Gasdermin D restricts Burkholderia cenocepacia infection in vitro and in vivo

Burkholderia cenocepacia (B. cenocepacia) is an opportunistic bacterium; causing severe life threatening systemic infections in immunocompromised individuals including cystic fibrosis patients. The lack of gasdermin D (GSDMD) protects mice against endotoxin lipopolysaccharide (LPS) shock. On the other hand, GSDMD promotes mice survival in response to certain bacterial infections. However, the role of GSDMD during B. cenocepacia infection is not yet determined. Our in vitro study shows that GSDMD restricts B. cenocepacia replication within macrophages independent of its role in cell death through promoting mitochondrial reactive oxygen species (mROS) production. mROS is known to stimulate autophagy, hence, the inhibition of mROS or the absence of GSDMD during B. cenocepacia infections reduces autophagy which plays a critical role in the restriction of the pathogen. GSDMD promotes inflammation in response to B. cenocepacia through mediating the release of inflammasome dependent cytokine (IL-1β) and an independent one (CXCL1) (KC). Additionally, different B. cenocepacia secretory systems (T3SS, T4SS, and T6SS) contribute to inflammasome activation together with bacterial survival within macrophages. In vivo study confirmed the in vitro findings and showed that GSDMD restricts B. cenocepacia infection and dissemination and stimulates autophagy in response to B. cenocepacia. Nevertheless, GSDMD promotes lung inflammation and necrosis in response to B. cenocepacia without altering mice survival. This study describes the double-edged functions of GSDMD in response to B. cenocepacia infection and shows the importance of GSDMD-mediated mROS in restriction of B. cenocepacia.

Previous studies have shown that the presence of cytosolic lipopolysaccharide (LPS) is responsible for the cleavage and auto-activation of CASP11 20,21 . Consequently, the liberation of LPS from the phagosome harboring the pathogens is necessary for CASP11 cleavage 22 . Whether the cleavage of CASP11 post-B. cenocepacia infection is influenced by GSDMD has yet to be determined. Notably, we found that the total (cell lysate and supernatants) amount of cleaved CASP11 in response to B. cenocepacia is reduced in gsdmd −/− macrophages compared to WT macrophages (Fig. 1C). It is important to note that due to the absence of GSDMD pores in gsdmd −/− macrophages, the cleaved form accumulates inside the cell (Fig. 1A). Together, these results indicate that GSDMD contributes to CASP11 cleavage.
Bacterial secretory systems (SS) release bacterial virulence factors 26 . These SS are essential for bacterial persistence within host cells. B. cenocepacia possesses different types (T) of SS including T3SS, T4SS, and T6SS 27 . The different SS of B. cenocepacia play a role in IL-1β activation 13 . To determine which SS contributes to GSDMD cleavage and inflammasome activation, WT macrophages were infected with WT B. cenocepacia and different SS mutants for 6 h at MOI of 10. Western blot analysis of the cell lysates combined with the cell culture supernatants (total) demonstrate that infection with SS −/− mutants led to decreased cleavage of GSDMD, CASP1, and IL-1β (Fig. 1D). T3SS −/− and T4SS −/− mutants exerted more cleavage of CASP11 (Fig. 1D). To further test whether different SS play a role in invasion or in the intracellular survival and replication within macrophages, we infected macrophages with the different B. cenocepacia mutants for 0.5, and 6 h then determined intracellular colony forming unit (CFU). Despite having fast growth rates in LB (Fig. S1B,C), all the mutants displayed defective replication within macrophages compared to the parental strain (Fig. S1D,E). This is independent of cell death as they cause less or equivalent cell death compared to WT B. cenocepacia (Fig. S1F). In contrast to the T6SS −/− , which shows increased invasion 12 , no difference in invasion between WT and T3SS −/− or T4SS −/− could be observed (Fig. S1D). Overall, these results indicate that, both host CASP11 and CASP1 in addition to bacterial T3SS, T4SS, and T6SS contribute to B. cenocepacia-induced GSDMD cleavage within macrophages. GSDMD contributes to the restriction of B. cenocepacia in vitro. The role of GSDMD during infection differs according to the pathogen. For instance, it enhances Escherichia coli replication 28 , while it restricts the replication of B. thailandensis and Legionella pneumophila 29,30 . However, its distinctive role during B. cenocepacia infection is unknown. To unveil this, macrophages from WT, gsdmd −/− , and casp11 −/− mice were infected with B. cenocepacia for 0.5, 3, and 6 h and macrophage associated colony forming units (CFU) were determined. Compared to WT macrophages, both gsdmd −/− and casp11 −/− macrophages had similar levels of uptake as indicated by similar CFU at 0.5 h, however, they are more permissive to B. cenocepacia at 6 h post-infection ( Fig. 2A). Additionally, using red fluorescent B. cenocepacia (K56-2), intracellular bacteria were enumerated after 2 and 6 h (Fig. 2B). Gsdmd −/− macrophages displayed significantly more B. cenocepacia intracellularly when compared to WT at 6 h post-infection (Fig. 2C)   www.nature.com/scientificreports/ We showed previously that casp11 −/− macrophages release less inflammasome-independent cytokines including KC (CXCL1) in their supernatants 6 h post-B. cenocepacia infection 12 , yet the mechanism of GSDMD involvement remains unknown. As a pore forming protein, GSDMD may affect the release of IL-1β in addition to inflammasome-independent cytokines. We examined if gsdmd −/− macrophages differentially release cytokines in comparison to WT and casp11 −/− macrophages post-B. cenocepacia infection. Indeed, supernatants from gsdmd −/− macrophages elicited significantly less IL-1α, IL-1β, IL-6, and TNF-α when compared to WT and casp11 −/− supernatants. Additionally, both casp11 −/− and gsdmd −/− macrophages had significantly less CXCL1 release in comparison to WT (Fig. 2D). These results demonstrate that GSDMD can restrict B. cenocepacia and mediate the release of different inflammasome-dependent and -independent, inflammatory cytokines. Together, in the absence of GSDMD, there is more B. cenocepacia yet less cytokine secretion.

GSDMD restricts B. cenocepacia infection irrespective of cell death.
Pyroptosis is an inflammatory, programmed cell death associated with excess GSDMD pore formation, and characterized by cytokine efflux, water influx, cell swelling, and finally osmotic lysis 31,32 . Cell death can deprive the pathogen from the niche required for its persistence and growth. To investigate if GSDMD-mediated restriction of B. cenocepacia is due to its function as an executer of pyroptotic cell death, the release of lactate dehydrogenase (LDH) was measured in supernatants of WT, gsdmd −/− , and casp11 −/− macrophages at 6 h post-B. cenocepacia infection. Gsdmd −/− and casp11 −/− macrophages exhibited lower levels of LDH released into supernatants when compared to WT (Fig. 3A), indicating that WT macrophage undergo more pyroptotic cell death compared to gsdmd −/− and casp11 −/− macrophages. To examine if increased cell death in WT macrophages is responsible for B. cenocepacia restriction and not through other potential functions of GSDMD, glycine was used to protect WT macrophages from cell lysis. Glycine has cytoprotective properties against pyroptosis as it buffers the extracellular media and prevents ion influx, thereby inhibiting swelling and cellular lysis [33][34][35][36][37] . When treated with glycine, B. cenocepacia growth or gentamicin sensitivity in LB were not affected (Fig. S2A). WT, gsdmd −/− , and casp11 −/− macrophages were infected for 6 h in the absence or presence of glycine (5 mM), and the release of LDH and GAPDH were determined. Infected WT macrophages, in the presence of glycine, displayed significantly lower levels of LDH and GAPDH release compared to those not treated with glycine. Thus, in the presence of glycine, there was no significant difference in the LDH and GAPDH release among WT, gsdmd −/− and casp11 −/− macrophages (Fig. 3B,C). To determine if protection against pyroptosis affects the production and secretion of inflammatory proteins, WT, gsdmd −/− and casp11 −/− macrophages were infected with B. cenocepacia in the absence or presence of glycine. Compared to the non-glycine treated macrophages, WT macrophages displayed decreased CASP1 and IL-1β in the cell supernatant. Conversely, they exhibited more expression of cleaved CASP11, CASP1 and IL-1β in cell lysate in the presence of glycine (Fig. 3C). These results indicated that glycine pre-treatment of macrophages can efficiently protect against pyroptosis during B. cenocepacia infection.
We then determined if gsdmd −/− macrophages would still be more permissive to B. cenocepacia than WT cells during glycine treatment. Macrophages from WT and gsdmd −/− mice were infected with B. cenocepacia and CFU were determined at 0.5, 3, and 6 h in the presence of glycine. Both WT and gsdmd −/− macrophages exhibited similar invasion of B. cenocepacia. However, gsdmd −/− macrophages were more permissive at 6 h post-infection compared to glycine-protected WT cells (Fig. 3D). These results confirm that GSDMD plays a role in hindering B. cenocepacia replication in addition to its function in mediating cell death.
WT macrophages release more cytokines than gsdmd −/− macrophages (Fig. 2D), however, WT cells exhibited higher levels of cell death during B. cenocepacia infection which may lead to increased inflammatory cytokine release (Fig. 3A). Therefore, we sought to decipher the contribution of GSDMD pore formation or cell lysis to inflammatory cytokine release. To answer this question, we added glycine, which does not block GSDMD pores but does protect against lytic cell death, and collected supernatant 6 h post-B. cenocepacia infection 38 . Glycine treatment and the prevention of cell lysis reduced the amount of IL-1α, IL-1β, and IL-6 in gsdmd −/− supernatants (Fig. S2B). Glycine-treated gsdmd −/− macrophage supernatants still had significantly less IL-1α, IL-1β, and CXCL1 in comparison to supernatants of WT macrophages, but not TNF-α or IL-6, indicating that the release of these cytokines was not mainly dependent on cell lysis (Fig. 3E). In addition, WT cells exhibited similar amounts of IL-1α, IL-1β, and TNF-α release with and without treatment, but lower levels of CXCL1 and IL-6 in infection supernatants (Fig. S2B). Together, these results confirm that GSDMD promotion of IL-1α, IL-1β, and CXCL1 release is not dependent on cell lysis in vitro.

GSDMD restricts the replication and dissemination of B. cenocepacia in vivo.
To test if GSDMD plays a role in B. cenocepacia replication in vivo, WT and gsdmd −/− mice were infected intratracheally with a sub-lethal dose (10 × 10 6 CFU/mouse) of B. cenocepacia. To verify if equal inoculums of bacteria were delivered to all mice, CFU enumeration was performed on the lung homogenates 4 h post-infection. There was no significant difference in the bacterial loads between WT and gsdmd −/− mice (Fig. 4A). However, after 48 h the bacterial burden was significantly increased in the lung, spleen, and liver of gsdmd −/− mice when compared to the corresponding WT mice (Fig. 4A). This indicates that GSDMD promotes the restriction of B. cenocepacia replication and prevents dissemination in vivo. To determine if the permissiveness of gsdmd −/− is associated with mortality, WT and gsdmd −/− mice were monitored after infection intratracheally with lethal and sub-lethal doses. Interestingly, no significant difference in the survival rate was detected between WT and gsdmd −/− mice in all the indicated doses ( Fig. S3A-C), indicating that higher bacterial burden in the absence of GSDMD is not associated with increased mortality.
To test if GSDMD plays a role in promoting inflammation in vivo, lung tissues were collected from WT and gsdmd −/− mice infected with 10 × 10 6 CFU/mouse for 48 h. Hemotoxylin and Eosin-staining was performed to assess morphological and histological changes. Overall, infection of WT lungs was accompanied In presence of glycine www.nature.com/scientificreports/ with severe inflammatory infiltrates and a larger percentage of the lung tissue inflammation when compared to gsdmd −/− lungs (Fig. 4B). Semi-quantitative histopathological scoring demonstrates that the infected WT lung tissues have significantly more areas of inflammation, edema, and necrosis than their gsdmd −/− counterparts. Additionally, they exhibited greater neutrophil infiltration, and clusters mixed with necrotic debris in the parenchyma, affecting alveoli, capillaries and small blood vessels (Fig. 4B,C). Myeloperoxidase (MPO) is one of the degrading enzymes typically produced by neutrophils 39,40 . To determine if decreased neutrophil infiltration, in the absence of GSDMD, correlates with decreased cytokine and/or MPO release, we collected the bronchoalveolar lavage fluid (BALF) from WT and gsdmd −/− mice infected intratracheally with B. cenocepacia (30 × 10 6 CFU/ mouse) for 24 h and measured various cytokines IL-1α, IL-1β, IL-6, TNF-α, CXCL1 in addition to MPO. There was no significant difference in all the measured cytokines in addition to MPO, except for CXCL1 which was less in gsdmd −/− BALF (Fig. 4D, S3D). Overall, these results show that GSDMD contributes to the restriction of B. cenocepacia replication, lung inflammation, and BALF CXCL1 production in vivo. However, the mechanism of GSDMD-mediated restriction of B. cenocepacia is still unclear.

GSDMD contributes to bactericidal mROS production during B. cenocepacia infection.
One of the main challenges to the intracellular lifestyle of B. cenocepacia is the production of host reactive oxygen species (ROS) [41][42][43][44] , however it is unclear if GSDMD contributes to the production of mROS during B. cenocepacia infection and if it contributes to the restriction of the organism. Therefore, we used the antioxidant N-acetylcysteine (NAC) to determine if decreasing ROS will allow more B. cenocepacia replication within macrophages. Since NAC (3 mM) was found to have a direct effect on B. cenocepacia replication in LB (Fig. S4A), it was added to macrophages post. B. cenocepacia infection. Despite increasing cell death (Fig. S4B), scavenging ROS significantly increased B. cenocepacia replication in all the infected genotypes (Fig. 5A). This indicates that ROS manipulation can significantly affect B. cenocepacia survival within macrophages. GSDMD inserts into mitochondrial membranes, when activated, contributing to mitochondrial ROS (mROS) release which is the major source of ROS in the cell 45 . Thus, we postulated that gsdmd −/− macrophages produce less mROS during B. cenocepacia infection, contributing to B. cenocepacia permissiveness. We treated WT and gsdmd −/− macrophages with MitoSOX dye which selectively targets the mitochondria where it is oxidized, producing a highly fluorescent product that is used for mROS measurement. We found that WT and gsdmd −/− macrophages induced comparable mROS during B. cenocepacia infection (Fig. 5B). Yet, the difference in mROS production between the macrophage genotypes could be masked by earlier death of WT macrophages. Therefore, glycine was added and the amount of mROS was determined. Notably, gsdmd −/− and casp11 −/− macrophages produced lower mROS in the presence of glycine when compared to non-pyroptotic WT cells (Figs. 5B, S4C). To further confirm if the defective mROS production associated with gsdmd −/− macrophages contributes to permissiveness to B. cenocepacia, Mitotempo (MT), a selective mitochondria-targeted antioxidant, was used 46 . MT does not induce macrophage death (Fig. S4D) and has no direct bactericidal effect against B. cenocepacia (Fig. S4E). WT, gsdmd −/− , and casp11 −/− macrophages were infected with B. cenocepacia for 0.5, 3, and 6 h in the absence and presence of MT (20 µM) then, macrophage associated bacterial CFUs were enumerated. B. cenocepacia replication increased in all the infected macrophage phenotypes, however this increase was more significant in MT-treated WT macrophages compared to gsdmd −/− macrophages at 6 h post-infection (Fig. 5C). Together, these results demonstrate that mROS contributes to the restriction of B. cenocepacia and that WT macrophages produce more mROS than gsdmd −/− and casp11 −/− macrophages during B. cenocepacia infection however, the mechanism is unclear.
GSDMD inserts pores in mitochondrial membranes which can increase the production of mROS by disrupting mitochondrial membrane potential (MMP) (ΔΨ m ) 45,[47][48][49][50][51][52] . To determine if GSDMD disrupts MMP (ΔΨ m ) during B. cenocepacia. TMRM dye (10 nM) was added 0.5 h before the designated time point (6 h post-B. cenocepacia infection). This dye accumulates in a negatively charged polarized mitochondria giving orange fluorescence. On the other hand, when MMP is disrupted, the dye is dispersed throughout the cell cytosol and fluorescence levels drop dramatically. Accordingly, MMP was found to be higher in gsdmd −/− macrophages during B. cenocepacia infection in comparison to WT counterparts (Fig. 5D). This indicates that activation and cleavage of GSDMD during B. cenocepacia infection leads to disruption of MMP.
Previous studies demonstrated that in human peripheral blood mononuclear cells, mROS promotes production of pro-inflammatory cytokines [53][54][55] . This led us to explore if defective mROS production in gsdmd −/− macrophages during B. cenocepacia infection contributes to lower cytokine levels (Fig. 3E). By inhibiting mROS via   www.nature.com/scientificreports/ MT, we detected significantly lower pro-inflammatory cytokines release (IL-1α, IL-1β, CXCL1, IL-6 and TNF-α) in the cell culture supernatants of WT murine macrophages 6 h post-B. cenocepacia infection (Fig. 5E). In addition, there was no significant difference in the release of cytokines, except for IL-6, in gsdmd −/− macrophages upon treatment with MT. These data indicate that GSDMD-induced mROS contribute to the secretion of proinflammatory cytokines from murine macrophages during B. cenocepacia infection.
GSDMD is recruited to the B. cenocepacia-containing phagosome but does not alter the integrity of the bacterial membrane. To determine if GSDMD co-localizes with B. cenocepacia, we purified phagosomes from B. cenocepacia-infected WT macrophages. We then used Western blot analysis to identify the presence of the cleaved GSDMD on the isolated phagosomes from WT macrophages infected with B. cenocepacia for 6 h (Fig. 5F). We found that cleaved but not full-length GSDMD is present on the phagosomal compartment of the B. cenocepacia-infected WT macrophages 56 Gsdmd −/− mice and their derived macrophages exhibit less autophagosome formation during B. cenocepacia infection. Autophagy is a conserved pathway among eukaryotic cells that degrades cytosolic non-functional organelles, protein aggregates, phagocytosed particles, and intracellular pathogens. It plays an important role in B. cenocepacia clearance from macrophages 59 . Several studies suggested that GSDMD contributes to the release of mROS which mediates the stimulation of autophagy and bacterial clearance [60][61][62] . To test this association, WT macrophages were treated with the selective mROS inducer (MitoPQ) and the mROS inhibitor (MT) for 4 h. Autophagy flux was accessed using Bafilomycin A1 (BafA1) which was added 2 h before the designated time point 17,63,64 . Then, the detection of the cleavage and lipidation of LC3-I into LC3-II was determined. Compared to the non-treated cells, MT treatment led to loss of the autophagic flux in WT macrophages. Accordingly, increasing mROS production by MitoPQ was accompanied by increase of the autophagy flux (Fig. S5A,B). These results demonstrate that mROS positively regulate autophagy in macrophages.
On the other hand, we and others have demonstrated that B. cenocepacia inhibits autophagy regulatory gene expression 59,65 . Additionally, the reduced activation and release of the pro-inflammatory cytokines negatively regulate autophagy 55,66 . In gsdmd −/− macrophages, there is higher B. cenocepacia loads along with defective mROS production in addition to defective cytokine release. Together, this led us to hypothesize that these macrophages elicit defective autophagy, which in turn reduces B. cenocepacia clearance. Functional autophagy is depicted by the formation of autophagosomes, called puncta, which can be monitored using confocal microscopy 12 . To test our hypothesis, WT and gsdmd −/− macrophages were infected with B. cenocepacia and autophagosomes were tracked using an antibody against the autophagy marker LC3. Remarkably, gsdmd −/− macrophages had less B. cenocepacia co-localization with LC3 at 2, and 6 h post-infection compared to WT cells (Fig. 6A,B). Additionally, GSDMD deficiency led to significantly less LC3 puncta formation at 6 h post-infection (Fig. 6C). Furthermore, we quantified the LC3-II level in response to B. cenocepacia via western blot. Compared to corresponding WT macrophages, LC3-II was significantly reduced within gsdmd −/− macrophages (Fig. 6D, S5C). Therefore, GSDMD is required for autophagosome formation during B. cenocepacia infection in vitro.
To determine if it plays a similar role in vivo, the lungs of WT and gsdmd −/− mice, infected intratracheally with B. cenocepacia (10 × 10 6 CFU/mouse) for 4 h, were homogenized and analyzed to determine the amount of LC3II. Compared to corresponding WT lungs, LC3-II was significantly reduced within gsdmd −/− lungs (Fig. 6E,F). Together these data show that during B. cenocepacia infection, autophagy is compromised in the absence of GSDMD both in vitro and in vivo. A schematic illustration of the interplay between GSDMD, mROS, autophagy, cytokines release and B. cenocepacia replication is suggested in (Fig. 7).

Discussion
B. cenocepacia is a classic opportunistic pathogen characterized by an inherent multidrug resistance 67 . To this end, better understanding of host-pathogen interaction and the discovery of new host-based approaches that can help control this opportunistic pathogen are clinically exigent. Our latest findings have characterized the crucial role of CASP11 in the restriction of B. cenocepacia 12 . CASP11 and/or CASP1 can cleave and activate GSDMD 11 , yet, its role in B. cenocepacia infection is not defined. In this study, we identify the requirements for GSDMD activation in response to B. cenocepacia and the downstream effect on B. cenocepacia growth in vivo and in vitro. GSDMD inhibits the replication of different bacteria including B. thailandensis 19,29,30 , but this is not the case for all bacteria. Particularly, GSDMD enhances the replication of other bacteria such as Escherichia coli 28 . As a pore forming protein that targets cardiolipin, GSDMD inserts into the mitochondrial membrane and disturbs the mitochondrial membrane potential, leading to mROS production during B. cenocepacia infection as reported for other microbes 45,68 . At present, several prominent reports demonstrated that mitochondria produce less mROS at high membrane potential which corroborates our findings with B. cenocepacia 45,[47][48][49][50][51][52] . Subsequently, mROS can be delivered to bacteria-harboring phagosomes via mitochondrial vesicles contributing to bacterial clearance 69,70 . The bactericidal effect of mROS against different Gram-negative and Gram-positive bacteria including Salmonella Typhimurium 71 and Staphylococcus aureus 70 , has been described, nevertheless, the contribution of GSDMD was not examined. www.nature.com/scientificreports/ On the other hand, mROS were shown to stimulate autophagy, which is an important macrophage defense mechanism against intracellular bacterial pathogens including B. cenocepacia 59,65,72 . The activation of autophagy promotes the clearance of B. cenocepacia 12 , Legionella pneumophila 18 , Listeria monocytogenes 73 , and Mycobacterium tuberculosis 74 . Increased mROS production which activates autophagy improves the clearance of Mycobacterium tuberculosis 75 , Streptococcus pneumoniae 76 and Listeria monocytogenes 77 . Here, we found that autophagy is defective in gsdmd −/− mice and their derived macrophages. Notably, increasing mROS production in macrophages stimulates autophagy and improved B. cenocepacia clearance. Remarkably, phagosomes isolated from B. cenocepacia-infected macrophages harbored cleaved GSDMD within their membranes. It is possible that the formation of GSDMD pore on phagosomal membranes recruits autophagy in an attempt to repair the pore as it happens at the cell membrane 78 . This will promote uptake by autophagosomes leading to bacterial degradation. Another possibility is that GSDMD pores contribute to the efflux of potassium K + 79 which promotes the activation of autophagy 80 . It is also possible that GSDMD pore, at the plasma membrane, mediates Ca 2 + influx 81 which may alter the activation of autophagy 82 . Therefore, the mechanisms by which GSDMD activation may promote autophagy activity are numerous.
On the other hand, GSDMD can exert direct bactericidal effect against the membranes of few Gram-negative bacteria within macrophages making them more susceptible to the effect of H 2 O 2 . These include B. thailandensis 30  Although gsdmd −/− mice exhibited increased bacterial load in different organs in our study, there was no significant difference in their survival compared to the WT control. Recent reports demonstrated that GSDMD similarly restricted bacteria, but in contrast to our study, GSDMD promoted survival of mice infected with B. thailandensis and Francisella novicida 30,84,85 . Although, increased bacterial loads are typically associated with decreased survival in mice 12,[86][87][88] , this association is not consistent with inflammation 89,90 . As such, GSDMD can play a conflicting role during in vivo infections by promoting cytokine production, immune cell recruitment and tissue damage while restricting bacterial infection 28 . This explains why the absence of GSDMD does not alter the survival of the mice in response to elevated B. cenocepacia loads.
In agreement with this hypothesis, we found that WT lungs exhibited, histologically, more inflammation, neutrophils clustering, and tissue necrosis in comparison to gsdmd −/− lungs. This is consistent with what we found before in casp11 −/− lungs in response to B. cenocepacia 12 . This can be attributed to the difference in cytokine release, especially the release of the pro-inflammatory cytokines which mediate recruitment of immune cells 91 . The role of CASP11 in the release of pro-inflammatory cytokines and chemokines in bronchoalveolar lavages (BALFs) varies upon infection with different bacterial pathogens. For instance, CASP11 deficiency during B. cenocepacia infection leads to less IL-1β, IL-1α, and CXCL1 production 12 . On the other hand, in B. thailandensis and Acinetobacter baumannii infection, IL-1β ' secretion in BALF is greatly affected by the lack of CASP11 30,92 . In a similar fashion, BALF from gsdmd −/− mice contains less IL-18 in response to B. thailandensis 30 . Here we found that, with the exception of CXCL1, which promotes neutrophil trafficking, IL-1β, IL-1α, IL-6, and TNFα were equivalently released in the lungs of the different mice strains which infected with B. cenocepacia 93,94 . Hence, it is plausible to conclude that the release of CXCL1 in the BALF during B. cenocepacia infection is GSDMD dependent, and promotes inflammation within the lungs. In vitro, GSDMD controls the release of not only the inflammasome-dependent cytokines (IL-1β and IL-1α) but also the independent one (CXCL1) in response to B. cenocepacia infection. We used glycine to rule out the possibility that cell death contributes to the increase release of cytokines in specific strains. This is in accordance with other studies, which used glycine as a cytoprotective agent, during Candida albicans 36,95 , Salmonella Typhimurium 34,37 , and B. thailandensis 30 infections. Intracellularly, the lysate of gsdmd −/− macrophages exhibited an accumulation of mature IL-1β. This can be explained by the absence of GSDMD pores and CASP8 activation, which can also cleave proIL-1β 96,97 . Notably, we found increased CASP8 activation in response to B. cenocepacia infection in gsdmd −/− macrophages as reported in previous studies [23][24][25]98 . The intracellular accumulation of mature IL-1β in gsdmd −/− macrophages www.nature.com/scientificreports/ can lead to feedback inhibition of CASP1 cleavage and IL1β processing in response to B. cenocepacia. The lower occurrence of their cleaved forms in the total supernatant and cell lysate of gsdmd −/− macrophages, despite the increased bacterial load, lend weight to this hypothesis of a negative feedback loop, however, it requires further validation. Besides IL-1β, we found less IL-1α in the supernatants of gsdmd −/− macrophages indicating that GSDMD pore also plays a role in IL-1α release. GSDMD activation in response to B. cenocepacia infection may increase CASP11 expression and cleavage by the released IL-1α, IL-1β as we previously reported 99 . Furthermore, the cytokines released from WT macrophages (IL-1α, IL-1β, and CXCL1) were found to be decreased in WT supernatants after reducing mROS production in response to B. cenocepacia. This emphasizes the role of mROS in bolstering cytokine production [53][54][55] . Consequently, the aforementioned lower secretion of IL-1α, IL-1β, and CXCL1 in gsdmd −/− macrophages could be attributed to the defective mROS in these cells during B. cenocepacia infection. Additionally, defected autophagy in gsdmd −/− macrophages may contribute in the decreased secretion of cytokines to the extracellular milieu as it has been proposed before 100 .
It is worth noting that GSDMD cleavage could be downstream of ROS production during canonical inflammasome activation 101 . On the other hand, in non-canonical inflammasome activation, mROS are believed to be downstream of GSDMD which provides a link by which the non-canonical pathway may activate the canonical inflammasome 45 . The decreased expression of CASP1 together with defective mROS production in gsdmd −/− macrophages favors the idea that GSDMD cleavage occurs upstream of mROS production during B. cenocepacia infection.  www.nature.com/scientificreports/ Inflammasome activation can be affected by B. cenocepacia secretory systems (SS) through which, B. cenocepacia is able to secrete effector proteins inside the host cytosol 12,56,102 . T6SS, but not T3SS, was identified to be responsible in eliciting pyrin-dependent, inflammasome activation within human monocytes and THP-1 cells 56 . Furthermore, T3SS plays a role in IL-1β maturation 13,103 , in addition to its role in maintaining B. cenocepacia survival and replication in murine macrophages 12 . Here, we have demonstrated that, T3SS, T4SS, and T6SS of B. cenocepacia contribute to the cleavage of GSDMD, CASP1, and IL-1β in B. cenocepacia infected macrophages. Moreover, all of these SSs promote intracellular bacterial persistence within murine macrophages. These findings clarify the different activators and factors that affect inflammasome activation in response to B. cenocepacia infection.
Taken together, GSDMD mediates a variety of double-edged functions that may affect bacterial infection and inflammation in a contrasting manner. The recognition that GSDMD exerts cellular functions independently of cell death is gaining momentum and warrants further research. Future directions. The interplay between mROS, GSDMD and autophagy during different infections may differ which licenses more investigation.

Materials and methods
Bacterial strains. Burkholderia cenocepacia (B. cenocepacia) K56-2 is a clinical isolate obtained from a CF patient who is not directly involved in the study 104 . Its derivate MH1K is a gentamicin-sensitive strain 105 . B. cenocepacia K56-2, used in immunofluorescence experiments, is complemented with a plasmid for red fluorescent protein (Ds-Red). B. cenocepacia ΔT3SS, ΔT4SS-1, and ΔT6SS mutants were kindly provided by Dr. Valvano at Queen's University, Belfast, UK. All bacterial strains were grown overnight in LB medium at 37 °C and 200 rpm as previously described 59,64,65,105 . Mice. C57BL/6 wild-type (WT) mice were obtained from the Jackson Laboratory (Bar Harbor, ME, USA). Bone marrow-derived macrophages. For the generation of primary bone marrow-derived macrophages (BMDMs) from mice, tibias and femurs were flushed with IMDM medium (Thermo Fisher Scientific, 12440053) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific, 16000044), 50% L cell-conditioned medium, 0.6 × MEM non-essential amino acids (Thermo Fisher Scientific, 11140050) and 0.1% penicillin and streptomycin (Thermo Fisher Scientific, 15140122). Cells were cultivated at least 6 days at 37 °C in a humidified atmosphere containing 5% CO 2 as previously described 12,17,19,105,107 . In vitro infection of primary macrophages. Prior to infection, macrophages were cultivated in IMDM medium supplemented with 10% fetal bovine serum for at least 1 h. In vitro infections were performed at MOI10:1, including centrifugation for 5 min at 30×g to synchronize the infection. Cells were infected with B. cenocepacia MH1K for 1 h followed by 30 min incubation with 50 µg/mL gentamicin (Thermo Fisher Scientific, 15750060) to avoid extracellular bacterial replication. To determine the macrophage associated colony forming unit (CFU), macrophages were lysed using 0.1% Triton X-100 (Fisher Scientific, BP151) in PBS at the indicated time points, and serial dilutions of the lysates were incubated on LB agar for 48 h. To protect against cell death, cells were incubated with 5 mM glycine (VWR Life Science, 76201-024) 1 h prior to infection and continuously until the end of the experiment. To inhibit cellular reactive oxygen species (ROS), macrophages were incubated with 3 mM N-Acetyl-l-cysteine (NAC) (ALX-105-005-G005) which was added 1 h post-infection and continuously until the end of the experiment. On the other hand, to specifically inhibit mitochondrial ROS, cells were incubated with 20 µM Mitotempo (Millipore Sigma, SML0737) which was added 1 h prior to infection and continuously until the end of the experiment. Additionally, to specifically induce mitochondrial ROS, cells were incubated with 1 µM MitoPQ (Abcam, ab146819). Furthermore, to inhibit autophagy in order to determine the autophagy flux, the cells were incubated with 100 nM Bafilomycin A1 (Baf A1) (Enzo, BML-CM110-0100) which was added 2 h before designated time point.

In vivo infection.
For intratracheal infection, mice were anesthetized with Isoflurane and inoculated with 100 µL of phosphate-buffered saline (PBS; Thermo Fisher Scientific, 14190144) containing 10 × 10 6 CFU/mouse (organ CFU, histology) unless otherwise noted. To determine the bacterial load in organs, mice were sacrificed at 4 h and 48 h post-infection to collect lung, liver, and spleen for homogenization in PBS as previously described 59,107 . Data are presented as CFU per gram of organ. For survival experiments, animals were monitored daily for 6-8 days.
Phagosome enrichment. Medium was removed from seeded BMDMs (30 × 10 6 cells) and replaced with ice-cold PBS. Cells were harvested and washed twice with ice-cold PBS, resuspended in homogenization buffer (250 mM sucrose, 0.5 mM EGTA, 20 mM Hepes/KOH (pH 7.2)), and lysed in a Dura Grind stainless-steel www.nature.com/scientificreports/ homogenizer (VWR, 22877-280). The resulting cell homogenate was cleared by centrifugation, and organelles were fractionated in a discontinuous sucrose density gradient by ultracentrifugation at 100,000×g. Phagosomes were recovered from the interface of 55% and 65% sucrose layers, separated through a 15% Ficoll cushion at 18,000×g, and concentrated in a final centrifugation step at 18,000×g 108,109 .
Histological analysis. Lungs were removed from infected mice, and fixed in 10% formalin at room temperature. Sample preparation, processing, hematoxylin & eosin staining, and semi-quantitative slide evaluation using ordinal grading scales was performed by the Histology laboratory within the Department of Veterinary Biosciences at The Ohio State University as previously described 59 .
Immunoblotting. Protein extraction from macrophages was performed using TRIzol reagent (Thermo Fisher Scientific, 15596026) according to the manufacturer's instructions. Briefly, after phase separation using chloroform, 100% ethanol was added to the interphase/phenolchloroform layer to precipitate genomic DNA. Subsequently, the phenol-ethanol supernatant was mixed with isopropanol to isolate proteins. The Bradford method was used to determine protein concentrations in the cell lysate. Equal amounts of protein were separated by SDS-PAGE and transferred to a polyvinylidene fluoride (PVDF) membrane. Membranes were incubated overnight with antibodies against GSDMD (Abcam, ab209845), CASP11 (abcam, ab180673), CASP1 (Adipo-Gen, AG-20B-0042-C100), murine IL-1b (R&D Systems, AF-401-NA), LC3A/B (Cell Signaling Technology, 12741) and GAPDH (Cell Signaling Technology, 2118). Corresponding secondary antibodies conjugated with horseradish peroxidase in combination with enhanced chemiluminescence reagent (Amersham, RPN2209,) were used to visualize protein bands. Densitometry analyses were performed by normalizing target protein bands to their respective loading control (GAPDH) using ImageJ software as previously described 17,19 .

Mitochondrial ROS assay.
To determine mitochondrial superoxide production, macrophages were incubated for 30 min with 2 µM MitoSOX dye (Thermo Fisher Scientific, M36008) diluted in cell imaging solution (formulated as previously described 17 ) at 37 °C in a humidified atmosphere containing 5% CO 2 . Then cells were washed with PBS and further incubated with cell imaging solution. The fluorescence was read using a Spec-traMax i3x microplate reader at 510/580 nm. Fluorescence values are reported as relative to the no treatment.
Mitochondrial membrane potential (MMP) assay. To determine mitochondrial membrane potential, macrophages were incubated for 30 min with 10 nM TMRM dye (Molecular Probes, Invitrogen, UK, T668) diluted in cell imaging solution (formulated as previously described 17 ) at 37 °C in a humidified atmosphere containing 5% CO2. Then cells were washed with PBS and further incubated with cell imaging solution. As a positive control, cells were treated with 10 µM of FCCP (MMP uncoupler) (Millipore Sigma, C2759). The fluorescence was read using a SpectraMax i3x microplate reader at 548/574 nm. Fluorescence values were normalized to cell number as measured per well using a SpectraMax MiniMax 300 Imaging Cytometer, then reported as fold change to the no treatment. www.nature.com/scientificreports/ on the data structure) followed by Holm's adjustment for multiple comparisons as indicated 12 . Adjusted P < 0.05 was considered statistically significant.