Human NAIP/NLRC4 and NLRP3 inflammasomes detect Salmonella type III secretion system activities to restrict intracellular bacterial replication

Salmonella enterica serovar Typhimurium is a Gram-negative pathogen that uses two distinct type III secretion systems (T3SSs), termed Salmonella pathogenicity island (SPI)-1 and SPI-2, to deliver virulence factors into the host cell. The SPI-1 T3SS enables Salmonella to invade host cells, while the SPI-2 T3SS facilitates Salmonella’s intracellular survival. In mice, a family of cytosolic immune sensors, including NAIP1, NAIP2, and NAIP5/6, recognizes the SPI-1 T3SS needle, inner rod, and flagellin proteins, respectively. Ligand recognition triggers assembly of the NAIP/NLRC4 inflammasome, which mediates caspase-1 activation, IL-1 family cytokine secretion, and pyroptosis of infected cells. In contrast to mice, humans encode a single NAIP that broadly recognizes all three ligands. The role of NAIP/NLRC4 or other inflammasomes during Salmonella infection of human macrophages is unclear. We find that although the NAIP/NLRC4 inflammasome is essential for detecting T3SS ligands in human macrophages, it is partially required for responses to infection, as Salmonella also activated the NLRP3 and CASP4/5 inflammasomes. Importantly, we demonstrate that combinatorial NAIP/NLRC4 and NLRP3 inflammasome activation restricts Salmonella replication in human macrophages. In contrast to SPI-1, the SPI-2 T3SS inner rod is not sensed by human or murine NAIPs, which is thought to allow Salmonella to evade host recognition and replicate intracellularly. Intriguingly, we find that human NAIP detects the SPI-2 T3SS needle protein. Critically, in the absence of both flagellin and the SPI-1 T3SS, the NAIP/NLRC4 inflammasome still controlled intracellular Salmonella burden. These findings reveal that recognition of Salmonella SPI-1 and SPI-2 T3SSs and engagement of both the NAIP/NLRC4 and NLRP3 inflammasomes control Salmonella infection in human macrophages.

Introduction Salmonella enterica serovar Typhimurium (referred to hereafter as Salmonella) is a Gram-negative bacterial pathogen that causes self-limiting gastroenteritis in immune-competent humans. Transmission of Salmonella typically occurs upon ingestion of contaminated food or water. Once inside the host, Salmonella uses specialized nanomachines known as type III secretion systems (T3SSs) to inject effectors into the host cell cytosol [1]. Subsequently, these effectors remodel host cellular processes to facilitate bacterial colonization and intracellular survival. Thus, Salmonella's T3SSs enable the enteric pathogen to successfully colonize the intestinal tract and infect a variety of cell types, including intestinal epithelial cells (IECs) and macrophages [1]. Specifically, Salmonella uses its first T3SS, located on Salmonella pathogenicity island 1 (SPI-1), to invade host cells, and its second T3SS, located on a second pathogenicity island, SPI-2, to persist and replicate within host cells [2][3][4][5][6][7][8][9][10]. Numerous other Gram-negative bacterial pathogens also use these evolutionarily conserved T3SSs to colonize the host [11]. While T3SSs are required for these bacterial pathogens to cause disease, they also translocate flagellin [12] and structural components of the T3SS into the cytosol, thus enabling the host to detect the invading pathogen. Unlike effectors, which display significant diversity across bacterial species, structural components of the T3SS or the flagellar apparatus retain significant structural homology across Gram-negative bacteria [13][14][15][16][17][18][19]. Thus, these ligands serve as ideal targets for host immune sensors.
Unlike mice, humans only express one functional NAIP [42,43]. In human macrophages, this single NAIP is sufficient to respond to the cytosolic delivery of bacterial flagellin as well as the SPI-1 T3SS inner rod (PrgJ) and needle (PrgI) proteins [44][45][46]. Interestingly, the SPI-2 T3SS inner rod protein (SsaI) fails to induce inflammasome activation in both murine and human macrophages [14,45], suggesting that the Salmonella SPI-2 T3SS evades NAIP detection to enable Salmonella replication within macrophages. The NAIP/NLRC4 inflammasome promotes control of intestinal Salmonella infection in mice by triggering pyroptosis and expulsion of infected intestinal epithelial cells [47,48]. Recently, both the NAIP/NLRC4 and NLRP3 (NLR pyrin domain-containing protein 3) inflammasomes have been shown to respond to Salmonella infection in human macrophages [49,50]. However, whether the NAIP/NLRC4 inflammasome recognizes the SPI-2 T3SS needle protein (SsaG), and whether the NAIP/ NLRC4 and NLRP3 inflammasomes contribute to the restriction of Salmonella replication within human macrophages is unknown.
In this study, we found that while human macrophages require both NAIP and NLRC4 for inflammasome responses to T3SS ligands, NAIP and NLRC4 are only partially required for the inflammasome response during Salmonella infection. Rather, we found that infection of human macrophages with Salmonella grown under SPI-1-inducing conditions activates multiple inflammasomes, including NAIP/NLRC4, CASP4/5, and NLRP3. Importantly, both the NAIP/NLRC4 and NLRP3 inflammasomes played a functional role in restricting Salmonella's intracellular replication, indicating that they contribute to host defense in a cell-intrinsic manner, as well as via release of inflammatory mediators. Finally, we found that the NAIP/NLRC4 inflammasome recognizes the SPI-2 T3SS needle protein SsaG, and that SPI-1 T3SS and flagellin-independent, NAIP/NLRC4-dependent recognition of Salmonella controls bacterial burden within human macrophages. Our studies uncover a multifaceted inflammasome response to Salmonella infection in human macrophages, and reveal that NAIP/NLRC4 inflammasomedependent sensing of the SPI-2 T3SS promotes control of intracellular Salmonella. Collectively, these findings yield important insight into how human macrophages use inflammasomes to sense and respond to intracellular bacterial pathogens.

NAIP and NLRC4 are necessary for inflammasome responses to T3SS ligands in human macrophages
In murine macrophages, multiple NAIPs are required for inflammasome responses to the Salmonella SPI-1 T3SS inner rod protein (PrgJ), the SPI-1 T3SS needle protein (PrgI), and flagellin [22,23,[26][27][28][29]. In addition, the murine NAIPs and NLRC4 contribute to the inflammasome response during in vivo Salmonella infection [14,29]. In human macrophages, PrgJ, PrgI, and flagellin all activate the inflammasome, while the Salmonella SPI-2 inner rod protein (SsaI) does not [44,45]. Using siRNA-mediated silencing of NAIP in human macrophages, we have previously shown that human NAIP is important for maximal inflammasome responses to PrgJ and flagellin [45]. However, siRNA-mediated knockdown of NAIP did not completely abrogate inflammasome activation, either due to incomplete knockdown, or the potential contribution of other inflammasomes. Therefore, it remained unclear whether human NAIP or NLRC4 is absolutely required for inflammasome responses to these bacterial ligands or whether additional host sensors also mediate sensing of these ligands.
To test the requirement of the NAIP/NLRC4 inflammasome in human macrophages, we used the Clustered Regularly Interspersed Palindromic Repeat (CRISPR) system, in conjunction with the RNA-guided exonuclease Cas9, to disrupt the NAIP and NLRC4 genes in the human monocytic cell line, THP-1 (S1A and S2A Figs). We selected one independent single cell clone of NAIP -/-THP-1s (NAIP -/-Clone #12) that exhibited reduced NAIP mRNA expression by qRT-PCR compared to WT THP-1s (S1C To test if NAIP and NLRC4 are necessary for inflammasome responses to bacterial T3SS ligands, we used the Bacillus anthracis toxin system to deliver bacterial T3SS ligands into the cytosol of THP-1s [51], and assayed inflammasome responses in wild type (WT), NAIP -/-, and NLRC4 -/-THP-1 macrophages. The B. anthracis system contains two subunits: a protective antigen (PA) that creates a pore in the host endosomal membrane and a truncated lethal factor (LFn) that is delivered through the PA pore into the cytosol. Our T3SS ligands of interest are fused to the N-terminal domain of the B. anthracis LFn. When the LFn is added to eukaryotic cells in conjunction with PA (collectively referred to as Tox), the bacterial ligand is delivered directly into the host cell cytosol. Using this system, we delivered a truncated version of Legionella flagellin (FlaTox), the Salmonella SPI-1 T3SS inner rod protein (PrgJTox), and the Burkholderia T3SS needle protein (YscFTox) into THP-1s. We then measured the release of the inflammasome-dependent IL-1 family cytokines IL-1α, IL-1β, and IL-18 and cell death as markers of inflammasome activation. Cells left untreated (Mock) or treated with PA alone or LFn fused to the bacterial ligand alone released negligible levels of IL-1β, IL-18, and IL-1α and exhibited minimal cell death (Figs 1A, 1C, S3A-S3C and S4A-S4C). In agreement with previous findings [45], WT THP-1s treated with both the PA and LFn subunits exhibited robust inflammasome activation, and released substantial levels of IL-1β, IL-18, and IL-1α and exhibited considerable cytotoxicity (Figs 1A, 1C, S3A-S3C and S4A-S4C), indicating that robust inflammasome activation requires cytosolic delivery of the ligands. In contrast, both NAIP -/-THP-1s and NLRC4 -/-THP-1s released negligible levels of inflammasome-dependent cytokines and did not undergo cell death when treated with FlaTox, PrgJTox, or YscFTox (Figs 1A, 1C, S3A-S3C and S4A-S4C). Importantly, the NAIP -/and NLRC4 -/-THP-1s released IL-1β at levels comparable to those released by WT THP-1s in response to the NLRP3 stimulus LPS + nigericin (Fig 1B and 1D), indicating that CRISPR/Cas9 editing was specific to the NAIP/ NLRC4 inflammasome pathway [52]. In addition, release of the inflammasome-independent cytokine TNF-α was unaffected in NAIP -/or NLRC4 -/-THP-1s (S3D and S4D Figs). Consistent with our prior results [45] and in agreement with recent studies [49], these results collectively demonstrate that NAIP and NLRC4 are required for inflammasome activation in response to the T3SS inner rod, T3SS needle, and flagellin proteins in human macrophages.

PLOS PATHOGENS
Human inflammasomes restrict Salmonella in macrophages

PLOS PATHOGENS
Human inflammasomes restrict Salmonella in macrophages observed a significant decrease but not complete abrogation of secreted IL-1β and IL-18 levels (Fig 2). Corroborating the ELISA data, we observed a decrease but not complete abrogation of cleaved and secreted IL-1β levels in NAIP -/-THP-1 macrophages infected with WT Stm by western blot analysis ( Fig 2E). Levels of IL-1α and cell death were largely unaffected (S5A- S5D  Fig). WT, NAIP -/-, and NLRC4 -/-THP-1s released similar levels of the inflammasome-independent cytokine TNF-α (S5E and S5F Fig).
This response was dependent on SPI-1 T3SS translocation into host cells, as cells infected with ΔsipB Stm, which lack a component of the translocon, failed to undergo robust inflammasome activation (Figs 2 and S5A-S5D). This may be partly due to reduced uptake of ΔsipB Stm into host cells (S6 Fig), given the role of the SPI-1 T3SS in mediating invasion of host cells [53]. Importantly, uptake of either WT Stm or ΔsipB Stm across THP-1 genotypes was consistent (S6 Fig), suggesting that the difference in inflammasome activation observed is not due to differential uptake of bacteria between THP-1 genotypes. Overall, these data indicate that NAIP and NLRC4 are partially required for inflammasome responses to Salmonella infection in human macrophages, in contrast to what we observe with individual T3SS ligand delivery (Figs 1, S3 and S4), where NAIP/NLRC4 is absolutely required for inflammasome activation. Thus, our data indicate that in addition to the NAIP/NLRC4 inflammasome, Salmonella also induces a NAIP/NLRC4-independent inflammasome response, in agreement with published studies [49,50].

PLOS PATHOGENS
Human inflammasomes restrict Salmonella in macrophages WT Stm-infected NAIP -/-THP-1 macrophages treated with MCC950 by western blot analysis ( Fig 3E). Release of the inflammasome-independent cytokine TNF-α was similar across the various THP-1 genotypes and treatments following infection (S7E and S7F Fig). Altogether, these data indicate that Salmonella infection induces both NAIP/NLRC4-and NLRP3-dependent inflammasome activation in human macrophages, in agreement with previous studies [49,50].

The NAIP/NLRC4 and NLRP3 inflammasomes control Salmonella burden within human macrophages
One of the mechanisms by which inflammasome activation leads to control of bacterial infection is by restricting intracellular bacterial replication [47,54,[71][72][73][74][75]. In mice, the NAIP/ NLRC4 inflammasome is important for controlling Salmonella replication in the intestine [47,48,72], whereas the NLRP3 inflammasome is dispensable for control of Salmonella infection in vivo [54,[72][73][74]. Caspases-1 and 11 restrict cytosolic Salmonella replication within murine macrophages [76]. Whether inflammasome activation restricts WT Salmonella replication in human macrophages is unknown. To test the hypothesis that inflammasome activation restricts Salmonella within human macrophages, we infected WT, NAIP -/-, and NLRC4 -/-THP-1 macrophages with WT Stm in the presence or absence of the NLRP3 inhibitor MCC950 and determined the bacterial colony forming units (CFUs) at various timepoints WT THP-1s treated with MCC950 also contained a significantly higher number of bacterial CFUs, comparable to those in NAIP -/and NLRC4 -/-THP-1s (Figs 5A, 5C, S9B and S9D). NAIP -/and NLRC4 -/-THP-1s treated with MCC950 had the highest bacterial burdens, which were significantly higher than the bacterial burdens in DMSO control-treated NAIP -/and NLRC4 -/-THP-1s or WT THP-s treated with MCC950 (Fig 5A and 5C). We then examined the fold-change in bacterial CFUs at 6 and 24 hours relative to 2 hours post-infection. The fold-change in bacterial CFUs was restricted the most effectively in WT THP-1s, moderately
To further assess bacterial burdens in human macrophages, we infected WT and NAIP -/-THP-1s with WT Stm expressing GFP in the presence and absence of MCC950 and quantified the number of Stm per cell (Fig 6). Microscopic analysis revealed that DMSO control-treated WT THP-1s contained the lowest number of bacteria per cell, while NAIP -/-THP-1s treated with MCC950 harbored the highest number of bacteria per cell at 6hpi (Fig 6A and 6C). WT THP-1s treated with MCC950 or DMSO control-treated NAIP -/-THP-1s contained similar and intermediate levels of bacteria per cell at 6hpi (Fig 6A and 6C). Comparable bacterial burdens were observed across THP-1 genotypes and treatment conditions at 2hpi (Fig 6B). Collectively, these data suggest that both the NAIP/NLRC4 and NLRP3 inflammasomes control intracellular Salmonella replication within human macrophages at both early (6 hours postinfection) and late (24 hours post-infection) timepoints.

Salmonella SPI-2 needle protein SsaG activates the NAIP/NLRC4 inflammasome in human macrophages
The Salmonella flagellin, SPI-1 T3SS inner rod (PrgJ), and needle (PrgI) proteins all activate NAIP in human macrophages, whereas the Salmonella SPI-2 T3SS inner rod protein (SsaI) is not sensed by NAIP [23,27,[44][45][46]. Similarly in mice, SsaI is not sensed by NAIP2 [14]. These findings have led to the model that the SPI-2 T3SS evades inflammasome detection to allow Salmonella to replicate or persist in both murine and human macrophages [14,45]. However, our data indicate that the NAIP/NLRC4 inflammasome restricts Salmonella replication within human macrophages even at late timepoints, when the SPI-1 T3SS and flagellin are thought to be downregulated [77][78][79][80][81][82][83][84][85]. As Salmonella requires the SPI-2 T3SS to replicate within macrophages [4,9], these data raise the possibility that the NAIP/NLRC4 inflammasome might detect a different SPI-2 T3SS structural component, such as the SPI-2 T3SS needle protein, SsaG. To address whether the human NAIP/NLRC4 inflammasome detects SsaG, we delivered bacterial ligands into the cytosol of primary human monocyte-derived macrophages (hMDMs) derived from anonymous healthy human donors using the Gram-positive bacterium Listeria monocytogenes. Following intracellular invasion, Listeria (Lm) escapes from its vacuole into the cytosol where it expresses the protein ActA on its surface. Fusing bacterial ligands of interest to the Nterminus of truncated ActA allows these ligands to be delivered into the host cytosol, where they trigger NAIP/NLRC4 inflammasome activation [45,86]. We infected hMDMs with WT control Lm or bacteria expressing PrgJ, SsaI, or SsaG and assayed for inflammasome activation (Figs 7A and S10). hMDMs infected with Listeria expressing the SPI-1 T3SS inner rod protein PrgJ induced robust inflammasome activation, indicated by significantly increased IL-18 secretion as well as robust IL-1α and IL-1β secretion compared to mock infection or WT Lm infection alone (Figs 7A and S10), in agreement with our previous findings [45]. In contrast, and as we previously observed [45], Listeria expressing the SPI-2 inner rod protein SsaI failed to induce IL-1β, IL-18, and IL-1α secretion or cell death in hMDMs (Figs 7A and S10). Intriguingly, we observed that Listeria expressing the SPI-2 needle protein SsaG induced significantly increased IL-18 and robust IL-1α and IL-1β secretion compared to mock infection or WT Lm infection alone (Figs 7A and S10).

PLOS PATHOGENS
Human inflammasomes restrict Salmonella in macrophages

PLOS PATHOGENS
Human inflammasomes restrict Salmonella in macrophages secretion and cell death (Figs 7B and S11A-S11C). Infection of WT THP-1s with Listeria expressing PrgI or SsaG led to robust release of IL-1 cytokines and cytotoxicity. In contrast, NAIP -/and NLRC4 -/-THP-1s infected with Listeria expressing PrgI or SsaG released significantly reduced levels of IL-1 cytokines and cell death relative to WT THP-1s that were comparable to the background levels secreted by THP-1s infected with WT Lm (Figs 7B and S11A-S11C). We observed a similar phenotype when we delivered SsaGTox into the cytosol of THP-1s. WT THP-1s released robust levels of IL-1β, IL-18, and IL-1α, whereas NAIP -/and NLRC4 -/-THP-1s released negligible levels of IL-1 cytokines (Figs 7C, S11D and S11E). Altogether, these data demonstrate that the SPI-2 needle protein activates the human NAIP/ NLRC4 inflammasome, providing evidence that human NAIP can sense and respond to the Salmonella SPI-2 T3SS needle.

PLOS PATHOGENS
Human inflammasomes restrict Salmonella in macrophages

NAIP/NLRC4 inflammasome recognition of the SPI-2 T3SS controls intracellular Salmonella in human macrophages
To determine if NAIP/NLRC4-mediated recognition of the SPI-2 T3SS needle controls intracellular Salmonella burdens, we generated a Salmonella mutant strain (ΔprgIfliCfljB) lacking flagellin and the SPI-1 T3SS needle protein, PrgI. This strain is therefore unable to assemble a functional SPI-1 T3SS, but still expresses a functional SPI-2 T3SS. We infected WT, NAIP -/-, and NLRC4 -/-THP-1 macrophages with ΔprgIfliCfljB and determined the CFUs at various timepoints to assay intracellular bacterial burdens (Figs 8A, 8B and S12). The bacterial burden of ΔprgIfliCfljB over a 24-hour post-infection time course was controlled the most effectively in WT THP-1s and was significantly less restricted in NAIP -/and NLRC4 -/-THP-1s (Figs 8A, 8B and S12). To confirm that inflammasome activation still occurs in response to ΔprgIfliCfljB, we infected WT, NAIP -/-, and NLRC4 -/-THP-1 macrophages with ΔprgIfliCfljB and measured IL-1β release at 24hpi (Fig 8C and 8D). We observed significant release of IL-1β in WT THP-1s. In contrast, we observed significantly reduced levels of IL-1β release in NAIP -/and

Discussion
Our data show that human macrophages engage multiple inflammasome pathways to sense and respond to Salmonella infection. Using NAIP -/and NLRC4 -/-THP-1s (S1 and S2 Figs), we found inflammasome activation in response to individual T3SS ligands to be entirely dependent on the NAIP/NLRC4 inflammasome in human macrophages (Figs 1, S3 and S4). In contrast, Salmonella infection induced activation of inflammasome responses that depended on the collective responses of NAIP/NLRC4, NLRP3, and CASP4/5 (Figs 2-4 and S5-S8). Our findings are consistent with previous studies demonstrating that both NLRC4 and NLRP3 are required for inflammasome responses to Salmonella in human macrophages [49,50]. Importantly, our data also reveal that both the NAIP/NLRC4 and NLRP3 inflammasomes contribute to control of Salmonella replication in human macrophages (Figs 5, 6 and S9). Furthermore, contrary to the prevailing model that the SPI-2 T3SS as a whole evades NAIP detection, our findings reveal that the NAIP/NLRC4 inflammasome can recognize the Salmonella SPI-2 T3SS needle protein SsaG (Figs 7, S10 and S11), and that NAIP/NLRC4-dependent detection of the SPI-2 T3SS controls intracellular Salmonella in human macrophages (Figs 8 and S12).
Many Gram-negative bacteria use evolutionarily conserved T3SSs to deliver virulence factors, or effectors, into host cells. We and others have previously shown that T3SS inner rod proteins from various Gram-negative bacteria activate the inflammasome in human macrophages [45,46]. In this study, we used T3SS inner rod, needle, or flagellin proteins from three different Gram-negative bacteria, Salmonella, Burkholderia, and Legionella, and observed that inflammasome activation in response to an isolated ligand is entirely dependent on NAIP/ NLRC4 (Figs 1, S3, S4, S7B and S7C). How human NAIP senses and responds to these diverse bacterial structures remains an open question. The Salmonella SPI-1 T3SS inner rod (PrgJ), SPI-1 T3SS needle (PrgI), and flagellin proteins exhibit low total sequence conservation, but they all retain several conserved hydrophobic amino acid residues within their structurally homologous C-terminal helices [14,[87][88][89][90][91][92]. Interestingly, alignment of the amino acid sequences of the SPI-2 T3SS needle protein (SsaG), PrgJ, and PrgI using Clustal Omega revealed that all three proteins contain conserved hydrophobic amino acid residues in their Cterminus (S13A Fig). Specifically, SsaG has C-terminal isoleucine residues like PrgI. These isoleucine residues are important for NAIP-mediated recognition of PrgI [27]. To compare these ligands at the structural level, we examined published three-dimensional structures of PrgJ and PrgI and used PHYRE2 Protein Fold Recognition Server to predict the structure of SsaG. Similar to PrgJ and PrgI, SsaG also displays an alpha-helical structure at its C-terminus (S13B Fig). Unlike these T3SS ligands, the Salmonella SPI-2 inner rod protein, SsaI, does not retain such conserved C-terminal residues. Perhaps this is why SsaI is not detected in murine or human macrophages [14,45]. In contrast, our findings indicate that the NAIP/NLRC4 inflammasome detects SsaG. Furthermore, we find that SPI-2-dependent NAIP/NLRC4 inflammasome activation contributes to intracellular control of Salmonella.
Salmonella infection induces NAIP/NLRC4-, CASP4/5-, and NLRP3-dependent inflammasome activation in human macrophages (Figs 2-4 and S5-S8). This suggests that there is redundancy in the inflammasome pathways when sensing and responding to Salmonella infection, such that loss or inhibition of just one inflammasome does not result in severe loss of inflammasome activation in human macrophages. Given our observations with individual
The NLRP3 inflammasome can be activated by a variety of different stimuli, including potassium efflux [52,[57][58][59][60]. We observed NLRP3 activation induced by Salmonella infection that is likely due to activity of the SPI-1 T3SS, since we do not observe inflammasome activation when we infect THP-1 macrophages with ΔsipB (Fig 3). Whether the NLRP3 activation is due to the activity of a specific SPI-1 effector or merely due to the collective activity of the SPI-1 effectors in promoting the uptake of Salmonella remains unknown. Of note, we observed reduced uptake of ΔsipB Salmonella into THP-1 macrophages (S6 Fig). Perhaps, NLRP3 activation is downstream of some Salmonella-induced activity that occurs when Salmonella successfully and abundantly gets taken up into the cell.
The NLRP3 inflammasome can also be activated downstream of the CASP4/5 inflammasome [64,68]. Given that we observed only partial loss of inflammasome activation in the NAIP -/-THP-1s treated with siRNA targeting CASP4 and CASP5, we hypothesize that at least part of the NLRP3-dependent response is due to canonical activation (Fig 4), although this partial loss may also be due to incomplete knockdown of CASP4 and CASP5. A recent study, which also found that Salmonella infection induces NLRC4-and NLRP3-dependent inflammasome activation in human macrophages, observed that full-length Salmonella flagellin can activate the NLRP3 inflammasome [49]. In contrast, we found the response to flagellin to be entirely dependent on the NAIP/NLRC4 inflammasome (Figs 1, S3 and S4). The precise reason for this apparent discrepancy remains to be determined. Notably, in contrast to Gram et al. who used full-length flagellin, our study used a truncated flagellin containing only the C-terminal D0 domain, which does not stimulate TLR5 signaling [49, 89,101]. It is possible that fulllength flagellin used by Gram et al., in addition to activating the NAIP/NLRC4 inflammasome, also stimulates TLR5 signaling, perhaps potentiating NLRP3-dependent responses.
We observed NAIP/NLRC4-and NLRP3-dependent restriction of Salmonella (Figs 5, 6 and S9), but the mechanism by which inflammasome activation promotes bacterial restriction is unclear. Inflammasome activation often triggers host cell death, thereby eliminating the pathogen's intracellular replicative niche. In vivo, pyroptosis can trigger formation of pore-induced intracellular traps (PITs). These PITs can trap intracellular bacteria that can subsequently be phagocytosed by neutrophils [102]. However, in murine macrophages, inhibition of Salmonella replication by caspase-1 and caspase-11 occurs prior to host cell death, indicating that caspase-1 and caspase-11 restrict Salmonella through a mechanism distinct from cell death [76]. Another mechanism of inflammasome-dependent restriction may be through promoting phagolysomal maturation. In murine macrophages infected with Legionella, NAIP5 activation results in increased colocalization of Legionella-containing vacuoles with the lysosomal markers cathepsin-D and Lamp-1 [103,104]. Perhaps a similar process occurs during Salmonella infection of human macrophages.
Overall, these data indicate that Salmonella infection of human macrophages triggers activation of multiple inflammasomes, and at least two of these inflammasomes, the NAIP/ NLRC4, and the NLRP3 inflammasomes, appear to be essential for controlling bacterial replication within macrophages. Furthermore, our data indicate that the human NAIP/NLRC4 inflammasome detects the SPI-2 needle protein SsaG, and that NAIP/NLRC4-mediated detection of the SPI-2 T3SS restricts Salmonella within macrophages. Collectively, our findings provide fundamental insight into how Salmonella is sensed and restricted by human macrophages. Moreover, these results offer a foundation for further understanding of how each of these pathways is activated and how these inflammasomes interact to mediate downstream responses that promote control of Salmonella infection in human macrophages.

Ethics statement
All studies involving primary human monocyte-derived macrophages (hMDMs) were performed in compliance with the requirements of the US Department of Health and Human Services and the principles expressed in the Declaration of Helsinki. hMDMs were derived from samples obtained from the University of Pennsylvania Human Immunology Core. These samples are considered to be a secondary use of deidentified human specimens and are exempt via Title 55 Part 46, Subpart A of 46.101 (b) of the Code of Federal Regulations.

Bacterial strains and growth conditions
Targeted deletion strains used in this study were made on the Salmonella enterica serovar Typhimurium SL1344 strain background. The ΔprgIfliCfljB strain was engineered using the ΔfliCfljB background [105], in which the SPI-1 T3SS needle, prgI, was deleted through a chloramphenicol resistance cassette insertion into prgI (fliCfljBprgI::CmR) using standard methods [106].
WT, ΔsipB [107], and ΔprgIfliCfljB isogenic strains were routinely grown overnight in Luria-Bertani (LB) broth with streptomycin (100 μg/ml) at 37˚C. For infection of cultured cells, overnight cultures were diluted in LB containing 300 mM NaCl and grown standing for 3 hours at 37˚C to induce SPI-1 expression [108].
Listeria monocytogenes WT and isogenic strains on the 10403S background were cultured in brain heart infusion (BHI) medium [86]. The Listeria strain encoding the heterologous bacterial ligand S. Typhimurium PrgJ translationally fused to the truncated N-terminus of ActA and under the control of the actA promoter was used [86]. The Listeria strains expressing S. Typhimurium SsaI and SsaG were constructed using codon-optimized gene fragments (IDT) cloned into the pPL2 vector and introduced into Listeria as previously described [86,109].

Cell culture of THP-1s
THP-1s (TIB-202; American Type Culture Collection) were maintained in RPMI supplemented with 10% (vol/vol) heat-inactivated FBS, 0.05 nM β-mercaptoethanol, 100 IU/mL penicillin, and 100 μg/mL streptomycin at 37˚C in a humidified incubator. Two days before experimentation, the cells were replated in media without antibiotics in a 48-well plate at a concentration of 2 × 10 5 cells/well and incubated with phorbol 12-myristate 13-acetate (PMA) for 24 hours to allow differentiation into macrophages. Macrophages were primed with 100 ng/mL Pam3CSK4 (Invivogen) for 16 hours prior to bacterial infections or anthrax toxin treatments. For experiments involving LPS, cells were pretreated with 500 ng/mL LPS (Sigma-Aldrich) for 3 hours. For experiments involving Nigericin, cells were treated with 10 μM Nigericin (EMD Millipore) for 6 hours. For experiments involving MCC950, cells were treated with 1 μM MCC950 (Sigma Aldrich) 1 hour prior to infection.

Cell culture of primary human monocyte-derived macrophages (hMDMs)
Purified human monocytes from de-identified healthy human donors were obtained from the University of Pennsylvania Human Immunology Core. Monocytes were cultured in RPMI

Bacterial infections
Overnight cultures of Salmonella were diluted into LB broth containing 300 mM NaCl and grown for 3 hours standing at 37˚C to induce SPI-1 expression [108]. Overnight cultures of L. monocytogenes were diluted and grown for 3 hours in BHI. All cultures were pelleted at 6,010 × g for 3 minutes, washed once with PBS, and resuspended in PBS. THP-1s were infected with S. Typhimurium or L. monocytogenes at a multiplicity of infection (MOI) of 20. hMDMs were infected with L. monocytogenes at an MOI of 5. Infected cells were centrifuged at 290 × g for 10 min and incubated at 37˚C. 1 hour post-infection, cells were treated with 100 ng/mL or 50 ng/ mL of gentamicin to kill any extracellular S. Typhimurium or L. monocytogenes respectively. Salmonella and Listeria infections in THP-1s proceeded at 37˚C for 6 hours. Listeria infection of hMDMs proceeded at 37˚C for 16 hours. For all experiments, control cells were mockinfected with PBS.

LFn-SsaG construct design and cloning
A construct encoding 6xHis-LFn-SsaG-6xHis was generated by sequential, ligation-free cloning. Briefly, the N-terminus of anthrax lethal factor (LFn) was amplified from FBDual-LFn-PrgJ and cloned into a linearized pOPIN-B E. coli expression vector [28,110]. Full-length SsaG was cloned into pOPIN-B-LFn from Salmonella enterica serovar Typhimurium SL1344 whole genomic DNA with addition of a C-terminal 6xHis tag.

Purification of His-tagged LFn-SsaG
Recombinant 6xHis-LFn-SsaG-6xHis was obtained by transforming pOPINB-LFn-SsaG into Rosetta DE3 competent E. coli. Cells were grown to an OD 600 of 0.7, at which point they were induced with 0.2 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) and grown overnight for 18 hours at 18˚C shaking at 180 rpm. Cells were pelleted and resuspended in a buffer solution containing 25 mM Tris (pH 8 at 4˚C), 200 mM sodium chloride, and 2 mM β-Mercaptoethanol (BME) and lysed by sonication. Crude lysate was allowed to bind to HisPur NI-NTA resin at 4˚C for 30 minutes and washed with a buffer solution containing 25 mM Tris (pH 8 at 4˚C), 400 mM sodium chloride, and 2 mM BME.
A step gradient elution was adapted from a Cold Spring Harbor protocol [111]. Briefly, 2 ml each of 25 mM Tris (pH 8 at 4˚C), 200 mM sodium chloride, 2 mM BME buffer with either 50 mM, 100 mM, 150 mM, 200 mM, 250 mM, 300 mM, or 500 mM imidazole were incubated sequentially on the resin for 3 minutes, and eluted into individual fractions. SDS-PAGE gel determined that the cleanest fractions were 200 mM and 250 mM imidazole. The selected fractions were dialyzed in 25 mM sodium phosphate buffer (pH 7.4) containing 0.5 mM EDTA over-night, concentrated by centrifugation, and flash frozen. Protein concentration was determined by NanoDrop.

siRNA-mediated knockdown of genes
All Silencer Select siRNA oligos were purchased from Ambion (Life Technologies). For CASP4, siRNA ID# s2412 was used. For CASP5, siRNA ID# s2417 was used. The two Silencer Select negative control siRNAs (Silencer Select Negative Control No. 1 siRNA and Silencer Select Negative Control No. 2 siRNA) were used as a control. Two days before infection, 30 nM of siRNA was transfected into macrophages using Lipofectamine RNAiMAX transfection reagent (Thermo Fisher Scientific) following the manufacturer's protocol. 16 hours before infection, the media was replaced with fresh antibiotic-free media containing 100 ng/ml Pam3CSK4. In parallel, siRNA-transfected cells were also transfected with 2 μg/ml of E. coli LPS strain W3110 (kindly provided by Robert Ernst) using FuGENE HD transfection reagent (Promega) for 6 hours.

Bacterial intracellular burden assay
Cells were infected with WT or ΔprgIfliCfljB S. Typhimurium as usual at an MOI of 20.1 hour post-infection, cells were treated with 100 μg/ml of gentamicin to kill any extracellular bacteria. 2 hours post-infection, the media was replaced with fresh media containing 10 μg/ml of gentamicin. At the indicated time points, cells were lysed with PBS containing 0.5% Triton to collect all intracellular bacteria. Harvested bacteria were serially diluted in PBS and plated on LB agar plates containing streptomycin (100 μg/ml) to enumerate colony forming units (CFUs). Plates were incubated at 37˚C overnight and then CFUs were counted.

LDH cytotoxicity assays
Harvested supernatants from infected cells were assayed for cytotoxicity by measuring loss of cellular membrane integrity via lactate dehydrogenase (LDH) activity. LDH release was quantified using an LDH Cytotoxicity Detection Kit (Clontech) according to the manufacturer's instructions and normalized to mock-infected cells.

Quantitative RT-PCR analysis
RNA was isolated using the RNeasy Plus Mini Kit (Qiagen) following the manufacturer's instructions. Cells were lysed in 350 μL RLT buffer with β-mercaptoethanol and centrifuged through a QIAshredder spin column (Qiagen). cDNA was synthesized from isolated RNA using SuperScript II Reverse Transcriptase (Invitrogen) following the manufacturer's protocol. Quantitative PCR was conducted with the CFX96 real-time system from Bio-Rad using the SsoFast EvaGreen Supermix with Low ROX (Bio-Rad). For analysis, mRNA levels of siRNAtreated cells were normalized to housekeeping gene HPRT and control siRNA-treated cells using the 2 −ΔΔCT (cycle threshold) method [112] to calculate knockdown efficiency. The following primers from PrimerBank were used. The PrimerBank identifications are CASP4 (73622124c1), and CASP5 (209870072c2), and HPRT (164518913c1); all 5 0 -3 0 : CASP4 forward: CAAGAGAAGCAACGTATGGCA CASP4 reverse: AGGCAGATGGTCAAACTCTGTA

Fluorescent microscopy of intracellular Salmonella
Two days prior to infection, 3 × 10 5 cells/well plated on glass coverslips in a 24-well plate. Cells were treated with PMA and primed with 100 ng/mL Pam3CSK4 as described above. Cells were infected with WT Salmonella pFPV25.1 (Salmonella constitutively expressing GFP) at an MOI of 20 as described above. At the indicated timepoints following infection, cells were washed 2 times with PBS and fixed with 4% paraformaldehyde for 10 min at 37˚C. Following fixation, cells were mounted on glass slides with DAPI mounting medium (Sigma Fluoroshield). Coverslips were imaged on an inverted fluorescence microscope (IX81; Olympus) and images were collected using a high-resolution charge-coupled-device camera (FAST1394; QImaging) at a magnification of 60×. All images were analyzed and presented using SlideBook (version 5.0) software (Intelligent Imaging Innovations, Inc.) and scale bars were added using ImageJ software. The proportion of infected cells containing GFP-expressing Stm (green) were scored by counting 50 infected cells per coverslip. 150 total infected cells were scored for each condition.

Statistical analysis
Prism 9.2.0 (GraphPad Software) was utilized for the graphing of data and all statistical analyses. Statistical significance for experiments with THP-1s and hMDMs was determined using the appropriate test and are indicated in each figure legend. Differences were considered statistically significant if the p value was <0.05.  Fig. (related to Fig 2). Uptake of Salmonella into THP-1 macrophages. WT, NAIP -/-, and two independent clones of NLRC4 -/-THP-1 monocyte-derived macrophages were primed with 100 ng/mL Pam3CSK4 for 16 hours. Cells were then infected with WT S. Typhimurium or ΔsipB S. Typhimurium at an MOI = 20. Cells were lysed at the 2 hours post-infection and bacteria were plated to calculate CFU. ns-not significant, � p < 0.05, ��� p < 0.001 by Tukey's multiple comparisons test. Data shown are representative of at least three independent experiments.

Supporting information
(TIF) S7 Fig. (related to Fig 3)  The three-dimensional structure of SsaG was predicted with high confidence and high coverage using the PHYRE2 server. The structure is colored from N to C terminus using the colors of the rainbow (red, orange, yellow, green, and blue).