ABSTRACT
ExoU is a potent phospholipase A2 effector protein secreted by the type III secretion system of Pseudomonas aeruginosa. By cleaving plasma membrane phospholipids, it causes rapid lysis of eukaryotic cells. However, ExoU does not exhibit activity on its own but instead requires eukaryotic cell cofactors for activation. Ubiquitin and ubiquitinated proteins have been shown to activate ExoU, but previous work suggested that other cofactors are also involved. In this study, we demonstrate that phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2] is another important coactivator of ExoU. PI(4,5)P2 works synergistically with ubiquitin to greatly enhance the phospholipase A2 activity of ExoU. Distinct residues of ExoU were critical for activation by PI(4,5)P2 and by ubiquitin, indicating that these factors activate ExoU by discrete mechanisms. In support of the biological relevance of PI(4,5)P2 coactivation, a yeast mutant with reduced PI(4,5)P2 levels was less susceptible to the cytotoxic activity of ExoU. Together, these findings further elaborate the molecular mechanism of ExoU.
INTRODUCTION
Pseudomonas aeruginosa is a Gram-negative bacterial pathogen that causes infection in immunocompromised hosts, such as mechanically ventilated patients, burn victims, and people with cystic fibrosis (1). The type III secretion system (T3SS) of P. aeruginosa, which injects toxic effector proteins directly into host cells, is particularly associated with disease in both humans and animal models of infection (2, 3). This secretion system transports four known effectors (ExoS, ExoT, ExoY, and ExoU), although individual strains of P. aeruginosa typically secrete only a subset of these proteins (4). ExoS and ExoT are bifunctional toxins with both ADP-ribosyltransferase and GTPase-activating protein domains (5). Once injected into host cells, each of these proteins is activated by a member of the 14-3-3 family of proteins (6). ExoY is an adenylate cyclase that causes cell rounding and also requires a cofactor for its activity, but the identity of this cofactor is currently unknown (7).
The final effector, ExoU, is a phospholipase A2 (PLA2) enzyme that is rapidly cytotoxic to eukaryotic cells (8, 9). Despite being secreted by only approximately one-third of clinical isolates, ExoU is the effector most associated with poor clinical outcomes in human patients and increased mortality in animal models (3, 10, 11). The recently solved structure of ExoU in complex with its cognate bacterial chaperone SpcU indicates that ExoU is a large protein with multiple independent domains (12, 13). The N-terminal 101 residues of ExoU comprise the secretion and SpcU-binding domains, although SpcU binds to other portions of ExoU as well (13). Residues 102 to 471 encompass the catalytic PLA2 domain of ExoU, which is responsible for cleavage of a wide variety of phospholipids following injection of ExoU into host cells (13, 14). Like the other P. aeruginosa T3SS effectors, ExoU requires a host cell cofactor to manifest this catalytic activity (15). Finally, functional and structural studies indicate that residues 503 to 687 of ExoU contain the membrane localization domain (MLD) of ExoU (13, 16). This C-terminal domain is necessary and sufficient for plasma membrane targeting in eukaryotic cells (16). The MLD is also important for the activity of ExoU, as both 5-amino-acid insertions and single-residue substitutions in this region attenuate PLA2 activity (17–19). The mechanism by which the MLD contributes to the catalytic activity of ExoU is unclear, although it has been postulated that it may bind cofactors (18, 19).
As mentioned, ExoU is inactive by itself and exhibits PLA2 activity only in the presence of eukaryotic cell lysate. Initially, the eukaryotic cofactor responsible for activating ExoU was thought to be superoxide dismutase 1 (SOD1) (20). However, it has recently been shown that ubiquitin and ubiquitinated proteins are the actual cofactors for ExoU. Earlier findings suggesting that SOD1 activated ExoU were the result of the presence of ubiquitin in some commercial preparations of SOD1 (21). However, ubiquitin may not be the only host cell activator of ExoU. We previously reported that ExoU containing specific amino acid substitutions demonstrated substantial activity in the presence of HeLa cell lysate but not SOD1 (19). Since it is now clear that the SOD1 preparations only activated ExoU because of the presence of contaminating ubiquitin, these results suggest that HeLa cell lysates contain a second factor (other than ubiquitin) capable of activating ExoU.
In the present study, we identify this second factor as the phospholipid phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2]. Although PI(4,5)P2 by itself minimally activated ExoU, in the presence of ubiquitin, it caused a synergistic increase in the PLA2 activity of ExoU. Additionally, PI(4,5)P2 activation was relevant in eukaryotic cells, as a yeast mutant with reduced PI(4,5)P2 levels was less susceptible to ExoU-mediated killing. Since PI(4,5)P2 is localized to the inner leaflet of the plasma membrane, where ExoU-mediated cleavage of host cell phospholipids is thought to occur, it is well positioned to serve as a coactivator of this toxin. Although several eukaryotic enzymes are regulated by PI(4,5)P2, to our knowledge this is the first demonstration of a bacterial protein regulated by this phospholipid.
MATERIALS AND METHODS
Cell lines, bacterial and yeast strains, and media.Escherichia coli strain BL21(DE3) Star was used for expression and purification of ExoU and was grown in Luria-Bertani broth (LB) (see Table S1 in the supplemental material). When appropriate, media were supplemented with 100 μg/ml ampicillin. Saccharomyces cerevisiae strains SEY6210 (wild type) and AAY202 (mss4ts) were grown in yeast extract peptone dextrose (YPD) medium (see Table S1). Yeast strains expressing pYC vectors were grown in synthetic complete medium lacking uracil, with either glucose or raffinose supplementation (SC−Ura+Glu or SC−Ura+Raf, respectively) (22). HeLa cervical carcinoma cells (ATCC, Manassas, VA) were grown in Eagle's minimal essential medium (ATCC) supplemented with 10% fetal bovine serum (Gemini Bio-Products, West Sacramento, CA).
Purification of recombinant ExoU.HN-tagged recombinant ExoU proteins were purified as previously described (19). Briefly, 1 liter of LB supplemented with 100 μg/ml ampicillin was inoculated with 10 ml of an overnight culture of strain BL21 containing plasmids expressing exoU (pExoU-HNC, pI609N-HNC, and pR661L-HNC) (see Table S1 in the supplemental material). Cells were grown to an optical density at 600 nm (OD600) of approximately 0.6, induced with 1 mM isopropyl-β-d-thiogalactopyranoside (Sigma-Aldrich, St. Louis, MO), and grown overnight at 25°C with shaking at 250 rpm. Cells were pelleted by centrifugation at 6,000 × g, resuspended in lysis buffer (10 mM Tris [pH 8.3], 500 mM NaCl, 10% glycerol, 0.01% Igepal CA-630, and 5 mM β-mercaptoethanol), and sonicated. The lysate was cleared by centrifugation at 18,000 × g and loaded onto an Akta purification system fitted with a HisTrap FF nickel column and a HiPrep 26/10 desalting column (GE Healthcare, Piscataway, NJ). Protein was eluted in 10 mM Tris (pH 8.3), 500 mM NaCl, and 5 mM β-mercaptoethanol. ExoU-containing fractions were concentrated using Vivaspin concentrators (GE Healthcare). Protein purity was assessed by SDS-polyacrylamide gel electrophoresis with Coomassie staining.
PLA2 assays.The activity of purified ExoU was assessed using a cPLA2 assay kit (Cayman Chemical, Ann Arbor, MI). To perform the assay, 5 μg of purified ExoU protein (65 pmol) was added to 200 μl of 1.5 mM arachidonoyl thiophosphatidylcholine (the assay substrate). The mixture was supplemented with HeLa cell fractions, ubiquitin, and/or purified lipids at room temperature. HeLa cell lysates were generated by washing 10-cm plates of HeLa cells with cold phosphate-buffered saline (PBS). Cells were collected into 1 ml cold PBS using a cell scraper and lysed by repeated passage through a 27-gauge needle. Following centrifugation at 18,000 × g at 4°C for 15 min, the supernatant was collected and used as the cytosolic fraction. The pellet was resuspended in PBS supplemented with 1% Triton X-100 and passed repeatedly through a 27-gauge needle. Following centrifugation at 18,000 × g at 4°C for 15 min, the supernatant was collected and used as the membrane fraction. Primary mouse cell lysate was similarly prepared, with mouse lungs from C57BL/6 mice homogenized in PBS followed by repeated passage through a 27-gauge needle. Following centrifugation at 18,000 × g at 4°C for 15 min, the supernatant was collected and used as the cytosolic fraction. The pellet was resuspended in PBS supplemented with 1% Triton X-100 and passed repeatedly through a 27-gauge needle. Following centrifugation at 18,000 × g at 4°C for 15 min, the supernatant was collected and used as the membrane fraction. Protease-treated lysate was prepared by treating fractions with 400 μg/ml proteinase K (Invitrogen, Grand Island, NY) at 55°C followed by addition of 2 mM phenylmethylsulfonyl fluoride (PMSF) (AppliChem, Darmstadt, Germany). For all samples with cell lysate, 5 μl was used for both the membrane and cytosolic fractions. When indicated, protein concentrations were normalized by absorbance at 280 nm.
When indicated, reactions were supplemented with purified monoubiquitin from bovine erythrocytes (Sigma-Aldrich) (65 pmol, or 230 pmol for assays with protease-treated lysate) or polyubiquitin (BostonBiochem, Cambridge, MA) (4 pmol). Purified lipids (Avanti Polar Lipids, Alabaster, AL) were prepared from liquid chloroform stocks that were dried under nitrogen gas followed by vacuum. Samples were resuspended in 20 mM HEPES (pH 7.5) and 100 mM KCl and sonicated. A total of 65 pmol of each lipid was added to the reaction for a final concentration of 283 nM unless otherwise indicated. For all reactions, activities were quantified by measuring the A405 at various time points after addition of 10 μl of 25 mM 5,5′-dithiobis(2-dinitrobenzoic acid) (DTNB). The PLA2 activity of ExoU was calculated using the following formula: A405/10.00 × 1/(nmol of ExoU), where 10.00 is the extinction coefficient for DTNB.
Immunoblots.Equal amounts of protein from the membrane and cytosolic fractions, as well as proteinase K-treated fractions, were run on 4 to 20% gradient gels (Bio-Rad, Hercules, CA) at 100 V for 1 h. Proteins were transferred to nitrocellulose membranes (Bio-Rad) using a wet-transfer apparatus at 100 V for 1 h. Membranes were incubated in blocking buffer (5% milk in PBS) for 2 h at room temperature with gentle shaking. The membranes were incubated overnight at 4°C with gentle shaking in the presence of anti-ubiquitin antibody (Santa Cruz Biotechnology, Santa Cruz, CA) diluted 1:1,000 in blocking buffer with 0.1% Tween 20. Membranes were washed and incubated with goat anti-mouse secondary antibody IR dye 680 (Li-Cor Biosciences, Lincoln, NE) diluted 1:10,000 in blocking buffer supplemented with 0.1% Tween 20 for 1 h at room temperature with gentle shaking. Blots were imaged using the Li-Cor Odyssey system, and bands were quantified by densitometry using Image J.
Yeast cloning.The galactose-inducible low-copy-number yeast vector pYC2/NT A (Invitrogen) was used for expression of the exoU gene in S. cerevisiae (22). The pYC and pYC-ExoU constructs have been described previously (22) (see Table S1 in the supplemental material). Constructs expressing ExoU-I609N and ExoU-R661L (designated pYC-I609N and pYC-R661L, respectively; see Table S1) were generated from the pYC-ExoU construct using site-directed mutagenesis with primers shown in Table S2 in the supplemental material as previously described (19). These vectors were transformed into wild-type (SEY6210) and mss4ts (AAY202) yeast as previously described (23).
Yeast viability assays.Wild-type (SEY6210) and mss4ts (AAY202) yeast expressing pYC constructs were grown in SC−Ura+Glu medium overnight at 30°C with shaking at 300 rpm. Yeast from overnight cultures were centrifuged at 18,000 × g, resuspended in water, and used to inoculate SC−Ura+Glu and SC−Ura+Raf cultures to an initial OD600 of 0.5. Cultures were grown at 30°C for 24 h with shaking at 300 rpm. After 24 h, yeasts were pelleted by centrifugation at 18,000 × g, resuspended in water, and plated onto SC−Ura+Glu agar. Colonies were enumerated after growth for 3 days at 30°C.
Statistical methods.Student's t test was used to compare data sets for PLA2 assays and yeast viability experiments. All assays were done in triplicate. Significance was defined as a P value of <0.05.
RESULTS
A membrane-associated cofactor activates ExoU.Recent studies indicated that ubiquitin and ubiquitinated proteins activate ExoU, and we sought to determine whether additional cofactors of ExoU existed (21). Prior to the identification of ubiquitin as a cofactor, a report indicated that the membrane-rich particulate fraction of HeLa cells activated ExoU substantially more than the soluble fraction (24). We used an in vitro PLA2 assay to verify this finding. Indeed, the membrane fraction of HeLa cells activated ExoU substantially more than the cytosolic fraction (Fig. 1A). Accordingly, we performed immunoblotting to determine the approximate relative levels of ubiquitin and ubiquitinated proteins in these fractions. Surprisingly, about equal amounts of ubiquitin and ubiquitinated proteins were present in the two fractions (Fig. 1B). Densitometry of the immunoblots confirmed this impression, indicating that 53% of the ubiquitinated proteins were in the membrane fraction, with the remainder in the cytosol (data not shown). That the membrane fraction activated ExoU better than the cytosolic fraction was further validated using primary mouse cell lysate from homogenized mouse lungs (data not shown). Together, these results suggest that a factor in the membrane fraction in addition to ubiquitin contributed to the activation of ExoU.
A protease-resistant component of HeLa cell membrane fractions coactivates ExoU. The membrane (MB) and cytosolic (Cyt) fractions of HeLa cell lysate were obtained and analyzed for activation of ExoU and the presence of ubiquitin. (A) Equal amounts of protein from HeLa membrane and cytosolic fractions were added to the in vitro PLA2 assay, and the PLA2 activity of ExoU was assessed at 24 h as described in Materials and Methods. Experiments were performed in triplicate; values are means, and error bars represent standard errors of the means (SEM). Supplementation with HeLa cell membrane fractions yielded statistically different levels of substrate hydrolysis than did supplementation with HeLa cell cytosolic fractions, and both values were statistically different from those observed with ExoU alone (P < 0.05). (B) Immunoblot analysis to detect ubiquitin in HeLa cell membrane and cytosolic fractions normalized for protein content. Ubiquitin was detected using a ubiquitin-specific antibody. “Prot-K” indicates treatment of the HeLa cell fraction with proteinase K. (C) ExoU PLA2 activity was assessed at 4 h following supplementation with ubiquitin, HeLa cell membrane fraction, or HeLa cell cytosolic fraction. In these experiments, a small amount of ubiquitin (230 pmol) was used that did not by itself result in appreciable activation of ExoU. “Prot-K” indicates treatment of the HeLa cell fraction with proteinase K. Experiments were performed in triplicate; values are means, and error bars represent SEM.
To determine whether the membrane-associated factor was protease-sensitive, we treated the membrane fraction of HeLa cells with proteinase K and assessed its ability to activate ExoU. Protease treatment degraded the ubiquitin in the membrane fraction (Fig. 1B) and substantially diminished activation of ExoU (Fig. 1C), indicating that ubiquitin was necessary for membrane fraction activation of ExoU (14). We next tested whether protease-treated membrane fraction could activate ExoU in the presence of small amounts of exogenous ubiquitin (i.e., concentrations of ubiquitin below those needed for detectable activation). Protease-treated membrane fraction dramatically increased ExoU activation in the presence of ubiquitin relative to ubiquitin alone (Fig. 1C). This same effect was not observed with the cytosolic fraction (Fig. 1C). We interpret these findings to indicate that a second cofactor exists in the membrane fraction, that it is protease resistant, and that it is capable of synergistically activating ExoU in the presence of ubiquitin.
PI(4,5)P2 acts as a potent coactivator of ExoU.Since our results indicated that the second cofactor was protease resistant and membrane associated, we examined whether lipid species were responsible for this activity. To detect even small amounts of activity, PLA2 assays were incubated up to 72 h. The phospholipid PI(4,5)P2 caused a detectable activation of ExoU but substantially less than did ubiquitin (Fig. 2). In fact, monoubiquitin caused 15-fold more activation of ExoU than did PI(4,5)P2 (Fig. 2). PI(4)P also slightly activated ExoU but even less than did PI(4,5)P2. That lipids only minimally activated ExoU was consistent with our earlier result (Fig. 1C) showing that protease-treated lysate, which still contains intact lipids, did not appreciably activate ExoU (Fig. 1C).
PI(4,5)P2 by itself is a poor activator of ExoU. Various purified lipids (65 pmol each) were added to ExoU in the PLA2 assay, and their activation of ExoU was compared to that of 65 pmol ubiquitin. Experiments were performed in triplicate; values are means, and error bars represent SEM. *, P < 0.05 at 72 h compared to ExoU only. PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PA, phosphatidic acid; PG, phosphatidylglycerol; PI, phosphatidylinositol; IP3, inositol triphosphate; DAG, diacylglycerol; PI(3)P, phosphatidylinositol 3-phosphate (and similar abbreviations for other phosphoinositides). The PI(4,5)P2-only sample did not contain ExoU.
Since protease-treated membrane fractions had little activity on their own but had substantial activity in combination with ubiquitin (Fig. 1C), we next examined whether combinations of lipids and ubiquitin activated ExoU more potently. We added purified lipids in combination with a small amount of ubiquitin insufficient to appreciably activate ExoU on its own and measured PLA2 activity using the in vitro assay. Although most lipids demonstrated little or no activity, PI(4,5)P2 caused a dramatic increase in activation of ExoU (Fig. 3). In fact, the addition of PI(4,5)P2 saturated the assay's ability to detect substrate cleavage within 6 h, whereas ExoU was not significantly activated by ubiquitin alone at this time point. No other lipid coactivated ExoU to the same extent as PI(4,5)P2, but ExoU was weakly activated by the addition of PI(4)P (Fig. 3). It had previously been shown that polyubiquitin activates ExoU to a greater extent than monoubiquitin (21), so we examined whether PI(4,5)P2 also increased ExoU activation in combination with polyubiquitin. As was observed with monoubiquitin, PI(4,5)P2 coactivated ExoU in the presence of polyubiquitin (see Fig. S1 in the supplemental material). Together, these data demonstrate that PI(4,5)P2 acts with ubiquitin to activate ExoU in vitro and that this activation is synergistic, since neither is a potent activator of ExoU on its own.
PI(4,5)P2 acts with ubiquitin to activate ExoU. ExoU, ubiquitin (Ub), and purified lipids were added in different combinations to the PLA2 assay and PLA2 activity was measured. For each factor, a concentration of 65 pmol was added. Experiments were performed in triplicate; values are means, and error bars represent SEM. *, P < 0.01 compared to ExoU plus ubiquitin.
PI(4,5)P2 is an essential lipid in eukaryotic cells and functions in many different cellular processes (25, 26). It is found predominantly at the inner leaflet of the plasma membrane, where it regulates motility, phagocytosis, exocytosis, and cell signaling (27–29). The localization of PI(4,5)P2 to the membranes of eukaryotic cells makes it a likely candidate for the protease-resistant membrane factor described in Fig. 1 (30). It is important to note that neither inositol triphosphate (IP3) nor diacylglycerol (DAG), the biologically active products of PI(4,5)P2 cleavage by phospholipase C, significantly activated ExoU (Fig. 2 and 3). Thus, ExoU coactivation requires intact PI(4,5)P2.
PI(4,5)P2 is a minor phospholipid, comprising only 1% of the total phospholipid content of mammalian cells (26, 30). It has an average cellular concentration of about 10 μM (25). We therefore tested whether physiologically relevant concentrations of PI(4,5)P2 were capable of coactivating ExoU. In the PLA2 assay, we combined a 1:1 molar ratio of ExoU and ubiquitin with different amounts of PI(4,5)P2 and measured ExoU activation over time. A final PI(4,5)P2 concentration of only 2.83 nM was sufficient to increase the activation of ExoU 13-fold over that observed with ubiquitin alone (Fig. 4). As expected, activation of ExoU increased with increasing amounts of PI(4,5)P2. Activation was saturated by 2.83 μM PI(4,5)P2, below physiological concentrations, suggesting that sufficient amounts of PI(4,5)P2 are present in vivo to activate ExoU.
Low concentrations of PI(4,5)P2 are sufficient to coactivate ExoU. PLA2 activity was assessed with ExoU (65 pmol) and ubiquitin (65 pmol) in the presence of various concentrations of PI(4,5)P2. Experiments were performed in triplicate; values are means, and error bars represent SEM. At 24 h, P < 0.05 for each sample compared to ExoU, except the value for ExoU + 28.3 μM PI(4,5)P2. At 2 h and later time points, the result for the sample with 2.83 nM PI(4,5)P2 and ubiquitin is significantly different (P < 0.05) compared to ExoU and ubiquitin only.
We also determined whether the paucity of PI(4,5)P2 explained the relative inability of the cytosolic fraction of HeLa cell lysates to activate ExoU. Addition of 283 nM exogenous PI(4,5)P2 to the cytosolic fraction of HeLa cell lysate increased the activation of ExoU to 3.7-fold relative to the activation by the cytosolic fraction alone (see Fig. S2 in the supplemental material). In contrast, the addition of a similar amount of exogenous PI(4,5)P2 to the membrane fraction caused only 1.2 times the activation of ExoU. These results are consistent with the supposition that endogenous PI(4,5)P2 in the membrane fraction accounts for its ability to synergistically activate ExoU.
Distinct ExoU residues are critical for coactivation by PI(4,5)P2 and activation by ubiquitin.An ExoU variant containing a single amino acid substitution at residue 609 (I609N) had been described previously as having attenuated but still significant residual activation by HeLa cell lysate (19). Interestingly, this I609N variant (designated ExoU-I609N) was not activated by SOD1 (19). Since previously observed SOD1 activation was likely due to contaminating ubiquitin, ExoU-I609N may be nearly “blind” to ubiquitin. We therefore used the in vitro PLA2 assay to examine whether the residual activity of ExoU-I609N was due to activation by PI(4,5)P2. As anticipated, ExoU-I609N was not significantly activated by ubiquitin alone but was activated by a combination of PI(4,5)P2 and ubiquitin (Fig. 5A). The requirement for ubiquitin to activate ExoU-I609N in combination with PI(4,5)P2 suggested that this ExoU variant still has a weak interaction with ubiquitin. To test this, we incubated ExoU-I609N with 45 times the amount of ubiquitin previously used. This large amount of ubiquitin allowed us to observe activation of ExoU-I609N, although the activation was relatively minor compared to that of wild-type ExoU (Fig. 5B). Together, these findings indicate that ExoU-I609N is activated by ubiquitin to a small degree but that the presence of PI(4,5)P2 considerably enhances this activation.
ExoU variants differ in their activation by ubiquitin and PI(4,5)P2. ExoU, ExoU-I609N, and ExoU-R661L were assessed for activation by ubiquitin (Ub) and PI(4,5)P2 in the in vitro PLA2 assay. (A) ExoU, ubiquitin, and/or PI(4,5)P2 were coincubated in equivalent molar ratios (65 pmol each), and PLA2 activity was assessed at 24 h. (B) Ubiquitin was added at 45-fold molar excess of ExoU (2.9 nmol), and the PLA2 activity was assessed at the indicated times. Experiments were performed in triplicate; values are means, and error bars represent SEM.
A second ExoU variant (ExoU-R661L, which contains a single amino acid substitution at residue 661) also exhibited an overall decrease in activation by HeLa cell lysates (19). However, in contrast to ExoU-I609N, ExoU-R661L was activated to a degree similar to that of wild-type ExoU when supplemented with SOD1 (and presumably ubiquitin) (19). Thus, the decreased activity of ExoU-R661L may be due to its inability to recognize PI(4,5)P2 in HeLa cell lysates. We examined this mutant in the in vitro PLA2 assay. Consistent with this hypothesis, ExoU-R661L was activated by ubiquitin similarly to wild-type ExoU, but its activity was not substantially enhanced by the addition of PI(4,5)P2 (Fig. 5A). Thus, ExoU-R661L is relatively blind to PI(4,5)P2 but not ubiquitin.
As shown in Fig. 1C, protease treatment of HeLa cell lysates did not destroy their ability to coactivate ExoU in the presence of exogenous ubiquitin. We used ExoU-R661L and ExoU-I609N as tools to investigate whether PI(4,5)P2 was indeed the component of protease-treated HeLa cell lysates responsible for residual coactivation of ExoU. We reasoned that if PI(4,5)P2 were indeed the protease-resistant coactivating factor in HeLa cell lysates, then these treated lysates should behave similarly to purified PI(4,5)P2 in PLA2 assays with these ExoU variants. In the presence of ubiquitin, ExoU-I609N was activated synergistically by the addition of protease-treated HeLa membrane fraction (Fig. 6). In contrast, the addition of protease-treated lysate did not increase the activation of ExoU-R661L beyond that observed with ubiquitin alone (Fig. 6). Since ExoU-R661L is PI(4,5)P2 blind, these observations are consistent with the protease-resistant factor in membrane fractions being PI(4,5)P2.
Coactivation of ExoU variants by protease-treated lysate. ExoU-I609N and ExoU-R661L were assessed for PLA2 activity at 24 h with the protease-treated membrane fraction [MB (Prot-K)] of HeLa cells, with and without 230 pmol ubiquitin (Ub). Experiments were performed in triplicate; values are means, and error bars represent SEM. NS, not significant.
PI(4,5)P2 coactivation of ExoU is biologically relevant.Due to the potent in vitro activation of ExoU by PI(4,5)P2 in combination with ubiquitin, we investigated the biological relevance of this effect in eukaryotic cells. The budding yeast Saccharomyces cerevisiae is a useful model for studying ExoU, as ExoU kills yeast similarly to mammalian cells (14, 22). Additionally, both ubiquitin and PI(4,5)P2 are found in all eukaryotes, including yeast (21, 31). The yeast enzyme Mss4 is a kinase that forms PI(4,5)P2 from PI(4)P. Since PI(4,5)P2 is essential for viability, we acquired a temperature-sensitive mss4 mutant yeast strain that had only about half the normal levels of PI(4,5)P2 even at the permissive temperature of 26°C (32). Accordingly, we examined whether mss4ts yeasts were less susceptible to ExoU killing by transforming wild-type and mss4ts yeasts with constructs expressing exoU under the control of a galactose-inducible and glucose-repressible promoter. The plasmid was selected for by growth in a minimal medium lacking uracil (SC−Ura medium) (22). Growth in galactose-supplemented medium resulted in the rapid death of both wild-type and mss4ts yeasts, preventing assessment of the importance of PI(4,5)P2 in killing (data not shown). Since even the small amount of ExoU produced in the absence of induction (growth in raffinose-containing medium) is sufficient to compromise yeast viability (22), we next grew yeast in medium supplemented with raffinose. As a negative control, yeasts were also grown in glucose-containing medium, which represses the exoU gene. Following growth in medium for 24 h, yeasts were plated onto glucose-containing agar for enumeration. Most likely due to a paucity of PI(4,5)P2, mss4ts yeast containing empty vector grew more slowly than wild-type yeast containing empty vector in both glucose-supplemented and raffinose-supplemented media (see Fig. S3 in the supplemental material). We therefore normalized all CFU counts to growth of the corresponding yeast strain containing empty vector in raffinose-supplemented medium. With exoU expression, a larger proportional drop in yeast viability was observed in wild-type yeast than in mss4ts yeast, suggesting that the decreased amounts of PI(4,5)P2 in mss4ts yeast renders them less susceptible to ExoU-mediated killing (Fig. 7). This effect was also seen with ExoU-I609N, which is consistent with this mutant still being synergistically coactivated by PI(4,5)P2 (Fig. 7). In contrast, wild-type and mss4ts yeasts were equally viable when producing ExoU-R661L, indicating that the cytotoxicity of this PI(4,5)P2-blind variant was not affected by the reduced PI(4,5)P2 levels present in mss4ts yeast. Furthermore, even wild-type yeasts were fully viable with production of ExoU-R661L in raffinose (Fig. 7). Since this mutant is activated by ubiquitin but not PI(4,5)P2 (Fig. 5), these data suggest that ubiquitin on its own is not sufficient to efficiently kill yeast under these conditions. In other words, coactivation by PI(4,5)P2 was necessary for biologically detectable levels of ExoU-mediated killing. It is important to note, however, that growth of R661L-expressing yeast in galactose did result in killing of the yeast (data not shown), so our results may be relevant only under conditions in which small amounts of ExoU are present. Overall, these results are consistent with a biologically significant contribution of PI(4,5)P2 to the killing of yeast by ExoU.
mss4ts yeasts are less susceptible to ExoU-mediated killing. Wild-type (WT) and mss4ts yeasts were grown for 24 h in medium containing raffinose and plated on SC−Ura+Glu agar for enumeration. We normalized all CFU to growth of the corresponding yeast strain containing empty vector in raffinose-supplemented medium. Experiments were performed in triplicate; values are means, and error bars represent SEM. *, P < 0.05.
DISCUSSION
Recently, ubiquitin and ubiquitinated proteins were implicated as cofactors necessary for the PLA2 activity of ExoU (21). However, preliminary findings indicated the possible presence of at least one additional cofactor for ExoU. Indeed, we found that the phospholipid PI(4,5)P2 acted with ubiquitin to synergistically increase the activity of ExoU. Importantly, even PI(4,5)P2 concentrations below physiological levels were capable of strongly coactivating ExoU in the presence of ubiquitin. In addition, yeasts with reduced PI(4,5)P2 levels were less susceptible to ExoU-mediated killing. Together these data indicate that PI(4,5)P2 acts as a cofactor for ExoU both in vitro and in eukaryotic cells.
It had previously been observed that the activator of ExoU in eukaryotic cell lysate was susceptible to protease treatment, so it was thought that this cofactor was a protein (14). We too found that protease treatment of the membrane fraction of HeLa cells drastically reduced activation of ExoU. These same protease-treated fractions, however, retained the ability to substantially enhance ExoU activity in the presence of exogenous ubiquitin. Our results offer an explanation for these observations: PI(4,5)P2 in the membrane fraction survives protease treatment and serves as a coactivator of ExoU but does not appreciably activate ExoU on its own, whereas ubiquitin in the membrane fraction is degraded. Stirling and colleagues previously reported that the membrane fraction of HeLa cell lysates more potently activated ExoU than did the cytosolic fraction (24). However, lipids were not identified as the factor responsible for this activation, perhaps because PI(4,5)P2 on its own is a poor activator of ExoU.
PI(4,5)P2 is derived from phosphatidylinositol, a lipid that can be modified with phosphates attached to the 3, 4, or 5 position of the inositol ring to form different phosphoinositides. Like ubiquitin, PI(4,5)P2 is conserved throughout eukaryotes and is absent from prokaryotes (31), ensuring that ExoU remains inactive within P. aeruginosa and manifests catalytic activity only upon injection into target cells. The wide distribution of PI(4,5)P2 among eukaryotic organisms is also consistent with the broad host range of P. aeruginosa and of the many eukaryotic cell types known to be susceptible to the cytotoxic action of ExoU (9, 10, 22). The role played by PI(4,5)P2 in crucial cellular functions such as phagocytosis and actin dynamics has made it an ideal target for a number of bacterial toxins (33). For instance, Shigella flexneri expresses the type III-secreted protein IpgD, which is a PI(4,5)P2 phosphatase that converts PI(4, 5)P2 into PI(5)P, allowing S. flexneri to invade host cells (34). Listeria monocytogenes uses the invasin InlB to invade eukaryotic cells by activating cellular PI(3)P kinases to convert PI(4,5)P2 into PI(3,4,5)P3 (35). In contrast, enteropathogenic E. coli (EPEC) uses type III secretion to inhibit these kinases and prevent the pathogen's internalization (36). Thus, several bacterial factors increase or decrease PI(4,5)P2 levels to subvert eukaryotic cells, but to our knowledge, ExoU is the first example of a bacterial protein that exploits PI(4,5)P2 to regulate its activity.
By using PI(4,5)P2 as an activator, ExoU coopts a strategy employed by eukaryotic proteins. PI(4,5)P2 is found predominantly at the inner leaflet of the plasma membrane, where it serves as a “signpost” to guide proteins destined for this location or to activate proteins once they arrive (30, 31). This is particularly true of eukaryotic phospholipases. For example, phospholipase C binds PI(4,5)P2 to target itself to the plasma membrane (37), and cytosolic PLA2 and phospholipase D are activated by PI(4,5)P2 (38, 39). In addition to activation, it is possible that PI(4,5)P2 also functions in the membrane localization of ExoU. This possibility is further substantiated by the recent finding that ExoU bound specifically to PI(4,5)P2 in protein-lipid overlays (12). Thus, by using PI(4,5)P2 as a coactivator and perhaps also as a targeting signal, ExoU may be appropriating a fundamental eukaryotic control and localization system. Interestingly, the ExoU variants R661L and I609N each exhibit cytoplasmic localization within mammalian cells, despite I609N being activated by and presumably interacting with PI(4,5)P2 (19). The significance of this is unclear, but it indicates that the role of PI(4,5)P2 in mediating localization of ExoU may be complex.
The existence of ubiquitin-blind ExoU variants such as ExoU-I609N has been previously suggested in the literature. In two separate studies, ExoU variants were described that were cytotoxic to eukaryotic cells but did not display detectable PLA2 activity with the addition of ubiquitin (in the form of SOD1) (18, 19). However, when a substantial excess of SOD1 (and presumably ubiquitin) was added, these variants were activated (18). This is similar to our results with ExoU-I609N, which required a large excess of ubiquitin for detectable activation. These data suggest that a subset of ExoU variants are not efficiently activated by ubiquitin alone but are still activated by the combination of PI(4,5)P2 and ubiquitin, which allows substantial cytotoxicity. Additionally, we found that residue I609 is critical for full activation by ubiquitin and residue R661 for full activation by PI(4,5)P2. Thus, the MLD (residues 503 to 687) appears to be important for activation of ExoU by each cofactor, but each appears to exert an independent effect on ExoU. Binding of one cofactor may induce a conformational state that facilitates binding of the second cofactor, which in turn causes the catalytic site of ExoU to assume a more open and active conformation. Alternatively, the binding of cofactors may facilitate modifications of ExoU that further regulate its enzymatic activity. Crystal structures of ExoU with one or both cofactors and additional biochemical and biophysical analyses will be necessary to further elucidate the transition states of ExoU and its molecular mechanism of action.
One interesting aspect of our findings is that although PI(4,5)P2 dramatically enhances the PLA2 activity of ExoU, it is not sufficient for activation; ExoU is only minimally activated by PI(4,5)P2 in the absence of ubiquitin. Whether activation by ubiquitin alone is sufficient to cause cell death is less clear. Expression of ExoU and ubiquitin in E. coli was sufficient to kill this bacterium, which lacks PI(4,5)P2 (21). However, both ExoU and ubiquitin were overexpressed in these experiments, which could compensate for the lack of PI(4,5)P2. Additionally, only 300 to 600 molecules of ExoU are required to kill eukaryotic cells, and cell death can occur within 10 min (19, 40). For ExoU to kill cells with this potency, it is likely that both ubiquitin and PI(4,5)P2 are required to fully activate ExoU. Indeed, we observed decreased killing of a yeast mutant with decreased PI(4,5)P2 levels, showing that PI(4,5)P2 is required for full activation of ExoU in eukaryotic cells.
Our data suggest that ExoU binds both PI(4,5)P2 and ubiquitin once injected into host cells, but it is unclear which interaction occurs first, since ExoU appears to have at least some independent affinity for each factor. ExoU rapidly localizes to the plasma membrane in host cells (40), so it may first bind to PI(4,5)P2. This binding could then cause a conformational change that allows ExoU to bind to membrane-localized ubiquitinated proteins or free ubiquitin and become fully activated (Fig. 8, model 1). However, since the majority of free ubiquitin is in the cytosol, ExoU could first bind ubiquitin and subsequently localize to the plasma membrane by binding PI(4,5)P2 (Fig. 8, model 2). In this scenario, ExoU would be partially active in the cytosol and become fully active at the plasma membrane, where binding of PI(4,5)P2 might allow ExoU to interact more directly with substrate membrane phospholipids, thereby increasing catalysis. Alternatively, cofactor interactions may not occur in a defined order; rather, ExoU could bind them simultaneously and in either order. Studies aimed at better understanding the relationship between PI(4,5)P2-mediated activation and PI(4,5)P2-mediated targeting of ExoU to the plasma membrane of eukaryotic cells are ongoing and may inform these models.
Two models of ExoU activation. (Model 1) Upon injection into host cells, ExoU localizes to the inner leaflet of the plasma membrane, interacts with PI(4,5)P2, and then binds to ubiquitin. (Model 2) Alternatively, ExoU may first bind ubiquitin in the cytosol to become partially activated and subsequently localize to the membrane and bind PI(4,5)P2.
Overall, it appears that the activation of ExoU when injected into host cells is a complicated process that has not been completely elucidated. It involves either sequential or simultaneous binding of PI(4,5)P2 and ubiquitin for maximal activation of ExoU both in vitro and in eukaryotic cells. Each of these molecules is evolutionarily conserved in eukaryotic cells, and this could account for the broad spectrum of eukaryotic organisms susceptible to the cytotoxic activity of ExoU. Together, these findings further demonstrate the ability of bacteria to subvert eukaryotic signaling systems for their own advantage.
ACKNOWLEDGMENTS
We thank Jeffrey Veesenmeyer for generating the yeast constructs and Brett Geissler for technical advice and help in lipid preparation. We also thank Andrei Halavaty and Wayne Anderson for providing biochemical expertise. We thank Scott Emr for generously providing the yeast strains used in this study. We also acknowledge Claire Knoten, Stephanie Rangel, and Jessica Tyson for critical reading of the manuscript.
This work was supported by the National Institutes of Health (grants AI053674, AI075191, AI099269, and AI088286 to A.R.H.) and by the American Heart Association (grant 12PRE8660003 to G.H.T.).
FOOTNOTES
- Received 1 April 2013.
- Returned for modification 1 May 2013.
- Accepted 19 May 2013.
- Accepted manuscript posted online 28 May 2013.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00414-13.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
