Characterization of YopT Effects on Rho GTPases in Yersinia enterocolitica-infected Cells*

Pathogenic yersiniae employ a type III secretion system for translocating up to six effector proteins (Yersinia outer proteins (Yops)) into eukaryotic target cells. YopT is a cysteine protease that was shown to remove the C-terminal isoprenoid group of RhoA, Rac, and CDC42Hs. Here we characterized the cell biological and biochemical activities of YopT in cells infected with pathogenic Yersinia enterocolitica. Bacterially injected YopT located to cell membranes from which it released RhoA but not Rac or CDC42Hs. In the infected cells RhoA was dissociated from guanine nucleotide dissociation inhibitor-1 (GDI-1) and accumulated as a monomeric protein in the cytosol, whereas Rac and CDC42Hs remained GDI-bound. Direct transfer of isoprenylated RhoA to YopT and RhoA modification could be reconstituted in vitro by guanosine 5′-3-O-(thio)triphosphate loading of a recombinant RhoA·GDI-1 complex. Finally, in macrophages infected with a Yersinia strain selectively translocating YopT podosomal adhesion structures required for chemotaxis as well as phagocytic cups mediating uptake of yersiniae were disrupted. These findings indicate that bacterially translocated YopT acts on membrane-bound and GDI-complexed RhoA but not Rac or CDC42, and this is sufficient for disruption of macrophage immune functions.

Rho GTP-binding proteins constitute ideal targets for pathogens because they are crucial regulators of the actin cytoskeleton as well as control cell cycle, intracellular vesicle transport, and gene transcription (5)(6)(7). Most Rho family proteins are molecular switches that are "on" when bound to GTP and "off " when bound to GDP. In the GTP-bound state Rho proteins associate with and stimulate a variety of effector proteins, which include protein kinases, lipid kinases, and multidomain scaffolds (5). The GDP/GTP cycling of Rho GTPases is regulated by (i) their intrinsic GTPase activity, (ii) GTPase-activating proteins, which greatly increase the intrinsic GTPase activity, and (iii) guanine nucleotide exchange factors, which promote the exchange of GTP for bound GDP. Furthermore, binding to the guanine nucleotide dissociation inhibitor (GDI) shields the hydrophobic isoprenoid moiety at the C terminus of Rho GTPases and thereby allows their cytosol/membrane cycling. GDI also "freezes" Rho GTPases in their actual guanine nucleotide bound state (GDP or GTP) and blocks their interaction with other regulators and effectors (8,9). Currently available data suggest that release of Rho GTPases from GDI is promoted by diverse molecules, including phosphatidylinositol bisphosphate (PIP2), proteins of the ezrin-radixin-moesin (ERM) family, and unsaturated fatty acids as well as by GDI phosphorylation (10 -12).
It has been intriguing to learn that many bacterial toxins and modulins exploit the regulatory mechanisms of Rho GT-Pases for their own purposes (13). For instance, YopE of Yersinia inactivates Rho, Rac, and CDC42 in vitro by acting as a GTPase-activating protein (14,15). Interestingly, when YopE is injected into cells by yersiniae, it is remarkably specific, inactivating selectively Rac-mediated signal pathways (15). We have previously reported that YopT of Yersinia induces covalent modification and inactivation of RhoA, as demonstrated by a more acidic isoelectric point of RhoA and its release from membranes (16). Subsequently, Shao et al. (17) provide conclusive evidence that YopT is a cysteine protease removing the C-terminal isoprenoid group of Rho, Rac, and CDC42. It was demonstrated biochemically that YopT cleaves just before the C-terminal cysteine, to which the isoprenoid moiety is linked (18).
The present study was undertaken to investigate the effects of YopT in a cell infection model. We show that bacterially injected YopT localizes to cellular membranes, from which it releases RhoA but not Rac or CDC42. YopT also specifically releases RhoA from GDI and thereby traps it in the cytosol. These YopT effects lead to disruption of actin structures required for phagocytosis and chemotaxis of macrophages.

EXPERIMENTAL PROCEDURES
Materials and Proteins-The expression plasmids for GST-YopT and GST-YopTC139S were generous gifts of Dr. J. Dixon and F. Shao (University of Michigan). GST-YopT and GST-YopTC139S were expressed in Escherichia coli and purified on glutathione-Sepharose beads as described (17). Purity of GST proteins was checked by SDS/ PAGE, and protein concentrations were determined with the Bio-Rad protein assay using bovine serum albumin as the standard. For production of recombinant RhoA⅐GDI complex the cDNAs of human RhoA and bovine RhoGDI (GenBank TM accession numbers X05026 and X52689) were ligated into the baculovirus transfer vector pVL1393. Recombinant baculoviruses were produced as described (19). Recombinant RhoA/RhoGDI heterodimers were produced by coinfection of Spodoptera frugiperda (Sf9) cells with recombinant baculoviruses encoding human RhoA and bovine RhoGDI, respectively. The soluble complex was purified by sequential chromatography on Superdex 75 prep grade and Mono Q (Amersham Biosciences) (19). The fractions were assayed for their reactivity with polyclonal antibodies reactive against RhoA (sc-179, Santa Cruz Biotechnology, Heidelberg, Germany) and RhoGDI (sc-359, Santa Cruz Biotechnology) on immunoblots. The purity of the heterodimer was at least 90%, as judged by analysis on silver-stained SDS-PAGE gels.
Cell Culture, Bacterial Infection, and Attachment of Invasin-coated Latex Beads-human umbilical vein endothelial cells (HUVEC) were isolated and cultured as described previously (22) with the following modifications. Cells harvested from human umbilical cords were seeded onto gelatin-coated plastic culture flasks and cultured in endothelial cell growth medium containing endothelial cell growth supplement and 2% fetal bovine serum (PromoCell, Heidelberg, Germany). For actin staining, HUVEC were plated at a density of 2 ϫ 10 4 cells/cm 2 onto gelatin-coated glass coverslips and grown for 2-3 days. Primary human macrophages were isolated and cultivated as described previously (23). HeLa cells were grown in RPMI 1640 medium, 10% fetal bovine serum in plastic culture flasks at 37°C, 5% CO 2 in a humidified atmosphere. For cell infection yersiniae were grown overnight in Luria Bertani (LB) medium at 27°C under appropriate antibiotic selection. Shortly before infection, HeLa cell growth medium was aspirated and replaced by prewarmed LB medium supplemented with 5% fetal bovine serum without antibiotics. Yersiniae in LB medium were added to HeLa cells for 2-3 h at a multiplicity of infection of 50 -100. HUVEC and human macrophages on glass coverslips were infected for 2 h by the addition of yersiniae to the respective cell growth medium followed by centrifugation (115 ϫ g, 37°C, 3 min) to obtain a multiplicity of infection of 50 -100. Invasin-coated latex beads were prepared and attached to the human macrophages as described previously (23).
Microinjection, Actin, and Bacterial Staining-Microinjection was performed using transjector 5246 (Eppendorf, Germany) and a Compic Inject micromanipulator (Cell Biology Trading, Hamburg, Germany). GST-YopT was injected at a concentration of 1 mg/ml. Cells were incubated for 1 h post-injection. Control injections were performed with GST alone, and injected cells were identified by detection of coinjected rat IgG (5 mg/ml) with fluorescein isothiocyanate-labeled goat anti-rat IgG antibody (both from Dianova, Hamburg, Germany). To double-stain actin and bacteria, infected HUVEC on glass coverslips were stimulated with thrombin (1 unit/ml for 2 min), fixed with 3.7% formaldehyde in PBS for 10 min, permeabilized for 5 min in ice-cold acetone, and airdried. Unspecific binding sites were blocked for 10 min with PBS/pH 7.4, 1% bovine serum albumin, and cells were reacted with rabbit polyclonal anti-Myf antibody (1:500) for 45 min and incubated for 30 min with fluorescein isothiocyanate-labeled anti-rabbit IgG (1:100; Molecular Probes) and rhodamine phalloidin (1:20; Molecular Probes) as described (15). Microinjected or infected human macrophages on glass coverslips were stained for actin using rhodamine phalloidin as described (23). Coverslips were mounted in Mowiol (Calbiochem) containing 1,4-diazobicyclo [2,2,2]-octane (DABCO) as an antifading agent. All steps were performed with 3 washes in PBS, pH 7.4, 1% bovine serum albumin between antibody incubations and before mounting in Mowiol. Samples were examined with a Leitz DM fluorescence microscope and recorded as digital images using a spot camera (Visitron Systems, Puchheim, Germany).
Cell Fractionation and in Vitro Assay for Membrane Release of Rho GTPases-Control or infected HeLa cells or HUVEC were recovered by scraping, washed once with ice-cold PBS, pH 7.4 and once with ice-cold buffer A containing 20 mM Tris-HCl, pH 7.5, 1 mM EGTA, 1 mM dithiothreitol, Complete protease inhibitors (1 tablet for 50 ml; Roche Applied Science), and disrupted using a Dounce homogenizer. The homogenate was centrifuged for 10 min at 400 ϫ g to remove cellular debris. The remaining homogenate was centrifuged for 30 min at 100,000 ϫ g in a TLA-100 bench-top centrifuge (Beckman Instruments). All procedures were performed at 4°C. The supernatant, containing the cytosolic fraction, and the pellet, containing the crude membrane fraction, were snap-frozen and stored at Ϫ80°C. Protein contents of cytosolic and membrane fractions were measured by Bio-Rad protein assay using bovine serum albumin as standard.
To determine the membrane/cytosol distribution of translocated YopT, cells were infected with WAC(pYLCRϩT) for 2 h followed by washing and further incubation with 100 g/ml gentamicin for 1 h. Thereafter, cells were treated with 100 g/ml of proteinase K (Sigma) in RPMI 1640 medium for 15 min at 37°C following an established protocol (24). Cells detached during the proteinase K treatment were collected by centrifugation, and cell fractionation was performed as described above.
For testing membrane release of Rho GTPases in vitro, HeLa cell membranes (200-g aliquots) were washed once in buffer A, resuspended in the same buffer, and incubated with 2 g of GST-YopT or GST-YopTC139S for 30 min at 37°C. The membrane suspension was centrifuged at 100,000 g for 30 min, and the resulting pellet and supernatant were subjected to anti-RhoA Western blot.
Non-denaturing Gel Electrophoresis-Samples of HeLa cytosol (10 l, 20 -100 g of protein) in buffer A were added to 2 l of loading buffer (25 mM Hepes, pH 7.4, 50% glycerol, bromphenol blue). Gels (6% of a 37.5:1 solution of acrylamide:bisacrylamide, 25 mM Hepes, pH 7.4, 5 mM NaCl) were prerun for 30 min at a constant voltage of 80 V using 25 mM Hepes, pH 7.4, 5 mM NaCl as running buffer. Samples were applied to gels and run for 150 min at 80 V.
Gel Filtration Chromatography-Cytosol of HeLa cells containing 4 -17 mg of protein in 200 l of buffer A (see above) was loaded onto a Superdex 75 HR 10/30 column (Amersham Biosciences) equilibrated with PBS, pH 7.4, 50 mM NaCl. Calibration of the column was performed with gel filtration molecular weight markers (Sigma) containing bovine serum albumin (66 kDa), bovine erythrocyte carbonic anhydrase (29 kDa), horse heart cytochrome c (12.4 kDa), and bovine lung aprotinin (6.5 kDa). The flow rate was 300 l/min, and the fraction size was 200 l. Approximately 10 l of each fraction was analyzed by Western blotting.
In Vitro Modification and GST Pull-down Assay Using RhoA⅐GDI Complex-For in vitro modification of RhoA, aliquots (3 g/l) of RhoA⅐GDI complex in buffer B containing 10 mM Tris, pH 7.5, 6 mM EDTA, 0.5 mM dithiothreitol, 0.05 mM phenylmethylsulfonyl fluoride, 2 mM MgCl 2 , 250 mM NaCl were incubated or not with 200 M GTP␥S or GDP␤S for 30 min at 25°C. The reaction was stopped by the addition of 20 mM MgCl 2 . Designated samples were treated with 100 M PIP2 and 2 g of YopT for 30 min at 37°C. For GST pull-down RhoA⅐GDI complex (28 g of protein/10 l) was loaded or not with 200 M GTP␥S and treated or not with 100 M PIP2 as described above. Thirty g of GST or 100 g of GST-YopTC139S were bound to 50 l of glutathione-Sepharose beads in a buffer containing 20 mM Tris, pH 7.5, 1 mM EGTA, 3 mM MgCl 2 , and 1 mM dithiothreitol for 2 h at 4°C and added to the samples of RhoA⅐GDI complex for 45 min at 37°C. Beads were washed 5 times with PBS containing 0.1% Nonidet P-40 and boiled in Laemmli sample buffer, and proteins were run on polyacrylamide gels. GST pull down from cytosol was performed identically except that 1 mg of HeLa cell cytosol was used as a source for RhoA⅐GDI complex.

A Yersinia Strain That Translocates Selectively YopT-We
have previously shown that infection of cells with YopT-expressing but not with YopT-knockout Y. enterocolitica causes modification and membrane release of the GTP-binding protein RhoA (16). To investigate YopT in the absence of other Yops known to affect Rho GTPases, yet under conditions whereby "life-like" amounts of YopT protein are injected by the Yersinia TTSS, we constructed a Yersinia strain named WAC (pYLCRϩT) that translocates YopT as sole effector (Table I). To this end, Yersinia strain WAC that had been cured of the virulence plasmid pYVO8 was recomplemented with two plasmids, one containing genes encoding the complete type III secretion and translocation machinery (lcr region) and the other carrying the genes encoding YopT and its chaperone SycT of pYVO8. To test WAC(pYLCRϩT), HUVEC were infected with this strain, stimulated with thrombin, and then analyzed for formation of actin stress fibers. As control, HUVEC were infected with strain WAC(pYLCR), which expresses the complete Yersinia secretion system (TTSS) but no YopT (Table I).
In HUVEC infected with WAC(pYLCRϩT) stress fiber formation was completely blocked, whereas HUVEC infected with WAC(pYLCR) formed abundant stress fibers (Fig. 1). Because stress fibers are controlled by RhoA in HUVEC, we conclude that WAC(pYLCRϩT) translocates functional YopT.
Bacterially Translocated YopT Localizes to Cell Membranes and Extracts RhoA but Not Rac or CDC42 from the Membranes-Within most cells 5-20% of the total cellular RhoA, Rac, or CDC42 proteins are found on membranes, whereas the remaining 80 -95% are cytosolic. It is well accepted that all of the cytosolic RhoA, Rac, and CDC42 molecules form tight 1:1 heterodimers with Rho GDIs. In mammals three different GDIs have been characterized and named GDI-1, -2, and -3. Of these GDI-1 is expressed ubiquitously and interacts with Rho, Rac, and CDC42 (9). Although YopT has been reported to modify and release Rho, Rac, and CDC42Hs from membranes upon overexpressing of these proteins in cells, the specificity of YopT toward endogenous Rho GTPases in infected cells has not been determined (17). Furthermore it remained unclear whether YopT can also act on cytosolic, GDI-complexed Rho proteins.
As a first step to answer these questions we determined the subcellular distribution of bacterially translocated YopT. HeLa cells were infected with Yersinia strain WAC(pYVO8)⌬TϩT, which expresses full-length YopT fused to the N-terminal 138 amino acids of YopE in addition to wild type YopH, -O, -P, -E, and -M (16). In contrast to strain WAC(pYLCRϩT), which is partly internalized by cells, WAC(pYVO8)⌬TϩT is completely extracellular due to the concerted action of the Yops. Moreover, the tagged YopT expressed by this strain can be detected by anti-YopE antibody. WAC(pYVO8)⌬TϩT-infected cells were subjected to Dounce homogenization, and the resulting membrane and cytosolic fractions were investigated by anti-YopE immunoblot. To exclude that whole bacteria or free YopT proteins adhered to the outside of the cell membrane, extracellular proteins were digested with proteinase K just before homogenization (24). As shown in Fig. 2A, bacterially translocated YopT could be detected in the membranes but not in the cytosolic fractions of HeLa cells and HUVEC. Next, we tested the membranes of the WAC(pYLCRϩT)-infected cells for the presence of endogenous RhoA, Rac, or CDC42Hs. As described before, RhoA was absent from the membranes of infected cells. Interestingly, however, Rac and CDC42Hs remained membrane-associated (Fig. 2B). Release of endogenous Rho GT-Pases was also tested by adding GST-YopT to freshly prepared cellular membranes. GST-YopT fully extracted endogenous RhoA, only a minute amount of endogenous CDC42, and no Rac under the conditions used here. The catalytically inactive GST-YopTC139S did not extract RhoA (Fig. 2C). These data suggest that bacterially translocated YopT associates mostly with cell membranes from which it extracts RhoA but not Rac or CDC42.
YopT Releases RhoA but Not Rac or CDC42Hs from GDI-1-To assay whether YopT releases Rho GTPases from GDI, we established an assay involving non-denaturing gel electrophoresis followed by Western blotting. In this assay RhoA of the control cytosol and of the cytosol of cells infected with the YopT-deleted strain WAC(pYVO8)⌬T was detected in one band. In comparison, an upper band of RhoA representing the YopT-modified form ap-   (Fig. 3A). It can be seen from the relative proportion between the lower and upper RhoA bands that WAC(pYLCRϩT) and WAC(pYVO8) induced a partial (about 70%) RhoA modification, whereas WAC(pYVO8)⌬TϩT modified all of the RhoA. This phenomenon is likely due to the modest overproduction of YopT by WAC(pYVO8)⌬TϩT. Non-denaturing gel electrophoresis should not dissociate the tight complex between Rho GTPases and GDI. To test for comigration of RhoA with GDI-1, the anti-RhoA immunoblot of the non-denaturing gel was reprobed with anti-GDI-1 antibody. GDI-1 was detected in a lower and an upper band. The lower GDI-1 band comigrated with the lower band of unmodified RhoA, representing the cytosolic RhoA-GDI complex. However, the upper GDI-1 band did not comigrate with YopTmodified RhoA (Fig. 3A). The upper GDI-1 band did comigrate with the upper Rac band and disappeared together with Rac1 from this position upon PIP2 treatment of cytosol, suggesting that it represents a Rac1⅐GDi complex (Fig. 3B and not shown). Next we tested the effect of WAC(pYLCRϩT) on GDI association of Rac1 and CDC42Hs. In the non-denaturing gels Rac1 was present in a lower and an upper band, which comigrated with the observed lower and upper bands of GDI, respectively. The two Rac1 bands were maintained, and no additional Rac1 band appeared upon infection of cells with WAC(pYLCRϩT) (Fig. 3B). Like RhoA, CDC42Hs was detected in a lower band comigrating with GDI-1. No additional CDC42Hs band appeared upon infection of cells with WAC(pYLCRϩT) (Fig. 3B). Together these results suggest that YopT leads to specific disruption of the RhoA-GDI-1 association.
To test whether the YopT-modified RhoA is monomeric or bound to another protein, cytosol of control or infected HeLa cells was subjected to gel filtration chromatography. Immunoblots of the gel filtration fractions confirmed that in control cytosol all of the RhoA coeluted with GDI at about 50 -60 kDa, consistent with a 1:1 RhoA⅐GDI complex. In contrast, in the cytosol of WAC(pYVO8)⌬TϩT-infected cells all of the RhoA eluted as a 23-25-kDa protein (Fig. 4). Consistent with the results of the non-denaturing gels, about 50 -70% of the RhoA in the WAC(pYLCRϩT)-infected cells eluted as a 23-25-kDa protein, whereas the remaining RhoA was still GDI bound (data not shown). In contrast to RhoA, Rac-1 and CDC42Hs remained complexed to the GDI-1 in the WAC(pYVO8)⌬TϩTinfected cells. These data indicate that RhoA is specifically released from GDI in the infected cells and accumulates as free, monomeric protein in the cytosol.
Reconstitution of YopT-induced RhoA Modification Using Purified RhoA⅐GDI Complex-As a first step to investigate whether YopT can directly act on RhoA complexed to GDI, we added recombinant GST-YopT to a membrane-free cytosol preparation of HeLa cells and employed non-denaturing gel electrophoresis. As demonstrated in Fig. 5A, GST-YopT was able to release RhoA from GDI, and this coincided with RhoA modification, as shown by a higher mobility in SDS-PAGE gels. GST-C139SYopT had no effect on GDI association or SDS/ PAGE mobility of RhoA (Fig. 5A). This suggests that RhoA⅐GDI complex from crude cytosol is a direct target for YopT.
To better define the conditions under which YopT works on RhoA bound to GDI we purified recombinant RhoA⅐GDI-1 complex from Sf9 insect cells. Initial experiments showed, however, that YopT could neither modify RhoA nor could YopTC139S bind to RhoA within the recombinant RhoA⅐GDI complex (Figs. 5B and 6). Furthermore, in contrast to the RhoA⅐GDI complex from crude cytosol, the complex partially purified from HeLa cell cytosol by gel filtration chromatography was no substrate for YopT (not shown). We therefore speculated that factors lost during purification may be required by YopT to access RhoA in the GDI complex. We estimated that those factors may include phosphoinositides such as PIP2, which among other lipids have been shown to release Rac from GDI complex (10,29). In fact, as demonstrated in Fig. 5B PIP2 partially restored the ability of YopT to modify RhoA in the recombinant RhoA⅐GDI complex. PIP2 alone was not able to release RhoA from GDI.
RhoA in complex with GDI can be GDP-or GTP-bound. Whereas the RhoA-GDP⅐GDI complex is highly stable, loading of GTP into the complex was shown to promote translocation of RhoA to liposomes and cell membranes (25,26). We therefore tested the effect of GTP␥S or GDP␤S loading on the ability of YopT to modify RhoA. The results in Fig. 5B clearly show that GTP␥S loading into RhoA⅐GDI complex enabled YopT to modify RhoA, whereas GDP␤S loading had no effect.
Last, we wanted to show direct transfer of RhoA from GDI to YopT. For this purpose we performed pull-down experiments using the "substrate trapping" GST-YopTC139S protein. The results depicted in Fig. 6 indicate that RhoA directly translocated from recombinant RhoA⅐GDI complex to YopT C139S in the presence but not in the absence of GTP␥S. It was also interesting to note that upon PIP2 treatment, RhoA did not transfer to YopTC139S. Moreover, when the complex was treated with both PIP2 and GTP␥S, only minor transfer of RhoA to YopTC139S occurred. Similar results were obtained when crude cytosol from HeLa cells was taken as a source of RhoA⅐GDI complex (Fig. 6). Unfortunately GDI-1 from the recombinant complex and GDI from cytosol shows unspecific binding to native glutathione beads and to GST-coated or GST fusion protein-coated beads irrespective of the washing procedure. However, because the amount of GDI associated with GST-YopTC139S-coated beads was not higher than that associated with GST-coated beads, one can conclude that YopT does not bind the RhoA⅐GDI-1 complex but to RhoA alone (data not shown). We conclude that GTP loading allows prenylated RhoA to transfer from GDI to YopT. In comparison, although PIP2 treatment allows YopT to modify RhoA in the GDI complex, high affinity binding of RhoA to YopTC139S does not seem to take place under these conditions.
Effects of YopT on Macrophage Immune Functions-Among the most important target cells of enteropathogenic yersiniae in vivo are macrophages. The concerted action of the effector Yops is thought to block macrophage phagocytosis and chemotaxis as part of the immune-evasive strategy of Yersinia (2). We therefore tested the effect of YopT on formation of cytoskeletal structures associated with phagocytosis and chemotaxis. It was shown that actin-rich phagocytic cups triggered by the surface adhesin invasin mediate phagocytosis of yersiniae (23). Using human macrophages we found that microinjection of GST-YopT or infection with WAC(pYLCRϩT) completely prevented formation of invasin-triggered phagocytic cups ( Fig. 7 and data not shown). We finally tested the effect of YopT on actin-rich adhesion structures called podosomes, which are required for the chemotactic and invasive capabilities of macrophages (27). Again, we found that bacterially translocated or microinjected YopT disrupted the podosomal adhesion structures (Fig. 7 and data not shown). Interestingly, the Yersinia-infected cells lacking podosomes displayed numerous actin ruffles. Because formation of actin ruffles requires Rac activity, this confirms that Rac is not affected by YopT. Collectively, these data suggest that YopT activities on RhoA are sufficient to block macrophage functions important for immune defense. DISCUSSION YopT is a cysteine protease removing the C-terminal isoprenoid group of Rho, Rac, and CDC42 in vitro and after overexpression of the reaction components in cells (17,18). The hydrophobic isoprenoid group of Ras-like small GTP-binding proteins mediates membrane attachment and interaction with effectors and regulators such as the GDI proteins. We show here that bacterially translocated YopT localizes to cell membranes, from which it removes RhoA but not Rac or CDC42Hs. Possibly, bacterially translocated YopT and Rac/CDC42 bind to different membrane compartments. It has been shown that RhoA is found in the plasma membrane, whereas the majority of Rac and CDC42 are on endosomal and Golgi membranes, respectively (28). In infected cells and upon the addition of recombinant YopT to cytosol, RhoA was released from GDI, but Rac and CDC42 remained GDI-bound. Considering that in our cell infection model the amount of translocated YopT is already at or above the values that could be obtained during human Yersinia infection, one may assume that YopT is specific for RhoA in vivo.
Using purified RhoA⅐GDI-1 complex we found that YopT could not cleave RhoA unless the complex was treated with PIP2 or GTP␥S. In an earlier study PIP2 was shown to greatly increase the ability of C3-transferase of Clostridium botulinum to ADP-ribosylate RhoA in GDI complex (29). Another study showed that loading of GTP␥S onto recombinant RhoGDI complex promoted translocation of RhoA to liposomes (25). However, because transfer of prenylated RhoA to YopT could only be detected after GTP␥S loading but not after PIP2 treatment, these two mechanism of "loosening" the RhoA⅐GDI association seem to be different. Potentially PIP2 displaces the prenyl group of RhoA from its binding pocket in the GDI molecule, and YopT can then attack the partially opened RhoA⅐GDI complex. Yet the affinity of YopTC139S for this complex may be too low for pull down.
Combining our data and data from recently published work (17,18), we propose the following scenario of YopT action. Upon infection of cells with yersiniae, the YopT protein is translocated to the plasma membrane, where it binds and cleaves RhoA, thereby releasing it from the membrane. Membranous YopT, in cooperation with the membrane lipid PIP2 and/or Rho guanine nucleotide exchange factors, can also attack the high affinity soluble Rho⅐GDI complex in the cytosol. The cleaved Upper panels, primary human macrophages were microinjected with GST-YopT and then reacted with Yersinia invasin-coated latex beads for 10 min. Left, phagocytic cups (arrows) were visualized by actin staining with rhodamine phalloidin. Right, the microinjected cell shows green fluorescence due to coinjected fluorescein isothiocyanate-dextran. Lower panels, primary human macrophages were infected with WAC(pYLCR) or WAC(pYLCRϩT) for 2 h and then stained for actin using rhodamine phalloidin. Podosomes appearing as actin dots were disrupted by infection with WAC(pYLCRϩT) but not with WAC(pYLCR). WAC(pYLCRϩT) infected cells demonstrated pronounced ruffling (arrows).
RhoA accumulates in the cytosol and is unable to associate with GDI or translocate to membranes. Moreover, after modification by YopT, the RhoA can no more associate with effector proteins (30). 2 Thus, as a consequence of the multiple functional defects of YopT-modified RhoA, the latter most likely cannot fulfill signaling functions. However, we can already notice a RhoA negative phenotype, i.e. disruption of actin stress fibers, when just a part (30 -50%) of RhoA is modified by YopT (16). Therefore, the YopT-modified RhoA may have dominant negative activity, as was proposed recently for a RhoA form in which the C-terminal polybasic region was cleaved off by calpain (31).
Recently the contribution of individual Yops to the antiphagocytic function of yersiniae was evaluated (32). It was found that bacteria lacking YopT were less resistant to phagocytosis. We show here that YopT can block the formation of actin-rich phagocytic cups induced by the Yersinia adhesin invasin. This YopT effect is rather slow, requiring at least 15-30 min of infection, but due the nature of the modification it is longlasting. In comparison, YopE also disrupts invasin-triggered phagocytic cups, which starts immediately after bacteria cell contact but is rather transient. It seems likely, therefore, that YopE and YopT act synergistically in antiphagocytosis by inducing immediate and prolonged inhibition of Rho GTPases, respectively. Considering that YopT, YopE, and YopO act on overlapping sets of Rho GTPases, more examples of their cooperation on the molecular level can be expected in the future. Last, we can detect in cells a form of RhoA that has an identical behavior in isolectric focusing as YopT-modified RhoA (33). It is, therefore, conceivable that there is an endogenous "YopTlike" activity in human cells that is capable of regulating RhoA. This exemplifies once more that investigation of bacterial virulence factors not only improves our understanding of bacterial pathogenicity but may also reveal hitherto unrecognized regulatory principles in eukaryotic cells.