Association between Pseudomonas aeruginosa type III secretion, antibiotic resistance, and clinical outcome: a review

Recently, multidrug-resistant (MDR) Pseudomonas aeruginosa has been identified as a major cause of nosocomial infections [1],[2]. P. aeruginosa is the most frequent Gram-negative pathogen to cause mortality of patients with ventilator-associated pneumonia (VAP) in intensive care units [3]-[5]. Better understanding of P. aeruginosa pathogenesis, and subsequent mortality, has been acquired by recent advances in knowledge regarding virulence mechanisms that lead to acute lung injury, bacteremia, and sepsis [6]. In common with other pathogenic Gram-negative bacteria, P. aeruginosa possesses a virulence mechanism known as the type III secretion system (TTSS). The TTSS allows the injection of toxins into the cytosol of target eukaryocytes [7],[8]. The type III secretory (TTS) toxin, ExoU, has been characterized as a major virulence factor in acute lung injury [9],[10]. The genomic organization of the ExoU gene, enzymatic activity of the ExoU protein, and mechanism of cell death induced by ExoU translocation have all been investigated. Among the various phenotypes of P. aeruginosa isolates, the ExoU-positive phenotype is a major risk factor for poor clinical outcomes. A correlation between the antimicrobial characteristics of the bacterium and an exoU-positive genotype has also been reported in recent clinical studies [11],[12].

This review summarizes progress with respect to basic research conducted on the TTS toxin, ExoU, to date. We have covered its genomic organization and biochemistry and its ability to cause acute lung injury in people. Additionally, we will discuss the findings of recent studies on the association between ExoU and poor clinical outcome in patients.

Isolates of P. aeruginosa show cytotoxicity in cultured epithelial cells and cause a high degree of acute lung injury in animal models of pneumonia [13]-[15]. Clinical isolates of P. aeruginosa display various genotypic and phenotypic variations that can affect the severity of an infection and its clinical outcome [9]. P. aeruginosa produces various exoproducts, among which exoenzyme S and its co-regulated proteins are candidates for cytotoxicity and acute lung injury in patients with P. aeruginosa pneumonia (Table 1) [16]-[18]. In the 1990s, based on genomic homology with its counterparts in other Gram-negative bacteria, P. aeruginosa exoenzyme S was identified as the effector protein that was injected into host cells via the TTSS (Figure 1) [19]. TTSSs, which are used by most pathogenic Gram-negative bacteria, including Yersinia, Salmonella, Shigella, Escherichia coli, and P. aeruginosa, function as molecular syringes, directly delivering toxins into the cytosol of eukaryotic cells [20]. The translocated toxins modulate eukaryotic cell signaling, a process that eventually causes disease [21],[22].

PA103 lacks the exoenzyme S gene (exoS) encoding the 49-kDa form of the toxin but possesses the exoenzyme T gene (exoT), which encodes the 53-kDa form. An isogenic mutant missing exoT was found to be cytotoxic to cultured epithelial cells and caused acute lung injury; therefore, it was concluded that neither ExoT nor ExoS was a major virulence factor for lung injury [18]. PA103 was found to secrete a unique unknown 74-kDa protein, the production of which was decreased when a transposon mutation in exsA was present. The gene encoding this protein was cloned, and a mutant lacking this protein was created in PA103. The isogenic mutant lacking the 74-kDa protein failed to cause acute lung injury in animal models [9]. This protein, regulated by ExsA, a transcriptional activator of P. aeruginosa TTSS, was designated ExoU [9],[23]. Along with other TTS toxins, such as ExoS and ExoT, ExoU is secreted through the TTSS and injected directly into the cytosol of targeted eukaryocytes. Clinical isolates with cytotoxic phenotypes in vitro were found to possess exoU, whereas non-cytotoxic isolates lacked exoU [24]. Additionally, cytotoxic clinical isolates secreting ExoU caused severe and acute epithelial injury in animal models of P. aeruginosa pneumonia (Figure 2) [24]. It was postulated that the ability of P. aeruginosa to cause acute lung epithelial injury and sepsis is strongly linked to TTS secretion of ExoU [10].

P. aeruginosa strain PAO1 was the first strain whose genome was completely sequenced in 2001 by the Pseudomonas Genome Project. A pathogenic gene cluster, the exoenzyme S regulon, encodes genes underlying the regulation, secretion, and translocation of the TTSS. In the exoenzyme S regulon, five operons (exsD–pscL, exsCBA, pscG–popD, popN–pcrR, and pscN–pscU) encode TTSS and translocation machinery. The exsCBA operon encodes the transcriptional activator protein ExsA, which regulates expression of exoenzyme S and co-regulated proteins. The PAO1 strain lacks exoU, whereas approximately 20% of clinical isolates possess exoU (Figure 3) [25]. The exoU gene was initially cloned from the PA103 strain, along with its cognate chaperone gene spcU [9]. The genomic organization of the ExoU-secreting clinical isolate PA14 was analyzed, and two insertional genomic islands, termed pathogenicity islands PAPI-1 and PAPI-2 (Pseudomonas aeruginosa pathogenicity island), were discovered (Figure 3) [26]. The 10.7-kb PAPI-2 region, which is probably derived via horizontal gene transfer, lies within the tRNA-Lys (PA0976.1) region (Figure 3); it encodes 14 open reading frames, including exoU, spcU, four transposases, one integrase, one acetyltransferase, and six hypothetical proteins. The exoU gene itself is 2,074 base pairs and encodes the 682 amino acid protein, ExoU [27] (Figure 3). Four nucleotides at the 3′ end, including the stop codon in exoU, overlap the start codon of the 324-base pair spcU gene, which encodes SpcU (137 amino acids). The promoter region of exoU has a binding motif (TXAAAAXA) for the transcriptional activator, ExsA [28],[29].

The N-terminal of ExoU starts at the secretory leader (MHIQS), the sequence of which is the same as the starter sequence for ExoS and ExoT. When ExoU was identified as a major virulence factor causing acute lung injury in 1997, little was known about its enzymatic mechanisms that were responsible for acute cell death. Analysis of the conserved domain of ExoU revealed a patatin-like domain, containing a glycine-rich nucleotide binding loop motif and a lipase motif with catalytically active serine and aspartate within its N-terminal primary sequence [31]. Patatin, a storage protein in potatoes, exhibits lipase activity and shares a catalytic dyad structure with mammalian phospholipase A₂ (PLA₂) (Figures 3 and 4) [32]-[35]. The catalytic domains of ExoU align with those of patatin, human calcium-independent PLA₂ (iPLA₂) and cytosolic PLA₂ (cPLA₂) [36]. The predicted active sites for ExoU PLA₂ activity are serine 142 (S142) and aspartate 344 (D344). Site-directional mutagenesis of the predicted catalytic residues (ExoUS142A or ExoUD344A) eliminated the cytotoxicity of PA103 [36],[37]. Inhibitors of iPLA₂ and cPLA₂, including bromoenol lactone (BEL), methyl arachidonyl fluorophosphate (MAFP), and arachidonyl trifluoromethyl ketone (AACOCF₃), reduced the cytotoxicity of PA103 in vitro. In the presence of a eukaryotic cell extract, recombinant ExoU displayed PLA₂ and lysophospholipase (lysoPLA) activities (Figure 5); these activities were inhibited by cPLA₂ or iPLA₂ inhibitors [31],[38]. The site-directional PA103 mutants lacking PLA₂ activity were tested by using an animal model of pneumonia. In PA103, either of the ExoUS142A or ExoUD344A mutations abolished virulence associated with acute lung injury and death. It was concluded that acute lung injury from cytotoxic P. aeruginosa is caused by the cytotoxic activity of the patatin-like phospholipase domain of ExoU.

ExoU displays serine acylhydrolase activity via a Ser/Asp catalytic dyad and can be classified as a group IV PLA₂ member. A major characteristic of serine acylhydrolases, such as PLA₂, PLA₁, and lysoPLA, is their ability to perform multiple lipase reactions [40]. Recently, more patatin-like PLA₂ proteins have been detected in various bacterial species [41]. It seems likely that bacteria use PLA₂ as a defense mechanism against predatory eukaryocytes such as phagocytes and environmental amoeba. Its presence allows them to attack a target cell to obtain nutrition, thereby increasing their population [40]. ExoU can kill eukaryotic predators, such as the amoeba Acanthamoeba castellanii [42],[43]. Intracellular expression of ExoU is cytotoxic to yeast, suggesting that fungi could be one of its potential targets [44]. In humans, P. aeruginosa targets phagocytic cells in the lungs and injects them with ExoU [45]-[47]. In an animal model of pneumonia, ExoU is produced during the early phase of infection; delaying exoU expression by as little as 3 hours enhanced bacterial clearance and survival of infected mice [48]. ExoU-mediated impairment of phagocytes probably allows P. aeruginosa to persist within the lungs, causing localized immunosuppression and facilitating superinfection with less pathogenic bacteria. This would explain not only why ExoU-secreting P. aeruginosa is associated with more severe pulmonary infections but also the tendency of hospital-acquired pneumonia to be polymicrobial [47].

Non-cytotoxic P. aeruginosa strains transformed with pUCP19exoUspcU, a plasmid that carries exoU and spcU, became cytotoxic to cultured epithelial cells in vitro and lethal in a mouse model of pneumonia [49]. Isogenic mutants, generated to secrete ExoU, ExoS, or ExoT, were evaluated for their relative contributions to pathogenesis in a mouse model of acute pneumonia [50]. In this study, measurements of mortality, bacterial persistence in the lungs, and dissemination of the bacteria indicated that ExoU secretion had the greatest impact on virulence but that secretion of ExoS had a moderate effect and ExoT a relatively minor effect.

ExoU translocation induces cell death by destroying cell membranes via PLA₂ activity. ExoU might also contribute to the induction of an eicosanoid-mediated inflammatory response in host organisms, as airway epithelial cells exposed to P. aeruginosa overproduce prostaglandin E₂ in an ExoU-dependent manner [51],[52]. A deleterious effect on phospholipid metabolism, in concert with caspase activation, was also reported to occur in an ExoU-dependent manner [53]. Another study reported that arachidonic acid-induced oxidative stress might cause cell damage during the course of an ExoU-producing P. aeruginosa infection. This is because endothelial cell death in cytotoxic PA103 infections was significantly attenuated by alpha-tocopherol [54]. ExoU could also contribute to the pathogenesis of lung injury as it induces a tissue factor-dependent procoagulant activity in airway epithelial cells [55], vascular hyperpermeability, platelet activation, and thrombus formation during P. aeruginosa pneumonia and sepsis [56].

TTS toxins use a unique mechanism for activating their enzymatic activities. These toxins are initially produced in the bacterial cytosol as inactive forms and, immediately after being injected into the cytosol of a target eukaryotic cell by the bacterial secretion apparatus, are activated by specific eukaryotic cell cofactors. As an example, ExoS ADP-ribosyltransferase activity is activated by the eukaryotic protein factor FAS (factor activating exoenzyme S), which is a member of the 14-3-3 protein family [57],[58]. In contrast, P. aeruginosa adenylate cyclase ExoY requires an unknown eukaryotic cell factor for its activation [59]. The PLA₂ activity of ExoU cleaves plasma membrane phospholipids and causes the rapid lysis of targeted eukaryotic cells. Similar to ExoS and ExoY, ExoU requires eukaryotic cell cofactors for its activation, whereas in vitro PLA₂ assays with recombinant ExoU require the addition of eukaryotic cell lysates. The patatin-like PLA₂ domain is located at the N-terminal region of ExoU; the C-terminal region, which includes a sequence corresponding to a conserved DUF885 domain, was reported to be important for the activation process and membrane localization of the protein [60]-[62]. In 2006, Sato and colleagues [63] reported that Cu²⁺, Zn²⁺-superoxide dismutase (SOD1) was a cofactor that activated the PLA₂ activity of ExoU. By this time, however, it had also been reported that ExoU localizes to the plasma membrane, where it undergoes modification in the cell by the addition of two ubiquitin molecules at lysine 178; five C-terminal residues (679 to 683) control membrane localization and ubiquitination [64]. Site-directed spin-labeling electron paramagnetic resonance spectroscopy revealed that the addition of SOD1 induced conformational changes in ExoU [65]. PLA₂ activity of ExoU was demonstrated by using ubiquitinated yeast SOD1 and other ubiquitinated mammalian proteins [66]. Therefore, it seems that ubiquitinated SOD1 works as a ubiquitin donor and that ubiquitination of the ExoU C-terminal activates the PLA₂ activity of ExoU.

The three-dimensional crystallographic structure of ExoU combined with its cognate chaperone SpcU was recently elucidated by two research groups [67],[68] (Figure 6). In one of these studies, the C-terminal membrane-binding domain of ExoU displayed specificity for phosphatidylinositol 4,5-bisphosphate (PI_4,5P₂); ubiquitination of ExoU resulted in its co-localization with endosomal markers [67]. The ubiquitin-binding domain was mapped to a C-terminal four-helix bundle in ExoU [69], with PI_4,5P₂ synergistically enhancing the PLA₂ activity of ExoU via a ubiquitin-related mechanism [70] (Figure 6). The Rickettsia prowazekii RP534 protein, a homologue of ExoU, possesses PLA₂ and lysoPLA activities and PLA₁ activity in the absence of any eukaryotic cofactors [71]. A structural comparison between ExoU and RP534 protein would help clarify the ubiquitin-associated mechanism of ExoU activation. Research into the mechanisms of ExoU activation has provided new insights into how bacteria manipulate eukaryotic cell signaling to facilitate their growth and pathogenesis.

Early studies on P. aeruginosa TTSS revealed an association between a cytotoxic or invasive phenotype and genotype of a strain. The invasive PAO1 strain and the cytotoxic PA103 strain harbor the exoS ⁺ exoT ⁺ exoU ⁻ and exoS ⁻ exoT ⁺ exoU ⁺ genotypes, respectively [9],[18]. This genetic variation in TTS toxin genes implies the presence of similar genotypic and phenotypic variations among clinical and environmental isolates [73]. Consequently, isolates from the respiratory tract or blood cultures of 108 patients were analyzed, and the relative risk of mortality was reported to be sixfold greater when expression of ExoS, ExoT, ExoU, or PcrV occurred (Table 2). The prevalence of the TTS-positive phenotype was significantly higher in acutely infected patients than in chronically infected cystic fibrosis (CF) patients [24]. When Schulert and colleagues [74] analyzed the virulence profiles of 35 P. aeruginosa isolates from patients with hospital-acquired pneumonia by using a cytolytic cell-death assay, an apoptosis assay, and a mouse model of pneumonia, they found that increased virulence was associated with the secretion of ExoU but not ExoS or ExoY secretion. These studies suggest that P. aeruginosa TTSS is present in nearly all clinical and environmental isolates. ExoU secretion could be used as a marker for highly virulent strains and could have some association with poor clinical outcome. It appears that isolates from acutely infected patients are genotypically different from those from chronically infected CF patients [73]. Other researchers have reported the presence of different P. aeruginosa genotypes in isolates from CF patients. The exoS ⁺ exoU ⁻ genotype is associated with chronic infection in CF patients, whereas the exoS ⁻ exoU ⁺ genotype is associated with bacterial strains isolated from blood [75]-[79].

Another important topic in P. aeruginosa biology that has recently emerged is the association of antibiotic resistance with TTSS virulence genotypes (Table 2). Mitov and colleagues [85] analyzed the antimicrobial resistance profiles and genotypes of 202 isolates from CF patients (n = 42) and non-CF in-patients (n = 160). The authors found that the prevalences of exoS and exoU were 62.4 and 30.2%, respectively, and that exoU was more prevalent among MDR than in non-MDR strains (40.2% versus 17.7%). Garey and colleagues [81] reported that 97.5% of bloodstream isolates harbored exoS or exoU genes and that exoS was the most prevalent (70.5%; n = 86). The prevalence of exoU was 25.4% (n = 31), and these isolates were significantly more likely to be resistant to multiple antibiotics, including cephems, carbapenems, fluoroquinolones, and gentamicin. Consistent with this, an analysis of 45 clinical isolates found that exoU ⁺ isolates were more likely to be fluoroquinolone-resistant than exoS ⁺ isolates (92% versus 61%, P <0.05). These isolates possessed a mutation in the gyrA gene and exhibited an efflux pump overexpression phenotype [12]. Agnello and Wong-Beringer [82] examined the relationship between the TTSS effector genotype and fluoroquinolone resistance mechanisms in 270 respiratory isolates and found that a higher proportion of exoU ⁺ strains was fluoroquinolone-resistant compared with exoS ⁺ strains (63% versus 49%) despite their lower prevalence (38% exoU ⁺ versus 56% exoS ⁺ ) [82]. Of epidemiological importance, Tran and colleagues [86] showed that 20 isolates (eight unique pulsed-field gel electrophoresis clusters) recovered from imported frozen raw shrimp sold in the US harbored TTS toxin genes and were resistant to quinolone with mutations in gyrA. These findings indicate co-evolution of resistance and virulence traits favoring a more virulent genotype in a quinolone-rich clinical environment [80].

There have been several studies in which associations between TTSS-associated virulence and poor clinical outcome for P. aeruginosa-infected patients have been observed. An analysis of TTS genotypes and phenotypes of isolates cultured from 35 mechanically ventilated patients with bronchoscopically confirmed P. aeruginosa-VAP showed a correlation between TTS phenotype, especially the ExoU phenotype, and severity of pneumonia [80]. More recently, El-Solh and colleagues [83] performed a retrospective analysis of 85 cases of P. aeruginosa bacteremia. Bacteremic patients with TTSS-positive isolates developed septic shock with high probability of death more frequently than patients with TTSS-negative isolates. The authors found that none of the TTSS-positive patients who survived the first 30 days of infection had a P. aeruginosa isolate that exhibited the ExoU phenotype; a higher frequency of antibiotic resistance was detected in TTSS-positive isolates. Jabalameli and colleagues [84] analyzed TTSS genotypes and antimicrobial resistance in 96 isolates collected from wound infections of burn patients. More than 90% of the isolates were MDR, and 64.5% of them carried exoU whereas 29% carried exoS. Their findings suggest that these genes, particularly exoU, are commonly disseminated among P. aeruginosa strains isolated from burn patients. Sullivan and colleagues [11] recently reported their analysis of antimicrobial resistance and TTSS virulence in P. aeruginosa isolates from hospitalized adult patients with respiratory syndromes. The authors studied 218 consecutive adult patients whose respiratory cultures were positive for P. aeruginosa, and reported that fluoroquinolone-resistant and MDR strains were more likely to cause pneumonia than bronchitis or colonization. The combination of fluoroquinolone resistance and the gene encoding the TTSS ExoU effector in P. aeruginosa was the strongest predictor of pneumonia development. Further investigations suggest that the fluoroquinolone-resistant phenotype and the exoU ⁺ genotype of P. aeruginosa might cause poor clinical outcomes in patients with P. aeruginosa pneumonia [87]. Although there is no clear genetic explanation and a less than convincing association between ExoU-associated virulence and antibiotic resistance, there is no doubt that bacterial strains possessing both virulent and MDR characteristics are more dangerous, especially for immunocompromised patients. Therefore, improved genotyping or phenotyping methods (or both) for analyzing TTS toxins of clinical isolates will enhance our understanding of this area.

Several prophylactic or therapeutic experimental strategies against the cytotoxic effects of TTS ExoU have been reported over the last decade. The P. aeruginosa V-antigen PcrV, a homolog of the Yersinia V-antigen LcrV, contributes to TTS toxin translocation [88]. In prophylactic strategies, active immunization against PcrV ensures the survival of challenged mice and decreases lung inflammation and injury [89]. DNA vaccination with pIRES-toxAm-pcrV has been proposed as a potential immunotherapy [90]. In passive immunization, the rabbit polyclonal anti-PcrV antibody and murine monoclonal anti-PcrV antibody mAb166 inhibit TTS toxin translocation [91]-[95]. For clinical use, the mAb166 was humanized [96], and the IgG antigen-binding (Fab′) fragment, KB001, is currently in use in phase II clinical trials for treating VAP in France and chronic pneumonia in CF patients in the US [97],[98].

In vitro experiments have shown that specific inhibitors against iPLA2, such as BEL, AACOCF3, and MAFP, decrease the cytotoxicity of ExoU. Several researchers have reported that small molecules, such as pseudolipasin A and arylsulfonamides, specifically inhibit the phospholipase activity of ExoU [99],[100]. More details regarding the activation mechanisms of ExoU have been recently reported; however, there is more potential in using small chemicals for the prevention of acute lung injury induced by P. aeruginosa.

P. aeruginosa ExoU, a toxin injected into the cytosol of target eukaryotic cells such as phagocytes and epithelial cells, is a major virulence factor in the cause of alveolar lung injury in patients with P. aeruginosa pneumonia. Virulent strains of P. aeruginosa possess the PAPI-2 pathogenic gene cluster region, which includes exoU. The PLA₂ activity exhibited by ExoU requires a ubiquitination-associated activation mechanism to operate in a eukaryotic cell factor-dependent manner. A combination of the exoU ⁺ genotype and fluoroquinolone-resistant phenotype in isolates was shown to correlate with poor clinical outcome. Cytotoxic and antimicrobial-resistant P. aeruginosa is a serious concern, especially for immunocompromised patients. Therefore, rapid diagnostic determination of isolate genotype and phenotype is important. Surveillance to determine the prevalence of cytotoxic and antibiotic-resistant isolates is needed if we are to reduce the risk of lethal P. aeruginosa outbreaks. Opportunities exist for improving the clinical outcome of patients infected with P. aeruginosa by identifying virulent and antimicrobial-resistant isolates that cause acute lung injury, sepsis, and mortality. Exploration of P. aeruginosa virulence apparatuses as potential antimicrobial targets is vital if we are to avoid the spread of dangerous super-resistant P. aeruginosa strains.

TS is a Professor of Anesthesiology. MS is a student in the Graduate School of Medicine at the Kyoto Prefectural University of Medicine in Japan. KM is an Associate Professor in the Anesthesia Department at Korin University in Japan. JWK is the Anesthetist-in-Chief (Henry Isaiah Dorr Professor of Anesthesia) at Harvard Medical School (Boston, MA, USA).