Guanylate binding proteins facilitate caspase-11-dependent pyroptosis in response to type 3 secretion system-negative Pseudomonas aeruginosa

Detection of bacterial ligands is a pre-requisite for inflammasome activation. During Pseudomonas aeruginosa infection, flagellin which is secreted through the T3SS is detected by the NLRC4 inflammasome. Activation of the NLRC4 inflammasome is believed to contribute to high IL-1β production and pathogenicity in cystic fibrosis patients with chronic P. aeruginosa infection. Interestingly, the majority of P. aeruginosa isolated from cystic fibrosis patients with chronic airway infection are non-motile and T3SS-negative, suggesting that yet un-characterized inflammasome pathways regulate IL-1β production in cystic fibrosis patients. Here we demonstrate the role of guanylate-binding proteins (GBPs) in regulating bacterial proliferation and inflammasome activation in response to T3SS-negative P. aeruginosa. Bacterial ligands liberated by the action of GBP2 and IRGB10 activate caspase-11 and regulate non-canonical NLRP3 inflammasome activation and IL-1β release. Overall, our results reveal the role of caspase-11 in inhibiting bacterial proliferation and promoting IL-1β secretion during T3SS-negative P. aeruginosa infection. This study suggests that non canonical inflammasomes might have co-evolved to detect Gram-negative bacterial pathogens that have evolved to bypass detection by canonical NLRs.


Introduction
P. aeruginosa is an extracellular opportunistic pathogen that is responsible for both acute and chronic infections in humans. P. aeruginosa can cause acute infection in patients with a variety of predisposing conditions such as immunodeficient subjects, burn victims and patients on mechanical ventilation 1 . P. aeruginosa can cause chronic airway infection and associated pulmonary damage in cystic fibrosis patients which leads to accumulation of pro-inflammatory cytokines including IL-1β in the sputum 2 . Even though detrimental to the host at high level, a low level of IL-1β has been shown to be beneficial for clearance of P. aeruginosa from the infected lungs 3,4 .
IL-1β secretion during bacterial infection is controlled by multimeric innate immune signaling complexes called inflammasomes 5 . During bacterial infection, bacterial ligands or host cell damage is sensed by the inflammasome sensors NLRP1, NLRP3, NLRC4, Pyrin, or AIM2. Recently, a new form of non-canonical NLRP3 inflammasome activation was described that is mediated by sensing of bacterial LPS by caspase-11 6,7 . Once activated, inflammasome sensors initiate the formation of an inflammasome complex containing ASC and caspase-1. Activated caspase-1 proteolytically processes inflammatory cytokines, pro-IL-1β and IL-18, and also activates a form of cell death called pyroptosis. Activation of inflammasomes thus restricts the dissemination of pathogen in two ways. Pyroptosis itself inhibits the spread of the pathogen by destructing the bacterial replication niche. In addition, secretion of inflammatory cytokines leads to the recruitment of immune cells that contain the pathogen at the site of infection 5 .
Flagellin is the monomer of flagellar apparatus associated with bacterial motility. During P.aeruginosa infection, flagellin is secreted through the T3SS into the cytosol of macrophage and is known to activate NLRC4 inflammasome 1 . Interestingly, majority of P. aeruginosa isolated from cystic fibrosis patients with chronic airway infection are either non-motile or are T3SS-negative [8][9][10] . These data suggest that NLRC4 inflammasome activation may not be the major contributor for IL-1β production during chronic airway infection seen in cystic fibrosis patients. This in turn indicates the activation of yet unidentified inflammasome sensors that detect the presence of P. aeruginosa and regulate the secretion of IL-1β. The above observations prompted us to investigate the role of innate immune sensors other than NLRC4 in mediating inflammasome activation during T3SS-negative P. aeruginosa infection.
In this study, we compared inflammasome activation by WT, non-motile and T3SS-negative P. aeruginosa and demonstrated for the first time that T3SS activates both NLRP3 and NLRC4 inflammasomes. The T3SSnegative P. aeruginosa failed to activate NLRC4 and NLRP3 inflammasomes during acute infection. However, T3SS-negative P. aeruginosa activated the noncanonical NLRP3 inflammasome during long term infection. We further demonstrate the requirement for interferon-inducible proteins IRGB10 and GBP2 in promoting non-canonical inflammasome activation. Additionally, GBP2 also regulates bacterial survival in macrophages. Overall, our findings help to explain the source of IL-1β secretion in cystic fibrosis patients with chronic airway infection due to P. aeruginosa where majority of the strains are T3SS-negative. The ability of GBPs in clearing T3SS-negative P. aeruginosa from macrophages also prompts the possibility of using interferons for activating cell autonomous immunity to promote clearance of P. aeruginosa from cystic fibrosis patients. Fig. 1 Non-motile P. aeruginosa activates caspase-1 through T3SS. a The motility phenotype of strains (PA14, motAB motCD, fliC, popB) confirmed in soft agar. b-e Caspase-1 activation, IL-1β, IL-18, and IL-6 release in unprimed bone marrow-derived macrophages (BMDMs) infected with indicated strains of P. aeruginosa (MOI 10) for 2 h. f Intracellular bacterial numbers in unprimed BMDMs infected with indicated strains of P. aeruginosa (MOI 1) for 30 min. Data are representative of two (a, f) or three (b-e) independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (Two-tailed T test)

Results
T3SS mediates flagellin-independent inflammasome activation during Pseudomonas infection P. aeruginosa strains isolated from chronically infected cystic fibrosis patients have either defective T3SS or are non-motile (aflagellated or flagellated, but with defective motor apparatus) 8,9 . To understand which of these mutations give a preferential advantage over bacterial detection and activation of inflammasome, BMDMs (Bone marrow-derived macrophages) were infected with either WT bacteria (PA14), T3SS mutant (popB), aflagellated bacteria (fliC) or flagellated but non-motile bacteria (motABmotCD). The motility defects in these strains were confirmed by motility assay (Fig. 1a). A substantial reduction in caspase-1 activation was observed in BMDMs infected with motABmotCD and fliC compared to BMDMs infected with PA14 (Fig. 1b). We also observed significant reduction in secretion of inflammasome-dependent cytokines IL-1β and IL-18 in BMDMs infected with motABmotCD and fliC compared to PA14-infected BMDMs (Fig. 1c, d). These results demonstrate that motility plays a major role in mediating inflammasome activation during P. aeruginosa infection. Interestingly, inflammasome activation was completely absent in popB-infected BMDMs (Fig. 1b-d). Overall, this suggests that PAMPs (Pathogen-associated molecular patterns) that are secreted through the T3SS can also activate inflammasomes even in the absence of flagellin.
Inflammasome activation is a two-step process that requires priming by bacterial ligands or cytokines and activation by PAMPs or DAMPS. The differences in inflammasome activation during infection with different bacterial strains could be due to decreased bacterial uptake or decreased priming 11 . In order to distinguish these pathways, cells were infected with P. aeruginosa and bacterial uptake was quantified within 30 min of infection. Bacterial internalization was significantly less in motAB-motCD and fliC-infected BMDMs as compared to PA14infected BMDMs suggesting reduced phagocytosis of motABmotCD and fliC (Fig. 1f). Interestingly, IL-6 levels were slightly increased in motABmotCD and fliC-infected BMDMs (Fig. 1e). These results suggest that decreased inflammasome activation during motABmotCD and fliC infection could be due to decreased bacterial uptake, and is not due to decreased priming of inflammasome. The decreased inflammasome activation in fliC mutant could also be due to reduced secondary signal (flagellin) that is necessary for NLRC4 inflammasome activation. In contrast, bacterial uptake as well as IL-6 levels were high in popB-infected BMDMs compared to PA14-infected BMDMs. Therefore, the lack of inflammasome activation during popB infection indicates the requirement for an active T3SS to secrete flagellin and other unidentified factors that facilitate inflammasome assembly and activation.
T3SS-mediated inflammasome activation leads to formation of single ASC foci containing caspase-1 and caspase-8 In order to investigate whether inflammasome assembly is impaired during T3SS mutant infection, we assessed the formation of "ASC specks" in the infected cells. "ASC specks" or pyroptosomes are complex multimeric protein structures formed in response to inflammasome activating stimuli 12 . These structures contain the inflammasome sensor, ASC and caspase-1. Recruitment of caspase-8 to the pyroptosome was reported during Salmonella and Aspergillus infection 13,14 . Immunofluorescent microscopy of PA14-infected BMDMs revealed formation of ASC foci containing both active caspase-1 and caspase-8 (Fig. 2a). The composition of pyroptosome and number of pyroptosomes varied among cells infected with different mutants (Fig. 2a, b). The number of ASC foci and ASC foci containing active caspase-1 was significantly less in fliCinfected BMDMs compared to PA14-infected BMDMs suggesting a key role of flagellin in mediating inflammasome assembly. Despite high bacterial numbers in the cytoplasm during popB infection, there was no detectable ASC foci in the cell (Fig. 1f, Fig. 2a, b). This further demonstrates that activity of T3SS is necessary for promoting inflammasome assembly.

P. aeruginosa T3SS activates NLRC4 and NLRP3 inflammasomes
During P. aeruginosa infection, flagellin secreted through the T3SS is sensed by NLRC4 and leads to inflammasome activation 15 . To understand whether inflammasome activation in fliC mutant is also dependent on NLRC4, we infected WT and Nlrc4 −/− BMDMs with PA14, fliC, motABmotCD, and popB. Interestingly, inflammasome activation and caspase-1 cleavage is observed even in the absence of NLRC4 during P. aeruginosa infection albeit at reduced levels (Fig. 3a). The inflammasome activation in the absence of NLRC4 during P. aeruginosa infection is not strain-dependent as the same was observed during infection with P. aeruginosa strains PA01 and PAK (Fig S1A-C).
To understand whether NLRP3 inflammasome activation regulates pyroptosome formation in the cytoplasm during P. aeruginosa infection, we quantified the ASC foci and components of ASC foci in BMDMs infected with PA14 ( Fig S2A). Even though there was no decrease in ASC foci formation in Nlrp3 −/− BMDMs, the number of ASC foci containing active caspase-1 and caspase-8 was less in Nlrp3 −/− BMDMs compared to WT BMDMs ( Fig  S2B). Similar to inflammasome activation in Nlrc4 −/− BMDMs, pyroptosomes were observed in Nlrc4 −/− BMDMs confirming that cells undergo NLRP3-dependent pyroptosis in the absence of NLRC4 during P. aeruginosa infection (Fig S2A,B). Although caspase-8 is recruited to the inflammasome complex, lack of caspase-8 did not affect inflammasome activation during P. aeruginosa infection as evidenced by similar caspase-1 cleavage in WT, Ripk3 −/− and Ripk3 −/− Cas3 −/− BMDMs infected with PA14 ( Fig S2C). To confirm the importance of ASC in inflammasome activation during P. aeruginosa infection, we infected Asc −/− BMDMs with PA14. P. aeruginosa mediated caspase-1 cleavage in BMDMs was completely dependent on ASC (Fig S2D). Overall, these results suggest that T3SS activates both NLRP3 and NLRC4 inflammasomes during P. aeruginosa infection.
Caspase-11-dependent pyroptosis inhibits formation of a replicative niche by T3SS mutant in macrophages P. aeruginosa T3SS mutants are associated with chronic airway infection 8 . We investigated whether the inability of the mutant to activate NLRC4 and NLRP3 inflammasome allows the mutant to form a replicative niche in macrophages. To address this question, we infected BMDMs with popB and bacterial numbers were quantified 4 h and 16 h post infection. Interestingly, we observed significant amount of pyroptosis in WT BMDMs during later time points of infection and bacteria were not able to proliferate inside macrophages (Fig. 4a, b). To understand whether NLRP3 and NLRC4 inflammasomes induce pyroptosis during chronic infection by popB, Nlrp3 −/− and Nlrc4 −/− BMDMs were infected with popB and caspase-1 cleavage was quantified (Fig. 4c). Nlrc4 −/− BMDMs showed similar levels of caspase-1 cleavage and pyroptosis compared to WT BMDMs (Fig. 4a, c). IL-1β and IL-18 secretion and bacterial survival was similar in Nlrc4 −/− and WT BMDMs (Fig. 4a-e). These results suggest that NLRC4 does not regulate pyroptosis during long term infection with popB. Nlrp3 −/− BMDMs also showed significant amount of pyroptosis during popB infection (Fig. 4a). Interestingly, caspase-1 cleavage and secretion of inflammasome dependent cytokines IL-1β and IL-18 was completely abrogated in Nlrp3 −/− BMDMs (Fig. 4c-e). These results point to caspase-11 dependent activation of non-canonical NLRP3 inflammasome in popB-infected BMDMs. In order to confirm this hypothesis, we infected caspase-11 −/− BMDMs with popB and quantified caspase-1 cleavage, IL-1β and IL-18 secretion. Pyroptosis, caspase-1 cleavage and secretion of inflammasome dependent cytokines IL-1β and IL-18 were completely abrogated in caspase-11 −/− BMDMs (Fig. 4a, c-e). IL-6 levels did not reduce in any of these conditions suggesting that the absence of inflammasome activation in caspase-11 −/− BMDMs is not due to a defect in priming of inflammasome (Fig. 4f). These results suggest that popB activates caspase-11-dependent non-canonical NLRP3 inflammasome during chronic infection.
GBPs are known to inhibit bacterial replication independent of inflammasome activation 21,22 . To understand whether GBPs contain popB replication in macrophages, Gbp2 −/− , Gbp5 −/− , Gbp chr3 -KO and Irgb10 −/− BMDMs were infected with popB and intracellular bacterial numbers were quantified 4 h and 16 h post infection. As shown before, popB cannot survive in WT BMDMs (Fig. 5f). However popB was able to survive in Gbp2 −/− BMDMs, suggesting that either GBP2 act as an anti-bacterial factor or can expose popB to other anti-bacterial molecules in the cytosol (Fig. 5f). Compared to WT BMDMs, there was no significant difference in survival of popB in Gbp5 −/− and Irgb10 −/− BMDMs suggesting a unique role for GBP2 in controlling popB proliferation in macrophages (Fig. 5f). These results suggest that even though GBP2 and IRGB10 release popB ligands leading to activation of caspase-11, GBP2 alone act as an anti-bacterial agent against popB in macrophages. Interestingly, despite having similar levels of pyroptosis in Irgb10 −/− Gbp chr3 -KO compared to caspase-11 −/− BMDMs, popB replication was significantly high in caspase11 −/− BMDMs (Fig. 5a-f, Fig S3). These results suggest that caspase-11 regulates bacterial proliferation independent of pyroptosis. Collectively, these results demonstrate that GBP2 and IRGB10 regulates caspase-11 activation and replication of Pseudomonas T3SS mutants (popB) which otherwise escape detection by NLRC4 and canonical NLRP3 inflammasomes. These findings point out to the importance of

Discussion
In the present study, we showed that P. aeruginosa infection leads to activation of both NLRC4 and canonical NLRP3 inflammasomes (Fig. 6). Multiple inflammasomes are known to be activated simultaneously during bacterial and fungal infections 14,16 . Activation of multiple inflammasomes might help to give robust inflammasome activation under conditions where a specific inflammasome component is defective or is absent in a particular cell or tissue. In our study, we have detected robust NLRP3 inflammasome activation only in Nlrc4 −/− BMDMs. Under normal conditions, high activity of NLRC4 inflammasomes mask the role of NLRP3 inflammasomes during P.aeruginosa infection. Indeed, previous studies have demonstrated NLRP3 inflammasome activation in response to Pseudomonas infection in human bronchial epithelial cells derived from cystic fibrosis patients 23 . However, the expression of NLRC4 inflammasome in these cells and the PAMPs that can activate NLRP3 inflammasome are not well characterized.
NLRP3 inflammasome activation in flagellin-deficient P.aeruginosa demonstrates that flagellin is not the PAMP for NLRP3 inflammasome activation. NLRP3 inflammasome activation is completely abrogated during acute infection (2 h) with T3SS mutant Pseudomonas. T3SS needle apparatus is known to create pores on cell membrane 24 . Pore formation during bacterial infection can activate NLRP3 inflammasome 25 . Whether T3SS itself or the secreted virulence factors are mediating NLRP3 inflammasome activation is yet to be established. Components of T3SS needle apparatus are known to activate NLRC4 inflammasome 26 . T3SS needle apparatus might have contributed to residual NLRC4 inflammasome activation in fliC-infected Nlrp3 −/− BMDMs.
P. aeruginosa uses type 3 secretion system (T3SS) to inject T3SS effectors ExoU and ExoS to inhibit inflammasome activation 27 . These effectors inhibit caspase-1 activation by unknown mechanisms. Therefore, during P. aeruginosa infection, T3SS plays a dual role as inhibitor and activator of inflammasomes 28 . Theoretically, during chronic airway infection, evolution to reduce inflammasome activation should have selected P. aeruginosa strains that are defective in inducing inflammasome activation (Non-motile but T3SS +ve strains). Interestingly, 39% of P. aeruginosa isolated from cystic fibrosis patients with chronic airway infection are non-motile and 88% are T3SS negative 8,9 . These data suggest that T3SS itself induce inflammasome activation independent of flagellin or Fig. 6 Model depicting inflammasome activation by WT and T3SS-negative P.aeruginosa during acute and chronic infection in macrophages. During acute infection (2 h), bacterial T3SS itself and flagellin secreted through T3SS activates NLRC4 inflammasome. T3SS also activates NLRP3 inflammasome by yet uncharacterized mechanisms. T3SS-negative P.aeruginosa escapes detection by both NLRP3 and NLRC4 inflammasomes and proliferates inside cytoplasm. During chronic infection (16 h), high numbers of T3SS-negative P.aeruginosa is detected by GBP2 and IRGB10. Bacterial lysis by action of GBPs release ligands that lead to activation of caspase-11. Caspase-11 inhibits the proliferation of bacteria and also activates NLRP3 inflammasome which leads to secretion of inflammatory cytokine IL-1β might have other inflammatory functions independent of inflammasome. The paradox of evolution of T3SS mutants (T3SS-ve) instead of non-motile but T3SS+ve strains during P.aeruginosa infection is not clear. Our study demonstrates that the evolution of T3SS mutants might be to help the bacteria to escape detection by both NLRP3 and NLRC4 inflammasomes (Fig. 6).
P.aeruginosa is known to cause chronic airway infections in cystic fibrosis patients. Even though majority of the strains isolated from these patients have defective T3SS, these patients still have significant amount of IL-1β in the sputum 2 . This suggests that during chronic airway infection, yet unknown PAMPs can contribute to inflammasome activation. Our study demonstrates that T3SS mutant Pseudomonas can be detected by IRGB10 and GBP2 (Fig. 6). Both GBP2 and IRGB10 are known to liberate ligands from bacteria 18 . Here, we demonstrated that during popB infection, only GBP2 contribute to bacterial killing whereas both GBP2 and IRGB10 were required to regulate inflammasome activation and cell death. This is different from inflammasome activation by Citrobacter or Fransicella, where the GBPs that regulate release of bacterial ligands can also regulate bacterial proliferation 19,22 . These data suggest that during popB infection, bacterial killing and generation of bacterial ligands are two separate processes. During Fransicella infection, GBP5 along with GBP2 and IRGB10 are important for activating AIM2 inflammasome 18 . Interestingly, GBP5 is completely dispensable for inflammasome activation during popB infection. These data further suggest that different GBPs can differentiate between different pathogens.
LPS liberated into the cytosol during Gram-negative bacterial infection is known to activate caspase-11 6,7 . During popB infection, we believe that LPS liberated by action of GBP2 and IRGB10 activates caspase-11dependent non-canonical inflammasomes (Fig. 6). Interestingly, neither GBP2 nor IRGB10 is involved in >regulating inflammasome activation during infection with wild type Pseudomonas. Our study demonstrates an unidentified role of GBPs in regulating inflammasome activation during Pseudomonas infection. The fact that GBPs are exclusively relevant in inducing immune response to T3SS-negative P.aeruginosa demonstrates the fact that these mechanisms have evolved to detect pathogens that escape detection by canonical inflammasomes. were generated as described previously 14,18 . Nlrp3 −/− Nlrc4 −/− mice were generated by crossing Nlrp3 −/− with Nlrc4 −/− mice. All mice were bred at the Animal Resource Center at SJCRH, and animal studies were conducted according to protocols approved by the SJCRH Animal Care and Use Committee.

Bacterial strains and growth conditions
The Pseudomonas aeruginosa strain PA14 is a non mucoidal clinical isolate and is the parental strain for all the mutants studied. The Pseudomonas aeruginosa strain PAK, PAO1, PA14 and isogenic mutants of PA14 (popB, fliC and motABmotCD) were provided by Brent Berwin (Department of Microbiology and Immunology, Geisel School of Medicine at Dartmouth, Lebanon). Bacteria were inoculated in Luria-Bertani media (3002-031, MP Biomedicals) and incubated overnight under aerobic conditions at 37°C. Bacteria were sub cultured (1:25) into fresh LB media for 2 1/2 h at 37°C to generate log-phase bacteria for infection.

Stimulation of bone marrow-derived macrophages
For acute bacterial infections, BMDMs were infected with indicated bacteria at an MOI of 10 for 2 h. For chronic infection with popB, BMDMs were infected at an MOI of 10 for 4 h. Cells were washed twice with PBS and were treated with 100 μg/ml gentamicin (15750-060, ThermoFisher Scientific) for 1 h to kill the extracellular bacteria. Cells were maintained in DMEM containing 10 μg/ml of gentamicin till 16 h. To activate the canonical NLRP3 inflammasome, BMDMs were primed using 500 ng/ml ultrapure LPS from Salmonella minnesota R595 (tlrl-smlps, InvivoGen) for 4 h and stimulated with 5 mM ATP (10127531001, Roche).

Bacterial phagocytic uptake and proliferation assay
Phagocytosis of bacteria by macrophages was calculated by plating the macrophage cell lysates 30 min postinfection. For quantifying bacterial survival in BMDMs, cells were infected with popB at an MOI of 10 for 3 h. Cells were washed twice with PBS and were treated with 100 μg/ml gentamicin for 1 h to kill the extracellular bacteria. Cells were maintained in DMEM containing 10 μg/ml of gentamicin and were lysed 4-and 16-h postinfection. Bacterial numbers in cell lysates were quantified by plating cell lysates in LB agar.

Immunofluorescence staining
Following infection, BMDMs were washed twice with PBS and incubated with media containing FAM FLICA caspase-1 for 1 h (ImmunoChemistry Technologies). Cells were then fixed in 4% paraformaldehyde for 15 min. Cells were incubated with a mouse anti-Asc antibody (1:500 dilution, clone 2EI-7; Millipore) overnight followed by incubation with a rabbit anti-caspase-8 (1:500 dilution, 8592; CST) for an additional 1 h. The secondary antibodies used were anti-rabbit Alexa Fluor 647 and antimouse Alexa Fluor 568. Cells were counterstained with DAPI. Cells and inflammasomes were visualized, counted, and imaged using a Nikon C2 confocal microscope at the Cell and Tissue Imaging Center Light Microscopy Facility (CTIC-LM) at St. Jude.

Real time quantification of cell death
For real time quantification of cell death, BMDMs were seeded into 12 well plates for overnight incubation. Following bacterial infection, 20 nM SytoxGreen (Molecular Probes) was added, and the cells were moved into an IncuCyte live cell imaging system. Cells were imaged every 30 min and the Sytox Green labeled cells (counted as dead cells) were quantified by the IncuCyte FLR or Zoom software (http://www.essenbioscience.com/en/ products/software/).

Cytokine measurement by ELISA
IL-1β, IL-18, and IL-6 were measured using ELISA kits (eBioscience) according to the manufacturers' instructions.