Ezrin/radixin/moesin proteins are high affinity targets for ADP-ribosylation by Pseudomonas aeruginosa ExoS.

Pseudomonas aeruginosa ExoS is a bifunctional type III-secreted cytotoxin. The N terminus (amino acids 96-233) encodes a GTPase-activating protein activity, whereas the C terminus (amino acids 234-453) encodes a factor-activating ExoS-dependent ADP-ribosyltransferase activity. The GTPase-activating protein activity inactivates the Rho GTPases Rho, Rac, and Cdc42 in cultured cells and in vitro, whereas the ADP-ribosylation by ExoS is poly-substrate-specific and includes Ras as an early target for ADP-ribosylation. Infection of HeLa cells with P. aeruginosa producing a GTPase-activating protein-deficient form of ExoS rounded cells, indicating the ADP-ribosyltransferase domain alone is sufficient to elicit cytoskeletal changes. Examination of substrates modified by type III-delivered ExoS identified a 70-kDa protein as an early and predominant target for ADP-ribosylation. Matrix-assisted laser desorption ionization mass spectroscopy identified this protein as moesin, a member of the ezrin/radixin/moesin (ERM) family of proteins. ExoS ADP-ribosylated recombinant moesin at a linear velocity that was 5-fold faster and with a K(m) that was 2 orders of magnitude lower than Ras. Moesin homologs ezrin and radixin were also ADP-ribosylated, indicating the ERMs collectively represent high affinity targets of ExoS. Type III delivered ExoS ADP-ribosylated moesin and ezrin (and/or radixin) in cultured HeLa cells. The ERM proteins contribute to cytoskeleton dynamics, and the ability of ExoS to ADP-ribosylate the ERM proteins links ADP-ribosylation with the cytoskeletal changes associated with ExoS intoxication.

Amino acids 51-72 of ExoS constitute a membrane localization domain (MLD) that traffics ExoS to the endoplasmic region of mammalian cells (29,30). Type III-delivered ExoS(⌬MLD) stimulates cell death but does not associate with the endoplasmic region yet ADP-ribosylates Ras, uncoupling the ADP-ribosylation of Ras from cytotoxicity (31). Recently, the ADP-ribosylation activity of ExoS has been linked to cellular adherence, the maintenance of filopodia, and enhanced lamellipodia and cellular ruffling (32). Different cell lines demonstrate different sensitivities to intoxication, with epithelial cells being the most affected by the action of ExoS (33). These data suggest that ExoS ADP-ribosylates an unidentified host protein(s) that is responsible for cytoskeletal changes and/or cell death. In this report, ezrin/radixin/moesin (ERMs) are identified as early, high affinity substrates of ExoS, which links reorganization of the actin cytoskeleton with ADP-ribosylation.

Cell Rounding
Confluent HeLa cells (85-mm dishes) were infected with P. aeruginosa expressing the appropriate ExoS construct (see above) at an m.o.i. of 8:1 (bacteria:HeLa cells). P. aeruginosa were quantified from the absorbance calculation 1 A 540 ϭ 4 ϫ 10 8 bacteria. At 3 h, 3 h 20 min, 3 h 40 min, and 4 h post-infection, HeLa cells were washed twice with phosphate-buffered saline and fixed in 1% paraformaldehyde in phosphate-buffered saline for 1 h at 4°C. Cell rounding was determined using a Nikon-inverted microscope (Roper CoolSnap ES camera) and quantified as % rounded by dividing the number of rounded cells by the total number of cells ϫ100 (5 random fields were assayed).

Tetanolysin Assay
Proteins ADP-ribosylated by type III-delivered ExoS in cultured HeLa cells were determined by a "tetanolysin permeability assay" that has been previously described with some modifications (21). Confluent HeLa cells (85-mm dishes) were infected with P. aeruginosa expressing the indicated form of ExoS at an m.o.i. of 8:1 (bacteria:HeLa cell). At 3 h, 3 h 20min, 3 h 40 min, or 4 h post-infection HeLa cells were washed with 10 ml of phosphate-buffered saline and incubated in 6 ml of ice-cold HGI buffer (20 mM PIPES, 2 mM NaATP, 4.8 mM Mg(CH 3 COO) 2 , 150 mM potassium glutamate, 2 mM EGTA, and KOH to obtain pH 7.0) containing 1 mM dithiothreitol for 20 min at 4°C with or without 4 g of tetanolysin and then incubated in 6 ml of HGI buffer containing 20 nM [ 32 P]NAD (10 Ci) at 37°C in 5% CO 2 . After 40 min HeLa cells were harvested in 0.5 ml of HB2 buffer (250 mM sucrose, 3 mM imidazole, pH 7.4, and 0.5 mM EDTA) and broken by passage (30 times) through a 25-gauge syringe; unbroken cells and nuclei were removed by centrifugation in a microcentrifuge at 2500 rpm for 5 min. The post-nuclear supernatant was either subjected to SDS-PAGE followed by autoradiography or Western blot.

In Vitro ADP-ribosylation of HeLa Lysates
ExoS, ExoS(⌬MLD), or ExoS(E381D) (0.04 -40 nM) isolated from culture supernatants of P. aeruginosa were incubated with a HeLa post-nuclear supernatant (20 g) and 10 M [ 32 P]NAD (0.2 Ci) in a 20-l reaction for 15 min at room temperature. Reactions were stopped with SDS-PAGE plus ␤-mercaptoethanol sample buffer, boiled for 5 min, and subjected to SDS-PAGE. Gels were silver-stained (35) and subjected to autoradiography.

Protein Expression
ExoS was purified as previously described (36). ExoS(⌬MLD), FAS, Ras⌬CAAX, (Ras⌬C), and ERMs were purified as His 6 -tagged proteins as previously described (27,37,38). Escherichia coli BL21(DE3) transformed with the appropriate construct were struck for confluency on LB agar plates containing 0.1 mg of ampicillin/ml and cultured overnight at 37°C. Bacteria from a single plate were collected, inoculated into 400 ml of LB (total volume cultured ϭ 1.6 liter) broth containing 0.1 mg of ampicillin/ml, and shaken at 250 rpm at 30°C. After 2 h isopropyl-B-D-thiogalactopyranoside was added to 0.6 mM, and cells were cultured for an additional 2 h. Cultures were centrifuged at 6000 ϫ g for 8 min and suspended in binding buffer (20 mM Tris-HCl, pH 7.9, 0.5 M NaCl, and 5 mM imidazole containing DNase I and RNaseA (at final concentrations of 12 g/ml) and protease inhibitor mixture (P-8340). Cells were broken with two passes through a French press. Broken cells were centrifuged at 30,000 ϫ g for 20 min, and the supernatant was passed through a 0.45-m pore-size cellulose filter. Filtrates were applied to an equilibrated 2-ml nitrilotriacetic acid/Ni 2ϩ agarose-affinity matrix (Qiagen, Valencia, CA). The column was washed with 40 ml of binding buffer. His 6 proteins were eluted with Binding buffer containing 0.25 M imidazole. Pooled fractions were subjected to Sephacryl-200 gel filtration (450 ml volume equilibrated in 10 mM Tris-HCl, pH 7.6, and 20 mM NaCl). The column was run at 0.3 ml/min, and 3.5-ml fractions were collected. Ras⌬C and moesin were stored at Ϫ20°C in 50% glycerol to prevent protein precipitation. ExoS(⌬MLD) and FAS were stored at Ϫ80°C in running buffer.

MALDI-TOF Analysis
The 70-kDa protein (ϳ1 g) was excised from SDS-PAGE gels and subjected to trypsin digestion (1 g, Promega, Madison, WI) in 50 l of 100 mM NH 4 HCO 3 , pH 8, at 37°C for 24 h. After digestion, gel slices were sonicated twice in 200 l of 80% acetonitrile and 1% formic acid (in H 2 O) for 10 min. Eluted material was combined and evaporated, and the pellet was dissolved in 15 l of 0.1% trifluoroacetic acid (in H 2 O). Peptide solutions were desalted with C18 Zip Tips (Millipore, Bedford, MA) that had been equilibrated successively in 15 l of 100% acetonitrile, 15 l of 50% acetonitrile (H 2 O), and 15 l of 0.1% trifluoroacetic acid in H 2 O. Resin was washed twice with 0.1% trifluoroacetic acid in H 2 O. Peptides were eluted in 2 l of 60% acetonitrile and 0.1% trifluoroacetic acid (H 2 O saturated with ␣-cyano-4-hydroxycinnamic acid) and applied to a sample plate to air dry. Samples were ionized by an N 2 UV laser using a PE-pro mass spectrometer (Applied Biosystems). Two hundred laser shots were conducted at an accelerating voltage of 25,000 V and laser intensity of 2075 (repetition rate 3 Hz). Scans were processed using Biosystems Voyager 6004 software. Peptide fingerprinting was used to identify proteins present in the 70-kDa band, using Protein Prospector (University of California at San Francisco).

Transfections
One microgram of DNA encoding pEGFP-C1 or pEGFP-C1-moesin was transiently transfected (LipofectAMINE Plus, Invitrogen) into 70% confluent HeLa cells (6-well plates in complete F-12 media with 5% CO 2 ) according to the manufacturer's instructions. Eighteen hours posttransfection cells were infected, and a tetanolysin assay was performed as described above. Cells were visualized using a Nikon-inverted microscope using a filter set for EGFP (HQ: F712, Nikon). Images were photographed with a Roper CoolSnap ES camera and cropped in Corel Photo-Paint 11.

Enzyme Activity
Linear Velocity Determination-ExoS and ExoS(⌬MLD) (4.5 nM) were incubated in 20-l reactions for 4, 8, of 16 min in the presence of FAS (250 nM), 10 M [ 32 P]NAD (0.2 Ci), ezrin, radixin, moesin, or Ras⌬C (500 nM) in the presence of 50 mM Tris, pH 7.6, containing 2.0 g of bovine serum albumin. Reactions were quenched with SDS sample buffer and subjected to SDS-PAGE. Gels were dried, and bands containing radiolabeled ERMs and Ras⌬C were subjected to scintillation spectrometry.
Stoichiometric Determination-Moesin stoichiometric analysis proceeded as described for linear velocity determination except that reactions were run for 5 h (determined to be saturating for ADPribose incorporated). Reactions were subjected to SDS-PAGE, gels were dried, and radiolabeled moesin bands were exposed to scintillation spectrometry.

Michaelis-Menten Kinetics
ExoS(⌬MLD) (4.5 nM) was incubated with 10 M [ 32 P]NAD (0.2 Ci) and FAS (250 nM) in 50 mM Tris, pH 7.6, containing 2.0 g of bovine serum albumin for 8 min with various concentrations of substrate (0.5-4 and 0.01-28 M for moesin and Ras⌬C, respectively). Incorporation of radiolabel into substrates was performed as described for linear velocity determination. Data were transformed to Lineweaver-Burk, and kinetic constants were determined using Ezfitter (Biosoft, Ferguson, MO). The utilization of NAD was less than 10% for velocity determinations.

ExoS ADP-ribosyltransferase Activity Is Sufficient to Induce HeLa Cell
Rounding-Recent studies demonstrated that the ADP-r activity of ExoS leads to a loss of cellular adherence upon infection with P. aeruginosa (33). To quantify the effects of the ADP-r activity on cell morphology, HeLa cells were infected with P. aeruginosa expressing either the wild-type ExoS, ExoS deficient in GTPase-activating protein activity (R146K), ExoS deficient for ADP-r and GTPase-activating protein activity (R146K/E381D), or a vector (pUCP) control. Cell rounding, characterized by a change in the morphology of HeLa cells from a flat, cubiodal state to a small, button-like appear-ance, was observed in cells infected with Wt and R146K-expressing P. aeruginosa (Fig. 1A). Examination of the kinetics of rounding from 3 to 4 h indicated ExoS-Wt rounded up to ϳ90% of the monolayer by 4 h, compared with ϳ70% for ExoS-R146K. The ADP-r/GTPase-activating protein-deficient ExoS and vector control-infected cells did not undergo significant rounding during this time frame (Fig. 1B). Similar amounts of ExoS were cell-associated for ExoS-Wt, ExoS-R146K, and ExoS-R146K/ E381D (inset, Fig. 1B). Together, these data indicated that the ADP-ribosylation activity of ExoS was sufficient to elicit cell rounding, which prompted an analysis into host proteins that were efficient and early targets for ADP-ribosylation.
ExoS ADP-ribosylates a 70-kDa Protein Early during Infection-A tetanolysin permeability assay was used to identify cellular proteins ADP-ribosylated by type III-delivered ExoS at early time points (21). At 3 h post-infection, several predominant radiolabeled bands were observed, notably at 70, 50, 33, and 26 kDa ( Fig. 2A). Previous studies identified the 50-and 26-kDa proteins to be ExoS and Ras, respectively (21,31). The 70-kDa protein was targeted for identification based upon its modification at the earliest time point assayed, suggesting that it was a high affinity target of the ExoS ADP-ribosyltransferase in cultured cells.
To determine whether the 70-kDa protein was a substrate in vitro, ExoS-Wt was incubated with [ 32 P]NAD and HeLa cell lysates and subjected to SDS-PAGE followed by autoradiography. At 40 nM ExoS-Wt several proteins were radiolabeled, including an intense 50-kDa protein (auto-ADP-ribosylated ExoS) and three substrates ranging in molecular mass from 70 to 90 kDa (Fig. 2B, left panel). To reduce the background of radiolabeling associated with ExoS-Wt in these reactions, ExoS(⌬MLD) was used in the assay, since ExoS(⌬MLD) can be purified to homogeneity from E. coli (37). Like ExoS-Wt, ExoS (⌬MLD) ADP-ribosylated several host proteins, including three proteins with molecular masses between 70 and 90 kDa (Fig.  2B, middle panel). The 70-kDa protein had the same relative mobility as the 70-kDa protein that ExoS-ADP-ribosylated in cultured HeLa cells. ExoS(E381D), which is deficient in ADP-r activity, did not ADP-ribosylate the 70-kDa protein (Fig. 2B,  right panel). These data demonstrate that the 70-kDa protein was a substrate of ExoS in cultured cells and in vitro, and prompted an analysis into the identity of this protein.
The 70-kDa Protein Is Moesin, a Member of the ERM Family of Proteins-To facilitate the identification of the 70-kDa protein, HeLa lysates were in vitro ADP-ribosylated by ExoS(⌬MLD), and proteins were separated using 11% resolving gels. Alignment of the autoradiogram with a silver-stained gel of the proteins allowed for the identification of a silver-stained protein corresponding to the 70-kDa radiolabeled protein (Fig.  3A, inset). Tryptic digestion and MALDI mass spectroscopy analysis of the 70-kDa protein identified it as moesin, a member of the ERM family of proteins (Fig. 3A, Table I).
To confirm that moesin is a substrate of ExoS, DNA encoding the ORF was cloned into the vector pET15b and expressed as an N-terminal His-tagged protein. Using Ni 2ϩ and gel filtration chromatography, moesin was expressed and purified as a soluble protein to near homogeneity (Fig. 3B). Using a linear velocity reaction to monitor the incorporation of labeled ADPribose into substrates, ExoS-Wt and ExoS(⌬MLD) were incubated with purified moesin. Monitoring ADP-ribose incorporation versus time demonstrated that ExoS-Wt and ExoS(⌬MLD) ADP-ribosylated moesin with equivalent rates (Fig. 4A). These data showed that moesin was a substrate for ExoS and was consistent with moesin being the 70-kDa protein ADP-ribosylated by ExoS in cultured cells.
Moesin Is a High Affinity Substrate for ExoS-Iglewski and Kabat (8)  ExoS, and the ExoS/Ras interaction has been well characterized (22). Ras ⌬CAAX (Ras⌬C) is a form of Ras with a deletion in the C-terminal four amino acids. Ras⌬C can be purified from E. coli and is a good substrate for ADP-ribosylation by ExoS (22). A kinetic analysis comparing Ras⌬C and moesin showed that ExoS ribosylated moesin at a 5-fold faster rate than Ras⌬C ( Fig. 4B and Table II). The K m for moesin was 0.4 MϮ 0.1, with a K cat /K m of 76 (Table II). Because of limitations in expressing Ras ⌬C to a concentration that approached the apparent K m , the affinity was estimated to be ϳ100 M, 2 orders of magnitude greater than moesin. Stoichiometric determination showed 0.5 mol of ADP-ribose incorporated per mol of moesin compared with 2 mol of ADP-ribose/mol of Ras ⌬C (Table II). This indicated that ExoS ADP-ribosylated moesin at a single, high affinity site.
Ezrin and Radixin Are Substrates for ADP-ribosylation by ExoS-Ezrin and radixin are homologs of moesin and may perform redundant functions in mammalian cells (Fig. 5A) (39). To determine whether the ADP-ribosylation was specific for moesin as opposed to the other ERMs, the ORFs of ezrin and radixin were cloned into Pet15b and purified as His-tagged proteins (Fig. 5A, right panel). Linear velocity analysis showed that ezrin and radixin were ADP-ribosylated at rates comparable with moesin (Fig. 5B), demonstrating each of the ERM proteins were high affinity targets for ADP-ribosylation by ExoS.
ERMs Are Substrates of ExoS in Vivo-Experiments were performed to test if moesin was in fact the 70-kDa protein ADP-ribosylated by ExoS in cultured cells, using a moesin monoclonal antibody to locate moesin relative to the 70-kDa ADP-ribosylated protein and by measuring the ability of ExoS-Wt to ADP-ribosylate a GFP-moesin fusion protein (GFP was fused to the N terminus of moesin). GFP-moesin could be expressed to similar levels as endogenous moesin, localized  Fig. 2 were subjected to 11% resolving gels to achieve maximal band separation followed by silver-staining to align radiolabeled bands with stained proteins. The arrow indicates the migration of the 70-kDa protein next to a faint silver-stained band (q). The protein indicated in the inset was subjected to tryptic digestion and MALDI-TOF analysis (see "Experimental Procedures"). Shown is a trace of the peptide profile with moesin-specific peptides numbered 1-16 (refer to Table I). B, moesin ORF was cloned into pET15b and purified using Ni 2ϩ /nitrilotriacetic acid and gel filtration chromatography. Ni 2ϩ /nitrilotriacetic acid and gel filtration (Gel. Fil.) lanes show 4 g of His 6 -moesin (arrow).  Fig. 3A was subjected to tryptic digestion and MALDI-MS analysis. Peptide numbers refer to the peaks indicated in Fig. 3A. Matched peptides covered 16 of the 28 peaks identified, corresponding to 57% of the moesin amino acid sequence. throughout the cell but excluded from the nucleus and was enriched at cell-cell contact points (arrow, Fig. 6A, left panel). In control cells, GFP was expressed diffusely in HeLa cells and was not excluded from the nucleus (Fig. 6A, right panel).
After GFP-moesin transfection, HeLa cells were infected with P. aeruginosa expressing ExoS-Wt and subjected to the tetanolysin permeability assay. Autoradiography of the cell lysates that had been subjected to SDS-PAGE and transferred to a polyvinylidene difluoride membrane showed three ADPribosylated bands in the range of 70 -92 kDa (70, 72, and 92 kDa (Fig. 6B, left panel)). After GFP transfection, HeLa cells that were infected with P. aeruginosa expressing ExoS-Wt and subjected to the tetanolysin permeability assay showed that the 70-and 72-kDa proteins were ADP-ribosylated but that a 92-kDa protein was not ADP-ribosylated. In GFP transfected cells GFP was not ADP-ribosylated by ExoS-Wt (data not shown). Western analysis using ␣-moesin antibody identified three bands in the GFP-moesin-transfected cells that were immunoreactive with molecular masses of 70, 72, and 92 kDa (Fig. 6, right panel), whereas in the GFP-transfected cells only the 70-and 72-kDa immunoreactive bands showed reactivity. The 72-kDa protein represents ezrin and/or radixin, since the moesin monoclonal antibody cross-reacts with these two proteins (data not shown). The 72-kDa protein also reacts with anti-ezrin antibodies (data not shown). Taken together, these data support the conclusion that moesin is the 70-kDa protein that is ADP-ribosylated by ExoS in cultured HeLa cells and suggests the ERM proteins collectively are early, high affinity targets for ADP-ribosylation by ExoS during infection. DISCUSSION One of the hallmarks of bacterial pathogenesis is the ability of certain pathogens to modulate host cell properties through the action of toxins or protein effectors. In recent years a paradigm has emerged whereby the enzymatic activity of many toxins is directed at components of the host cytoskeleton. This action can disrupt the actin networks to subvert normal cell clearance mechanisms (phagocytosis) or tissue persistence (barrier formation) or to promote specific uptake of the pathogen, as is the case with intracellular replicating bacteria. For example, P. aeruginosa, Yersinia, and Salmonella produce GTPase-activating proteins and proteases (ExoS/T, SptP/YopE, and YopT, respectively) that inactivate small monomeric GTPases such as Rho, Rac, and Cdc42 (7,40,41). Salmonella also produces effectors, which act as guanine exchange factors to activate GTPases (SopE/E2), and actin polymerizing/depolymerizing (SipA/ADF, cofilin) enzymes, to control actin filament assembly/disassembly (42). Yersinia translocates a tyrosine phosphatase (43), YopH, which acts at focal adhesions (44), whereas ExoT ADP-ribosyltransferase of P. aeruginosa targets the Crk proteins (12), also components of focal adhesion complexes.
ExoS ADP-ribosyltransferase targets a diverse group of proteins including members of the Ras family of GTPases such as Rap, Ral, and Rho (17,18,19,23). Different cell lines exhibit different profiles of GTPase modification, making interpretation into the effects of the ADP-r domain on cell morphology difficult to determine (34). ExoS (⌬MLD), a form of ExoS that does not efficiently modify Ras in vivo but induces actin rearrangements is cytotoxic (31). This indicated that the ADP-ribosylation of Ras was not responsible for either cell rounding or the cytotoxic phenotypes elicited by ExoS. The finding that the ADP-r domain alone induced cellular changes prompted the identification of early host targets of ExoS ADP-ribosylation. Although several proteins are ADP-ribosylated early, only three or four represent proteins previously unidentified, one of which is labeled at 70 kDa. This protein was also labeled in Chinese hamster ovary cells (data not shown), indicating it was a common substrate of ExoS. MALDI mass spectroscopy identified the 70-kDa protein as moesin. Purified moesin was a high affinity substrate for ExoS, and both moesin and GFP-moesin were ADP-ribosylated in cultured cells by ExoS. The finding that ezrin and/or radixin was also an in vivo target of ExoS suggests the ERM proteins rather than moesin alone are substrates for ExoS ADP-r activity. Indeed, linear velocity analysis showed that the ERMs were ADPribosylated at equivalent rates.
The affinity of ExoS for the ERMs approaches that of well characterized substrates of other bacterial ADP-ribosyltransferases. For example, the K m values for the ADP-ribosylation of elongation factor 2 by exotoxin A from P. aeruginosa and G␣␤␥ complex by pertussis toxin from Bordetella pertussis are ϳ1 and 0.3 M, respectively, compared with 0.4 M for moesin (45,46). The finding that moesin was a higher affinity target than Ras yet both moesin and Ras are intensely labeled early in an infection suggests protein localization or other undefined factors influence the ADP-ribosylation of these substrates in cultured cells. Indeed, membrane-bound Ras, as opposed to cytosolic Ras, is ADP-ribosylated by ExoS (31), suggesting the intense labeling of Ras in an infection, despite its lower affinity compared moesin, may be due to the high local concentrations of ExoS and Ras on membranes. Similar to Ras, the ADPribosylation of the 70-kDa protein is observed at each time point in the tetanolysin assay. This indicates that a fraction of the moesin pool was ADP-ribosylated and that the complete ADP-ribosylation of the pool may not be required to elicit affects on the actin cytoskeleton.
Ezrin, radixin, and moesin are 585-, 582-, and 576-amino acid proteins, respectively, consisting of two major domains, an N-terminal ERM domain that is Ͼ80% homologous between each protein and interacts with a wide variety of proteins, including Rho guanine disassociation inhibitor, Dbl (a guanine exchange factor for Rho), and phosphatidylinositol 4,5-diphos- phate, and a C-terminal domain that binds to actin (47,48,49). ERM proteins function in many actin processes, including the formation of cell shape, microvilli, motility, cell adhesion, and phagocytosis. Mouse knockouts of moesin do not show a phenotype, suggesting the ERM proteins are functionally redundant (50). ERMs exist in two states; in an inactive, actin unbound form, where the N terminus masks the actin binding site on the C terminus, and an active "open" form, where the C terminus is accessible for F-actin binding (51). ERM activation may proceed through several mechanisms, including phosphorylation of Thr-558 by Rho kinase (52) and phosphatidylinositol 4,5-diphosphate binding (53). Phosphorylation on Thr-558 is thought to prevent the N-C-terminal interaction, leaving ERM in an open form that can associate with F-actin cables. Localization of moesin and ezrin to the plasma membrane in extending lamellipodia and microvilli as well as their association with membrane adaptor proteins like EBP50 (54) has led to a model whereby activated ERMs serve as a bridge between receptor complexes and the actin cytoskeleton (47,55). Placing the ERMs at a critical point in actin regulation leads to the hypothesis that ADP-ribosylation may induce some of the observed cytoskeletal rearrangements by disrupting ERM interaction with upstream or downstream proteins.
ExoS ADP-ribosylates at 0.5 mol of ADP-ribose/mol of moesin, suggesting there is one preferred site of ADP-ribosylation of the ERMs. ADP-ribosylation at the N terminus may prevent the ERMs from interacting with known interacting proteins, such as Rho guanine disassociation inhibitor or Dbl. The ERMs are thought to facilitate the exchange of guanine disassociation inhibitor for Dbl bound to Rho, promoting Rho activation through GDP/GTP exchange (47). An ADP-ribose moiety located at the interface of the guanine disassociation inhibitor/ ERM interaction may prevent Dbl exchange, keeping Rho inactive. Because Rho is inactivated by ExoS GTPase-activating protein activity, the ADP-ribosylation of the ERMs may be an additional and additive inactivation mechanism to inactive Rho-mediated signal transduction. A second possibility is that ADP-ribosylation at the C terminus prevents the ERMs from interacting with F-actin. This could eliminate actin-specific b Moesin stoichiometric analysis proceeded as described for linear velocity determination except that reactions were quenched after 5 h (determined to be saturating for radiolabel incorporation).
c Kinetic constants were determined using reaction conditions described for linear velocity determination. Substrate concentrations varied as follows: moesin (0.5-4 M) and Ras⌬C (0.01-28 M). Data were transformed to Lineweaver-Burk, and kinetic constants were determined using Ezfitter. The utilization of NAD was less than 10% for velocity determinations. d From Ganesan et al. (38). e ND, not determined.
FIG. 5. ExoS ribosylates ezrin and radixin at rates comparable with moesin. A, the ERM (ezrin/radixin/moesin) proteins are greater than 80 and 60% homologous in their N-and C-terminal ERM domains, respectively. Ezrin and radixin contain a proline-rich region that is absent from moesin. Ezrin and radixin were cloned into pET15b and purified as His 6 proteins (see "Experimental Procedures"). Shown is a Coomassie gel of 5 g of purified ezrin and radixin (right panel). B, the linear velocity analysis was carried out as described in the legend to Fig. 4. Insets show autoradiographs (A) and Coomassie-stained gels (P) of ERM reactions with time. ERMAD, ezrin, radixin, and moesin association domain.
FIG. 6. Moesin is an in vivo target of ExoS ADP-ribosyltransferase. A, HeLa cells (80% confluent) were transfected with 1 g of GFP-moesin or GFP (control) for 18 h and visualized by fluorescence or phase contrast microscopy. B, transfected HeLa cells in A were infected with P. aeruginosa (m.o.i. 8:1) expressing ExoS-Wt followed by a tetanolysin assay to label ExoS substrates. HeLa post-nuclear supernatant (20 g) was subjected to SDS-PAGE, and proteins were transferred to polyvinylidene difluoride membranes. Membranes were exposed to autoradiography followed by anti-moesin Western blotting to locate GFPmoesin (GFP-M) and endogenous moesin relative to radiolabeled bands. The analysis represents one of two independent determinations. E, ezrin, M, moesin. structures for which ERM function is required, such as microvilli, membrane ruffles, or cell-cell/matrix contact points. Such a model is consistent with the finding that ExoS has been described as disrupting cell adherence (33), a possible survival mechanism aimed at disrupting epithelial barriers during infection. Another possibility is that ERMs in immune cells are primary targets. ERMs play a role in T-cell-antigen presenting cell association (56). The ADP-ribosylation of ERMs in these cells may be a mechanism of P. aeruginosa anti-immunity, perhaps down-regulating the adaptive or innate immune response. It will be interesting to determine whether other ERM domain-containing proteins, such as merlin and protein 4.1, are ADP-ribosylated by ExoS.Defining how ADP-ribosylation effects ERM localization and activation may resolve the mechanism responsible for cytoskeletal reorganization by ExoS.