An rhs gene of Pseudomonas aeruginosa encodes a virulence protein that activates the inflammasome

Abstract

The rhs genes are a family of enigmatic composite genes, widespread among Gram-negative bacteria. In this study, we characterized rhsT, a Pseudomonas aeruginosa rhs gene that encodes a toxic protein. Expression of rhsT was induced upon contact with phagocytic cells. The RhsT protein was exposed on the bacterial surface and translocated into phagocytic cells; these cells subsequently underwent inflammasome-mediated death. Moreover, RhsT enhanced host secretion of the potent proinflammatory cytokines IL-1β and IL-18 in an inflammasome-dependent manner. In a mouse model of acute pneumonia, infection with a P. aeruginosa strain lacking rhsT was associated with less IL-18 production, fewer recruited leukocytes, reduced pulmonary bacterial load, and enhanced animal survival. Thus, rhsT encodes a virulence determinant that activates the inflammasome.

The rhs genes are a widely distributed, enigmatic family of horizontally acquired genes. First described in Escherichia coli in the 1980s (1), rhs genes have subsequently been found in a broad range of Gram-negative bacteria, including other members of the Enterobacteriaceae, such as Salmonella, Yersinia, and Photorhabdus, as well as the Pseudomonadaceae, Flavobacteriaceae, Neisseriaceae, Myxoccaceae, and Vibrionaceae. An observed chromosomal rearrangement following recombination between distinct rhs genes in E. coli led to the name “rearrangement hot spot (rhs)” (1). However, recombination events are no longer thought to be a biologically relevant aspect of rhs genes (2). Despite their ubiquity and the many years since their discovery, rhs genes have not been assigned a definitive function, and even clear evidence of gene expression has been elusive (3).

Interest in rhs genes has been fueled by their unique structure. These genes, which generally range from 2 to 12 kb in size, exhibit a bipartite structure consisting of two distinct sequences: a long core followed by a short tip (Fig. S1A). Core sequences are GC-rich and display a high degree of intra- and interspecies sequence conservation. In contrast, tip sequences are relatively GC-poor and are highly variable even between closely related rhs genes. Corresponding to the rhs gene core and tip sequences, the proteins predicted to be encoded by rhs genes also contain two regions: a large core domain and a short C-terminal tip domain. Rhs core domains are defined by a variable number of tyrosine-aspartate (YD) repeats and are separated from their cognate tip domains by a 61-amino acid hyperconserved region ending in the consensus sequence PXXXXDPXGL (2). Some predicted Rhs proteins also contain a large N-terminal domain. Putative Rhs proteins are predicted to be hydrophilic, and the products of some rhs genes have features of bacteriocins or capsule transport proteins (4–6). Furthermore, Rhs protein YD-repeats and hyperconserved regions display sequence similarity with toxins produced by bacteria that infect insects (7). These observations suggest that some rhs genes encode surface exposed or secreted proteins.

rhsT is an rhs gene found within the genomic island PAGI-9 of Pseudomonas aeruginosa (Fig. S1A). This gene was first identified in a screen for genetic elements present in the highly virulent P. aeruginosa strain PSE9 but absent in the relatively avirulent strain PAO1 (8). For this reason, we examined whether rhsT contributed to the highly virulent phenotype of PSE9.

Results

Levels of rhsT mRNA Are Growth Phase-Dependent and Increase in the Context of Infection.

In previous studies with bacteria grown in routine laboratory media, demonstration of rhs gene expression has been difficult (3). To examine the expression of rhsT, we performed quantitative RT-PCR analysis on bacteria grown under a variety of conditions. rhsT transcripts were detected in PSE9 grown in either rich (Luria-Bertani, LB) or minimal (Vogel-Bonner minimal) medium, with a trend toward higher mRNA levels in stationary phase compared with exponential phase (Fig. S1B). Interestingly, rhsT transcripts were increased over 30-fold (compared with exponential phase LB cultures) during growth in the presence of THP-1 cells, a human monocytic cell line. Growth of PSE9 bacteria in THP-1–conditioned medium alone (RPMI taken from THP-1 cells following 16 h of growth) did not promote rhsT transcription beyond that observed with laboratory media, suggesting that rhsT expression is induced during bacterial infection of human cultured phagocytic cells.

rhsT Is a 6672-bp ORF with σ⁵⁴ −12/−24 Sequences.

To better characterize the rhsT gene and its regulation, we determined its transcriptional start site by 5′ RACE. We obtained the 5′ sequence of the rhsT transcript (Fig. S1A) and experimentally validated the rhsT transcriptional start site predicted by bioinformatic approaches. This start site was located 38-bp upstream of a predicted RhsT start codon. Inspection of sequences upstream of the rhsT transcriptional start site revealed consensus −12 and −24 sequences recognized by the alternative σ factor σ⁵⁴, which regulates diverse processes ranging from metabolism to virulence (9, 10). In addition to suggesting σ⁵⁴ regulation, mapping of the rhsT transcriptional start site defined the length of the rhsT ORF. Two different rhsT translational start sites were predicted by bioinformatic approaches. EasyGene 1.2 and GeneMark.hmm for Prokaryotes 2.8 predicted a 6,672-bp ORF, and National Center for Biotechnology Information ORF Finder predicted a 6,837-bp ORF. Our 5′ RACE results exclude the possibility of a 6,837-bp ORF because the beginning of the ORF precedes the rhsT transcriptional start site.

RhsT Is on the Bacterial Cell Surface.

To further characterize the protein encoded by the rhsT gene, we generated an isogenic mutant (PSE9ΔrhsT) in which the majority of the rhsT ORF is replaced by a gentamicin-resistance cassette. Furthermore, an inducible overexpression strain PSE9ΔrhsT(pRHST) was generated by cloning rhsT into a P_BAD-based arabinose inducible expression plasmid. Growth curves at 37 °C with shaking indicated no difference in growth rate between the parental, mutant, and overexpression strains. Immunoblot analysis using an antibody generated against a predicted peptide of RhsT was performed to determine if RhsT protein was produced in P. aeruginosa. Although no protein was detected in PSE9 lysates, a faint band was observed between the 238- and 268-kDa molecular weight markers in the cell lysate of PSE9ΔrhsT(pRHST) induced with arabinose (Fig. 1A). This band is consistent with the predicted molecular weight of full-length RhsT (240 kDa), and was identified as RhsT by tandem mass spectrometry. Because rhs genes have been predicted to encode bacterial cell surface-associated and -secreted proteins (3), we next determined whether RhsT could be detected in culture supernatant and on the bacterial cell surface. A 240-kDa band was detected in concentrated PSE9ΔrhsT(pRHST) culture supernatant (Fig. 1A), and examination of fixed but nonpermeabilized PSE9ΔrhsT(pRHST) bacteria grown under inducing conditions by indirect immunofluorescence revealed staining around the bacterial cell circumference (Fig. 1B). Because these bacteria were not permeabilized (Fig. S2A), this staining pattern indicated that RhsT was on the bacterial cell surface, either as a surface protein or in the process of being secreted.

Fig. 1.

Detection of RhsT by immunoblot and indirect immunofluorescence microscopy. (A) Immunoblot analysis of cell lysates and culture supernatants from PSE9ΔrhsT(pRHST) and PSE9ΔrhsT(pHERD) (PSE9ΔrhsT containing the empty arabinose inducible vector). RNAPβ was used as a loading control and a control for cell lysis in culture superntants. (B) Visualization by indirect immunofluorescence microscopy of RhsT in PSE9ΔrhsT(pHERD) and PSE9ΔrhsT(pRHST) bacteria that were fixed but not permeabilized.

RhsT Is Translocated into Eukaryotic Cells.

Given that the YD-repeats and the hyperconserved region within the RhsT core domain are similar in sequence to the Photorhabdus TccC insecticidal toxin (7), we hypothesized that RhsT is a secreted protein that is translocated into eukaryotic cells. To determine if this is indeed the case, we used the CCF2 reporter system, which is highly sensitive because it is based on enzymatic signal amplification (11). In this approach, eukaryotic cells are loaded with the fluorogenic substrate CCF2, which consists of a β-lactam ring modified with two fluorophores that upon excitation undergo FRET to fluoresce green. In the presence of β-lactamase activity, however, the β-lactam ring is cleaved, FRET is lost, and the substrate fluoresces blue upon excitation. Thus, cells intoxicated with β-lactamase will exhibit blue fluorescence but those not intoxicated will fluoresce green. We generated a strain of PSE9 that expressed RhsT fused at its C terminus to β-lactamase (designated PSE9rhsT-bla) and used it to determine whether the RhsT protein could target the fused β-lactamase domain to the cytosol of CCF2-treated J774 cells (a macrophage-like cell line). RhsT-β-lactamase translocation into J774 cells was detected by fluorescence microscopy, which showed blue fluorescent J774 cells (Fig. S2B). These qualitative results were confirmed and quantified using blue and green fluorescence intensity measurements. As expected, cells infected with PSE9 producing RhsT-β-lactamase exhibited a relatively high blue-to-green fluorescence ratio (Fig. 2A). In contrast, low blue-to-green ratios were observed with cells infected with PSE9 producing RhsT lacking the β-lactamase domain, as well as cells infected with the control strain PSE9gst-bla, which expresses the cytoplasmic protein GST fused to β-lactamase. Thus, the elevated blue-to-green fluorescence ratio observed with PSE9rhsT-bla infection was not because of β-lactamase release following lysis of bacteria internalized by J774 cells. Consistent with this observation, RhsT translocation was not abrogated in J774 cells pretreated with cytochalasin D, an inhibitor of phagocytosis, indicating that bacterial internalization was not necessary for RhsT translocation (Fig. 2A). Interestingly, cytochalasin D treatment did eliminate the background blue fluorescence observed with the PSE9 and PSE9gst-bla controls, suggesting that phagocytosis followed by release of endogenous P. aeruginosa β-lactamases, such as AmpC, contributed to these background levels (12). Taken together, these results indicate that RhsT is translocated into J774 cells.

Fig. 2.

RhsT translocation and RhsT-mediated cytotoxicity and IL-1β release. (A) RhsT translocation into J774 murine macrophages in vitro. Fluorometric plate reader quantification was performed on J774 cells treated with CCF2-AM and infected with bacteria expressing RhsT-β-lactamase fusion protein. (B) Comparison of cytotoxicity (as measured by LDH release) of J774 cells infected for 4 h at an MOI of 20 with PSE9 or PSE9ΔrhsT in the presence or absence of YVAD pretreatment. (C) IL-1β release from THP-1 cells following infection with PSE9 or PSE9ΔrhsT. Cells were infected at an MOI of 15 (P. aeruginosa strains PSE9 or PSE9ΔrhsT) or 5 (Shigella strains M90T or BS176). IL-1β levels were measured at 3 h postinfection. All data are means ± SEM (n = 3). *P < 0.05, two-tailed paired Student t test.

RhsT Translocation Is Associated with Death and Inflammasome Activation in Monocyte/Macrophage-Like Cells.

We next examined the fate of J774 macrophage-like cells following infection with RhsT⁺ bacteria by using lactate dehydrogenase (LDH)-release as a marker for cell death. PSE9 killed J774 cells, and disruption of rhsT reduced this killing by ∼50% (Fig. 2B). This RhsT-dependent cytotoxicity was abrogated by pretreatment of J774 cells with the caspase-1 inhibitor YVAD, suggesting that host cell death caused by RhsT involves inflammasome signaling.

Mammalian cells detect bacterial factors within their cytosol by multiprotein complexes known as inflammasomes. Inflammasomes assemble around NOD-like receptor or HIN-200 proteins and cause the activation of caspase-1 by a process requiring the adaptor protein ASC. To further explore the relationship between host cell death caused by RhsT and inflammasome activation, we used THP-1 human monocytic-like cells, which have become a model for studying inflammasome activation. THP-1 cells with intact or disrupted inflammasomes because of a deficiency in ASC were infected with PSE9 or PSE9ΔrhsT. In THP-1 cells with intact inflammasomes, PSE9 caused twice as much cell death as PSE9ΔrhsT (Fig. S3A). In contrast, both strains of bacteria caused similar amounts of cell death in THP-1 cells deficient in ASC (Fig. S3 A and B). Next, we determined whether RhsT by itself was sufficient to mediate host cell killing. Transfection of an rhsT-expressing construct indicated that RhsT was indeed sufficient to cause THP-1 cell lysis (Fig. S3 C and D). Significantly less cell lysis was observed in ASC-deficient THP-1 cells transfected with rhsT, indicating that in the context of transfection, RhsT-mediated host cell lysis is dependent in part on the inflammasome adapter protein ASC.

Inflammasome activation is associated with activation of caspase-1, which results in the processing and subsequent release of cytokines IL-1β and IL-18. To further evaluate whether RhsT activates host cell inflammasomes, we used ELISAs to measure release of IL-1β from infected THP-1 cells. The Shigella strains M90T and BS176 were used as controls. M90T, which contains a plasmid encoding a type III secretion system that has been shown to activate caspase-1, caused robust release of IL-1β. BS176, which lacks the type III secretion plasmid, did not cause release of detectable amounts of IL-1β. PSE9 caused significantly more IL-1β release than PSE9ΔrhsT (Fig. 2C). When inflammasome activation was blocked either by the caspase-1–specific inhibitor YVAD or by a deficiency of the adaptor protein ASC, this difference was no longer evident. Similar results were observed with IL-18 (Fig. S4A) and with transfection of an rhsT-expressing construct (Fig. S4B), indicating RhsT by itself is sufficient to induce ASC-dependent IL-1β release. Taken together, these results demonstrate that RhsT causes release of cytokines consistent with activation of host cell inflammasomes.

The preceding results suggest that RhsT is translocated into host cells and causes inflammasome activation under in vitro conditions. We next wished to determine whether similar translocation and inflammasome activation occurred in vivo. To accomplish this, we used an intranasal aspiration mouse model of acute pneumonia. After 15 h of infection, cells in the airways were recovered by bronchoalveolar lavage (BAL), incubated with CCF2, and analyzed by flow cytometry. Similar to what we observed in vitro, significantly more blue cells were detected in lavage fluid from mice infected with PSE9rhsT-bla than from mice infected with PSE9gst-bla (Fig. S2 C and D). To assess in vivo inflammasome activation, we chose to examine IL-18 levels, because in vitro measurements of this cytokine were more consistent than those of IL-1β. Mice were infected with PSE9 or PSE9ΔrhsT and killed after 4 or 15 h of infection. Mouse lungs were removed and homogenized, and ELISAs were performed to quantify IL-18 levels in the lungs. Significantly higher levels of IL-18 were detected in the lungs of PSE9-infected mice relative to those of PSE9ΔrhsT-infected mice at both time points (Fig. S4C). These results suggest that RhsT-mediated inflammasome activation is not simply an in vitro phenomenon but also occurs in vivo.

RhsT Is Associated with Enhanced Lung Inflammation in a Mouse Model of Acute Pneumonia.

RhsT causes increased release of IL-1β and IL-18, both of which are potent inducers of inflammation during pneumonia (13, 14). To determine whether RhsT enhanced inflammation in vivo, we used a mouse model of acute pneumonia to compare inflammation associated with infection by PSE9 and PSE9ΔrhsT. Histological examination of mouse lungs at 26 h postinfection with PSE9 revealed marked inflammatory cell infiltrates involving both the alveolar space and the bronchioles (Fig. 3A and Fig. S5A). In contrast, modest inflammatory infiltrates were observed in the lungs of mice infected with PSE9ΔrhsT, and the alveolar air spaces were largely preserved. Infection with the complemented mutant PSE9ΔrhsT+RHST, which contained a single copy of rhsT under control of its endogenous promoter at a neutral chromosomal site, resulted in inflammation similar to that of PSE9. Quantification of pulmonary leukocytes (CD45⁺ cells) by flow cytometry verified these impressions and demonstrated that inflammation began to resolve in PSE9ΔrhsT-infected mice by 26 h postinfection, whereas the inflammatory infiltrate continued to increase in PSE9-infected mice (Fig. 3B). Although the absolute number of leukocytes in the lungs of mice infected with the parental and rhsT mutant strains was highly disparate by 26 h postinfection, the relative proportions of cell types comprising these leukocytes were similar (Fig. S5B). For both PSE9- and PSE9ΔrhsT-infected mice, the predominant cell types in the lungs were macrophages and monocytes at 4 h postinfection. At 15 h postinfection, the predominant cell types shifted to neutrophils, and then at 26 h postinfection shifted back to macrophages and monocytes. These results indicate that infection with an RhsT⁺ strain of P. aeruginosa is associated with not only higher IL-18 levels, but also enhanced inflammation and loss of normal alveolar architecture during acute pneumonia. The results also suggest that robust recruitment of inflammatory cells associated with RhsT more than compensates for any loss of these cells because of inflammasome-mediated cell death.

Fig. 3.

Pulmonary inflammation associated with RhsT in a mouse model of acute pneumonia. (A) H&E-stained mouse lung sections taken at 26 h postinoculation with PSE9, PSE9ΔrhsT, PSE9ΔrhsT+RHST, or PBS. Representative images from similar anatomical sites were chosen. n = 3 for each strain. See Fig. S5A for scoring. (Scale bars, 50 μm.) (B) Total numbers of CD45⁺ cells in the lungs of mice infected with PSE9 or PSE9ΔrhsT over a 37-h time course. *P < 0.05, two-tailed paired Student t test.

RhsT Is Associated with Enhanced Virulence in a Mouse Model of Acute Pneumonia.

The inflammasome is a part of the host innate-immune system. Thus, activation of the inflammasome and subsequent recruitment of inflammatory cells to the site of infection is in many situations beneficial to the host and results in eradication of the infectious agent (15). In other cases, however, excessive inflammation may not only fail to eliminate the pathogen but may cause collateral damage to host tissues. In these cases, activation of the inflammasome may actually worsen infection and be a virulence mechanism of the pathogen (16). To determine which of these scenarios occurs with RhsT, we infected mice with P. aeruginosa and determined the bacterial loads in the lungs at subsequent time points. Whereas PSE9 bacteria multiplied to high numbers during the first 26 h of infection, PSE9ΔrhsT bacteria were slowly cleared (Fig. 4A). Similar trends were observed in mixed infections in which deletion of the rhsT gene was associated with a competitive survival disadvantage relative to the parental strain (Fig. S6A). Interestingly, the effect of RhsT in competition assays was much reduced relative to that observed in mice infected with either PSE9 or PSE9ΔrhsT individually (compare Fig. 4A and Fig. S6A), consistent with the notion that bacteria producing RhsT aided the survival of bacteria that did not produce this factor. Such a “transcomplementation” effect has been observed with bacterial toxins acting on phagocytic cells in the lungs (17). Other measures of disease severity also indicated that PSE9 was more virulent than PSE9ΔrhsT. LDH, a marker for tissue destruction during pneumonia (18), was present in higher amounts in BAL fluid from PSE9-infected mice compared with PSE9ΔrhsT-infected mice (Fig. S6B). Similarly, PSE9 was also associated with enhanced dissemination to the spleen by 26 and 37 h postinfection (Fig. S6C). Finally, in survival assays, mice infected with parental PSE9 or the complemented strain did not survive beyond 48 h (Fig. 4B). In contrast, mice infected with PSE9ΔrhsT survived through the entire 7-d course of the experiment. These results indicate that RhsT is a virulence determinant that plays an important role in the pathogenesis of P. aeruginosa pneumonia.

Fig. 4.

Virulence of RhsT in a mouse model of acute pneumonia. (A) Bacterial load in the lungs over time. Mice were infected with 3.0 × 10⁵ CFU of either PSE9 or PSE9ΔrhsT and killed at the indicated time points to determine the CFU in the lungs. Data shown are means ± SEM (n = 6 for each time point). At 26 h postinfection, PSE9 persistence was significantly greater than that of PSE9ΔrhsT (P < 0.001, two-tailed paired Student t test). (B) Survival of mice over time following infection. Mice were infected with 3.0 × 10⁵ CFU of PSE9, PSE9ΔrhsT, or PSE9ΔrhsT+RHST and monitored for 7 d. Each group contained 10 mice combined from two independent experiments. PSE9ΔrhsT survival was significantly better than that of PSE9 and PSE9ΔrhsT+RHST (for both, P < 0.00001, log-rank test).

Discussion

The rhs genes are present in a broad spectrum of Gram-negative bacteria and have been studied for 25 y, yet their function has remained poorly understood. Here we demonstrate that rhsT, a P. aeruginosa rhs gene, is induced during infection of monocyte/macrophage-like cells, is translocated into these cells, and causes cell death in vitro. Moreover, we demonstrate that RhsT intoxication is associated with features of inflammasome activation, such as ASC-dependent cytotoxicity and release of the potent proinflammatory cytokines IL-1β and IL-18. Although we have not demonstrated a causal relationship, our data show that the release of these proinflammatory cytokines is associated with enhanced pulmonary inflammation, increased bacterial numbers, and decreased survival in a mouse model of acute P. aeruginosa pneumonia.

To the best of our knowledge, this report of an rhs gene encoding a virulence determinant against mammals is unique. Prior studies have suggested that rhs genes could encode factors that facilitate bacterial-host or bacterial-bacterial interactions. rhsA, an E. coli rhs gene, was previously identified in a transposon insertion mutagenesis screen for genes required for calf intestine colonization; however, the rhsA insertion mutant was not further characterized (19). An endosymbiotic bacterium carrying an rhs gene was associated with enhanced protection of insect hosts from parasitic wasp larvae (20), and an rhs gene encoded protein in Xenorhabdus bovienii was toxic to nematodes (GenBank accession no. CAC19493). Other members of the rhs gene family appear to interact with bacteria instead of eukaryotic cells. Early work by Vlazny and Hill indicated that rhsA of E. coli had features of a bacteriocin-encoding gene (21). Sisto et al. identified an rhs gene in Pseudomonas savastonoi with similar properties (4). Consistent with these reports, subsequent bioinformatics analysis has linked some rhs genes to a large family of nucleases and nucleic acid deaminases, some of which function as bacteriocins (22). McNulty et al., however, have linked the rhsA gene of E. coli to polysaccharide transport (6), an observation supported by the similarity between the YD-repeats of putative Rhs proteins and those of teneurin-1, a eukaryotic protein that binds heparin (23). Taken together, these findings suggest that rhs genes encode a large family of proteins with a variety of activities and cell targets. Whether additional Rhs proteins act as virulence factors in mammals is currently being investigated.

Our results indicate that at least the C-terminal tip of RhsT is translocated into monocyte/macrophage-like cells, but it remains to be determined whether full-length RhsT is translocated into eukaryotic cells. Some large toxins, such as MARTX toxins and the large clostridial toxins, are secreted as full-length proteins that are proteolytically processed subsequent to entering host cells (24). Other large toxins, such as the contact-dependent inhibition proteins, are thought to be expressed as full-length proteins on the bacterial cell surface but function to deliver only a C-terminal domain to susceptible cells (25). We were unable to detect the RhsT protein inside eukaryotic cells by immunoblotting with antibodies to RhsT or β-lactamase, suggesting the amount of RhsT translocation is below the limit of antibody detection. The secretion mechanisms responsible for RhsT transport to the bacterial cell surface and translocation into eukaryotic cells is also unclear. As RhsT does not have an identifiable N-terminal Tat or Sec signal sequence, secretion is unlikely to proceed through the type II or type V secretion systems. Secretion may be through one of the one-step secretion systems present in P. aeruginosa (type I, III, or VI) or through outer membrane vesicles. Alternatively, as mentioned above, RhsT may not be secreted at all but rather be directly inserted into host cells by a mechanism similar to that described for contact-dependent inhibition proteins. In this regard, it is interesting that rhs genes and genes encoding toxins of contact-dependent inhibition systems have been linked by bioinformatic approaches (22).

RhsT-dependent inflammasome activation serves as further evidence of RhsT translocation into host cells. RhsT caused increased release of the cytokines IL-1β and IL-18 in an ASC- and caspase-1–dependent manner. It is unclear whether RhsT itself directly activates inflammasome signaling by gaining access to the cytosol and interacting with a sensor protein, or if inflammasome activation is indirect, resulting from the sensing of RhsT-induced injury to the host cell. Some speculations on the mechanism of RhsT-dependent inflammasome activation can be made by comparing RhsT to other inflammasome-activating bacterial toxins. RhsT bears neither primary sequence nor secondary structure similarity to the majority of characterized inflammasome-activating bacterial toxins, which are pore-forming toxins thought to indirectly trigger inflammasome activation through K⁺ efflux (26). RhsT does, however, share some similarities with the inflammasome-activating toxins TcdA and TcdB produced by Clostridium difficile. Like RhsT, TcdA/B are large multidomain toxins, and the YD-repeats in the RhsT core domain are similar in sequence to the TcdA/B C-terminal repeats thought to mediate host cell binding. A recent study found that purified TcdB protein is sufficient for inflammasome activation, and interestingly this activation is independent of TcdB enzymatic activity (27). Although it is intriguing to speculate that a common fold shared between the RhsT core domain and the TcdA/B C-terminal domain is involved in inflammasome activation, the RhsT domains necessary for inflammasome activation are not currently known.

Although the inflammasome is a component of the host innate immune system and inflammasome activation often functions to eliminate bacteria during early infection, certain microbes can co-opt the inflammasome to cause excessive inflammation. For example, Shigella spp. and Salmonella enterica secrete IpaB and SipB, respectively, to activate caspase-1 and cause IL-1β release. The subsequent excessive inflammation that develops is thought to be a critical step in the pathogenesis of these bacteria (16, 28, 29). A similar mechanism has been proposed for the clostridial toxins TcdA/B (27). We found that RhsT⁺ P. aeruginosa strains were associated with robust inflammation in the lungs of infected mice. Whether enhanced inflammation led to increased bacterial numbers or vice versa cannot be directly determined from our data. However, RhsT-associated elevation of IL-18 in the lungs occurred as early as 4 h postinfection, when equivalent numbers of PSE9 and PSE9ΔrhsT bacteria were present, indicating that proinflammatory cytokine levels were not solely driven by differences in bacterial numbers. Rather, excessive inflammation driven by inflammasome activation and subsequent IL-1β and IL-18 release may have been at least partly responsible for the increased numbers of RhsT⁺ P. aeruginosa bacteria in the lungs. Increased levels of these cytokines have been shown to be detrimental to the clearance of P. aeruginosa in animal models (13, 14), and elevated IL-1β in BAL fluid correlates with bacterial persistence in the lungs of human intensive-care unit patients (30). Thus, in our mouse model of acute P. aeruginosa pneumonia, RhsT-dependent enhancement of proinflammatory cytokine production and leukocyte infiltration may function as a virulence mechanism. Alternatively, RhsT may possess undefined intrinsic activity that causes tissue injury independent of inflammasome activation, and this activity may be responsible for the bacterial persistence and mortality observed in the mouse model of pneumonia. Finally, it remains possible that RhsT is acting indirectly to potentiate the activity of another P. aeruginosa virulence determinant. Experiments are underway to distinguish these possibilities.

In summary, the rhsT gene of P. aeruginosa encodes a unique virulence determinant that activates the inflammasome and plays an important role in the pathogenesis of pneumonia. This finding suggests that other members of the widely distributed rhs family of genes may also contribute to pathogenesis.

Materials and Methods

Bacterial strains, culture conditions, cell lines, 5′ RACE, immunoblotting, tandem mass spectrometry, transfections, and cytokine assays are described in detail in SI Materials and Methods and Tables S1–S3.

Quantitative Real-Time PCR.

PSE9 cultures were grown at 37 °C with shaking to either stationary or exponential phase. Stationary phase cultures were defined as cultures grown for 16 h, and exponential phase cultures were grown by subculturing overnight cultures in fresh medium to an OD of 0.300. For infection, bacteria were incubated with THP-1 cells for 2.5 h at a multiplicity of infection (MOI) of 20.

For all conditions, total RNA was used to generate cDNA, and quantitative RT-PCR was performed with B-R SYBR Green SuperMix for iQ (Quanta Biosciences). See SI Materials and Methods for more detail.

Indirect Immunofluorescence Microscopy.

Bacteria were grown for 17 h in LB medium at 37 °C with shaking and then subcultured with 0.02% (wt/vol) arabinose (Sigma) for 2 h. Bacteria were then either fixed with PBS containing 2.4% (vol/vol) formaldehyde, 0.04% (vol/vol) glutaraldehyde, or fixed and permeabilized with cold acetone. RhsT was detected with a rabbit antibody generated against an RhsT peptide (21st Century Biochemicals). RNA polymerase (RNAPβ) was detected with the mouse monoclonal antibody 8RB13 (Abcam). See SI Materials and Methods for more detail.

Cytotoxicity Assays.

Cytotoxicity was measured using the CytoTox 96 Nonradioactive Cytotoxicity Assay (Promega). Where indicated, cells were pretreated with 100 μM Ac-YVAD-CMK (Cayman) for 45 min before infection. For all experiments, percent cell lysis = [(LDH in sample well − LDH in background well)/(LDH in Triton X-100 treated wells − LDH in background well)] × 100. Background was taken as uninfected cells, cells transfected with empty vector (pcDNA3.1D/V5-His), or BAL fluid from PBS-infected mice.

Translocation Assay.

J774 cells seeded in 24-well black tissue culture treated plates (Wallac, PerkinElmer) were incubated with DMSO carrier control or 5 μg/mL cytochalasin D (Sigma) for 15 min, loaded for 1 h at room temperature and 15 min at 37 °C with 1 μM CCF2-AM (Invitrogen), and then infected at an MOI of 20. Infections proceeded in a bottom read SpectraMax M5 fluorescence plate reader at 37 °C. Excitation was set at 410 nm; emissions at 450 nm and 520 nm were recoded. Blue-to-green fluorescence ratio = (RFU_{450nm, t = 30min} − RFU_{450nm, t = 0})/(RFU_{520nm, t = 0} − RFU_{520nm, background}), where RFU is relative fluorescence units and background fluorescence is fluorescence emission from cells that were infected but not loaded with CCF2-AM. Microscopy was performed as previously described (31). For detection of RhsT-β-lactamase translocation in vivo, mice were infected with 3.0 × 10⁶ CFU PSE9gst-bla or PSE9rhsT-bla. At 15 h postinfection, immune cells in the airways were collected by performing BAL as previously described (17). Following quantification of Trypan blue-excluding cells using a hemocytometer, 2.0 × 10⁵ cells were loaded with 1 μM CCF2-AM in PBS. After a 1-h incubation at room temperature protected from light, cells were fixed and analyzed by flow cytometry using a Becton Dickinson LSRFortessa instrument.

Mouse Model of Acute Pneumonia.

BALB/c mice were infected as previously described (31). BAL and lung histopathology at 26 h postinfection were performed as previously described (17). Images were captured using the TissueGnostics image acquisition software TissueFAXS. For flow cytometry analysis of immune cell recruitment to the lungs, mice were killed at the indicated time points, and immune cells in the lungs were quantified as previously described (17). See SI Materials and Methods for more detail.