Pseudomonas aeruginosa ExoT ADP-ribosylates CT10 Regulator of Kinase (Crk) Proteins*

Pseudomonas aeruginosa ExoT is a type III cytotoxin that functions as an anti-internalization factor with an N-terminal RhoGAP domain and a C-terminal ADP-ribosyltransferase domain. Although ExoT RhoGAP stimulates actin reorganization through the inactivation of Rho, Rac, and Cdc42, the function of the ADP-ribosylation domain is unknown. The present study characterized the mammalian proteins that are ADP-ribosylated by ExoT, using two-dimensional SDS-PAGE and matrix-assisted laser desorption ionization/time of flight (MALDI-TOF) analysis. ExoT ADP-ribosylated two cytosolic proteins in cell lysates upon type III delivery into cultured HeLa cells. MALDI-TOF mass spectrometry analysis identified the two proteins as Crk-I and Crk-II that are Src homology 2–3 domains containing adaptor proteins, which mediate signal pathways involving focal adhesion and phagocytosis. ExoT ADP-ribosylated recombinant Crk-I at a rate similar to the ADP-ribosylation of soybean trypsin inhibitor by ExoS. ExoS did not ADP-ribosylate Crk-I. ADP-ribosylation of Crk-I may be responsible for the anti-phagocytosis phenotype elicited by ExoT in mammalian cells.

Exoenzyme S (ExoS) 1 was originally isolated from spent culture medium of Pseudomonas aeruginosa as a protein aggregate that had ADP-ribosyltransferase activity distinct from exotoxin A (1). The ExoS aggregate consisted primarily of two proteins with molecular masses of 53 and 49 kDa (2). The 53and 49-kDa proteins were subsequently shown to be encoded by separate genes, termed exoT and exoS, respectively (3,4). ExoT and ExoS share 76% amino acid identity and are bi-functional proteins with two independent enzymatic activities, which include a RhoGAP domain and an ADP-ribosyltransferase domain. The RhoGAP domain is encoded within the N terminus of ExoT and ExoS and causes actin reorganization as a GTPaseactivating protein for the Rho GTPases (5)(6)(7). Both ExoT and ExoS utilize an active site Arg (Arg-149 for ExoT and Arg-146 for ExoS) for expression of RhoGAP activity (5,8,9). Expression of RhoGAP activity has been implicated in the inhibition of phagocytosis in polarized epithelial cells and macrophage-like cells (8).
The ADP-ribosyltransferase domain is encoded within the C terminus of ExoT and ExoS. ExoS ADP-ribosylates numerous proteins, including members of the Ras protein family (10,11). ADP-ribosylation of Ras at Arg-41 interferes with the ability to bind the guanine nucleotide exchange factor, uncoupling Ras signal transduction (12). Expression of ADP-ribosyltransferase activity is dependent upon the binding of a eukaryotic protein termed FAS (Factor Activating ExoS), later shown to be a 14-3-3 protein (13,14). Scanning mutagenesis showed that Glu-381 contributed to expression of ADP-ribosyltransferase activity of ExoS (15,16). Subsequent studies showed that ExoS was a bi-glutamic acid transferase, where Glu-381 functioned in a catalytic capacity and Glu-379 was required for efficient ADP-ribosyltransferase activity (16).
In contrast, ExoT does not efficiently ADP-ribosylate proteins that are ADP-ribosylated by ExoS (17). Thus, ExoT was considered to have limited capacity to express ADP-ribosyltransferase activity. Engel and co-workers (8) recently observed that the mutation of catalytic Arg (149) of ExoT partially diminished the anti-internalization activity, and P. aeruginosa expressing ExoT(R149K) retained some capacity to stimulate cell rounding and disruption of the actin cytoskeleton. Sundin et al. (18) also reported that ExoT reorganized the actin cytoskeleton without interfering with Ras signal transduction and that ExoT(R149A) stimulated a morphological change of cultured cells. These data suggested that expression of ADP-ribosyltransferase activity by ExoT contributed to modulation of the actin cytoskeleton. This promoted an investigation on the ADP-ribosyltransferase activity of ExoT, which resulted in the identification of two mammalian adaptor proteins that ExoT efficiently ADP-ribosylates.

Site-directed Mutagenesis and Plasmid Construction
ExoT Mutagenesis-DNA encoding ExoT-HA was engineered into pEGFPN1 vector (6). HA was used as a reporter epitope tag. Mutations were engineered into ExoT by Quick-change Mutagenesis (Stratagene), using pEGFP-ExoT-HA as template. The following primers were utilized: ExoT(R149K)-HA, 5Ј-gcg acg gcg ccc tga aat cgc tgg cca ccg c-3Ј and 5Ј-gcg gtg gcc agc gat ttc agg gcg ccg tcg c-3Ј; and ExoT(E383D/ E385D)-HA, 5Ј-ctt gtc gta gag gat atc ctg atc atc gcc ctc gat cga-3Ј and 5Ј-tcg atc gag ggc gat gat cag gat atc ctc tac gac aag-3Ј. ExoT(R149K/ E383D/E385D)-HA was engineered using the primers for ExoT(R149K) and using DNA encoding ExoT(E383D/E385D)-HA as template. Mutated DNA was sequenced to confirm the presence of the mutation and that additional mutations were not introduced into the template. Subsequently, DNA encoding ExoT-HA and mutated ExoT-HA in the pEG-FPN1 vector were subcloned into pUCP vector at NsiI and BamHI sites, which allowed for expression from the exoS promoter (19).
Crk Expression Vector-pCMV-Sport6, containing Crk cDNA, was * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  purchased from American Type Culture Collection (GenBank TM code BC008506). cDNA encoding Crk-I was amplified by PCR using the primers 5Ј-gtt ccg cgt gga tcc cat atg gcg ggc aac ttc ga-3Ј and 5Ј-atc tgc agg cgg ccg cgt cga cga ctc aaa gct tcc gac ctc caa tca ga-3Ј to introduce a 5Ј BamHI site in-frame with the GST fusion component of pGEX4T and a SalI site 3Ј to the tga stop codon. After restriction endonuclease digestion, the PCR product was cloned into the BamHI and SalI sites of pGEX4T. pGEX-Crk-I was transferred into Escherichia coli BL21(DE3) for protein expression.
Expression of ExoT, ExoS, and Mutated Forms of ExoT in P. aeruginosa P. aeruginosa PA103 (⌬exoU,exoT::Tc) transformed with pUCP encoding ExoS-HA, ExoT-HA, or the mutated forms of ExoT-HA were cultured at 37°C in low calcium media (20) to A 540 ϭ 4 -5. Cultures were centrifuged at 10,000 rpm for 10 min (SS-34) to pellet bacteria, and the proteins in spent culture medium were precipitated (56% final concentration of saturated ammonium sulfate, v/v). The precipitate was collected and suspended in HG-1 buffer (20 mM PIPES, 2 mM Na-ATP, 1 mM MgSO 4 , 150 mM potassium glutamate, and 2 mM EGTA). Secreted ExoS-HA, ExoT-HA, or the mutated forms of ExoT-HA were detected by ECL-Western blot analysis (SuperSignal, Pierce), using mouse ␣-HA IgG (Covance) as the primary antibody and goat ␣-mouse-IgG-horseradish peroxidase (Pierce) as the secondary antibody. The density of HA reactive bands were measured in an Alpha Imager. Concentrations of the secreted ExoS-HA, ExoT-HA, and the mutated forms of ExoT-HA were calculated, using known concentration of recombinant ExoS(⌬51-72)-HA as standard in an Alpha Imager.

In Vitro ADP-ribosylation Assay
Concentrated culture supernatants containing 3 ng of the secreted toxins were incubated with 20 -40 g of CHO cells or HeLa cell lysates with 1 M NAD ϩ ([ 32 P]adenylate phosphate-NAD ϩ ). The reaction was stopped by either adding SDS sample buffer or 90% acetone (final concentration, v/v). Samples treated with SDS sample buffer were subjected to one-dimensional SDS-PAGE followed by autoradiography of dried gels, whereas acetone-precipitated samples were subjected to analysis by two-dimensional SDS-PAGE.
Preparative Analysis for Protein Identification-0.6 mg of a cytosolic fraction of HeLa cells was incubated with ExoT with 1 M NAD ϩ ([ 32 P]adenylate phosphate-NAD ϩ ) at 37°C for 1 h. Proteins in the reaction mixture were precipitated with 90% acetone. Precipitated proteins were suspended in 300 l of IEF rehydration buffer and loaded onto an Immobiline DryStrip (18-cm pH 4.5-5.5 strips). After 12 h of passive rehydration, proteins were focused in PROTEAN IEF Cell (Bio-Rad) with the following program: step 1, 300 V for 1 h; step 2, 400 V for 1 h; step 3, 500 V for 1 h; step 4, 3500 V for 3 h; step 5, 5000 V for 72 h; and step 6, holding at 500 V.
Analysis of Two-dimensional SDS-PAGE by Two-dimensional Software-Silver-stained two-dimensional SDS-PAGE gel containing twodimensional SDS-PAGE standards (Bio-Rad) or corresponding autoradiographic images with marked two-dimensional SDS-PAGE standards were scanned into PD-quest two-dimensional software (Bio-Rad) or Phoretix two-dimensional software (Nonlinear Dynamics) to determine molecular weight and pI of proteins of interest and to align proteins that were ADP-ribosylated by ExoS and ExoT.

ADP-ribosylation of Proteins in Cultured Cells by Type III Delivered Cytotoxins
Tentanolysin Assay-CHO or HeLa cells were infected at 37°C with the indicated strain of P. aeruginosa at m.o.i. ϭ 8:1 (bacteria/mammalian cell) for 3.5-5 h. The cell monolayer was washed and then incubated with tetanolysin (0.4 g/ml) in 6 ml of HG-1 buffer at 4°C for 7-10 min to permeabilize the plasma cell membrane (21). After permeabilization, cells were incubated with HG-1 buffer containing 20 nM NAD ϩ ([ 32 P]adenylate phosphate-NAD ϩ ) at 37°C for 20 -25 min. Cells were washed, harvested in HB2 buffer, and fractionated (see below) to obtain membrane and cytosolic fractions. The subcellular fractions were subjected to one-dimensional SDS-PAGE or two-dimensional SDS-PAGE followed by Western blot or autoradiography.
Subtraction Assay-CHO or HeLa cells were infected at 37°C with various strains of P. aeruginosa at m.o.i. ϭ 8:1 (bacteria/mammalian cell) for 3.5-5 h. Infected cells were harvested, lysed, and fractionated. The cell fractions from infected cells were incubated with secreted ExoT with 1 M NAD ϩ ([ 32 P]adenylate phosphate-NAD ϩ ) for 30 min. Samples were subjected to two-dimensional SDS-PAGE followed by autoradiography.

Cellular Fractionation of Mammalian Cells
CHO or HeLa cells (85-mm dishes) were harvested in 0.6 ml of HB2 buffer (250 mM sucrose, 3 mM imidazole (pH 7.4), 0.5 mM EDTA, 1% protease inhibitor mixture) and lysed by passage through a 1-ml syringe, using a 25-gauge needle 14 -20 times. A post-nuclear supernatant (PNS) was obtained by centrifugation of cell lysate at 250 ϫ g for 5 min. Membrane and cytosolic fractions were obtained by centrifugation of the PNS at 100,000 ϫ g for 30 min. One day prior to infection, cultured cells were transfected with 50 ng of pEGFPN1 (Lipofect-AMINE-Plus, Invitrogen), and enhanced green fluorescent protein was used as a cytosolic marker for cellular fractionation.

Purification of GST-Crk-I
The GST-Crk-I fusion protein was affinity-purified from E. coli using glutathione-Sepharose 4B as described by the manufacturer (Amersham Biosciences). GST-Crk-I was cleaved by thrombin, and the cleaved GST and uncleaved GST-Crk-I were removed by incubating with glutathione-Sepharose 4B. Thrombin was removed in the same incubation using p-aminobenzamidine-agarose (Sigma). Resins were pelleted by centrifugation at 250 ϫ g for 10 min. The soluble fraction, Crk-I, was stored in aliquots at Ϫ80°C. Purified Crk-I and soybean trypsin inhibitor (SBTI) were quantified by measuring Coomassiestained samples subjected to SDS-PAGE, using bovine serum albumin as a standard.

Enzyme Assay
Specific Activity-The ADP-ribosylation activity of ExoS for SBTI and ExoT for Crk-I was determined by a linear velocity assay. Secreted ExoS or ExoT (5 nM) was incubated with SBTI or Crk-I (3 M), respectively, with 50 mM Tris-HCl (pH 7.4), 0.1 mM NAD ϩ ([ 32 P]adenylate phosphate-NAD ϩ ), 500 nM FAS, and 0.2 g/l bovine serum albumin. The reactions were stopped at 2, 4, 8, and 16 min by adding SDS sample buffer. Samples were applied to SDS-PAGE, followed by Coomassie staining. The radioactive bands were excised and subjected to scintillation counting.
Stoichiometry-ExoT or ExoS (45 nM) was incubated with Crk-I or SBTI (3 M), respectively, with 500 nM FAS and 0.1 mM NAD ϩ ([ 32 P]adenylate phosphate-NAD ϩ ) for 1, 3, and 5 h. Stoichiometry of the NAD ϩ /target protein was determined at the time where incorporation of NAD ϩ to target protein did not increase with extended reaction time.

P. aeruginosa Type III Delivered ExoT-stimulated Morphological Changes in HeLa
Cells-Several forms of ExoT were engineered for type III expression in P. aeruginosa from the pexoS promoter in the broad host range plasmid, pUCP. P. aeruginosa PA103 (⌬exoU, exoT::Tc) was used as the host for these experiments, because it does not express known type III cytotoxins. Initial experiments showed that P. aeruginosa expressed and secreted similar amounts of ExoT, ExoT(R149K) (RhoGAP-deficient), ExoT(E383D/E385D) (ADP-ribosyltransferase-deficient), and ExoT(R149K/E383D/E385D) (RhoGAPand ADP-ribosyltransferase deficient) (data not shown). Whereas type III delivered ExoT and ExoT(E383D/E385D) stimulated HeLa cell rounding, type III delivered ExoT(R149K) stimulated an intermediate change in HeLa cell morphology relative to ExoT and ExoT(E383D/E385D) (Fig. 1, upper panel). The morphological change stimulated by ExoT(R149K) was dependent on ADP-ribosyltransferase activity, because type III delivered ExoT(R149K/E383D/E385D), which was defective in both RhoGAP and ADP-ribosyltransferase activities, did not stimulate morphological changes in HeLa cells. Expression and subcellular distribution of the type III delivered ExoT and the ExoT mutants in HeLa cells (Fig. 1, lower panel) were similar and found in both the membrane and cytosolic fractions. This indicated that aberrant expression or subcellular localization was not responsible for the morphological changes elicited by the expression of ADP-ribosyltransferase activity of ExoT and prompted an evaluation of host proteins that ExoT ADP-ribosylated. Type III delivered ExoT(R149K) also stimulated an intermediate morphological change in CHO cells relative to ExoT and ExoT(E383D/E385D) (data not shown).
ExoT ADP-ribosylated a Subset of Proteins in HeLa Cells-Proteins that ExoT ADP-ribosylated were determined in HeLa cell lysates by one-dimensional SDS-PAGE ( Fig. 2A) and twodimensional SDS-PAGE (Fig. 2B). Incubation of the HeLa cell lysate with [ 32 P]adenylate phosphate NAD ϩ identified one 43-kDa protein that was ADP-ribosylated by an endogenous ADPribosyltransferase in the lysate. In addition to the 43-kDa endogenously ADP-ribosylated protein, HeLa cell lysates incubated with supernatants from P. aeruginosa or supernatants from P. aeruginosa transformed with the pUCP vector control contained a 97-kDa ADP-ribosylated protein, which represented the ADP-ribosylated elongation factor-2 by exotoxin A. HeLa cell lysates incubated with supernatants containing either ExoT or ExoT(R149K) displayed identical profiles of ADPribosylated proteins in one-dimensional and two-dimensional SDS-PAGE with ExoT being auto-ADP-ribosylated and three proteins (labeled T1, T2, and T3) being ADP-ribosylated. In contrast, HeLa cell lysates incubated with the ADP-ribosyltransferase-deficient form of ExoT, ExoT(E383D/E385D), did not display auto-ADP-ribosylated ExoT or radiolabeled T1, T2, or T3. Subcellular fractionation of the HeLa cell lysates in in vitro ADP-ribosylation assay showed that T1, T2, and T3 were present in the cytosolic fraction (data not shown).
A subtraction assay was utilized to determine whether T1, T2, and T3 were ADP-ribosylated by type III delivered ExoT in HeLa cells (Fig. 3). In the subtraction assay, HeLa cells were infected with P. aeruginosa PA103 (⌬exoU, exoT::Tc) expressing the indicated form of ExoT for 4 h. Cell lysates prepared from the infected cells were then incubated with ExoT and [ 32 P]adenylate-phosphate NAD ϩ to [ 32 P]ADP-ribosylate host proteins that were not ADP-ribosylated during the infection phase of the assay. Two-dimensional SDS-PAGE analysis showed that three radiolabeled proteins had electrophoretic mobility identical to T1, T2, and T3 in cell lysates from the mock infection or infection with P. aeruginosa expressing ExoT(E383D/E385D), suggesting that these three proteins had not been ADP-ribosylated during the infection with P. aeruginosa expressing ExoT(E383D/E385D). In contrast, T1 and T2 were not available for in vitro ADP-ribosylation in cell lysates from the cells infected with P. aeruginosa expressing ExoT or ExoT(R149K), indicating that both T1 and T2 had been ADPribosylated during the infection. Only a portion of T3 was available for in vitro ADP-ribosylation in these lysates, indicating that only a portion of T3 had been ADP-ribosylated during the infection. Additional experiments showed that extending the infection time to 6 h reduced the amount of T3 that was available for in vitro ADP-ribosylation, suggesting that T3 was a late target for ADP-ribosylation by type III delivered ExoT (data not shown). Therefore, T1 and T2 appear to be early targets for ADP-ribosylation by type III delivered ExoT.
T1 and T2 Are Crk-I and Crk-II, Respectively-HeLa cell lysates were incubated with ExoT and [ 32 P]adenylate phosphate NAD ϩ and subjected to preparative two-dimensional SDS-PAGE, using 18-cm IEF strips with a pH gradient between 4.5 and 5.5. Overlaying the autoradiogram (Fig. 4B) with the silver-stained gel (Fig. 4A) identified three protein spots that were radioactive, representing T1, T2, and T3. The silverstained spots for T1 and T2 from 12 gels were collected, tryptic  (Table I). Crk-I and Crk-II are the translation products of alternative splicing of the human CRK gene (22). Whereas the predicted molecular mass was 23 kDa for Crk-I and 34 kDa for Crk-II (Table I) Reactions were stopped by addition of 90% acetone (final concentration v/v). Acetone precipitates were subjected to twodimensional SDS-PAGE (1st dimension IEF, 7-cm Immobiline dry strips with pH 4 -7 linear gradient; 2nd dimension, 13.5% SDS-PAGE). Gels were fixed, silver-stained, and subjected to autoradiography. Images of the exposed x-ray film are shown. In the left lane of each gel, an aliquot of each respective reaction was subjected to one-dimensional SDS-PAGE. Arrows within the gels indicate three unique ADP-ribosylated proteins (labeled T1, T2, and T3) in mockinfected cell lysates (mock infection) and cell lysates from cells infected with PA103 (⌬exoU, exoT::Tc) expressing ExoT(E383D/ E385D)-HA (ExoT(EE/DD)) but were absent in lysates from cells infected with P. aeruginosa that express ADP-ribosyltransferase activity, ExoT and ExoT(R149K).

FIG. 4. Separation of T1, T2, and T3 by two-dimensional SDS-PAGE.
A HeLa cytosolic cell lysate (0.6 mg) was incubated with ExoT (30 ng) with the presence of 1 M [ 32 P]adenylate phosphate NAD ϩ for 2 h. The reaction was stopped by addition of 90% acetone (final concentration v/v). Acetone precipitates were subjected to preparative twodimensional SDS-PAGE (18-cm (pH 4.5 to 5.5) linear gradient strip) followed by 10% SDS-PAGE (20-cm separating gel). The gel was silverstained (A) and then subjected to autoradiography (B). The determined pI and molecular mass for T1, T2, and T3 are indicated within the gels in parentheses. T1, T2, and T3 (circled protein spots in A) were identified by overlapping the autoradiogram on the silver-stained gel.

FIG. 2. ExoT ADP-ribosylated eukaryotic proteins in HeLa cells in vitro.
A cytosolic fraction of a HeLa cell lysate (30 g) was incubated alone (leftmost Ϫ) or with concentrated supernatants from P. aeruginosa PA103 (⌬exoU, exoT::Tc) (2nd to left Ϫ), transformed with the vector control (pUCP) or expressing equivalent amounts (3 ng) of the indicated forms of ExoT: pUCPExoT-HA (ExoT), pUCPExoT(R149K)-HA (ExoT(R149K)), or pUCPExoT(E383D/E385D)-HA (ExoT(EE/DD)) with [ 32 P]adenylate phosphate-NAD ϩ for 30 min. Reactions were stopped with the addition of 90% acetone (final concentration v/v). Acetone precipitates were collected and subjected to 13.5% one-dimensional SDS-PAGE (A) or two-dimensional SDS-PAGE (B) (1st dimension IEF, 7-cm Immobiline Dry Strips with pH 3-10 linear gradient; 2nd dimension, 13.5% SDS-PAGE). Gels were fixed, silver-stained, and subjected to autoradiography. Images of the exposed x-ray films are shown. To the left of the gel in A are the migrations of molecular weight marker proteins. Arrows, along the right of the gel in A and within the gels of B indicate the auto-ADP-ribosylation of ExoT (ExoT) and three unique proteins that are ADP-ribosylated (labeled T1, T2, and T3). molecular mass has been reported to be 28 kDa for Crk-I and 42 kDa for Crk-II (22), similar to the molecular mass determined in the present study (28 kDa for Crk-I and 40 kDa for Crk-II). The predicted pI values for Crk-I and Crk-II were 5.3 and 5.4, respectively, whereas the determined pI values for ADP-ribosylated Crk-I and Crk-II were 4.9 and 5.0, respectively. Because one ADP-ribosylation decreases a protein pI by ϳ0.2 pH unit, the difference between predicted pI and determined pI suggested that there were 1-2 ADP-ribosylations per Crk protein. Crk-I and Crk-II have one SH2 and one SH3 domains, whereas Crk-II has an additional SH3 domain at its C terminus. Presumably, the ADP-ribosylation sites of ExoT are localized in the common regions between Crk-I and Crk-II.
T3 was resolved on 18-cm IEF strips with pH gradient 5-6 followed by preparative two-dimensional SDS-PAGE. Silverstained T3 was collected, tryptic digested, and subjected to MALDI-MS/PSD analysis. MALDI-TOF analysis did not generate sufficient peptides (two major peptides) to identify T3 from data bases, but the MS/MS data derived from post-source decay analysis of the two peptides identified T3 as phosphoglycerate kinase 1 (PGK-1 or primer recognition protein 2) ( Table I). The determined molecular weights and pI values for ADP-ribosylated T3 in two-dimensional SDS-PAGE are 28.6 kDa and 5.21, respectively, whereas the predicted molecular weights and pI values for PGK-1 are 44.7 kDa and 8.3, respectively. This indicates that T3 may be a proteolytic product of PGK-1. The location of the two peptides within the primary amino acid sequence of PGK-1 suggests that T3 is not a known splice variant. In addition to catalyzing the conversion of 3-phospho-D-glycerate and ATP to 1,3-diphospho-D-glycerate and ADP, PGK-1 also plays a structural role through association with microtubules (23). Whereas a majority of PGK-1 is in the cytoplasm of HeLa cells, PGK-1 complexes with annexin II (primer recognition protein 1) on the nuclear matrix to form a primer recognition complex, which plays a role in lagging strand DNA synthesis (24).
Recombinant Crk Was Specifically ADP-ribosylated by ExoT but Not ExoS-To address the specificity of ExoT for the ADPribosylation of Crk, recombinant Crk-I was expressed as a GST fusion protein (Fig. 5A). Following proteolytic digestion, affinity purification was used to obtain a purified form of Crk-I (Fig.  5A). SBTI was used as substrate for ExoS. Previous studies (17) had shown that ExoT ADP-ribosylated SBTI at ϳ0.2% the rate of ExoS. Consistent with previous results, ExoS ADP-ribosylated SBTI at a velocity that was much greater than ExoT (Fig.  5B), whereas ExoT, but not ExoS, ADP-ribosylated Crk-I in a FAS-dependent reaction (Fig. 5B). ADP-ribosylation of Crk-I required ExoT ADP-ribosyltransferase activity because the ADP-ribosyltransferase-deficient mutated protein, ExoT-(E383D/E385D), did not ADP-ribosylate Crk-I.
Linear Velocity for ADP-ribosylation of Crk-I by ExoT Is Similar to ADP-ribosylation Rate of SBTI by ExoS-By using Crk-I as ExoT substrate and SBTI as ExoS substrate, the specific activities of ExoS and ExoT in a linear velocity assay (Table II) were determined to be similar. At saturation, ϳ0.3 mol of NAD ϩ had been incorporated per mol of Crk-I, which did not increase with extended incubations. In contrast, under similar conditions, ϳ1 mol of NAD ϩ was incorporated per mol of SBTI by ExoS and continued to increase over extended incubations. The accuracy for determining the number of ADPribose bound to each protein is limited by several assumptions that were made during the determination, including the use of Coomassie staining to establish protein concentrations. These data suggest that although there appeared to be a limited number of sites for ExoT to ADP-ribosylate Crk-I, ExoS was capable of ADP-ribosylating multiple sites on SBTI.
Type III Delivered ExoT ADP-ribosylates a Limited and Unique Set of Mammalian Proteins Relative to ExoS-A teta-FIG. 5. Crk-I was ADP-ribosylated by ExoT but not ExoS. A, GST-Crk-I was purified by affinity chromatography and, then cleaved by thrombin. The cleaved GST and uncleaved GST-Crk-I were removed by glutathione-Sepharose 4B, and thrombin was removed by p-aminobenzamidine-agarose. Lane 1, purified GST-Crk-I. Lane 2, purified Crk-I after cleavage by thrombin. Lane M, molecular weight marker proteins. B, SBTI or Crk-I (3 M) was incubated with ExoT or ExoS (5 nM) with the presence of 0.1 mM [ 32 P]adenylate phosphate NAD ϩ alone or with FAS (500 nM) for 30 min. Reactions were stopped by addition of SDS sample buffer and boiling, and reactions were then subjected to SDS-PAGE followed by Coomassie Blue staining (stained gel is shown in lower panel) and autoradiography (exposed x-ray film is shown in upper panel). Migrations of known proteins are shown to the right of the gels. nolysin permeabilization assay has been used to detect the direct ADP-ribosylation of proteins in cultured cells by ExoS (21). CHO cells were used in this assay, because more radiolabel was incorporated into host proteins by type III cytotoxins in CHO cells relative to HeLa cells. In the present study, we observed that type III delivered ExoS ADP-ribosylated numerous host proteins, including Ras family proteins, and that ExoS was auto-ADP-ribosylated (Fig. 6A). In contrast, whereas type III delivered ExoT was also auto-ADP-ribosylated, only 4 -5 host proteins were ADP-ribosylated. Most of the ADP-ribosylated proteins by ExoS were in membrane fraction, whereas most proteins ADP-ribosylated by ExoT were in cytosolic fraction. A processed form of auto-ADP-ribosylated ExoS and ExoT was observed in the one-dimensional SDS-PAGE (Fig. 6A) and two-dimensional SDS-PAGE (Fig. 6B), which may occur during the tetanolysin permeabilization assay. As controls, CHO cells infected by P. aeruginosa transformed with pUCP vector or expressing ExoT(E383D/E385D) did not contain ADP-ribosylated proteins under the same assay conditions (Fig. 6A). Analysis of proteins that were ADP-ribosylated by ExoS and ExoT in two-dimensional SDS-PAGE did not identify proteins that were ADP-ribosylated by both ExoS and ExoT, indicating that type III delivered ExoS and ExoT ADP-ribosylate different host proteins. The T1, T2, and T3 proteins that were ADP-ribosylated by type III delivered ExoT in CHO cells had identical electrophoretic mobility with ADP-ribosylated Crk-I, Crk-II, and PGK-1 in HeLa cells, respectively. This indicates that Crk-I, Crk-II, and PGK-1 are also ADP-ribosylated by type III delivered ExoT in CHO cells. Similar to the results obtained in cultured HeLa cells (Fig. 3), the ADP-ribosylation of PGK-1 was less efficient than that of Crk-I and Crk-II by type III delivered ExoT in CHO cells.

DISCUSSION
Anti-internalization is an important mechanism for P. aeruginosa virulence. In most clinical isolates of P. aeruginosa, cytotoxicity is accompanied with anti-internalization activity. Current models propose that either ExoU or ExoS is responsible for the cytotoxicity elicited by P. aeruginosa, whereas the anti-internalization activity is attributed to ExoT (25). One strategy used by bacteria to inhibit internalization by eukaryotic cells is to reorganize the actin cytoskeleton through modulating Rho GTPase activity. The N terminus of ExoT is a RhoGAP for Cdc42, Rac1, and RhoA. The inactivation of Rho GTPases by ExoT contributes to anti-internalization of P. aeruginosa (7)(8)(9). Whereas the RhoGAP activity of ExoT was important for anti-internalization activity, it did not contribute to full anti-internalization capacity of ExoT because the ExoT RhoGAP mutation, ExoT(R149K), still displayed partial antiinternalization activity and induction of actin reorganization. This suggested that ExoT ADP-ribosyltransferase was also involved in anti-internalization and actin reorganization (8,18). By using two-dimensional SDS-PAGE combined with MALDI-MS, Crk-I and Crk-II were shown to be ADP-ribosylated by ExoT, which links the anti-internalization activity to ExoTmediated ADP-ribosylation. As adaptor proteins, Crk proteins contain both SH2 and SH3 domains (22). The SH2 domain of Crk proteins specifically interacts with the phosphotyrosine motif containing proteins such as paxillin, p130Cas, and FAK that are phosphorylated by Src kinase upon integrin activation. The SH3 domain of Crk proteins specifically interacts with proline-rich domain containing proteins that include DOCK180 and C3G. Although the biological function is not completely a ExoS and ExoT (5 nM) were incubated with SBTI or Crk-I (3 M), respectively, in the presence of 0.1 mM [ 32 P]-adenylate phosphate NAD ϩ with or without FAS (500 nM) for 2, 4, 8, and 16 min. The reactions were stopped with SDS sample buffer and subjected to SDS-PAGE followed by Coomassie Blue staining and autoradiography. The radioactive bands were excised and subjected to scintillation counting to measure the incorporation of radiolabel. The specific activity was determined as mol of NAD ϩ incorporation per mol of enzyme per min.
b ExoS and ExoT (45 nM) was incubated with SBTI or Crk-I (3 M), respectively, with the presence of NAD ϩ (0.1 mM) and FAS (500 nM). The stoichiometry NAD ϩ /target protein was determined at saturation points where incorporation of NAD ϩ did not increase with extended reaction time for ExoT/Crk-I. c Not detected. ADP-ribosylation activity was not greater than a control reaction that did not contain indicated toxin.

FIG. 6. Type III delivered ExoS and ExoT ADP-ribosylated unique proteins in CHO cells.
A, CHO cells were infected with P. aeruginosa PA103 (⌬exoU,exoT::Tc) transformed with pUCP vector (pUCP), or expressing ExoS-HA (ExoS), ExoT-HA (ExoT), or ExoT(E383D/E385D)-HA (ExoT(EE/DD)) at m.o.i. ϭ 8:1 (bacteria/CHO cells). At 3.5 h post-infection, infected cells were subjected to tetanolysin permeabilization assay followed by subcellular fractionation. Postnuclear supernatant (P), membrane (M), and cytosolic (C) fractions were applied to SDS-PAGE. The gel was transferred to polyvinylidene difluoride filter and subjected to autoradiography and Western blot using ␣-HA IgG as primary antibody. (Exposed x-ray film from autoradiography is shown.) B, post-nuclear supernatant from cells infected with P. aeruginosa PA103 ⌬exoU,exoT::Tc expressing ExoS-HA (ExoS) or ExoT-HA (ExoT) was applied to two-dimensional SDS-PAGE, using 7-cm Immobiline IEF strips with a pH 3-10 linear gradient, followed by autoradiography. Auto-ADP-ribosylated ExoS and ExoT, and proteins that were ADP-ribosylated by ExoT, are labeled with indicated arrows. In addition to the two major forms of ExoT, several lower molecular weight degradation products of ExoT (labeled with arrowheads) were observed. Arrows indicate the migration of five radiolabeled spots. Spot T1, T2, and T3 correspond with T1, T2, and T3 that were ADP-ribosylated by ExoT in HeLa cells. characterized, Crk proteins appear to play a central role in mediating phagocytosis, focal adhesion, and cell migration (26 -28). Upon activation of integrin, Src kinase phosphorylates a series of upstream factors of Crk, paxillin, FAK, and p130Cas, which subsequently recruit Crk proteins and activate phagocytosis through Crk-DOCK180-Rac1 signal cascade. Interestingly, the Crk-DOCK180-Rac1 is a conserved pathway mediating phagocytosis from Drosophila to human (29,30), whereas Crk-C3G-Rap1 appears to be a central signal cascade in regulation of integrin-matrix adhesion (31). Fleiszig and co-workers (32) showed that the deletion of csk (csk Ϫ/Ϫ ) in fibroblasts, a negative regulator of Src kinase, attenuated ExoT-dependent anti-internalization activity, which first correlated ExoT antiinternalization activity to Src-mediated phagocytosis.
In addition to inactivation of Rho GTPases by RhoGAP activity, the present study implicates the inhibition of the Crkmediated phagocytosis pathway by ADP-ribosylation as a mechanism for ExoT-dependent anti-internalization of P. aeruginosa. Similar mechanisms of anti-internalization were observed in Yersinia whose anti-internalization activity requires two type III secreted effectors, YopE and YopH. Like ExoT, YopE inhibited internalization through inactivation of Rho GTPases by its RhoGAP activity (33), whereas YopH, a protein-tyrosine phosphatase, inhibited internalization through de-phosphorylating focal adhesion proteins FAK, paxillin, and p130Cas, which then blocked the recruitment of Crk to the focal adhesion complex (34,35). This indicated that interference with Crk-mediated phagocytosis may be a common mechanism for bacterial anti-internalization activity.
Although ExoS and ExoT share 76% amino acid identity in their ADP-ribosylation domains, ExoS and ExoT appear to ADP-ribosylate different host proteins. ExoT elicited actin reorganization in mammalian cells without interfering Ras signaling, whereas the small molecular weight Rho GTPases that are ADP-ribosylated by ExoS were not ADP-ribosylated by ExoT (18,36). Consistent with previous results, the present study showed that ExoT displayed a different ADP-ribosylation target profile in vivo from ExoS in two-dimensional SDS-PAGE. Note that ExoS and ExoT have similar subcellular distribution in fractionation study, and their N-terminal RhoGAP domains target Cdc42, Rac, and RhoA both in vitro and in vivo (7,37). Therefore, the different in vivo target specificity of the ADP-ribosylation domains of ExoS and ExoT seems to be due to their different intrinsic affinities for substrate proteins, not due to different subcellular localization. This is supported by in vitro data showing that ExoS did not ADP-ribosylate Crk, and ExoT did not ADP-ribosylate SBTI. Therefore, it is reasonable to predict that ExoS and ExoT have different mechanisms for substrate recognition.
By using SBTI as a substrate, ExoT was characterized as a defective ADP-ribosyltransferase relative to ExoS. Identification of Crk as targets of ExoT ADP-ribosyltransferase allowed measuring the ADP-ribosylation activity of ExoT on a natural substrate. By using purified Crk-I as substrate, the rate of ADP-ribosylation catalyzed by ExoT was similar to ExoS for SBTI. Both ExoT and ExoS retained the requirement for FAS as an activator of catalysis. Interestingly, a previous study (17) showed that ExoT had lower NAD ϩ glycohydrolyase activity compared with ExoS, which indicates that ExoS and ExoT may use different mechanisms for enzymatic catalysis. The co-crystal structures of NAD ϩ complexed with ADP-ribosylating toxins including diphtheria toxin, pertussis toxin, exotoxin A of P. aeruginosa, and C3 toxin of Clostridium botulinum support a stepwise model for ADP-ribosylation, in which NAD ϩ binding to ADP-ribosylating enzymes induces a conformational change in ADP-ribosylating toxins, which allows protein substrate binding (38 -41). Recent biochemical studies showed that the NAD ϩ glycohydrolase activity of C3stau2 of C. botulinum is closely correlated with its NAD ϩ binding capacity (42). Therefore, current data suggests that ExoT might use a different mechanism for NAD ϩ and protein substrate binding in the ADP-ribosylation reaction.