|Home | About | Journals | Submit | Contact Us | Français|
Background.The role of toxins secreted by the type II secretion system (T2SS) of Pseudomonas aeruginosa during lung infection has been uncertain despite decades of research.
Methods.Using a model of pneumonia in Toll-like receptor (TLR) 2,4−/− mice, we reexamined the role of the T2SS system. Flagellin-deficient mutants of P. aeruginosa, with mutations in the T2SS and/or T3SS, were used to infect mice. Mice were followed up for survival, with some killed at different intervals to study bacterial clearance, inflammatory responses, and lung pathology.
Results.Strains carrying either secretion system were lethal for mice. Double mutants were avirulent. The T3SS+ strains killed mice within a day, and the T2SS+ strains killed them later. Mice infected with a strain that had only the T2SS were unable to eradicate the organism from the lungs, whereas those infected with a T2SS-T3SS double deletion were able to clear this mutant. Death caused by the T2SS+ strain was accompanied by a >50-fold increase in bacterial counts and higher numbers of viable intracellular bacteria.
Conclusions.The T2SS of P. aeruginosa may play a role in death from pneumonia, but its action is delayed. These data suggest that antitoxin strategies against this organism will require measures against the toxins secreted by both T2SS and T3SS.
Pneumonia due to Pseudomonas aeruginosa carries a high mortality rate  attributed to its secreted exotoxins  and possibly the inflammatory response to bacterial products . This bacterium is known to possess 5 protein secretion systems of which the type II secretion system (T2SS) and type III secretion system (T3SS) secrete the majority of known toxins. The T2SS secretes exotoxin A, LasA and LasB proteases, type IV protease, and phospholipase H, as well as lipolytic enzymes . The T3SS secretes exotoxins U, S, T, and Y . The latter system uses a membrane-spanning structure and needle to inject toxins into mammalian cells  whereas the T2SS is composed of multiprotein secretons encoded by the xcp and hxc operons  as well as an additional secretin, XqhA, that functions in T2 secretion when xcpQ, the major secretin, is mutated .
The biologic roles of the exoproducts of these systems have been under study for decades with clarity only about the role of the T3SS. Its major toxins, ExoU and ExoS cause death in animal models of pulmonary infection  and possible humans . However, most studies with few exceptions have failed to demonstrate that the T2SS toxins are important virulence factors during pulmonary infections [11–13]. There are several possible reasons for this: (1) no studies have addressed the role of the T2SS as a whole in an acute pneumonia model; (2) multiple toxins may be involved in death confounding mutational analysis when the roles of single toxins were addressed; (3) the full repertoire of T2 secreted toxins is not known, and other unidentified toxins may be effectors in mortality; (4) studies were done in the presence of a fully functional T3SS; and (5) any contribution of inflammation as a cause of death was not circumvented.
In prior studies, we demonstrated that Toll-like receptor (TLR) 2,4−/− mice are hypersusceptible to low inocula of a flagellin-negative mutant of P. aeruginosa, owing to the lack of recognition of lipopolysaccharide and flagellin . In those studies, bronchoalveolar lavage (BAL) samples of infected mice demonstrated a grossly defective early inflammatory response , Thus, we hypothesized that death was due to bacterial toxins and not inflammation. Using this model, we conclusively demonstrate that both the T2SS and the T3SS play independent roles in death due to Pseudomonas lung infection and that the XcpQ secretin is the major outer membrane pathway used by the T2SS effectors.
TLR2−/− and TLR4−/− mice obtained from S. Akira were backcrossed 8 times with C57BL/6 to ensure similar genetic backgrounds. TLR2,4−/− mice were generated by breeding TLR2−/− and TLR4−/− mice. Male mice were used for the experiments. Mice were fed and housed under standard conditions with air filtration and cared for in accordance with Pasteur Institute guidelines and in compliance with European Animal Welfare regulations.
Mice were anesthetized by intraperitoneal injection of a mixture of ketamine-xylazine, and infected via the intratracheal route, as described elsewhere . A 50-μL bacterial suspension containing 0.5–1.0 × 107 colony-forming units (CFU) was administered. In some mice survival was observed for 1 week after infection, and in others BAL was performed on groups of three mice after pentobarbital euthanasia at 6, 24, or 44 h after infection. The BAL fluids were diluted and plated on LB agar plates to obtain total viable bacterial counts. In some animals viable intracellular bacteria were enumerated by incubating washed BAL cells in tobramycin for 30 min, to kill extracellular bacteria, and lysing the cells with .1% triton X - 100 before plating on L-agar plates. Cell counts were measured in the BAL fluids and cell differential counts were determined after cytospin centrifugation and staining with Diff-Quik products. Murine cytokine concentrations in BAL fluid were determined using DuoSet enzyme-linked immunosorbent assay kits (R&D Systems).
All bacterial strains and plasmid vectors used in this study are described in Table1. For mutant construction, P. aeruginosa was grown in Luria broth at 37°C with shaking at 250 rpm or on 1.5% L-agar plates with or without antibiotics. For sucrose selection, plates containing 10% sucrose were used. To detect secretion of proteases, wild-type or mutant bacteria were spotted on 1.5% Casein-Milk agar plates. For the animal challenges, bacteria were grown overnight in Luria broth, transferred to fresh medium and grown for 4–5 h to mid–log phase. The cultures were centrifuged at 4000 g for 15 min, and the cell pellets were washed twice with phosphate-buffered saline and suspended in its original volume. The optical density was adjusted to give the approximate desired inoculum and verified by plate counts.
P. aeruginosa mutant strains PAKΔpscC (T3SS−), PAKΔxcp (T2SS−), and PAKΔpscCΔxcp (T2SS−/T3SS−) were obtained from Stephen Lory [16, 17]. These well-characterized mutants were used in studies of the role of the T3SS in toxicity to cells . The fliC mutants of these strains were engineered as described elsewhere, using a disrupted allele of the fliC gene of strain PAK . This provided 1 set of fliC mutant strains for the first series of experiments.
An in-frame partial gene deletion of 138 bp was made in pscF, the gene encoding the major needle protein of the T3SS, by cloning 1-kb DNA fragments up and downstream of the region to be deleted into the vector pEX18Gm . Plasmid pEXpΔpscF was used to introduce a partial deletion of the pscF gene in strain PAKΔfliC, using sucrose selection, as described elsewhere . An in-frame partial deletion in xcpQ was made by first cloning a 3.2-kb DNA fragment containing the entire xcpQ gene into pEX18Gm and then excising a 1109-bp Sal I fragment from xcpQ. This plasmid, pEXpΔxcpQ, was used to generate PAK mutants in PAKΔfliC and PAKΔfliCΔpscF, providing a second independent mutation in the T2SS (PAKΔfliCΔxcpQ) as well as a T2SS-T3SS mutant (PAKΔfliCΔpscFΔxcpQ). Next, we engineered these latter mutations in strain PA14 to examine whether our observations held with another P. aeruginosa strain, and lastly, the same T2SS and T3SS mutations were engineered in the wild-type flagellated strain PAK.
Chromosomal complementation of the PAKΔfliCΔpscFΔxcpQ mutant was achieved at the att site by cloning a 5.2-kb BamHI fragment from cosmid PAX24 that contains the xcp operon of P. aeruginosa strain PAO1  into the mini-CTX1 vector . This construct was transformed into Escherichia coli S17 and mated into PAKΔfliCΔpscFΔxcpQ mutant. The tetracycline-resistant colonies were resolved through flp excision . The mutant with an insertion of the mini-CTX1 vector into the att site was also constructed and used as a control. Complementation was verified by polymerase chain reaction and examining for restoration of protease secretion on casein-milk agar plates.
Mice were infected with 0.5–1 × 107 CFU of the mutants to be studied and euthanized with pentobarbital at different intervals after infection, and their lungs were fixed in formalin, sectioned, and stained with hematoxylin-eosin.
Cytokine levels, polymorphonuclear neutrophil (PMN) counts, and pathogen counts were expressed as means ± standard errors of the mean. Differences between groups were assessed for statistical significance, using analysis of variance followed by Fisher's test or t test, as appropriate. Differences were considered statistically significant at P < .05.
TLR2,4−/− mice were infected with the T2SS− and/or T3SS−defective strains and observed for mortality. A PAKΔfliC mutant that caused 100% mortality  served as the control. In most experiments, PAK mutants were flagellin defective; therefore, the notation “PAKΔfliC” will not be repeated unless required for clarification. Δxcp is a T2SS mutant, in which there is a deletion of the xcp operon, including xcpQ encoding the secretin XcpQ. ΔpscC is a T3SS mutant, in which part of the gene encoding PscC, an outer membrane protein, is deleted. ΔpscCΔxcp is a double mutant of the 2 secretion systems. The PAKΔfliC mutant rapidly killed all mice within 24 h (Figure 1A). This dose of bacteria, 0.5–1.0 × 107 CFU, does not cause death in wild-type mice (data not shown), because the innate host defense eradicates the infection . Unequivocally, the Δxcp mutant (T3SS+) killed all the mice within 24 h. The ΔpscC mutant (T2SS+) killed 8 of 9 mice, but deaths began after 24 h, with most occurring between days 2 and 3. In contrast, ΔpscCΔxcp (T3SS−/T2SS−) was avirulent (Figure 1A). Thus, each toxin secretion system played an independent role in causing death.
P. aeruginosa possesses 3 secretins, XcpQ, HxcQ and XqhA, that are known to be involved in T2 secretion . We deleted xcpQ to provide another mutation in the T2SS as well as to ascertain whether this was the secretin involved in secretion of the factors responsible for death. To test whether another independent mutation in the T3SS demonstrated a similar phenotype as the pscC mutant, we engineered a mutation in pscF, encoding the major protein of the T3SS needle and a double-secretion mutant, ΔpscFΔxcpQ. TLR2,4−/− mice were infected as described earlier. This second set of mutants had the same virulence phenotypes as the first, demonstrating that each secretion system had an independent effect (Figure 1B), and that XcpQ was the critical secretin, with no measurable contribution from the alternate secretion pathway through HxcQ or XqhA. The ΔxcpQ mutant (T3SS+) rapidly killed all mice, the ΔpscF mutant (T2SS+) killed 9 of 10 mice, again with the delay seen with the previous T2SS+ strain, and ΔpscFΔxcpQ (T3SS−/T2SS−) was avirulent (Figure 1B). To confirm that this specific secretin deletion led to the loss of virulence rather than a secondary mutation, we complemented xcpQ in ΔpscFΔxcpQ and challenged mice with this strain. As a control for these experiments, the mini-CTX vector was inserted into the att site of ΔpscFΔxcpQ. The complemented secretin mutation restored virulence, confirming the important role of the T2SS and XcpQ in virulence (Figure 1B).
Strain PA14 is believed to be one of the most virulent P. aeruginosa strain studied. We therefore engineered flagellin and secretion system mutants in this strain and examined the virulence of these mutants at 2 × 107 CFU per animal. Again, we observed the same independent effects of the T2SS and T3S, with the double mutant being less avirulent (Figure 2).
To characterize the role of the toxin secretion systems at the pathologic level, histopathologic analysis of the lungs was performed in mice 6, 24, or 44 h after infection with strains having mutations in the secretion systems, as indicated in Figure 3. Gross examination of the lungs at 6 h after infection (Figure 3A top panel) showed an intense inflammatory response among mice infected with strains that had a functioning T3SS (both T2SS+/T3SS+ and T2SS−/T3SS+). No data were obtained beyond this time with these 2 strains, because mice were dead within 24 h. In contrast, infection with a strain having only a functioning T2SS demonstrated gross hemorrhagic lesions only at 24 h after infection, which progressed with time (Figure 3B top panel). Mice infected with the T2SS−/T3SS− mutant demonstrated some gross inflammation on the surface of the lungs at 24 h, but this did not progress. Microscopic examination of the lungs (Figure 3 bottom panels) demonstrated changes that correlated with the survival data and the gross pathologic findings. At 6 h after infection only mice infected with strains having an intact T3SS (T2SS+/T3SS+ and T2SS−/T3SS+) showed pathologic changes, characterized by randomly distributed infiltrates of neutrophils, located in bronchiolar and alveolar spaces, often associated with focal necrosis of the bronchiolar overlying epithelium and alveolar walls. At this time, mice that were infected with the strain having only the T2SS did not demonstrate histopathologic changes, nor did mice infected with the T2SS−/T3SS− strain. At 24 and 44 h after infection, mice infected with T2SS+/T3SS− and T2SS−/T3SS− strains displayed similar inflammatory lesions—infiltrates of neutrophils located in the bronchiolar and alveolar spaces—but the extent and severity of the lesions differed between the 2 mutants (Figure 3B bottom panel). At 44 h after infection, the changes were more extensive and severe for the T2SS+/T3SS− mutant, whereas the mild inflammatory response seen in the double mutant was receding. Thus, the pathologic effects of the T2SS on the lungs are delayed, and they correlated with the survival studies, whereas the effects of the T3SS were seen 6 h after the challenge, and the animals died within 24 h.
Infected mice were euthanized, and BAL performed. Mice infected with T3SS+ mutants could be studied only at 6 h after infection. Mice infected with mutants that were T2SS+/T3SS− or T2SS−/T3SS− were studied at 6, 24, and 44 h after infection. Regardless of the mutant used, the number of bacteria in the BAL increased 5–10-fold at 6 h after challenge, compared with the administered inoculum (Figure 4A). Interesting differences were observed at later time points between mice infected with ΔpscFΔxcpQ (T2SS−/T3SS−) and those infected with ΔpscF (T2SS+/T3SS−). In mice infected with ΔpscFΔxcpQ, BAL bacterial counts fell rapidly in the next 20 h, but in those infected with ΔpscF, counts rose to >50-fold more than the initial inoculum (Figure 4A), implying a defect in the host defenses in the presence of the T2SS. The number of PMNs recovered from the BAL fluids did not differ significantly at 24 h (Figure 4B) from that at other time points (data not shown). Thus, the difference in colony counts cannot be explained by differences in the number of PMNs recruited. We therefore examined whether there was a defect in phagocytic killing when the T2SS was present by measuring viable intracellular bacterial counts in total BAL PMNs (Figure 4B). There were 4-fold more viable intracellular bacteria in the cells of the T2SS+ infected group than in the T2SS− group (P = .0186).
We also analyzed the synthesis of cytokines under the same experimental conditions (Figure 5 A–F). Globally, there was a tendency for a higher production in the presence of the T2SS, except for the tumor necrosis factor α response at 24 h, ruling out the subversion of these responses by the T2SS. However, there was one finding of possible significance in the host response to the T2SS competent mutant, elevated levels of the anti-inflammatory cytokine interleukin-10 (Figure 5F), which has been shown to inhibit P. aeruginosa clearance from the lungs and to worsen survival rates [23, 24], as well as affecting survival in other types of infections .
Given the clear-cut distinction between the action of the secretion systems in TLR2,4−/− mice, we examined whether this difference held for mice that had functioning TLRs 2, 4, and 5 by infecting wild-type mice with flagellated secretion system mutants (Figure 6). Even in these mice, we note that both secretion systems function independently, but 5–10-fold more bacteria were required to kill 100% of wild-type mice with the wild-type strain or the T3SS+ strain of P. aeruginosa. Deaths were even more delayed under these circumstances.
These studies demonstrate unequivocally that the T2SS of P. aeruginosa is capable of causing death in lung infections, an issue that has been unsettled. They also demonstrate that the onset of action differs between the T2SS and the T3SS , with the T3SS acting rapidly within 24 h, and the T2SS causing death at a slower rate, differences that were not hitherto appreciated and no doubt contributed to the lack of an appreciation of the role of the T2SS. It is not known whether this difference in the timing of lung injury is due to the fact that activation of the T2SS requires that bacteria reach high concentrations to achieve quorum sensing  or due to the fact that pathologic lesions are late in appearing; however, because the bacterial count in the lungs had increased significantly by 24 h, the observations are consistent with a role of quorum sensing in mediating lung injury and death.
These studies also narrow down the possible pathways used for T2 toxin secretion. It has long been suspected that the Xcp secreton was responsible for secretion of the better-known T2 secreted toxins of P. aeruginosa , but the discovery of the Hxc secreton  has raised the possibility that this pathway may also be used for novel toxin secretion. Additionally, there is XqhA , an alternative secretin that enables low-level secretion of XcpQ substrates. Neither of these, however, appear to play significant roles in virulence in this model, because deletion of xcpQ alone renders the organism avirulent in the absence of T3 secretion.
The ability of P. aeruginosa to avoid phagocytic clearance by PMNs is a major virulence determinant. Successful evasion of phagocytosis by P. aeruginosa is believed to be primarily dependent on the presence of a functional T3SS, because this system has been demonstrated to kill neutrophils . However, it also appears that a functional T2SS causes a clearance defect, as demonstrated in the present study by rising bacterial counts when the T2SS is present and more viable bacteria within the neutrophils. The sole possible insight into this defect involved the elevated concentrations of the anti-inflammatory cytokine interleukin-10 in BAL fluid. This cytokine has been demonstrated to impair both neutrophil and alveolar macrophage bactericidal activity [29, 30], and antibody against it improves survival in P. aeruginosa lung infection in a cecal ligation puncture model of immunosuppression . Thus, there may be 2 independent actions of the T2SS as a whole that are not fully explained by our knowledge of the secreted products—a toxic effect on the mouse lung and an independent one on host defenses. In vitro, none of the major T2 secreted enzymes (ExoA, LasB, and phospholipase H) kill neutrophils , even though subtle effects on function have been reported . In vivo, a mutant in a regulatory gene for ExoA, has been shown to impair host defenses in the lungs of wild-type mice by allowing bacterial counts to reach higher levels, but this mutation did not affect survival of mice and it had no effect on neutrophil migration into the airways . Elucidating which toxins are having these effects(s) will require a detailed characterization of the secreted products that use the XcpQ secretin and testing of mutants in the secreted products.
It may be argued that the T3SS is the first system to act, and because it kills more rapidly, there may not be a role for the T2SS. However, not all P. aeruginosa strains are competent for T3 secretion. Sokol et al  examined 124 clinical isolates from burns and bacteremia and found that only 38% of the isolates secreted ExoS, whereas 80% secreted ExoA. Assuming that strains having ExoU were missed (approximately one-third of those that are ExoS positive), this suggests that only half of these strains were T3 secretion competent. Roy-Burman et al  examined 71 non CF lung isolates and found that only 66% secreted ExoS or ExoU. Hauser et al  examined 35 selected strains from patients with pneumonia and found that 74% were T3SS competent, but they further pointed out that among these same isolates, secretion of ExoS was not consistently associated with virulence in a mouse model of pneumonia , suggesting that other virulence factors played a role in death. We have also examined 100 blood isolates for T2 secretion, using elastolytic activity as a proxy for T2 secretion, and have found that 99% of these isolates are T2 secretion competent (data not shown). Thus, these 2 systems may be viewed together as comprising a fail-safe system for defense against whatever host the organism encounters and are both integral to pathogenesis, neither more important than the other.
The findings in this study are also of practical significance in the development of active or passive vaccines or treatments for P. aeruginosa if one wishes to target the toxin-producing systems. Targeting either the T3SS or the T2SS, or any of their products alone, will not be optimal. Although it has been demonstrated that targeting PcrV of strain PA103 is protective , it should be pointed out that this strain demonstrates defects in T2 secretion; it does not secrete the major protease LasB and lacks a flagellum  and is therefore not representative of a large number of P. aeruginosa strains. Thus, some surface-exposed component of the T2SS or a specific toxic secretion product needs to be included in a vaccine designed to target secretions.
Besides demonstrating a role for the T2SS, this study also demonstrates the potential of using TLR-knockout mice to elucidate pathogenesis under circumstances where there is host compromise. A reasonable assumption is that for a significant infection caused by this opportunistic bacterium  to occur there must be failure of innate immunity which we have replicated by using these mice. Although challenge with larger numbers of bacteria will overcome the innate immune response of normal mice, this is also likely to cause an inflammatory response that may confound an analysis of the effects on individual toxins, especially if multiple toxins act together to cause death.
This work was supported in part by grants from Vaincre la Mucoviscidose and Legs Poix to M.C. and National Institutes of Health grant 1R01AI078770-01A1 to R.R.