Guanylate-Binding Protein-Dependent Noncanonical Inflammasome Activation Prevents Burkholderia thailandensis-Induced Multinucleated Giant Cell Formation

ABSTRACT Inflammasomes are cytosolic multiprotein signaling complexes that are activated upon pattern recognition receptor-mediated recognition of pathogen-derived ligands or endogenous danger signals. Their assembly activates the downstream inflammatory caspase-1 and caspase-4/5 (human) or caspase-11 (mouse), which induces cytokine release and pyroptotic cell death through the cleavage of the pore-forming effector gasdermin D. Pathogen detection by host cells also results in the production and release of interferons (IFNs), which fine-tune inflammasome-mediated responses. IFN-induced guanylate-binding proteins (GBPs) have been shown to control the activation of the noncanonical inflammasome by recruiting caspase-4 on the surface of cytosolic Gram-negative bacteria and promoting its interaction with lipopolysaccharide (LPS). The Gram-negative opportunistic bacterial pathogen Burkholderia thailandensis infects epithelial cells and macrophages and hijacks the host actin polymerization machinery to spread into neighboring cells. This process causes host cell fusion and the formation of so-called multinucleated giant cells (MNGCs). Caspase-1- and IFN-regulated caspase-11-mediated inflammasome pathways play an important protective role against B. thailandensis in mice, but little is known about the role of IFNs and inflammasomes during B. thailandensis infection of human cells, particularly epithelial cells. Here, we report that IFN-γ priming of human epithelial cells restricts B. thailandensis-induced MNGC formation in a GBP1-dependent manner. Mechanistically, GBP1 does not promote bacteriolysis or impair actin-based bacterial motility but acts by inducing caspase-4-dependent pyroptosis of the infected cell. In addition, we show that IFN-γ priming of human primary macrophages confers a more efficient antimicrobial effect through inflammasome activation, further confirming the important role that interferon signaling plays in restricting Burkholderia replication and spread.

suggested that in mouse macrophages, GBP coating of Burkholderia cells restricts cell fusion by preventing bacterial actin-based motility and spread (33). Unlike the mouse infection model, where data clearly support a role for IFNs in mediating Burkholderia restriction, little is known about the role of this cytokine in human cells, specifically in epithelial cells, in response to B. thailandensis.
Here, we provide evidence that in human epithelial cells, GBP1 restricts B. thailandensis-induced MNGC formation and that GBP1-dependent restriction is mediated by caspase-4-induced pyroptosis of infected cells and independent of restricting bacterial motility. Moreover, we show that the IFN-mediated restriction of B. thailandensis expands to several physiologically more relevant human cell lines, such as keratinocytes (HaCaT) and bronchial epithelial cells (HBEC3-KT). Finally, we observe that even in primary human macrophages, IFN priming confers a much more efficient clearance of Burkholderia infections.

RESULTS
Interferons restrict MNGC formation in epithelial cells during B. thailandensis infection. In a mouse model of infection, type I and type II IFNs have been described to participate in the immune response against B. thailandensis (29,33,34). However, it has been suggested that they play differential roles in epithelial cells and macrophages (35). Moreover, how IFNs regulate defense against this bacterium in the human system remains poorly understood. To gain more insights into the role of IFN priming in protecting human epithelial cells against B. thailandensis infection, we infected HeLa cells with B. thailandensis and monitored the formation of MNGCs, a hallmark of B. thailandensis spread and replication. Microscopy-based analysis showed that by 20 h postinfection (p.i.), naive HeLa cells formed large cell clusters with tightly packed nuclei ( Fig. 1a; see also Fig. S1a in the supplemental material), consistent with the formation of cell aggregates known as MNGCs (6,36). Through image-based quantification, we estimated that approximately 50% of all nuclei belonged to giant cells (Fig. 1b). Strikingly, B. thailandensis-induced MNGC formation was almost completely restricted in IFN-g-primed HeLa cells ( Fig. 1a and b and Fig. S1a). Importantly, IFN-g did not interfere with bacterial uptake as assayed by CFU counting (Fig. S1b).
Two main classes of IFNs have been described: type I, which includes many IFN types, such as IFN-a and IFN-b, and type II, e.g., IFN-g. Following cognate receptor binding, both classes trigger a downstream signaling pathway through STAT1 that culminates in the transcription of interferon-stimulated genes (ISGs) (37). In order to confirm the role of IFN signaling in restricting MNGC formation, we used the STAT1 inhibitor fludarabine. IFN-g-primed HeLa cells pretreated with fludarabine lost the ability to restrict MNGC formation upon B. thailandensis infection, almost to the levels found in naive cells ( Fig. 1c and d). Immunoblotting confirmed the inhibition of the IFN signaling pathway, as the expression of hGBP1 (an ISG product selected as a marker to assess STAT1 inhibition) was partially reduced by fludarabine (Fig. S1c).
To better understand the cell-cell fusion dynamics and further confirm the IFN-g-dependent restriction of MNGC formation during B. thailandensis infection, we used a coculture model of HeLa cells expressing doxycycline (Dox)-inducible enhanced green fluorescent protein (eGFP) (HeLa-eGFP) or Dox-inducible mCherry (HeLa-mCherry). We then performed time-lapse fluorescence confocal microscopy to track cell fusion and MNGC formation, which is characterized by the mixing and colocalization of both cytosolic fluorescent proteins (Fig. S1d). We found that infected naive cells started to fuse at about 6 to 8 h p.i. (Fig. 1e, top; Fig. S1e; and Movie S1), resulting in decreases in eGFP-or mCherry-positive cells ( Fig. 1f and g) and concomitant increases in eGFP/mCherry-double-positive cells (Fig. 1h). Eventually, MNGCs were formed as a result of the fusion of several cells (Fig. 1e Altogether, these findings point out that in human epithelial cells, B. thailandensis spread and the resulting multinucleated cell formation are impaired in an IFN-dependent manner. Human GBPs restrict MNGC formation. GBPs are well-known ISGs and have been shown to be crucial for the proper activation of innate immune defense mechanisms against Gram-negative bacteria, protozoan parasites, and viruses (12,14,15,38). It was recently proposed that during B. thailandensis infection of mouse bone marrowderived macrophages (BMDMs), mouse GPBs (mGBPs) contribute to restricting actinbased bacterial spread and cell-cell fusion, thus also reducing bacterium-induced pathology in vivo (33). Although it has been reported that B. thailandensis is targeted by GBP1 in human epithelial cells (16), the role of GBPs during Burkholderia infection of human cells, and specifically epithelial cells, is largely unknown. We first assessed if intracellular B. thailandensis is targeted by GBPs in HeLa cells ectopically expressing Nterminally eGFP-tagged GBPs (eGFP-GBPs). In accordance with a previous study (16), we observed that B. thailandensis was targeted by eGFP-GBP1 in IFN-g-primed cells, where around 90% of the intracellular bacteria are GBP1 coated at 3 h p.i. (Fig. 2a and  b). In naive cells, GBP1 was also associated with a high percentage of intracellular B. thailandensis bacteria, suggesting that it also senses cytosolically exposed LPS on the surface of this bacterium, similar to its role in Salmonella or Shigella infections (17)(18)(19)(20)(21). On the other hand, GBP2, -3, and -4 coated 40 to 20% of intracellular Burkholderia bacteria only in IFN-g-primed but not in naive HeLa cells, corroborating the notion that their recruitment to cytosolic Gram-negative bacteria is driven by additional effectors that act upstream, namely, GBP1 (16)(17)(18)(19)(20)(21). Similar to what has been shown upon Shigella infection (16), a very low percentage of bacteria positive for eGFP-GBP5, -6, and -7 was detected in both naive and IFN-g-primed HeLa cells ( Fig. 2a and b), suggesting that these three GBPs do not play a major role in recognizing this pathogen.
In order to investigate the possible role of hGBPs in restricting MNGC formation upon IFN-g priming, we used GBP1 -/-HeLa cells previously generated by CRISPR-Cas9 genome editing (17). We found that hGBP1 is involved in restricting MNGC formation upon B. thailandensis infection, as IFN-g-primed GBP1-deficient HeLa cells formed MNGCs in a manner comparable to that of wild-type naive cells ( Fig. 2c and d and Fig. S1f and g) without affecting bacterial entry into cells (Fig. S1h). Collectively, these data suggest that hGBPs are the IFN-dependent downstream effectors responsible for restricting MNGC formation and B. thailandensis spread.
GBP1 promotes caspase-4-dependent pyroptosis and restricts MNGC formation and B. thailandensis replication. Recent studies have shown that in human epithelial cells and macrophages, GBPs are required for noncanonical inflammasome activation by targeting LPS and assembling a caspase-4-activating platform on the surface of cytosolic Salmonella and Shigella bacteria (17)(18)(19)(20)(21). Polymerized GBP1 on the bacterial surface was also proposed to act as an LPS surfactant that increases bacterial susceptibility to antimicrobial effectors by destabilizing the bacterial outer membrane (21). Moreover, the GBP coat assembled on Shigella cells appears to have an additional function of inhibiting actin-based motility and consequent bacterial cell-to-cell spread (16,39). This same mechanism has recently been proposed to prevent B. thailandensis invasion of neighboring cells in murine BMDMs (33). Therefore, we speculated that one or several of these mechanisms might be responsible for the GBP1-dependent restriction of MNGC formation in human epithelial cell lines in response to Burkholderia infection (Fig. S2a).
Upon B. thailandensis infection, IFN-g-primed HeLa cells showed signs of cell ballooning and blebbing, which are hallmarks of pyroptosis (Fig. 1e). Furthermore, a closer analysis of time-lapse confocal microscopy images of B. thailandensis-infected IFNg-primed HeLa cells showed that GBP1 targeting to cytosolic bacteria was followed by pyroptotic cell death in the majority of cases (as observed by nuclear condensation and plasma membrane swelling), with a concomitant restriction of bacterial replication (Movie S3 and Fig. S2b and c). The activation of the noncanonical inflammasome leads to caspase-4 activation and autoprocessing and the subsequent cleavage of the pyroptotic executor GSDMD (40). The cleaved GSDMD N-terminal domain forms pores that result in propidium iodide (PI) uptake, which can be used as a marker of lytic cell death. In the early stages of B. thailandensis infection, IFN-g-primed wild-type HeLa cells showed a higher percentage of PI-positive cells than did naive cells ( Fig. 3a and Fig. S2d). Furthermore, we observed that B. thailandensis infection resulted in caspase-4 activation only in IFN-g-primed cells (Fig. 3b, p32 fragment), which is in accordance with what has been observed during infection of epithelial cells with other cytosolic Gram-negative bacteria (17,18). According to recent studies, coating of the surface of cytosolic Gram-negative bacteria by GBPs facilitates caspase-4 recruitment and activation, initiating the downstream pathway that culminates in the lytic death of the infected cell (17)(18)(19)(20)(21). Caspase-4 localization during B. thailandensis infection of IFNg-primed HeLa cells was assessed by fluorescence confocal microscopy. As expected,  caspase-4-eGFP was found to be recruited to cytosolic B. thailandensis in a GBP1-dependent manner ( Fig. 3c and d, Fig. S3e, and Movie S4) but not to the same levels as GBP1 targeting ( Fig. 2a and b). Furthermore, the recruitment of caspase-4 to bacteria correlated with pyroptosis of the infected cell, as determined by the appearance of the typical pyroptotic morphology (Fig. S2e and f and Movie S4). To test if this cell death was caused by noncanonical inflammasome activation, we used wild-type, CASP4 -/-, GSDMD -/-, and GBP1 -/-HeLa cells, which have been previously generated and verified (17). We observed that caspase-4, GSDMD, and GBP1 were required for PI influx, indicating that the cell lysis observed upon B. thailandensis infection was triggered by noncanonical inflammasome activation ( Fig. 3e and Fig. S2g).
Together with the observation that the IFN-dependent expression of GBPs is important for preventing MNGC formation ( Fig. 2c and d), these data show that GBP1 induces rapid death of the infected cells by triggering caspase-4-dependent pyroptosis, thus restricting B. thailandensis-induced cell-to-cell fusion and bacterial spread and replication.
GBP1 does not impair actin-based motility or promote direct bacteriolysis of cytosolic B. thailandensis. After demonstrating that GBP1 triggers noncanonical inflammasome activation upon sensing cytosolic B. thailandensis in human epithelial cells, we also tested additional antimicrobial GBP-induced mechanisms that have been proposed previously (16,33,39,41) (Fig. S2a). Cytosolic Burkholderia spp. are known to coopt the host actin polymerization machinery in order to spread from cell to cell (7). We first evaluated if IFN-g priming and GBP1 affect actin tail polymerization on intracellular B. thailandensis cells by confocal microscopy. For this, we used CASP4-deficient HeLa cells to avoid IFN-g-induced noncanonical activation and cell death. Surprisingly, we found that IFN-g priming did not affect actin tail formation on intracellular B. thailandensis bacteria (Fig. 4a and Fig. S3a), contrary to the IFN-g-and GBP1-dependent restriction of Shigella flexneri actin tail formation (39) (Fig. S3b). To test if GBP1-positive B. thailandensis cells were able to form actin tails, we infected naive or IFN-g-primed CASP4 -/-HeLa cells expressing iRFP703-GBP1 with B. thailandensis-mCherry for 5 h. Confocal microscopy analysis showed that in IFN-g-primed cells, GBP1-positive bacteria formed actin tails to the same extent as in naive cells (Fig. 4b and c). The same was observed when we performed time-lapse fluorescence microscopy on LifeAct-GFPexpressing GSDMD -/infected HeLa cells; i.e., GBP1 coating of cytosolic B. thailandensis did not inhibit comet tail formation and actin-based bacterial motility in both naive and IFN-g-primed cells (Fig. 4d and e and Movies S5 and S6). This suggests that GBP1, directly or in combination with other GBPs that oligomerize on the surface of cytosolic B. thailandensis, cannot impair B. thailandensis cell-to-cell spread via actin tails. To confirm this, we used GBP1-deficient HeLa cells and found that in infected wild-type cells, B. thailandensis formed actin tails to the same extent as in GBP1 -/cells, both with and without IFN-g priming (Fig. 4f). Interestingly, this is in contrast to what is observed in the case of Shigella flexneri infection, where GBP1 partially blocks actin tail formation (39) (Fig. S3b).
Furthermore, in naive HeLa cells infected with B. thailandensis-mCherry, the rapid oligomerization of GBP1 on the bacterial surface did not prevent bacterial replication in the host cytosol ( Fig. 4g and Movie S7), and cell fusion and the formation of MNGCs were still observed (Fig. 4g, DIC [differential interference contrast]). This shows that GBP1 by itself does not appear to display antimicrobial activity when recruited to the bacteria in cells despite the previous observation that in vitro, the direct binding of GBP1 alone to bacteria disrupts cell envelope functions (21). (Continued on next page) Human GBP1 Restricts Burkholderia-Induced MNGCs ® Taken together, we conclude that GBP1-dependent restriction of MNGC formation in HeLa cells is not due to bacteriolysis or impaired bacterial actin-based motility within the host cytosol.
Interferon has a protective role against B. thailandensis infection in human bronchial epithelial cells, keratinocytes, and primary macrophages. Since bacteria of the Burkholderia genus employ different routes of infection (subcutaneous infection, inhalation, ingestion of contaminated particles, and aerosol) (1), we tested if IFN-g priming restricts MNGC formation in human cell lines that are physiologically more relevant for Burkholderia-induced melioidosis, such as HBEC3-KT cells (human bronchial epithelial cells), HaCaT cells (human keratinocytes), and human primary monocytederived macrophages (hMDMs). Similar to HeLa cells ( Fig. 1a and b), naive HBEC3-KT and HaCaT cells formed MNGCs after B. thailandensis infection, which was almost completely blocked upon IFN-g priming ( Fig. 5a and b and Fig. S3c). IFN-g priming did not reduce bacterial uptake in these cells (Fig. S3d and e). We next determined whether the IFN-g-dependent restriction of giant cell formation was promoted by noncanonical inflammasome activation, as shown for HeLa cells (Fig. 3e). In accordance with what we observed in HeLa cells, small interfering RNA (siRNA)-mediated knockdown of CASP4, GSDMD, or GBP1 in HBEC3-KT and HaCaT cells abrogated the IFN-g-mediated restriction of MNGC formation ( Fig. 5c and Fig. S3f and g).
Burkholderia can invade both phagocytic and nonphagocytic cells. Among phagocytic cells, mainly macrophages and neutrophils take part in the immune response against this pathogen. Briefly, in a murine model of infection, Burkholderia is initially detected by macrophages through the Naip/Nlrc4 inflammasome (32), and the consequent IL-18 release triggers the production of IFN-g whereby in neutrophils and macrophages, caspase-11 is upregulated (32). The subsequent noncanonical inflammasome activation in both cell types represents the critical step at which the B. thailandensis intracellular niche is removed. Therefore, we evaluated the role of IFN-g priming in hMDMs during B. thailandensis infection. Interestingly, while unprimed murine BMDMs form MNGCs upon B. thailandensis infection (33), we did not detect the formation of MNGCs in either naive or IFN-g-primed hMDMs (Fig. S3h). Instead, we observed robust induction of host cell death under both conditions, although the percentage of cell death, assessed by PI influx, was significantly higher in IFN-g-primed hMDMs than in naive hMDMs ( Fig. 5d and e). These results implied that analogously to IFN-g-primed HeLa cells, the induction of cell lysis prevents MNGC formation in hMDMs. This cell death can be driven by IFN-independent (most likely via the NLRC4-caspase-1 axis) or IFN-dependent mechanisms, yet IFN signaling promotes a faster and more efficient way to activate GSDMD-induced pyroptosis and clear the bacteria. The latter most likely depends on the GBP-induced activation of the noncanonical inflammasome, as GBP1 expression in hMDMs was observed only after IFN-g priming, whereas caspase-4 was constitutively expressed (Fig. S3i). NLRP3 can be activated downstream of caspase-4-induced GSDMD activation and cell death, further amplifying pyroptotic cell death via ASC (apoptosis-associated speck-like protein containing a CARD) speck formation and caspase-1 (Fig. S3l). To corroborate the role of B. thailandensis-induced noncanonical inflammasome activation in hMDMs, we treated naive or IFN-g-primed cells with  experiments (b, d, e, and g). Graphs show the means 6 SD, and data are pooled from three (c and f) or five (a) independent experiments performed in duplicate. ns, not significant (by a parametric t test).
MCC950, a known selective NLRP3 inhibitor (42,43) (Fig. S3l), and quantified the percentage of ASC specks by fluorescence microscopy. Inhibition of NLRP3 activation reduced ASC speck formation only in IFN-g-primed and not in unprimed hMDMs ( Fig. 5f and g), indicating that the noncanonical inflammasome is activated only in IFNg-primed hMDMs and that unprimed cells induce inflammasome activation by canonical inflammasomes. We also hypothesize that MNGCs were not observed in hMDMs Human GBP1 Restricts Burkholderia-Induced MNGCs ® because pyroptotic cell death occurred too quickly in response to B. thailandensis infection, even in naive cells. In conclusion, we demonstrate that the IFN-g-dependent signaling axis described for HeLa cells upon Burkholderia infection is also found in other human epithelial cells as well as in human primary macrophages.

DISCUSSION
This study provides the first evidence that in human epithelial cells, GBP-dependent noncanonical inflammasome activation prevents B. thailandensis-induced MNGC formation in the early stages of infection. Our results suggest that GBP1 impairs B. thailandensis cell-to-cell spread by triggering caspase-4-dependent pyroptosis of infected cells.
Previous work in mouse models of infection reported that both the Naip/NLRC4 inflammasome and the caspase-11 noncanonical inflammasome participate in the immune response against Burkholderia (29,32,35). Interestingly, these reports showed that both inflammasomes are connected given that caspase-1 activation in macrophages mediates IL-18 release to drive IFN-g-dependent caspase-11 activation in epithelial cells. In agreement with this, we show that in human epithelial cells, IFNs and caspase-4-dependent pyroptosis provide protection against Burkholderia infection, whereas pyroptosis in primary hMDMs is driven by IFN-dependent and -independent mechanisms. The latter observation correlates with data in murine BMDMs that suggest that both the canonical and noncanonical pathways can be activated upon Burkholderia infection (29)(30)(31). The importance of noncanonical inflammasome activation in response to Burkholderia is particularly evident in HeLa cells, which lack canonical inflammasome pathways, but even human bronchial epithelial cells and keratinocytes mainly activate the noncanonical inflammasome in response to Burkholderia, suggesting that also in the human system, canonical inflammasome activation is restricted to professional immune cells.
Our work provides further support for the notion that human GBP1 acts as a cytosolic pattern recognition receptor that binds LPS in order to activate caspase-4 and restrict bacterial replication (17)(18)(19)(20)(21). Since GBP1 recruitment alone was not sufficient to restrict Burkholderia replication, we propose that GBP1-dependent caspase-4 activation is linked to its ability to recruit GBP2-4 to bacteria, which might amplify its effects. Whether the GBP1-4 coat exerts a strong LPS surfactant effect in cells to disrupt bacterial membranes or whether the coat directly interacts with and activates caspase-4 will need to be addressed by additional studies. It is also conceivable that GSDMD pores amplify noncanonical inflammasome activation by disrupting bacterial envelopes, as a previous study showed that recombinant GSDMD reduces bacterial viability in vitro upon caspase-1 processing and that bacteria were more susceptible to microbicidal effectors when harvested from mouse wild-type macrophages rather than Gsdmd -/macrophages, which suggests that GSDMD can directly kill bacteria (44).
Disruption of the bacterial membrane was proposed to be the main mechanism by which mouse GBPs promote inflammasome activation, as GBPs were found to induce bacteriolysis by recruiting Irgb10 (27), a member of the IRG family of GTPases that are found in mouse but not human cells. However, a more recent study by the same authors suggested that during B. thailandensis infection, mouse GBPs do not lyse bacteria or activate the inflammasome but rather restrict B. thailandensis infection by inhibiting bacterial actin-based motility (33). Specifically, the higher number of multinucleation events observed in Gbp2 -/-, Gbp5 -/-, and Gbp Chr3 knockout (Gbp Chr3 -KO) BMDMs than in wild-type macrophages has been associated with the ability of GBPs to inhibit the host Arp2/3-dependent actin polymerization machinery and, consequently, Burkholderia actin tail formation (33). A similar mechanism has been reported in human cell lines infected with Shigella flexneri, in which the hierarchical targeting of GBPs on the bacteria, reliant on GBP1, impairs Shigella actin-based motility, delaying its spread (16,39,41). Parting ways with the literature, we did not observe any impairment in the polymerization of the Burkholderia actin tails upon GBP targeting in human cells, confirming that the main function of GBPs in response to B. thailandensis infection is to serve as a signaling platform for caspase-4 recruitment and activation. Furthermore, GBP1-deficient cells show a level of MNGC formation similar to those of CASP4and GSDMD-deficient cells, arguing that in human cells, GBPs restrict replication via pyroptosis induction and not by additional inflammasome-independent mechanisms. It is possible that this discrepancy results from species-dependent differences since additional IFN-induced factors that are not expressed in human epithelial cells might account for the restriction of Burkholderia actin dynamics.
In summary, our study is the first to report that interferon restricts the multinucleation events induced by B. thailandensis through the pyroptosis of infected cells. It is important to keep in mind, though, that B. thailandensis is less pathogenic than other species of the B. pseudomallei complex that cause severe disease in humans (1)(2)(3)(4). Therefore, further studies are needed to understand whether interferon improves the clearance of Burkholderia species that are most adapted to infect humans or whether more-pathogenic species have found ways of escaping GBP/inflammasome-mediated immune surveillance.

MATERIALS AND METHODS
Bacterial and mammalian cell culture. All bacteria were grown at 37°C in an orbital shaker. B. thailandensis strain E264 and its isogenic strain expressing mCherry2 were kindly provided by Marek Basler (Biozentrum, Basel, Switzerland) and were grown in lysogeny broth (LB) medium supplemented with 5 g/liter NaCl. Shigella flexneri M90T expressing the adhesin AfaI was provided by Jost Enninga (Institut Pasteur, Paris, France) and was grown in tryptic soy broth (TSB) supplemented with ampicillin (50 mg/ ml). Wild-type HeLa (ATCC CCL-2) and CRISPR-Cas9 knockout HeLa cell lines, generated as previously described (17), were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco) supplemented with 10% fetal calf serum (FCS; BioConcept). HaCaT cells, obtained from CLS (Cell Lines Service) GmbH, were grown in RPMI 1640 (Gibco) supplemented with 10% FCS. HBEC3-KT cells (ATCC) were grown in bronchial/tracheal epithelial cell growth medium (Cell Applications, Inc.). Human primary monocytederived macrophages (hMDMs) were purified from buffy coats obtained from the Swiss Red Cross and cultured as described previously (45). Cells were grown at 37°C with 5% CO 2 .
Infection assays. When indicated, cells were primed for 16 h with human IFN-g (Peprotech) at a concentration of 10 ng/ml for HeLa cells and hMDMs or 2.5 ng/ml for HBEC3-KT and HaCaT cells. B. thailandensis cultures grown overnight were adjusted to an optical density at 600 nm (OD 600 ) of 1, subcultured 1:20, and grown until mid-exponential phase (OD 600 = 0.4 to 0.6). S. flexneri cultures grown overnight were subcultured 1/100 and grown until mid-exponential phase (OD 600 = 0.4 to 0.6). Before infection, bacteria were collected by centrifugation, washed, and resuspended in Opti-MEM (Gibco). Bacteria were added to confluent cells in 96-well plates (HeLa, 5 Â 10 4 cells/well; HBEC3-KT, 2.5 Â 10 4 cells/well; HaCaT, 1 Â 10 5 cells/well; hMDMs, 8 Â 10 4 cells/well) at different multiplicities of infection (MOIs), as described in the figure legends. For B. thailandensis infections, plates were then centrifuged at 300 Â g for 5 min at 37°C and incubated for 1 h at 37°C. For S. flexneri infections, plates were just incubated at 37°C for 30 min. Noninternalized bacteria were then removed by washing cells three times with prewarmed medium, and cells were incubated with Opti-MEM containing 250 mg/ml kanamycin, in the case of B. thailandensis infections, or 100 mg/ml gentamicin, in the case of S. flexneri infections, in order to kill extracellular bacteria. At the desired time points postinfection (p.i.), cells were either processed for CFU analysis (CFU), multinucleated giant cell (MNGC) quantification, propidium iodide (PI) uptake, or Western blot analysis or fixed for immunofluorescence assays. To determine CFU, infected cells were gently washed with phosphate-buffered saline (PBS) and lysed with water containing 0.2% Triton X-100 at the indicated time points. Bacteria were then serially diluted and plated onto LB agar.
MNGC quantification assay. Starting at 20 h p.i., HeLa, HBEC3-KT, and HaCaT cells were stained with Hoechst stain (1:1,000) and examined by fluorescence microscopy. The extent of multinucleation was measured by counting nuclei in 6 fields of view under each experimental condition using Fiji software.
Plasmids. Plasmids expressing N-terminally fluorescently tagged GBPs were generated by inserting the GBP coding sequences at the XhoI/HindIII sites of pEGFP-C1 (Clontech) (17). Doxycycline-inducible eGFP and doxycycline-inducible mCherry plasmids were generated by amplifying eGFP and mCherry generated as described above by PCR and inserting the coding sequences at the BamHI site of the pLVX-Puro vector (Clontech). Plasmids expressing LifeAct-eGFP were generated by amplifying eGFP from pLJM1-eGFP (Addgene) and inserting the sequence into the NheI and BstBI cloning sites of a LifeAct-iRFP670 (Addgene) plasmid. All cloning was performed using In-Fusion cloning technology (Clontech), and plasmids were verified by sequencing. When required, HeLa cells were transfected with expression plasmids as previously described (17).
Lentiviral particle production and HeLa cell transduction. Lentiviral particles were produced by transfecting HEK293T cells. Cells seeded into a 6-well plate at a density of 1 Â 10 6 cells/well 24 h prior to transfection were transfected with expression plasmids (pLVX-GFP and pLVX-mCherry), packaging plasmid psPax2 (1.9 mg), and envelope plasmid pVSV-G (0.2 mg) using jetPRIME (Polyplus), according to the manufacturer's instructions. After a 24-h incubation, HEK293T medium containing lentiviral particles was transferred to HeLa cells seeded at a density of 0.8 Â 10 6 cells/well in a 6-well plate. HeLa cells were centrifuged at 2,900 rpm for 90 min and incubated for 48 h (medium was changed after incubation Human GBP1 Restricts Burkholderia-Induced MNGCs ® overnight). Puromycin (5.0 mg/ml; InvivoGen) was added to the medium for 6 to 8 days in order to positively select transduced cells.
Microscopy, time-lapse imaging, and image analysis. Fluorescence and phase-contrast images of nonfixed samples were obtained using a Leica DFC3000G instrument (40Â objective) for MNGC quantification. For fluorescence microscopy of fixed samples, infected HeLa cells and hMDMs were washed twice with PBS and fixed for 20 to 30 min with 4% paraformaldehyde (Electron Microscopy Sciences). Cells were washed four times with PBS and incubated with Hoechst stain (1:1,000) and, when indicated, with CellMask green actin tracking stain (catalog number A57243; Thermo Fisher Scientific) to label F-actin. For ASC speck formation assays, hMDMs were permeabilized with 0.05% saponin and blocked with 1% bovine serum albumin (BSA). Coverslips were then incubated with anti-ASC antibody (catalog number sc-22514-R; Santa Cruz Biotechnology) (1:1,000), washed four times with PBS, and incubated with Hoechst stain (1:1,000). Samples were then analyzed by confocal microscopy by imaging with a Zeiss LSM800 confocal laser scanning microscope using a 63Â/1.4-numerical-aperture (NA) oil objective by acquiring Z-stacks with a 300-nm step size. For live imaging, HeLa cells plated onto 8-well m-slides (Ibidi) at a density of 1 Â 10 5 cells/well were infected as described above. Extracellular B. thailandensis bacteria were removed by washing with warm Opti-MEM, and time-lapse microscopy of living cells was performed in Opti-MEM supplemented with kanamycin (250 mg/ml) at 37°C using a motorized xyz stage with autofocus. Samples were imaged with a Zeiss LSM800 confocal laser scanning microscope using a 63Â/1.4-NA oil objective by acquiring Z-stacks with a 600-nm step size. Data were further analyzed and processed using Fiji software, and all fluorescence-derived images shown correspond to maximum three-dimensional (3D) projections.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. FIG S1, TIF