoes a radical conformational change compared to its uncleaved counterpart. While the globular domain remains largely unchanged, the β-strands reorganize into two anti-parallel β-hairpins, whose length is approximates 50 Å, sufficient to transverse a lipid bilayer [12]. Electrostatic analysis of the transmembrane domain revealed the amphipathic nature of this region, in which the side facing the membrane is mainly hydrophobic while the inner side of the pore conduit contains positively and negatively charged patches.

Further analyses identified three oligomerization interfaces, including both the globular and β-hairpin domains. The first interface contains the helix α3 of one subunit and the region around α2 and β11 of its neighboring subunit, which interact with each other by hydrophobic and electrostatic interactions. The second interface is mediated by the head-to-tail interactions between the neighboring α1 helices to form a stabilizing helical belt. The third major interface is the interaction between the β-strands of the transmembrane domain [12]. The structure of the GSDMD pore shows similar features to NT-GSDMA3 except for the larger pore size (31 to 34 subunits with an inner diameter of 215 Å) and slight differences in the relative position between the globular and the β-barrel domain [31]. These findings implied the conserved structures of different gasdermin pores, which may vary in oligomerization and size of the conduit.

Modifications at the oligomerization interfaces may regulate gasdermin pore formation and cell fate. Phosphorylation of the threonine at amino acid 6 in the α1 helix in GSDME (equivalent to threonine 8 in GSDMA3) inhibits its ability to oligomerize and induce pyroptosis [32]. The kinase responsible for these phosphorylation events is unknown. However, GSDMA is a potential substrate of polo-like kinase 1 (PLK1), a proto-oncogene [35]. Further studies are required to investigate if the ability of PLK1 to suppress cell death is mediated through phosphorylation of gasdermins.

Using SDS-PAGE, non-reducing conditions revealed the existence of multimers of NT-GSDMD that form upon pyroptosis [30]. These multimers collapse into monomers when SDS-PAGE is performed under reducing conditions. Cys192 in murine GSDMD (Cys191 in human GSDMD) was identified as important for NT-GSDMD oligomerization [13, 30, 36]. Interestingly, all recently discovered GSDMD-targeting drugs covalently modify human Cys191 or mouse Cys192 of GSDMD, resulting in defects in NT-GSDMD oligomerization. For example, the chemical disulfiram covalently modifies human/mouse Cys191/192 and blocks NT-GSDMD oligomerization to prevent pyroptosis and IL-1β release [37]. Necrosulfonamide (NSA) also covalently binds to Cys191 and inhibits NT-GSDMD oligomerization and cell death [36]. Beyond the scope of synthesized chemicals, cellular metabolites also control GSDMD oligomerization by covalently modifying Cys191/Cys192. Exogenous dimethyl fumarate (DMF) modifies Cys192 to generate 2-(succinyl)-cysteine, which prevents GSDMD cleavage and oligomerization, thus limiting pore formation and pyroptosis [38]. The accumulation of endogenous fumarate can be forced experimentally, by blocking fumarate hydratase with chemical FHIN1. This chemical results in GSDMD modification at Cys192 and impairs cell death. The finding that endogenous fumarate can modify GSDMD illustrates how cells can regulate pyroptosis in response to changes in metabolism [38]. It is possible that Cys192 forms disulfide bonds between subunits that enhance or stabilize the pore forming oligomer. However, the structure of the GSDMD pore shows no adjacent cysteine residues in neighboring NT-GSDMD subunits [13]. Additional work is needed to define how posttranslational modifications impact NT-GSDMD oligomerization and pore forming activity.

The modification of cysteines within gasdermins may be a conserved mechanism to control the oligomerization. For example, in addition to targeting GSDMD, DMF covalently binds to Cys45 of GSDME and prevents cell death in Gsdmd-deficient cells [38]. Moreover, Cys3 is highly conserved among bGSDMs and is required for the oligomerization of these bacterial proteins. Interestingly, Cys3 of the Bradyrhizobium gasdermin is palmitoylated. Mutant proteins lacking Cys3 are not defective for protein cleavage, but are defective for pore forming activities [8]. It is possible that the 16-carbon palmitoyl group may facilitate lipid binding and oligomerization of the pore-forming fragment.

Finally, recent data has highlighted ROS as a regulator of gasdermin oligomerization. A genetic screen identified the Ragulator-Rag complex as a factor required for NT-GSDMD oligomerization. NT-GSDMD oligomerization and pore-forming activity was defective in macrophages lacking the genes RagA or RagC. These activities could be restored by treatments with exogenous H₂O₂ or chemicals that flux mitochondrial ROS, such as Rotenone and Antimycin A [39]. This finding suggests that the primary function of Ragulator-Rag in regulating NT-GSDMD oligomerization may be in ROS generation. Further studies are needed to define the precise role of ROS in GSDMD oligomer formation and whether ROS modification of GSDMD itself impacts its activities.

Substrate selectivity of gasdermin pores

The formation of gasdermin pores changes the permeability of the plasma membrane by creating channels that directly connect the intracellular and extracellular compartments. Recent studies have shown that pyroptosis is not an inevitable consequence of gasdermin pore formation. Rather, gasdermin pores can serve as conduits for the secretion of select molecules from living cells (or dead cells whose plasma membrane has not ruptured).

Two molecular features determine which cytoplasmic molecules can traverse gasdermin pores and exit cells—size and charge. Based on the structures of the GDSDMA3 and GSDMD pores, the membrane-spanning channels created are approximately 20nm in diameter [12, 31]. In a minimal system, when liposomes were loaded with dextrans of variables size and exposed to cleaved GSDMD, only dextrans with a hydrodynamic radii of less than 4 nm escaped into the extra-liposomal space. In contrast, dextrans with hydrodynamic radii of 27nm showed minimal ability to escape the lumen of gasdermin-permeabilized liposomes [40]. Thus, the size of a given molecule can determines its ability to traverse membrane via gasdermin pores. Notably, the GSDMD diameter is sufficiently large to permit the release of IL-1β and IL-18 [40, 41]. Besides IL-1β and IL-18, other small cytosolic proteins such as RhoGTPase Rac1 and galectin-1 are secreted through gasdermin pores [41, 42]. Similarly, ions or small molecules can passively traverse gasdermin pores. Potassium may follow its gradient and diffuse into the extracellular space, by a process known as potassium efflux. This process can activate the NLRP3 inflammasome and create a feed-forward loop to further cleave GSDMD and pyroptosis [43].

Recent work indicates that IL-1β can pass through GSDMD pores more efficiently that pro-IL-1β, yet both of these molecules are smaller than the diameter of the membrane-spanning channel created by GSDMD. This observation led to the identification of charge as a second mechanism of GSDMD pore selectivity. The GSDMD pore has four solvent-exposed acidic patches near the conduit, which creates a negative potential within the membrane-spanning channel. The negative charge within the GSDMD pore acts as an electrostatic filter that allows positively charged molecules to pass through most effciently. This mechanism explains why IL-1β which contains net positive charge, is preferred to be secreted through GSDMD pore over neutrally charged pro-IL-1β [31]. Consistent with this idea are findings that small dextrans of positive charge are released from GSDMD permeabilized liposomes more efficiently than neutral or negatively charged dextrans [31]. The electrostatic filter activity therefore provides the second layer of control to define which substrates can traverse GSDMD pores.

Membrane repair activities remove gasdermin pores from the plasma membrane

Various stimuli can trigger activities that do not induce pyroptosis, but rather enable living murine phagocytes to secrete IL-1β [40, 41, 44-48]. Similarly, extracellular bacterial lipopolysaccharide (LPS) signaling via TLR4 can promote IL-1β secretion from living human and porcine monocytes [46-48]. Studies in recent years have examined how cells remove gasdermin pores from the plasma membrane in order to maintain viability. GSDMD pore formation enables the high concentration of extracellular calcium to flow into the cytosol. This process, triggers removal of the GSDMD pores by the ESCRT-III complex, a central regulator of plasma membrane repair. Based on the structure of GSDMD pores, the GSDMD pore may bend the membrane toward the extracellular space to create a concave surface [31], which is the preferential membrane topology to recruit the ESCRT-III machinery [49, 50]. These membrane bending changes may provide a mechanism for the GSDMD pore to prime its own removal.

In addition to GSDMD pores, the recruitment of the ESCRT-III machinery to the perforated membrane is observed upon formation of other membrane pores, such as the necroptosis-inducing protein MLKL [50]. These findings suggest that the ESCRT-III machinery is a conserved mechanism in response to membrane perforation caused by pore-forming proteins. Such a mechanism allows cells to maintain viability by balancing the number of newly formed pores and the number of pores removed through membrane repair mechanisms. This dynamic equilibrium may keep the number of gasdermin pores at the sublytic level at which the cells can secrete IL-1β but still maintain viability. Consistent with this idea, studies from Broz and colleagues demonstrated that disruption of ESCRT-III activities results in an increase in GSDMD-dependent membrane permeability and pyroptosis [51]. Cell fate may therefore be decided by the number of gasdermin pores present on the plasma membrane, which depends on the cell type and the nature of the stimuli [5].

Membrane rupture after GSDMD pore formation during pyroptosis

When gasdermin pore abundance overwhelms the membrane repair machinery, the excessive perforated membrane may lead to pyroptosis. Pyroptotic cells are characterized by initial ionic fluxing, followed by cell swelling, mitochondrial depolarization, lysosome leakage, and eventual membrane rupture [52]. Analysis of pyroptosis using high-resolution live-cell imaging showed that this process is initiated by a GSDMD-dependent influx of Ca²⁺ into the cytosol. Pyroptotic macrophages have a Ca²⁺ influx approximately 10 minutes before the plasma membrane becomes permeable to larger cell-impermeant fluorescent dye molecules. The initial Ca²⁺ influx disappears in GSDMD-deficient macrophages suggesting that GSDMD pores are responsible for the increase of the cytosolic Ca²⁺ [52]. The early size-exclusion selectivity prompted speculation that the pore-forming fragments insert into the plasma membrane as small oligomers to form pores that increase in diameter over time [52]. This mechanism is similar to the mechanism used by Bax to form pores on the mitochondrial outer membrane [53]. This speculation is consistent with findings that NT-GSDMD oligomers that have been assembled in vitro can form pre-pores that display variable diameters. These pre-pore states include arcs, slits, and rings with the potential to fuse and form larger and more stable pores [54].

The initial GSDMD-dependent Ca²⁺ flux is not necessarily from the extracellular compartment, as the endoplasmic reticulum and lysosomes also contain Ca²⁺. Live-cell imaging found that lysosomal leakage happens in pyroptotic macrophages approximately 10 minutes before plasma membrane rupture, which correlates with the initial Ca2+ influx [52]. In neutrophils, which contain less caspase-1 than macrophages [55], NT-GSDMD predominantly associates with the membrane of organelles rather than the plasma membrane. NT-GSDMD within neutrophils associates with LC3+ autophagosomes and mediates an autophagy-dependent IL-1β secretion [56].

Plasma membrane rupture is the final event of pyroptosis. Recent findings have shown that this process is regulated by Ninj1, a plasma membrane protein containing two transmembrane domains [58]. In response to inducers of pyroptosis such as nigericin and LPS electroporation, Ninj1 oligomerizes and induces membrane rupture by an unknown mechanism. Notably, Ninj1 acts downstream of GSDMD pore forming activity, as Ninj1−/− macrophages contain GSDMD pores that serve to release IL-1β in the absence of membrane rupture. Much additional work is necessary to understand the relationship between Ninj1 and GSDMD activities.

Gasdermin-dependent effects on cell behavior

The ability to gasdermins to permit the secretion of IL-1β from living cells is notable, as it has long-been recognized that IL-1 release from cells occurs independent of the biosynthetic vesicle mediated pathways. In macrophages and dendritic cells, common agonists of inflammation (e.g. TLR ligands) are able to induce pro-IL-1β production, but not cleavage or release of IL-1β. As such, TLR ligand stimulated cells release inflammatory mediators whose expression and secretion are coupled events, such as TNFα and IL-6, but they do not typically release IL-1β. These cells are classically referred to as activated cells. Gasdermin pore forming activities that are non-lytic will enable IL-1β to be added to the repertoire that cytokines that viable cells can secrete. This addition of a potent pyrogen renders these cells more inflammatory than their activated counterparts, and they are therefore referred to as hyperactive cells [45, 59]. Several stimuli have been described that hyperactivate cells (described below), all of which act together with TLR ligands to promote IL-1β release while maintaining viability.

Based on the dynamic equilibrium model, the hyperactive cell state can be achieved by maintaining the pore formation rate less than or equal to the pore removal rate. Low doses of pyroptotic stimuli potentially induces pore formation at a slower rate, which may lead to the state of hyperactivation. Indeed, immortalized bone marrow-derived macrophages when treated with low-dose nigericin sustain IL-1β secretion while showing minimal membrane rupture events [31]. Low-dose nigericin-treated cells secrete even more IL-1β than high-dose nigericin-treated cells, which is probably due to the continuous IL-1β release over a more extended time in the absence of cell death [31]. In addition, cells control the rate of pore formation by regulating the activity of proteases that cleave and activate gasdermin. The duration and magnitude of caspase-1 activity varies between different cell types based on the concentration of caspase-1 [55]. Neutrophils continuously secrete IL-1β up to 8 hours after nigericin stimulation. This prolonged IL-1β secretion may be due to the lower concentration of caspase-1 per cell, which may lower the magnitude of caspase-1 activation to limit pore formation and maintain cell viability [55]. Finally, the nature of the stimuli also contributes to the ability of the cells to acquire the hyperactive state. A mixture of oxidized phosphorylcholine derivatives (oxPAPC) or an isolated component 1-palmitoyl-2-glutaryl-sn-glycero-3-phosphocholine (PGPC) can induce dendritic cells (DCs) to continuously secrete IL-1β without causing cell death [40, 45, 59, 60]. The mechanism of how these stimuli elicit cleavage of GSDMD at the sublytic level to maintain cell viability is still unknown.

The advantage of the hyperactive cell state is that they can add IL-1β and IL-18 to the repertoire of secreted cytokines, while preserving other cellular functions to modulate host immune responses. For example, DCs treated with hyperactivating stimuli (LPS and PGPC) can be loaded with tumor antigens and transferred to recipient mice [59]. These cells migrate to the draining lymph node and induce long-lived T cell-mediated anti-tumor responses by an IL-1β dependent process [59]. In contrast, stimuli that induce pyroptosis of DCs are unable to induce anti-tumor effects. This observation is probably explained by the inability of pyroptotic DCs to migrate to the draining lymph nodes to prime T cell activation [59]. Similarly, the resistance of neutrophils to pyroptotic death when exposed Salmonella enterica subsp. Typhimurium, ATP, or nigericin, may be an adaptation to ensure neutrophils can secrete IL-1β without compromising other antimicrobial functions [46, 61]. The hyperactive state may not be an exception of gasdermin activation, but is actually the predominant mode activation in vivo, where the cells may be activated with much lower concentrations of stimuli than are likely used in vitro.

Perspectives

This review was focused on described the molecular events that mediate gasdermin pore formation and the consequences of these pores on cell fate and inflammation. Structural analysis has been key to our understanding of the intrinsic activities of gasdermins, yet much more work is needed to understand the regulation of gasdermin behaviors within cells. For example, the recently discovered GSDMD inhibitors are all recognized to target the conserved Cys192 to block GSDMD oligomerization and pore formation, but the role of this cysteine in mediating GSDMD oligomerization remains unknown. Similarly, ROS has been shown to promote GSDMD oligomerization, but how ROS regulatory pathways in cells impact gasdermin activities at steady state or during infection are unknown. In addition to these molecular considerations, it is worth noting that most of our knowledge derives from studies of gasdermins in macrophages and other phagocytes. However, gasdermins are expressed by T cells, B cells, epithelial cells, and neurons [7]. Mechanisms that regulate gasdermin functions in these cell types remain largely unknown. Based on studies focusing on phagocytes, the outcomes of gasdermin activation are cell type and context dependent. Can we apply what we know about gasdermin activation in phagocytes to other cell types? In epithelial cells, both GSDMD and GSDME have been linked to pyroptosis and the release of IL-1α and IL-1β [62, 63]. Further studies are needed to understand the roles of gasdermins in other cell types.

The origin of the gasdermin family also poses an interesting question. The versatility of gasdermin cleavage by various proteases suggests that gasdermins can act as a protease-sensing system that respond to endogenous and exogenous proteases to trigger cell death. We speculate that this system may be initially selected to respond to viruses, many of which utilize proteases to assist their entry and replication. Supporting this speculation, Seneca Valley virus (SVV), a swine-specific virus that causes vesicular diseases, carries a protease 3C (SVV 3C^pro) that cleaves multiple host proteins [64-66]. SVV 3Cpro also cleaves porcine GSDMD to generate a pore-forming fragment that triggers pyroptosis. SVV 3C^pro does not cleave human and mouse GSDMD, suggesting a species-specific selection toward host-specific pathogens [67]. Gasdermins can also be targeted by viral proteases to inhibit pyroptosis and favor viral replication. 3C^pro of enterovirus 71, a virus causing hand-foot-and-mouth diseases in young children, cleaves GSDMD to generate a shorter N-terminal fragment that has no pore-forming activity [68]. These findings exemplify the coevolution between host gasdermins and viral proteases, suggesting that the gasdermin system may be an ancient antiviral mechanism by sensing viral proteases.

Finally, although gasdermins were first discovered in mammalian cells in the context of cell death, recent studies have described homologs of gasdermins across multiple kingdoms including fungi and bacteria. Cell death is an unusual activity of unicellular organisms. How is gasdermin-dependent pore forming activity beneficial to bacteria, for example? It is possible that gasdermins may kill the host cells upon bacteriophage infection and release alarmins to alert neighboring cells about the potential threats to the population. This “altruistic behavior” may inform the other colony members to alter gene expression to cope with potential infection. The cost of dying is compensated by the benefit of protecting other cells which carry the same genetic information. This line of thinking suggests that programmed cell death processes may have evolved together with the transition from unicellular organism to multicellular organism. Another possibility is that gasdermins may be remnants of an ancient transport system, which may enable bacteria to transport relatively large proteins into the periplasm or extracellular space. This hypothetical transport system would therefore be designed not to trigger cell death, but to create pores that can be removed by membrane repair mechanisms. Interestingly, homologs of ESCRT-III were found across all domains of life including archaea and bacteria, suggesting that removing perforated membrane may be a conserved mechanism that predates the last universal common ancestor [69, 70]. The secretion of IL-1β mediated by gasdermins without causing cell death in hyperactive phagocytes and activated neutrophils is an example of how gasdermins can function as pore-forming factors to secrete leaderless secretory proteins while preserving other functions of the cells. This secretory mechanism may still be relevant in primary organisms like bacteria and archaea or in certain cell types under specific conditions. This secretory system might be wired into a protease-sensing system to trigger cell death, which may be the origin of inflammasome pathways. Therefore, examining gasdermins from an evolutionary perspective gives us a glimpse into the primordial immune system, which would help us better understand the origin and functions of gasdermins in human health and disease.

Highlights.

• Gasdermins are ancient pore forming proteins

• Gasdermins function as channels for cytokine secretion or mediators of pyroptosis

• Multiple levels of regulation control gasdermin pore forming activities