Research ArticleCELL DEATH

Noncanonical inflammasome signaling elicits gasdermin D–dependent neutrophil extracellular traps

See allHide authors and affiliations

Science Immunology  24 Aug 2018:
Vol. 3, Issue 26, eaar6676
DOI: 10.1126/sciimmunol.aar6676

Casting NETs

Gasdermin D (GSDMD), a pore-forming protein, has emerged a key downstream effector in pyroptosis, a form of cell death induced by intracellular lipopolysaccharide (LPS). Here, by examining the role of GSDMD in the neutrophil response to LPS and cytosolic Gram-negative bacteria, Chen et al. have uncovered an important role for GSDMD in the generation of neutrophil extracellular traps (NETs). NETs are composed of chromatin and antimicrobial proteins and are cast by dying neutrophils in a process termed NETosis. The authors report that GSDMD is directly cleaved by caspase-11 and that intracellular LPS–induced NETosis is dependent on both caspase-11 and GSDMD. In the same issue, Sollberger et al. also report a role for GSDMD in NETosis.


Neutrophil extrusion of neutrophil extracellular traps (NETs) and concomitant cell death (NETosis) provides host defense against extracellular pathogens, whereas macrophage death by pyroptosis enables defense against intracellular pathogens. We report the unexpected discovery that gasdermin D (GSDMD) connects these cell death modalities. We show that neutrophil exposure to cytosolic lipopolysaccharide or cytosolic Gram-negative bacteria (Salmonella ΔsifA and Citrobacter rodentium) activates noncanonical (caspase-4/11) inflammasome signaling and triggers GSDMD-dependent neutrophil death. GSDMD-dependent death induces neutrophils to extrude antimicrobial NETs. Caspase-11 and GSDMD are required for neutrophil plasma membrane rupture during the final stage of NET extrusion. Unexpectedly, caspase-11 and GSDMD are also required for early features of NETosis, including nuclear delobulation and DNA expansion; this is mediated by the coordinate actions of caspase-11 and GSDMD in mediating nuclear membrane permeabilization and histone degradation. In vivo application of deoxyribonuclease I to dissolve NETs during murine Salmonella ΔsifA challenge increases bacterial burden in wild-type but not in Casp11−/− and Gsdmd −/− mice. Our studies reveal that neutrophils use an inflammasome- and GSDMD-dependent mechanism to activate NETosis as a defense response against cytosolic bacteria.


Newly described forms of nonapoptotic, programmed cell death in myeloid cells are emerging as critical drivers of both inflammatory pathologies and host defense. Inflammatory cell lysis by pyroptosis is extensively characterized in macrophages (1, 2), whereas neutrophils can die by expelling neutrophil extracellular traps (NETs) (3). Pyroptosis is initiated by inflammasome-activated inflammatory caspases (e.g., caspase-1, caspase-4, caspase-5, and caspase-11), which cleave gasdermin D (GSDMD) to trigger plasma membrane pores (1, 2, 48). For example, in macrophages, active caspase-1 cleaves GSDMD to elicit GSDMD-p30 pores and drive pyroptosis during signaling by canonical inflammasomes, such as those assembled by nod-like receptor (NLR) pyrin domain-containing 3 (NLRP3) and NLR CARD (caspase activation and recruitment domain) domain-containing 4 (NLRC4) (1, 2, 8). Macrophage pyroptosis can also be triggered by the noncanonical inflammasome, which confers host defense against Gram-negative bacteria in the cytosol. In this pathway, murine caspase-11 or human caspase-4/5 directly recognize cytosolic bacterial lipopolysaccharide (LPS) from Gram-negative bacteria, triggering caspase-4/5/11 activation, GSDMD cleavage, GSDMD-p30 plasma membrane pores, NLRP3 signaling, and cell lysis (1, 2, 4, 912). Macrophage pyroptosis is an important innate immune defense mechanism, because it exposes vacuole- and cytosol-resident pathogens to neutrophil-mediated extracellular destruction (13, 14). We, and others, previously reported that active caspase-1 does not trigger pyroptosis in neutrophils (1517), suggesting that GSDMD is unable to elicit neutrophil cell death in this context. The ability of neutrophils to resist caspase-1–directed pyroptosis allows these cells the life span to perform their classic antimicrobial functions [e.g., degranulation and generation of reactive oxygen species (ROS)] to kill the pathogen at a site of infection (15). The capacity of neutrophils to signal via the noncanonical inflammasome is unknown.

Under specific conditions, neutrophils can also neutralize extracellular pathogens via a nonclassical mechanism, in which neutrophils expel NETs, web-like structures of DNA embedded with antimicrobial proteins, which ensnare and neutralize microbes (3). Neutrophil death by NET extrusion (NETosis) is a multistage process, involving first disintegration of the nuclear envelope and granule membranes and chromatin relaxation, leading to nuclear delobulation and DNA expansion within the intact cell. During this early stage of NETosis, proteins, such as myeloperoxidase (MPO) and neutrophil elastase (NE) that normally reside in granules, mix with the DNA (18). Chromatin relaxation is mediated by histone inactivation, for example, by protein arginine deiminase–4 (PAD4)–dependent histone citrullination (19, 20) or by histone degradation by the granule protease NE (18). In the final stage of NETosis, the plasma membrane ruptures, leading to the extrusion of DNA from the neutrophil. The signaling mechanisms underlying NETosis are only partially characterized (21).

Here, we investigated the capacity of neutrophils to undergo cell death downstream of canonical and noncanonical inflammasomes. Although neutrophils are resistant to cell death downstream of canonical inflammasomes, we found that noncanonical inflammasome signaling triggered GSDMD-induced cell death and the extrusion of host-protective NETs. Our data thus reveal an unexpected mechanism of neutrophil-mediated host defense against cytosolic intruders.


Neutrophils resist caspase-1–mediated pyroptosis but are sensitive to caspase-11–driven death

We previously reported that canonical inflammasome activation in neutrophils triggered caspase-1 activation and the secretion of mature interleukin-1β (IL-1β) but not pyroptosis (15). To investigate the mechanisms by which neutrophils resist caspase-1–driven pyroptosis, we performed parallel assays in bone marrow–derived macrophages and mature bone marrow neutrophils from wild-type (WT) versus Gsdmd-deficient mice. We used a well-established protocol to activate caspase-1, in which cells are primed with a toll-like receptor (TLR) ligand (e.g., LPS) to up-regulate the expression of the inflammasome sensor and pro–IL-1β, before challenge with log-phase Salmonella enterica serovar Typhimurium (22) or nigericin (23) to activate the NLRC4 or NLRP3 inflammasomes, respectively. Salmonella and nigericin induced rapid caspase-1–dependent pyroptosis of macrophages, which indeed requires GSDMD (fig. S1), as previously reported (1, 2). In neutrophils, Salmonella and nigericin induced hallmarks of caspase-1 activation, such as caspase-1 self-processing, to generate the inactive p20 fragment (Fig. 1, A and B) (24) and IL-1β release (fig. S2). Both Salmonella and nigericin induced some neutrophil death; however, death in these settings was not pyroptotic, because it was not driven by caspase-1 (15) or GSDMD (Fig. 1, C and D). Thus, although canonical inflammasomes signal in neutrophils to drive IL-1β production, these cells do not die by pyroptosis.

Fig. 1 Activated caspase-11 but not caspase-1 triggers neutrophil cell death.

(A to D) Neutrophils were primed with LPS (100 ng/ml) for 4 hours before stimulation with Salmonella Typhimurium (MOI 25, log phase) or nigericin (5 μM) for the indicated times: (A and B) Caspase-1 (cleaved p20 or full-length p46) in cell supernatants (Sup.) or cell lysates (Lys.) were measured by immunoblot. (C and D) LDH release was quantified to monitor cell lysis. ns, not significant. (E to K) Neutrophils were primed with Pam3CSK4 (1 μg/ml) for 4 hours and transfected without (mock) or with ultrapure LPS (10 μg/ml) for 4 hours. LDH release (E, F, and H), GSDMD cleavage to p30 (G), or IL-1β secretion (I to K) was monitored. (L) Neutrophils were primed with Pam3CSK4 (1 μg/ml) for 4 hours and transfected without (mock) or with ultrapure LPS (10 μg/ml) for 4 hours, with and without KCl supplementation to the medium immediately before cell transfection. IL-1β release was quantified. Immunoblots are representative of data from at least three biological replicates. Graphs are means + SEM from data pooled from three (K), four (H and L), six (I), seven (E), six to eight (F and J), or eight (C and D) biological replicates. Data were analyzed for normality using the Shapiro-Wilk normality test. Statistical analyses were performed using parametric t tests (two-sided) for normally distributed data sets and nonparametric Mann-Whitney t tests (two-sided) for data sets with non-normal distributions. Data were considered significant when *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, or ****P ≤ 0.0001.

To examine whether the noncanonical inflammasome is able to induce neutrophil pyroptosis, we primed neutrophils with the TLR1/2 agonist Pam3CSK4 for 4 hours to suppress basal neutrophil apoptosis and induce the expression of caspase-11, NLRP3, and pro–IL-1β. We then transfected LPS into the cytosol to activate caspase-11. Surprisingly, cytosolic LPS induced a modest but significant increase in neutrophil lactate dehydrogenase (LDH) release, in a caspase-11– and GSDMD-dependent manner (Fig. 1, E and F), similar to the macrophage response (fig. S3) (1, 2, 9). Cytosolic LPS also induced caspase-11–dependent cleavage of GSDMD in neutrophils (Fig. 1G). In macrophages, caspase-11–dependent GSDMD signaling triggers a loss of cytosolic K+, which in turn activates the NLRP3 inflammasome, leading to caspase-1–dependent IL-1β maturation and secretion, whereas cell death proceeds independently of NLRP3 (9, 25). Similar to macrophages, NLRP3 was dispensable for LPS-induced death of neutrophils (Fig. 1H) but was required for caspase-11/GSDMD–dependent IL-1β release (Fig. 1, I to K). Culturing neutrophils in high extracellular potassium to prevent potassium efflux during caspase-11–driven neutrophil death suppressed NLRP3-dependent IL-1β cleavage and release (Fig. 1L and fig. S4). These data indicate that the caspase-11–GSDMD–NLRP3 signaling axis is broadly conserved between neutrophils and macrophages, and neutrophils are susceptible to caspase-11–driven death.

Caspase-11, but not caspase-1, efficiently cleaves GSDMD

The observation that neutrophils are capable of dying via caspase-11 and GSDMD raises the question of how these cells resist caspase-1–dependent pyroptosis during canonical inflammasome signaling. To investigate the responsible mechanisms, we first examined LPS- and Pam3CSK4-primed macrophages and neutrophils for expression of key components of the pyroptosis pathway. Neutrophils expressed GSDMD at equivalent levels to macrophages but expressed substantially less caspase-1 and caspase-11 (Fig. 2A). We next compared the ability of caspase-1 versus caspase-11 to cleave GSDMD to generate the p30 fragment that executes pyroptosis. GSDMD was effectively cleaved by recombinant caspase-11, but GSDMD cleavage by recombinant caspase-1 was markedly less efficient (Fig. 2B). Caspase-1 but not caspase-11 cleaved pro–IL-1β (Fig. 2B). To determine whether these differences in the capacity of caspase-1 versus caspase-11 to cleave GSDMD were also reflected in inflammasome-signaling neutrophils, we transfected neutrophils with flagellin or LPS, to activate the canonical or noncanonical inflammasomes, respectively. Although both stimuli induced IL-1β secretion (Fig. 1, I to L, and fig. S5A), LPS but not flagellin triggered strong GSDMD cleavage (Fig. 2C), and flagellin did not induce GSDMD-dependent cell death (fig. S5B). We propose that the extent of GSDMD cleavage dictates the number of GSDMD-p30 pores that form in the membrane, and the latter must reach a certain threshold to overwhelm the neutrophil’s natural membrane repair mechanisms and induce cell death. Given that neutrophils express substantially less caspase-1 than macrophages and that GSDMD is most efficiently cleaved by caspase-11, our data suggest that the abundance of these caspases and their relative efficiency for GSDMD cleavage dictates whether or not a cell dies by pyroptosis downstream of canonical and noncanonical inflammasomes. It remains possible that other mechanisms also contribute to the specific suppression of caspase-1–mediated GSDMD cleavage and pyroptotic cell death in neutrophils.

Fig. 2 Neutrophils express limited caspase-1, and GSDMD is more efficiently cleaved by caspase-11 than caspase-1.

(A) Macrophages (M) and neutrophils (N) were exposed to LPS (100 ng/ml) or Pam3CSK4 (1 μg/ml) for 4 hours, and cell extracts were analyzed by immunoblot for expression of inflammasome pathway components. (B) Cytosolic lysates of HEK293T expressing GSDMD-V5 or pro–IL-1β–V5 were incubated with recombinant caspase-1 or recombinant caspase-11 for 30 min at 37°C, and substrate cleavage was assessed by anti-V5 immunoblot. (C) Neutrophils were primed with Pam3CSK4 (1 μg/ml) for 4 hours and transfected without agonist (mock) or with ultrapure LPS (10 μg/ml) or flagellin (1 μg/ml) for 4 hours. Cell lysates and supernatants were harvested and pooled for immunoblot analysis of GSDMD cleavage to p30. Blots are cropped from the same film. All data are representative of at least three biological replicates (A to C).

Neutrophil pyroptosis evokes the extrusion of NETs

With the hypothesis that pyroptosis and NETosis may be mechanistically linked, we examined LPS-transfected neutrophils for characteristics of NETosis. Cytosolic LPS triggered substantial NETosis in WT but not in Casp11- or Gsdmd-deficient neutrophils, as observed by nuclear delobulation, DNA extrusion, DNA-MPO colocalization, and histone H3 citrullination (Fig. 3, A and B, and fig. S5C). By contrast, Pam3CSK4-primed and mock-transfected neutrophils were generally negative for citrullinated histone, exhibited MPO staining in cytosolic granules, and the lobular nuclear structure characteristic of healthy neutrophils (Fig. 3A and fig. S5, C and D). Blinded image quantification scored cells as healthy or undergoing NETosis (which was defined as neutrophils with delobulated nuclei or diffuse or extruded DNA), and objective image quantification in ImageJ also scored the extent of histone H3 citrullination. Cytosolic LPS induced NETosis in ~90% of WT neutrophils (Fig. 3C) and elicited caspase-11– and GSDMD-dependent histone H3 citrullination (H3Cit; Fig. 3D). In line with earlier observations that active caspase-1 does not trigger neutrophil pyroptosis or strong GSDMD cleavage (Figs. 1, A to D, and 2C) (15), transfection of the NLRC4 inflammasome agonist flagellin did not induce LDH release, NETosis, or histone H3 citrullination (Fig. 3, C and D, and fig. S5). In monitoring the kinetics of early versus late NETosis and histone citrullination in neutrophils stimulated with cytosolic LPS, we noted that histone citrullination occurs concurrently with early NETosis, as early as 2 hours and peaking at 4 hours, whereas DNA extrusion continues to rise throughout the 8-hour time course (fig. S6). Casp11- and Gsdmd-deficient neutrophils failed to citrullinate histones or induce NETosis, even at late time points (e.g., 8 hours) of cytosolic LPS exposure (fig. S6). Thus, active caspase-11, but not active caspase-1, triggers neutrophil death and NET extrusion. Pyroptosis is defined as gasdermin-dependent cell death (2) but is morphologically distinct from NETosis, because the nuclear membrane remains intact in pyroptosis, but becomes compromised during NETosis. NETosis induced by classical stimulants proceeds independently of caspases (26) but shares the morphological features of NETosis induced by caspase-11/GSDMD signaling. We thus termed this GSDMD-induced mode of neutrophil death as caspase-11–driven NETosis.

Fig. 3 LPS-activated caspase-11 induces NETosis via GSDMD.

(A to D) Neutrophils were primed [Pam3CSK4 (1 μg/ml) for 4 hours] and then transfected with either ultrapure LPS (10 μg/ml) or flagellin (1 μg/ml) for 4 hours. Cells were stained for DNA (DAPI), MPO, and citrullinated histone H3 (H3Cit). NETosis was examined by deconvolution immunofluorescence microscopy (A to D). Images shown are representative of data from at least three biological replicates. (C) Microscopy images were blinded and quantified for NETosis, where cells with delobulated nuclei or diffuse/spread DNA were counted as cells undergoing NETosis. (D) Microscopy images were quantified using ImageJ for the percentage of cells that are positive for citrullinated histone H3. Graphs show means + SEM of data from three biological replicates (C and D). Data were analyzed for normality (Shapiro-Wilk normality test) and significance (parametric two-sided t test; ***P ≤ 0.001 or ****P ≤ 0.0001).

Caspase-11–induced NETosis proceeds independently of MPO, NE, and PAD4

Although the signaling pathways driving and executing NETosis are not fully elucidated and may not be universally deployed by all NET-inducing stimuli (27), functions for NE (18), MPO (18), and PAD4 (19, 20, 28) are implicated in histone inactivation and resultant chromatin relaxation during the early stage of NETosis induced by specific classical stimuli. We took a genetic approach to examine the signaling requirements for caspase-11–induced NETosis. Casp11 and Gsdmd deficiency blocked LPS-induced granule rupture and the resultant redistribution of MPO (Fig. 3B), suggesting that GSDMD pores may directly target and lyse granule membranes, which may provide a mechanism for NE and MPO to be liberated from granules to mediate histone degradation and chromatin decondensation. However, NE or MPO deficiency did not affect caspase-11–driven NETosis, histone citrullination, or GSDMD cleavage (fig. S7, A to C). Chromatin decondensation can also be mediated by PAD4, which is activated by a spike in cytosolic calcium (28). PAD4 deficiency abrogated LPS-induced histone citrullination but did not affect caspase-11–induced NETosis (fig. S7, A to C). The calcium chelator EGTA also blocked LPS-induced histone citrullination (fig. S7D). Because GSDMD pores enable calcium influx (29), this suggests that GSDMD pores facilitate calcium-dependent PAD4 activation and resultant histone citrullination; this accompanies, but is not absolutely required, for caspase-11–induced NETosis. Because PAD4, NE, and MPO were not required for LPS-induced NET extrusion, we thus sought to establish alternative mechanisms that may underlie the cellular changes required for caspase-11–driven NETosis.

Caspase-11 and GSDMD function coordinately to drive NETosis

Caspase-11 triggers macrophage pyroptosis by cleaving GSDMD to induce GSDMD-p30 pores in the plasma membrane. However, GSDMD-p30 is also inserted into internal cell membranes (e.g., endosome, lysosome, and Golgi) (5), with unknown functions. NETosis is associated with the sequential breakdown of multiple cell membranes; first, the neutrophil granule and nuclear membranes dissolve, allowing granular proteins and DNA to mix within the intact cell (early NETosis), and lastly, the plasma membrane ruptures to allow DNA extrusion (late NETosis). To determine which of these stages may be directed by GSDMD pores, we quantified the proportion of neutrophils in early versus late NETosis from immunofluorescence images. Caspase-11 and GSDMD were required not only for DNA extrusion through the plasma membrane during late NETosis but also for early-stage NETosis (Fig. 4A). The number of WT neutrophils in late-stage NETosis 4 hours after LPS transfection (~14% with ruptured plasma membranes and extruded NETs; Fig. 4A) agreed well with LDH release at the same time point (Fig. 1, E and F), consistent with a requirement for plasma membrane rupture for LDH release during neutrophil caspase-11–driven NETosis, as is the case during macrophage pyroptosis (30). Because Casp11- and Gsdmd-deficient neutrophils showed a profound loss of both early and late NETosis (Fig. 4A) and their granules remained intact (Fig. 3B), this suggests that GSDMD-p30 pores may target the nuclear, granule, and plasma membranes.

Fig. 4 The coordinated functions of GSDMD and caspase-11 drive early and late NETosis.

(A) Murine neutrophils were primed [Pam3CSK4 (1 μg/ml) for 4 hours] and then transfected without (mock) or with ultrapure LPS (10 μg/ml) for 4 hours. NETosis was examined by deconvolution immunofluorescence microscopy, and the percentage of cells undergoing early versus late NETosis was scored by blind counting for cells with delobulated nuclei or diffuse DNA (early) versus spread extracellular DNA (late). Data are means + SEM of three biological replicates. Data were analyzed for normality (Shapiro-Wilk normality test) and significance (parametric two-sided t test; **P ≤ 0.01 or ****P ≤ 0.0001). (B) Murine neutrophils were primed [Pam3CSK4 (1 μg/ml) for 4 hours] and then transfected without (mock) or with ultrapure LPS (10 μg/ml) for 4 hours. Cells were then fractionated to give cytosolic versus nuclear extracts, which were then analyzed by immunoblot. Data are representative of three biological replicates. (C and D) Nuclei of murine neutrophils were incubated for 15 min with recombinant full-length GSDMD, with or without recombinant caspase-11 to enable GSDMD cleavage to p30. Image analysis of (C) SYTOX uptake assessed permeabilization of the nuclear membrane and (D) SYTOX volume assessed chromatin decondensation. Data are means + SEM of n > 890 cells, representative of at least three biological replicates. Statistical analyses were performed using nonparametric Mann-Whitney t tests (two-sided; ****P ≤ 0.0001). (E) Nuclear extracts were prepared from murine neutrophils and incubated with 100 nM recombinant caspase-11 for up to 2 hours. Histone H3 cleavage was assessed by immunoblot and quantified by densitometry using ImageJ. Data are representative of three biological replicates.

To determine whether GSDMD may exert functions in the nuclear membrane, we stimulated neutrophils with cytosolic LPS and prepared nuclear extracts. Immunoblot analysis confirmed that GSDMD-p30 was indeed recruited to the nucleus in a caspase-11–dependent manner (Fig. 4B). We hypothesized that GSDMD and caspase-11 may function in concert to promote nuclear delobulation and DNA expansion. For example, GSDMD-p30 pores may permeabilize the nuclear membrane, allowing caspase-11 to access the nuclear contents to perform an analogous function to NE, in which caspase-11 cleaves and inactivates histones to relax chromatin. When recombinant caspase-11 was added to GSDMD to generate GSDMD-p30, and this mix was added to purified neutrophil nuclei, it permeabilized the nuclear membrane, leading to uptake of the DNA dye SYTOX (Fig. 4C) and caused nuclear delobulation and DNA expansion (Fig. 4D), features of early NETosis. Full-length GSDMD (i.e., not coadministered to nuclei with caspase-11) did not induce such features in neutrophil nuclei. Furthermore, when neutrophil nuclear extracts were exposed to recombinant caspase-11, caspase-11 specifically degraded histone H3 (Fig. 4E). These results suggest that caspase-11 and GSDMD together coordinate early and late NETosis. We propose that GSDMD-p30 pores dissolve the nuclear membrane and in so doing facilitate caspase-11–mediated nuclear changes, such as histone H3 degradation, leading to DNA expansion. GSDMD-p30 pores may then eventually trigger plasma membrane rupture as they do in macrophages, leading to the extrusion of the neutrophil’s decondensed DNA.

A cytosolic Gram-negative bacterium induces NETosis of human neutrophils

We considered that LPS recognition by caspase-11 and resultant NETosis may be a mechanism for neutrophils to defend themselves against cytosolic invasion by Gram-negative bacteria. To test this and provide proof of principle that this pathway operates in humans, we challenged primary blood neutrophils with Citrobacter rodentium that is well established to escape the phagosome and activate caspase-11 (9), which we used here to activate the caspase-11 human ortholog caspase-4. Similar to the well-established NETosis activator, phorbol myristate acetate (PMA; fig. S8), C. rodentium indeed triggered NETosis of human neutrophils, as detected by SYTOX uptake and chromatin externalization, and this was blocked by cell exposure to the caspase-1/4 inhibitor VX765 (Fig. 5). Thus, human neutrophils launch an inflammasome-driven NETosis pathway in response to cytosolic invasion by Gram-negative bacteria.

Fig. 5 A cytosolic Gram-negative bacterium induces NETosis of human neutrophils.

Human neutrophils were primed for 2 hours with Pam3CSK4 (1 μg/ml) and then infected with C. rodentium in the presence and absence of the caspase-1/4 inhibitor VX765 (50 μM). (A) NETosis for cells infected at an MOI of 25 was measured as SYTOX uptake over a 3-hour time course. Data are means ± SEM of infections of neutrophils from six healthy donors (****P ≤ 0.0001, two-way ANOVA). (B) NETosis was imaged by confocal microscopy, with staining for MPO and extracellular chromatin relative to the DAPI DNA stain.

Caspase-11–driven NETosis protects neutrophils from cytosolic bacterial invasion

NETs are well characterized to entrap and neutralize extracellular pathogens. We hypothesized that caspase-11–induced NETosis may be a defense pathway to restrict neutrophil cytosolic infection. To test this, we infected murine neutrophils with the Salmonella Typhimurium mutant ΔsifA, which is well characterized to localize to the cytosol of infected macrophages, where it activates macrophage caspase-11 (14). As anticipated, neutrophil infection with ΔsifA Salmonella triggered NETosis and histone citrullination via caspase-11 and GSDMD (Fig. 6, A to C). Image quantification of plasma membrane–labeled neutrophils revealed that GSDMD protected neutrophils from cytosolic infection, because Gsdmd  −/− neutrophils were more frequently infected with bacteria and exhibited strikingly higher numbers of bacteria per infected cell compared with WT neutrophils (Fig. 6D). Caspase-11–driven NETosis is thus a host defense mechanism that prevents bacteria from surviving, and potentially replicating, in the neutrophil cytosol.

Fig. 6 Neutrophil infection with cytosolic bacteria triggers caspase-11/GSDMD-driven NETosis and in vivo host defense.

(A to C) Neutrophils were primed for 4 hours with Pam3CSK4 (1 μg/ml) and then challenged with Salmonella Typhimurium ΔsifA (MOI 50, stationary phase) for 4 hours, and NETosis was examined by deconvolution immunofluorescence microscopy. (A) Microscopy images, representative of three biological replicates. Microscopy images were quantified for (B) NETosis and (C) the percentage of cells that were positive for citrullinated histone H3. Data are means + SEM of three biological replicates. (D) Microscopy images were also quantified for the percentage of cells infected (mean + SEM of 18 fields of view from three biological replicates) and the number of bacteria per infected cell (>100 infected cells per genotype from three biological replicates). (E) WT, Casp11−/−, and Gsdmd−/− mice were challenged with stationary-phase Salmonella Typhimurium ΔsifA (1 × 105 CFU ip), with and without in vivo 150-U DNase I application at 4 hours after infection. Bacterial burden in the spleen was quantified at 24 hours after infection. Data are geometric means and individual values of five individual mice per treatment group. Data were analyzed for normality using the Shapiro-Wilk normality test. Statistical analyses were performed using two-sided parametric t test (B to D, upper) or nonparametric Mann-Whitney t test (D, lower, and E). Data were considered significant when *P ≤ 0.05, **P ≤ 0.01, or ****P ≤ 0.0001.

Caspase-11/GSDMD-induced NETs mediates in vivo host defense

To determine whether caspase-11–induced NETosis provides host defense in vivo, we challenged mice with Salmonella ΔsifA via intraperitoneal injection. To test whether NETs kill bacteria and/or prevent bacterial dissemination to a secondary infection site in this context, we used established methods (31) to degrade extracellular DNA as NETs are extruded in vivo by injecting mice with recombinant deoxyribonuclease I (DNase I) at 4 hours after infection. Bacterial dissemination to the spleen was assessed 24 hours after infection. Casp11 or Gsdmd deficiency resulted in increased splenic bacterial burden (Fig. 6E), confirming that noncanonical inflammasome signaling provides host defense against Salmonella ΔsifA, as was previously indicated by ΔsifA challenge in Casp11−/− mice (14). Consistent with an essential role for caspase-11–driven NETosis in host defense, DNase I application to dismantle NETs in vivo increased splenic bacterial burden in WT but not in Casp11−/− or Gsdmd  −/− mice (Fig. 6E). NETosis was also evident in the spleens of infected WT but not of Casp11−/− mice at 48 hours after infection (fig. S9). In all, these data demonstrate that a major in vivo function of caspase-11 and GSDMD is to prevent bacterial dissemination by driving NETosis and NET-dependent bacterial killing.


NETosis is well appreciated to provide host defense against extracellular pathogens that cannot be efficiently phagocytosed (32). Until now, mechanisms protecting neutrophils from cytosolic invasion by pathogens were unclear. Here, we demonstrate the unexpected finding that cytosolic LPS and cytosolic Gram-negative bacteria trigger neutrophil caspase-11/GSDMD-dependent NET extrusion, and this protects neutrophils from cytosolic infection in vitro. Furthermore, caspase-11 provides in vivo host defense against cytosolic bacteria via the extrusion of GSDMD-dependent NETs. We thus reveal that caspase-11–driven NETosis is an antimicrobial pathway deployed against cytosolic Gram-negative pathogens that have evaded classical neutrophil defense mechanisms. Host defense by caspase-11–induced NETosis is likely mediated by multiple coordinated antimicrobial programs (fig. S10). First, like pyroptosis, caspase-11–driven NETosis allows an infected cell to produce IL-1β and undergo cell lysis, leading to bacterial release. This will prevent bacterial survival and replication in the cytosol and render bacteria susceptible to extracellular defense mechanisms, such as killing by IL-1β–recruited and activated neutrophils (33). Second, extruded NETs entrap bacteria to prevent their dissemination and can be directly antibacterial (3). Caspase-11–induced NETosis thereby allows cytosolic bacteria to be neutralized upon the NET as they are released from their host neutrophil. Last, GSDMD pores may directly target the bacterial cell wall to mediate bacteriolysis (6). Caspase-11–driven NETosis thus offers a multilayered defense strategy against bacteria that have invaded the neutrophil cytosol.

We, and others, previously reported that neutrophils are resistant to caspase-1–dependent pyroptosis (1517), unlike macrophages, which readily die by caspase-1–driven pyroptosis. We show here that neutrophils are not intrinsically resistant to inflammasome-mediated lysis, because caspase-11 triggers robust GSDMD cleavage and neutrophil NETosis. Functions for GSDMD in mediating neutrophil lysis are also supported by a recent study, in which in vitro neutrophil ageing for 24 hours triggered NE-dependent GSDMD cleavage and resultant neutrophil death (34). Although it is possible that caspase-1 triggers a pyroptosis protection program in neutrophils (e.g., membrane repair mechanisms), our data suggest that the block in caspase-1–driven pyroptosis primarily occurs upstream of GSDMD pores, because caspase-1 appears unable to efficiently cleave GSDMD into the pore-forming p30 fragment in neutrophils. This is likely to be a consequence of relatively low caspase-1 expression in neutrophils and the relatively poor efficiency of caspase-1 in cleaving GSDMD. In addition, the canonical inflammasome signaling adaptor ASC is expressed at much lower levels in neutrophils compared with macrophages, so inflammasome-activated neutrophils assemble a smaller inflammasome complex with reduced caspase-1 activity (24). In all, this suggests that caspase-1 activity in neutrophils is fine-tuned to enable sublytic GSDMD cleavage, presumably to facilitate IL-1β release (35), while resisting pyroptosis. This suggests that neutrophil viability during canonical inflammasome signaling is critical for host defense. This makes intuitive sense because infections that trigger canonical inflammasome activation are generally effectively cleared by the classical defense mechanisms of IL-1β–recruited neutrophils (e.g., degranulation, phagocytosis, and ROS); such host-protective functions of neutrophils would be lost if neutrophils died by caspase-1–dependent pyroptosis. If, under specific circumstances, neutrophil caspase-1 can induce sufficient GSDMD cleavage for cell death induction, then it remains an interesting question as to whether the cell would die by pyroptosis or NETosis. Because caspase-1 only remains active when it is bound to the inflammasome (24), it is unlikely to be able to enter the nucleus to drive the nuclear changes required for NETosis, such as histone inactivation and DNA decondensation. Thus, we speculate that NETosis may be a cell death mode specific to neutrophils signaling via the noncanonical inflammasome.

NETosis is a multistage process involving nuclear and granule membrane breakdown, chromatin relaxation and mixing with granule proteins, and lastly, the expulsion of decondensed chromatin from the cell. NETosis induced by classic stimuli (e.g., extracellular pathogens, PMA, and ionomycin) proceeds independently of caspase activity and thus caspase-11 (26) and can be induced by at least two pathways that await full characterization. One pathway involves the ROS-dependent translocation of NE from granules into the nucleus, where NE induces histone degradation and chromatin decondensation (18). This pathway is elicited by mitogens (e.g., PMA), cholesterol crystals, and extracellular pathogens (e.g., Candida albicans and group B Streptococcus) (18), and the accompanying report shows that this pathway triggers NE-dependent GSDMD cleavage and requires GSDMD-p30 pores for rupture of the granule and plasma membranes and subsequent NET extrusion (36). In an alternative pathway, the enzyme PAD4 is activated by a spike in cytosolic calcium and triggers histone citrullination to promote chromatin decondensation (19). Here, we describe an unexpected signaling pathway leading to NET extrusion (fig. S10). In contrast to NETosis induced by classic stimulants such as extracellular pathogens, caspase-11–driven NETosis is elicited by the LPS of cytosolic Gram-negative bacteria and proceeds independently of NE, MPO, and PAD4. However, caspase-11–driven NETosis does converge with the NE-driven NETosis signaling pathway because both lead to the cleavage and activation of the NETosis executioner protein GSDMD. LPS binds to, and activates, caspase-4/11 within the noncanonical inflammasome complex, leading to GSDMD cleavage, pore formation, and membrane perforation. Previous studies have characterized GSDMD pores in the macrophage plasma membrane, where they interact with the phosphoinositide, phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2], in the inner leaflet (5, 6) and mediate cell rupture (1, 2, 48, 29). Here, we show in neutrophils previously uncharacterized functions for GSDMD in targeting the nuclear envelope. The phospholipid content of the nuclear membrane is ill-defined, but several reports suggest that it contains PI(4,5)P2, including in neutrophils (37, 38), suggesting a potential mechanism for GSDMD-p30 nuclear membrane recruitment in neutrophils. Our data indicate that the combined actions of GSDMD and caspase-11 drive nuclear permeabilization, chromatin relaxation, and plasma membrane rupture during LPS-induced NETosis. Mechanistically, our data suggest that GSDMD pores in the nuclear membrane allow nuclear membrane breakdown and caspase-11 access to chromatin, wherein caspase-11 performs an analogous function to NE, mediating histone clipping and inactivation to enable DNA expansion. During caspase-11–driven NETosis, PAD4 becomes activated, likely via calcium influx through GSDMD pores, and PAD4 citrullinates histone H3; this accompanies but is not required for NETosis. Thus, although caspase-11–induced NETosis and classical NE-driven NETosis are morphologically similar, they occur in response to distinct challenges via different signaling mechanisms that converge on a common executioner protein, GSDMD.

In addition to their host-protective functions, NETs can drive pathology in a multitude of inflammatory, autoimmune, and other diseases, such as sepsis, lupus, rheumatoid arthritis, vasculitis, type 1 diabetes, gout, preeclampsia, periodontitis, and cancer (21, 39). A gasdermin family member closely related to GSDMD, GSDME, was recently shown to execute pyroptosis during chemotherapy (40) and uncontrolled apoptosis (41). NETs can promote cancer progression (39), and lupus patients show defects in apoptotic cell clearance (42). Characterizing functions for GSDMD and GSDME in driving NETosis during autoimmune and other inflammation-linked diseases could thus reveal these molecules as previously unappreciated therapeutic targets for such conditions.


Study design

This study aimed to define inflammasome signaling pathways leading to GSDMD-dependent neutrophil death by NETosis and antibacterial host defense. To do this, we challenged murine and human neutrophils with inflammasome agonists and bacteria in vitro and monitored inflammasome signaling outputs (cleavage and/or release of caspase-1, GSDMD, and IL-1β), cell lysis (LDH release), and NETosis (microscopy and SYTOX uptake). The contribution of caspase-11–dependent NETs to host defense was assessed by murine in vivo challenge with ΔsifA Salmonella. Signaling pathways were mapped through the use of gene-deficient mice or the application of specific inhibitors. Where image quantification required manual image scoring, images were blinded and randomized before their analysis by a researcher not involved in image acquisition. No other formal randomization or blinding method was used elsewhere in the study. The number of independent replicates for each experiment is outlined in the figure captions.

Bacterial strains

S. enterica serovar Typhimurium SL1344 strains and C. rodentium were grown overnight at 37°C in Luria-Bertani broth at 200 rpm (stationary phase). To induce Salmonella SPI1-T3SS expression, overnight bacteria were diluted 1:40 and grown for 3 hours (log phase). The SL1344 ΔsifA mutant was generated using the λ red–mediated homologous recombination system (43). mCherry-expressing SL1344 ΔsifA was generated by electroporation of the low-copy plasmid pFPV-mCherry (Addgene) and used for some imaging experiments.


All experiments were performed with age- and sex-matched mouse cohorts. C57BL/6, Casp11−/− (9, 44), Gsdmd  −/− (1, 45), Ne−/− (46), Mpo−/− (47), and Pad4−/− (48) mice were housed in specific pathogen–free facilities at the University of Queensland or the University of Lausanne. All animal experiments were performed with approval from the University of Queensland’s animal ethics committee or the veterinary office of the Canton de Vaud and Swiss animal protection law (license VD3257).

Murine primary macrophage and neutrophil preparation and stimulation

Mature neutrophils were purified from murine bone marrow using magnetic-assisted cell sorting and used for experiments immediately after purification, while bone marrow progenitors were differentiated with recombinant human colony-stimulating factor–1 (CSF-1) over 7 days to generate macrophages, as previously described (15).

Inflammasome assays

Neutrophils were seeded at a density of 1 × 107 cells/ml in Opti-MEM (Life Technologies) supplemented with aprotinin (4 μg/ml; Sigma). Differentiated macrophages were seeded at a density of 1 × 106 cells/ml in RPMI 1640 supplemented with 10% heat-inactivated fetal calf serum (FCS), GlutaMAX (Life Technologies), and recombinant human CSF-1 (150 ng/ml; endotoxin-free, produced in insect cells by the University of Queensland Protein Expression Facility) on day 6, and cell culture medium was replaced with CSF-1–supplemented Opti-MEM on day 7 of differentiation. Cells were primed with ultrapure LPS (100 ng/ml) or Pam3CSK4 (1 μg/ml; Invivogen) for 4 hours. Cells were then transfected with ultrapure Escherichia coli K12 LPS (2 to 10 μg/ml; Invivogen) or ultrapure flagellin (1 μg/ml; Invivogen) using 0.25% FuGENE HD (Promega) and centrifuged at 700g for 10 min at room temperature or stimulated with 5 μM nigericin. IL-1β in cell-free supernatants was quantified by enzyme-linked immunosorbent assay (eBioscience). Cell extracts and methanol/chloroform-precipitated cell-free supernatants were analyzed by immunoblot as previously described (15), using antibodies against caspase-1 (Casper-1, AdipoGen), caspase-11 (EPR18628, Abcam), IL-1β (AF-401-NA, R&D Systems), GSDMD (EPR19828, Abcam), histone H3 (96C10, Cell Signaling Technology), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH; polyclonal rabbit, BioScientific). LDH release into the cell culture supernatant was quantified using CytoTox 96 nonradioactive cytotoxicity assay (Promega).

Murine neutrophil and macrophage in vitro infection assays

Cells were primed with ultrapure E. coli K12 LPS (100 ng/ml; Invivogen) or Pam3CSK4 (1 μg/ml; Invivogen) for 4 hours. Cells were then infected with WT Salmonella SL1344 or isogenic ΔsifA at a multiplicity of infection (MOI) of 25 (neutrophils) or 10 (macrophages) and centrifuged immediately at 700g for 10 min at room temperature. Cells were then incubated at 37°C for 25 min to allow bacterial uptake, before adding gentamicin (100 μg/ml; Life Technologies) to the culture medium for the remainder of the assay to kill extracellular bacteria. For microscopy, stationary-phase grown bacteria were opsonized for 15 min at 37°C with 10% rat serum and added to neutrophils at an MOI of 50. Infection was allowed to proceed for 40 min, and then, cells were washed twice with Opti-MEM, and cell culture medium was replaced with medium containing gentamicin (20 μg/ml) for the remainder of the assay. The cells were then further incubated for the indicated times before harvesting cell-free supernatants and cell lysates for cytokine and LDH release assays or fixed in paraformaldehyde (PFA) for imaging.

Deconvolution immunofluorescence microscopy

Neutrophils were plated on 0.0001% poly-l-lysine–coated glass coverslips at 1 × 106 cell/ml and fixed in 2% PFA for 15 min at 37°C. Cells were permeabilized in 0.1% Triton X-100 for 10 min, blocked in 0.5% bovine serum albumin (BSA) for 30 min, and incubated with primary and respective secondary antibodies for 1 hour. Primary antibodies used were H3Cit (ab5103; Abcam) and MPO (goat polyclonal; R&D Systems). Alexa Fluor 488, Alexa Fluor 594, and Alexa Fluor 647 were purchased from Molecular Probes. Coverslips were mounted with ProLong Gold Antifade Reagent (Life Technologies). Z-stacks were imaged using a DeltaVision Olympus IX80 inverted wide-field deconvolution microscope equipped with Olympus U-Apochromat ×40/1.35, Plan-Apochromat ×60/1.35, or UPLS-Apochromat ×100/1.4 oil lenses and a Lumencor 7 line light-emitting diode lamp. Images were captured using a Photometrics CoolSNAP HQ2 camera. Images were deconvolved using 10 cycles of conservative iterative deconvolution using softWoRx. Image analysis was performed using ImageJ software. The percentage of neutrophils with citrullinated histone H3 and undergoing NETosis was quantified via image analysis of five to seven nonoverlapping fields of view (60×; in which each field of view contained about 20 to 25 cells) for each treatment and genotype for each of three biological replicates (>300 cells total for each condition) unless otherwise indicated. To quantify histone H3 citrullination, deconvolved images were subjected to quantitative analysis with a custom ImageJ macro script. Images were maximum intensity–projected, split into individual channels, and subject to 100-pixel rolling-ball background subtraction. Total cell number in each field of view was determined by counting intensity-thresholded (250-65,535) 4′,6-diamidino-2-phenylindole (DAPI) (nuclear) labeling and using size exclusion to eliminate noise, whereas H3-positive cells were subjected to 100-pixel rolling-ball background subtraction and median filtering (radius, 2 pixels). High-intensity labeled cells were selected as positive via thresholding (750:65,535). The percentage of cells undergoing NETosis was determined by blinded manual counting, in which those cells with delobulated nuclei or diffuse DNA (early NETosis) or spread extracellular DNA (late NETosis) were identified from DAPI-stained blinded images. For analysis of intracellular bacteria, deconvolved images of WGA-680/DAPI-stained, infected neutrophils were split into individual channels. Total number of cells per field of view was determined in ImageJ using DAPI as above, and infected cells and numbers of bacteria per cell were enumerated by analysis of mCherry-positive bacteria that were confined within wheat germ agglutinin–stained cells.

Isolation and NETosis of human primary neutrophils

Blood was taken from healthy human volunteers with written informed consent, with the approval of the Charité–Universitätsmedizin Berlin Ethics Committee and in accordance with the Declaration of Helsinki. Neutrophils were isolated as previously described (26) and seeded at 106/ml in RPMI 1640 (without phenol red) supplemented with 10 mM Hepes and 0.02% human serum albumin in a 96-well plate. Neutrophils were primed for 2 hours with Pam3CSK4 (1 μg/ml). Where indicated, 50 μM VX765 (Invivogen) was added to cells during the final 30 min of priming. Neutrophils were challenged with stationary phase C. rodentium at the indicated MOI and centrifuged immediately for 10 min at 300g and incubated at 37°C for a further 30 min to allow phagocytosis. Gentamicin (100 μg/ml) was then added to the culture medium to kill extracellular bacteria. After 30 min, the culture medium was replaced with fresh media containing 50 nM SYTOX Green (Invitrogen). Neutrophils were also stimulated with 100 nM PMA as a positive control for NETosis and analyzed as above. Cell death was quantified over time by measuring fluorescence intensity using a fluoroscan plate reader (excitation/emission, 485/518) and acquiring pictures at an EVOS microscope. Pam3CSK4-primed neutrophils infected at an MOI of 10 for 2.5 hours were also assessed for NETosis by confocal microscopy. Cells were fixed by pH shift fixation [3% PFA k-Pipes (80 mM; pH 6.8) for 5 min followed by 3% PFA borax (100 mM) for 10 min]. After washing with phosphate-buffered saline (PBS), cells were permeabilized with 0.1 % Triton X-100 for 5 min at 4°C and blocked with 3% BSA for 1 hour. Cells were then incubated overnight at 4°C with primary antibodies [anti-MPO, A0398, Dako; antichromatin PL2.3 (49)]. After three washes with PBS, secondary antibodies were added for 30 min in 3% BSA, and DNA was counterstained with DAPI (1 μg/ml). After three further PBS washes, images were acquired using a Leica SP8 confocal microscope.

In vivo infection

Mice were challenged intraperitoneally (ip) with 1 × 105 colony-forming units (CFU) of the Salmonella Typhimurium ΔsifA mutant grown to stationary phase. At 4 hours after infection, mice were administered 150 U of cell culture–grade DNase I (Roche) intraperitoneally, to degrade NETs. Mice were sacrificed at 24 hours after infection for determination of splenic bacterial burden by serial dilution or at 48 hours after infection for spleen fixation in 4% PFA for 24 hours at 4°C, followed by paraffin embedding.

Immunofluorescence staining of fixed tissue

Immunofluorescence was performed on 7-μm spleen sections. Sections were dewaxed using xylene and rehydrated in a gradient of ethanol (100, 95, 80, and 70%). Antigen unmasking was performed by boiling the slides in sodium citrate buffer [10 mM trisodium citrate and 0.05% Tween (pH 6.0)] for 10 min. Slides were then blocked (5% FCS and 0.3% Triton X-100 in PBS). Slides were incubated in primary antibodies overnight (rabbit anti-H3Cit, goat anti-MPO) in blocking buffer and then incubated for 1 hour with secondary antibodies (anti-rabbit Alexa Fluor 594 and anti-goat Alexa Fluor 488). Slides were then stained with DAPI for 10 min, mounted using Dako mounting media, and sealed. Images were captured at 63× using oil immersion with Carl Zeiss Meta Inverted LSM 510 microscope (Carl Ziess) and processed using ImageJ.

Recombinant protein expression

Murine caspases were expressed with C-terminal His8 tags in E. coli BL21 Gold. ΔCARD_caspase-1 was expressed as a GST-DmrB fusion protein, and expression was induced with 200 μM isopropyl-β-d-thiogalactopyranoside (IPTG) for 8 hours at 18°C. Caspase-11 expression was induced with 200 μM IPTG for 6 hours at 30°C. ΔCARD_caspase-1 and caspase-11 were purified using Ni Sepharose 6 Fast Flow resin as described (50). Active caspase concentration was quantified via active site titration using zVAD-fmk (50). Murine GSDMD was expressed as a glutathione S-transferase (GST) fusion protein in E. coli BL21, and expression was induced with 200 μM IPTG for 12 hours. GST-GSDMD was purified using MagneGST gluthatione beads (Promega), where GSDMD was cleaved from GST using 3C protease (expressed in house) and eluted in PBS. GSDMD concentration was determined using the Beer-Lambert law.

Neutrophil fractionation

Neutrophils were fractionated into cytosolic versus nuclear extracts using standard techniques. Briefly, 4 million neutrophils were incubated with 150 μl of hypotonic buffer [10 mM Hepes (pH 8.0), 10 mM KCl, 3 mM NaCl, 3 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 2 mM dithiothreitol (DTT), 2 mM phenylmethylsulfonyl fluoride (PMSF), cOmplete mini protease inhibitor cocktail; Roche)] and incubated on ice for 15 min. IGEPAL was then added to a final concentration of 0.5%, and the cells were gently vortexed and then centrifuged at 500g for 10 min at 4°C. The cytoplasmic (soluble) fraction was removed, and the pellet (nuclear fraction) was washed once more with hypotonic buffer before resuspension in Western loading dye.

Assays with neutrophil nuclei

Nuclei were prepared from murine neutrophils primed for 4 hours with Pam3CSK4 (1 μg/ml). Cells were centrifuged at 400g for 5 min, resuspended in lysis buffer (100 mM KCl, 1 mM NaCl, 1 mM ATP.Na2, 3.5 mM MgCl2, 10 mM Pipes, 0.5 mM PMSF, and cOmplete mini protease inhibitor cocktail; Roche) at 5 × 107 cells/ml and disrupted by nitrogen cavitation at 380 psi for 5 min at 4°C before centrifugation at 500g for 5 min at 4°C. The pellet was washed with lysis buffer and again centrifuged. Nuclei were gently resuspended at 1 × 106 cells/ml in assay buffer [200 mM NaCl, 50 mM KCl, 50 mM Hepes (pH 8.0), and 10 mM DTT] and plated in a 96-well plate. SYTOX green nucleic acid stain (100 nM; Thermo Fisher Scientific) was added to each well 15 min before imaging. Recombinant proteins (2 μM full-length GSDMD and/or 100 nM caspase-11) were added to wells immediately before live-cell imaging using a Nikon deconvolution microscope in a 37°C chamber. Imaris (Bitplane) software quantified SYTOX volume and sum mean intensity of each cell.

Caspase cleavage assays

Active site–titrated, recombinant caspase-1 and recombinant caspase-11 (2 μM) were incubated in high-salt buffer [50 mM Hepes (pH 7.4), 50 mM NaCl, 1 M sodium citrate, 0.01% CHAPS, and 10 mM DTT] for 30 min to promote dimerization. Caspases were then serially diluted in caspase buffer [100 mM NaCl, 10 mM Pipes (pH 7.3), 10% (w/v) sucrose, 0.1% (w/v) CHAPS, and 10 mM DTT] and incubated for 30 min at 37°C with cytosolic extracts from human embryonic kidney (HEK) 293T cells overexpressing V5-tagged murine GSDMD or pro–IL-1β. Alternatively, caspase-11 was added to neutrophil nuclear extracts. For this, nuclei were isolated as above and then incubated for 20 min on ice in nuclear lysis buffer [20 mM Hepes (pH 7.5), 25% glycerol, 0.8 M KCl, 1 mM MgCl2, 1% IGEPAL CA-680, 0.5 mM EDTA, 2 mM DTT, and 0.5 mM PMSF]. The nuclear extract was cleared by centrifugation at 14,000g for 15 min at 4°C and then diluted in caspase buffer. SDS–polyacrylamide gel electrophoresis loading dye was added to quench cleavage reactions, and samples were boiled for 5 min before analysis by immunoblot.

Statistical analyses

Statistical analyses were performed using GraphPad Prism software. Data were analyzed for normality using the Shapiro-Wilk normality test. Statistical analyses on normally distributed data sets were performed using parametric t tests (two-sided), whereas data sets with non-normal distributions were analyzed using nonparametric Mann-Whitney t tests (two-sided). For repeated measures over time, two-way analysis of variance (ANOVA) was performed. In vivo experiments were analyzed using the Mann-Whitney test. Data were considered significant when *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, or ****P ≤ 0.0001.


Fig. S1. GSDMD mediates rapid macrophage pyroptosis.

Fig. S2. Salmonella Typhimurium and nigericin induce IL-1β secretion from WT neutrophils.

Fig. S3. Cytosolic LPS triggers caspase-11–dependent cell death and IL-1β production from macrophages.

Fig. S4. Extracellular potassium blocks noncanonical NLRP3 activation and resultant release of mature IL-1β.

Fig. S5. Cytosolic LPS but not mock or flagellin transfection triggers GSDMD-dependent cell death.

Fig. S6. Time course analysis profiles the kinetics of caspase-11/GSDMD-induced NETosis.

Fig. S7. NE, MPO, and PAD4 are dispensable for GSDMD cleavage and caspase-11–driven NETosis.

Fig. S8. PMA induces NETosis of human neutrophils.

Fig. S9. Cytosolic bacteria triggers caspase-11–dependent NETs in vivo.

Fig. S10. Model for LPS-driven NETosis via GSDMD during cytosolic infection.

Table S1. Raw data file.


Acknowledgments: We thank V. Dixit (Genentech) for GSDMD reagents, L. Luo for 3C protease, and A. Zychlinsky for advice and useful discussions. Funding: This work was supported by the National Health and Medical Research Council of Australia (project grants 1064945 and 1122240 to K.S., Senior Research Fellowship 1107914 to M.J.S., and Career Development Fellowship 1141131 to K.S.) and the Australian Research Council (Future Fellowship FT130100361 to K.S.). D.B. was supported by a University of Queensland Postdoctoral Fellowship. P.B. is supported by a grant from the European Research Council (ERC) (ERC-2017-CoG-770988-InflamCellDeath). Imaging was performed in the Australian Cancer Research Foundation–funded Institute for Molecular Bioscience Cancer Biology Imaging Facility. Author contributions: K.W.C., M.M., and D.B. designed and performed experiments and analyzed data. J.B.v.P. performed immunoblots. G.S. designed and performed experiments with human neutrophils. D.R. performed immunofluorescence on tissue sections. N.D.C. assisted with microscopy analysis. P.B. and M.J.S. provided reagents and expert advice. K.W.C. and K.S. designed the study and wrote the manuscript with input from all authors. K.S. supervised the study. Competing interests: K.S. is a co-inventor on patent applications for NLRP3 inhibitors that have been licensed to Inflazome Ltd., a company headquartered in Dublin, Ireland. Inflazome is developing drugs that target the NLRP3 inflammasome to address unmet clinical needs in inflammatory disease. The authors declare that they have no other competing interests. Data and materials availability: Nalp3/NLRP3 knockout mice were obtained by a materials transfer agreement (MTA) with N. Fasel, University of Lausanne. The Gsdmd−/− mice were obtained by MTA with Genentech Materials.

Stay Connected to Science Immunology

Navigate This Article