SopE and SopE2 from Salmonella typhimurium activate different sets of RhoGTPases of the host cell.

The bacterial enteropathogen Salmonella typhimurium employs a specialized type III secretion system to inject toxins into host cells, which trigger signaling cascades leading to cell death in macrophages, secretion of pro-inflammatory cytokines, or rearrangements of the host cell cytoskeleton and thus to bacterial invasion. Two of the injected toxins, SopE and the 69% identical protein SopE2, are highly efficient guanine nucleotide exchange factors for the RhoGTPase Cdc42 of the host cell. However, it has been a puzzle why S. typhimurium might employ two toxins with redundant function. We hypothesized that SopE and SopE2 might have different specificities for certain host cellular RhoGTPases. In vitro guanine nucleotide exchange assays and surface plasmon resonance measurements revealed that SopE is an efficient guanine nucleotide exchange factor for Cdc42 and Rac1, whereas SopE2 was interacting efficiently only with Cdc42, but not with Rac1. Affinity precipitation of Cdc42.GTP and Rac1.GTP from lysates and characteristic cytoskeletal rearrangements of infected tissue culture cells confirmed that SopE is highly efficient at activating Cdc42 and Rac1 in vivo, whereas SopE2 was efficiently activating Cdc42, but not Rac1. We conclude that the translocated effector proteins SopE and SopE2 allow S. typhimurium to specifically activate different sets of RhoGTPase signaling cascades.

The bacterial enteropathogen Salmonella typhimurium employs a specialized type III secretion system to inject toxins into host cells, which trigger signaling cascades leading to cell death in macrophages, secretion of pro-inflammatory cytokines, or rearrangements of the host cell cytoskeleton and thus to bacterial invasion. Two of the injected toxins, SopE and the 69% identical protein SopE2, are highly efficient guanine nucleotide exchange factors for the RhoGTPase Cdc42 of the host cell. However, it has been a puzzle why S. typhimurium might employ two toxins with redundant function. We hypothesized that SopE and SopE2 might have different specificities for certain host cellular RhoGTPases. In vitro guanine nucleotide exchange assays and surface plasmon resonance measurements revealed that SopE is an efficient guanine nucleotide exchange factor for Cdc42 and Rac1, whereas SopE2 was interacting efficiently only with Cdc42, but not with Rac1. Affinity precipitation of Cdc42⅐GTP and Rac1⅐GTP from lysates and characteristic cytoskeletal rearrangements of infected tissue culture cells confirmed that SopE is highly efficient at activating Cdc42 and Rac1 in vivo, whereas SopE2 was efficiently activating Cdc42, but not Rac1. We conclude that the translocated effector proteins SopE and SopE2 allow S. typhimurium to specifically activate different sets of RhoGTPase signaling cascades.
The RhoGTPase subfamily of the Ras superfamily of small GTPbinding proteins comprises more than 10 different proteins (1). They act as molecular switches and cycle between GDP-bound (inactive) and GTP-bound (active) conformations (2). Activation and inactivation of RhoGTPases is controlled by guanine nucleotide exchange factors (GEFs), 1 GTPase-activating enzymes, and guanine dissociation inhibitors (3,4). The eukaryotic GEFs for Rho-GTPases share two common sequence motives, the DH (Dbl homology) and the PH (pleckstrin homology) domains, which are responsible for targeting, binding the RhoGTPases, and catalysis of RhoGTPase⅐GDP 3 RhoGTPase⅐GTP exchange (5). Only the active GTP-bound RhoGTPases can interact with downstream elements of signal transduction cascades mediating the cellular responses. GTPases of the Rho subfamily are central switches in the signaling cascades regulating motility, cellular adhesion, cell shape, cytokinesis, cell contraction, and gene expression (6,7). Each of the RhoGTPases can regulate a specific set of downstream signaling cascades, leading to activation of specific cellular functions (4). For example, in Swiss 3T3 cells, activation of Cdc42 leads to the formation of filopodia, activation of Rac1 induces formation of lamellipodia, and activation of RhoA leads to the formation of stress fibers and focal adhesions (8)(9)(10)(11). Thus, selective activation of specific RhoGTPases leads to specific cellular responses.
It has been shown that Cdc42 is a key element in the Salmonella-induced signaling cascades leading to transcriptional activation and bacterial invasion: disruption of Cdc42 signaling by transfection with dominant negative Cdc42 N17 alleles interferes with bacterial invasion and activation of c-Jun kinase and p21-activated kinase (PAK) signaling (23)(24)(25). Rac1 plays a less prominent role, and disruption of Rac1 signaling merely leads to reduced S. typhimurium invasion rates, whereas inhibition of RhoA signaling does not affect invasion at all (23). This suggests that the translocated effector proteins of S. typhimurium may preferentially address certain RhoGTPase signaling pathways.
Recent work has demonstrated that S. typhimurium (strain SL1344) relies mainly on a set of three translocated effector proteins to trigger invasion. A triple mutant S. typhimurium strain lacking SopB, SopE, and SopE2 is non-invasive, even though the SPI1 type III secretion system is still fully functional (22, 24, 26 -30). SopB has phosphatidyl inositol phosphatase activity and a sopB Ϫ mutant is less invasive than wild type S. typhimurium (28,31). Until now, however, the molecular mechanism explaining the role of SopB in triggering invasion has been unclear.
SopE and SopE2 of S. typhimurium are 69% identical (21,22,27). Besides the lack of any recognizable sequence similarity to proteins with DH or PH domains, both proteins are highly efficient GEFs for Cdc42 in vitro (22,24,32). In fact, the catalytic parameters of SopE-mediated guanine nucleotide exchange of Cdc42 are similar to those reported for the active domains of eukaryotic GEFs for members of the Ras superfamily (32). The specificity of SopE and SopE2 for other GTPases of the Rho subfamily has not been studied in detail. If both bacterial virulence factors had different preferences for different RhoGTPases, SopE and SopE2 might provide S. typhimurium with a means to differentially activate specific signaling pathways inside the host cell.
In the present study we have analyzed the specificity of SopE and SopE2 for the RhoGTPases Cdc42 and Rac1. Biochemical analyses of purified recombinant proteins and analysis of RhoGTPase activity in infected tissue culture cells revealed that SopE is an efficient activator for both Cdc42 and Rac1 in vitro and in vivo. In contrast, SopE2 efficiently activates Cdc42, but not Rac1. This demonstrates for the first time that expression of two homologous translocated effectors with GEF activity for RhoGTPases allows S. typhimurium to differentially activate specific signaling pathways within host cells.

MATERIALS AND METHODS
Bacterial Strains-All S. typhimurium strains used in this study have been described (26) and are isogenic derivatives of the virulent wild-type strain SL1344 (33). M516 (SL1344, ⌬sopB, sopE::aphT, sopE2::tet r ) lacks the three major effector proteins necessary for tissue culture cell invasion (26). Transformation of M516 with pM136 (pBAD24, which expresses SopE 1-240-M45 under the control of the native sopE promotor; Ref. 22) or with pM226 (pBAD24, which expresses SopE2 1-240-M45 under the control of the native sopE2 promotor; Ref. 22) complements the invasion defect. For tissue culture cell infection experiments, the bacteria were grown in high salt media as described (22).
Preparation of Recombinant Proteins-Preparation of recombinant proteins was performed essentially as described (32,34). Briefly, all proteins used in this study were overexpressed as GST fusion proteins, recovered from bacterial extracts by binding to glutathione-Sepharose 4B (Amersham Pharmacia Biotech) and either eluted with 20 mM glutathione (GST, GST-SopE 78 -240 , GST-SopE2 69 -240 , GST-Cdc42Hs  , and GST-Rac1  or cleaved off the column by digestion with thrombin protease (SopE 78 -240 and SopE2 69 -240 , Cdc42Hs  and Ha-Ras) or with factor Xa (Rac1  ). Proteins were concentrated by ultrafiltration (M r cut-off 8000), snap-frozen in liquid nitrogen, and stored at Ϫ80°C.
Due to design of the expression vectors, the proteins carry the following additional amino acids. GST-SopE 78 -240 carries PGISGGGGG-ILEFEM between the thrombin cleavage site and Leu-78 of SopE; GST-SopE2 69 -240 carries PGISGGGGGIL between the thrombin cleavage site and Gly-69 of SopE2; SopE 78 -240 carries the additional Nterminal amino acids GSPGISGGGGGILEFEM; SopE2 69 -240 carries the additional N-terminal amino acids GSPGISGGGGGIL; GST-Rac1  carries GIDPGAT between the factor Xa recognition site and Met-1 of Rac1; Rac1  carries the N-terminal amino acids GIDPGAT; Ha-Ras carries the additional N-terminal amino acids GS; GST-Cdc42Hs  carries RRASVGSKIISA between the thrombin recognition site and Met-1 of Cdc42. Cdc42Hs  carries the N-terminal amino acids GSRRASVGSKIISA.
The expression vector for the GST fusion protein with the Rac1 and Cdc42 binding region (aa 56 -272) of human PAK1B was generously provided by E. Sander and J. G. Collard (35), and purification of the protein bound to glutathione-Sepharose beads was performed as described (35).
Preparation of mGDP⅐Cdc42Hs  Complex-To remove the associated GDP, GST-Cdc42Hs  bound to a glutathione-Sepharose 4B column was washed with 10 ml of buffer D (50 mM Tris-HCl, pH 7.6, 100 mM NaCl, 2 mM EDTA, 2 mM DTT). Afterward, beads were incubated as a batch in the presence of a 2.5-fold molar excess of fluorescent mGDP (Molecular Probes, Netherlands) for 10 min at 22°C in buffer D. After addition of excess MgCl 2 , mGDP⅐Cdc42Hs  was cleaved off the column material using thrombin (Amersham Pharmacia Biotech) in buffer A (4°C; overnight) and unbound mGDP was removed by gel filtration chromatography. Fractions containing the mGDP⅐Cdc42Hs 1-192 complex were identified by fluorescence spectroscopy, pooled, concentrated by ultrafiltration (Millipore Ultrafree-15, M r cut-off 8000), snap-frozen, and stored at Ϫ80°C.
Preparation of mGDP⅐Rac1  Complex-For preparation of mGDP⅐Rac1  , we devised a new method based on the high stability of the SopE⅐Rac1 complex (32). 2 mg of GST-SopE 78 -240 was bound to a glutathione-Sepharose 4B column. Rac1  (in buffer A: 50 mM Tris-HCl, pH 7.6, 100 mM NaCl, 5 mM MgCl 2 , 2 mM DTT) was applied to the column, and unbound Rac1  was removed by washing with 10 ml of buffer A. The bound Rac1  was eluted as mGDP⅐Rac1  complex using buffer A (20°C) supplemented with 200 M mGDP and purified and concentrated as described above for mGDP⅐Cdc42Hs  .

SopE and SopE2 Display Differential Specificities for
Cdc42Hs and Rac1-Previous work had shown that SopE and SopE2 are efficient guanine nucleotide exchange factors for Cdc42 (22,32). Guanine nucleotide exchange factor activity for other RhoGTPases has not been studied in detail. Here, we have compared the GEF-activity of SopE 78 -240 and SopE2 69 -240 on Rac1 1-191 and Cdc42Hs      (Fig. 1, a (q) and d (ϫ)). In contrast, [ 3 H]GDP release from Rac1  was much faster in the presence of SopE 78 -240 (Fig. 1b, q) than in the presence of 1 M SopE2 69 -240 (Fig. 1e, ϫ). Therefore, SopE 78 -240 is a highly efficient GEF for Cdc42Hs  and Rac1  , whereas SopE2 69 -240 acts equally efficient on Cdc42Hs  , but is much less active on Rac1  in vitro.
Association/Dissociation Kinetics of the Complexes between SopE, SopE2, Rac1, and Cdc42Hs-To analyze the binding specificity of SopE and SopE2, we have measured the kinetics of formation and dissociation of the complexes between Cdc42 (or Rac1) and SopE2 (or SopE) using surface plasmon resonance. This technique allows one to study binding/dissociation kinetics by measuring the change in mass on the surface of a sensor chip.
We have also analyzed the dissociation of the complexes (Table I). However, in the absence of GDP, dissociation was slow and the dissociation rate constants are prone to experimental error and should be regarded as rough estimates. The GST-Cdc42Hs 1-192 ⅐SopE 78 -240 complex and the GST-Rac1 1-191 ⅐SopE 78 -240 complex are roughly equally stable. In contrast, dissociation of the GST-Rac1 1-191 ⅐SopE2 69 -240 complex is 6-fold faster than dissociation of the GST-Cdc42-Hs 1-192 ⅐SopE2 69 -240 complex (Table I). Overall, SopE 78 -240 binds with very similar equilibrium binding constants (K D ϭ k off /k a ) to GST-Cdc42Hs  and to GST-Rac1 1-191 (K D ϭ 3.1 ϫ 10 Ϫ10 M), whereas equilibrium binding of SopE2 69 -240 to GST-Cdc42Hs 1-192 is 40-fold stronger than equilibrium binding to GST-Rac1   (Table I).
In line with previous results for the GST-SopE 78 -240⅐Cdc42 ⌬C complex (32), dissociation of all complexes between GST-RhoGTPases and SopE 78 -240 or SopE2 69 -240 was acceler-ated more than 1000-fold in the presence of 20 M GDP and the dissociation reactions were completed in less than 5 s (data not shown). Identical dissociation curves were obtained when we employed 20 M GTP (data not shown). However, the dissociation kinetics in the presence of guanine nucleotides were too fast to allow an accurate analysis in order to detect differences between the dissociation rates of the complexes with GST-Cdc42Hs 1-192 and GST-Rac1  . of unbound mGDP (32,38,39). The kinetics and concentration dependence of mGDP dissociation from Cdc42Hs 1-192 ⅐mGDP or Rac1 1-191 ⅐mGDP was followed by fluorescence spectrometry (Fig. 3). In the SopE 78 -240 -mediated nucleotide exchange reactions, Cdc42Hs 1-192 ⅐mGDP nucleotide dissociation rate constants (v) reached a plateau at 20 -40 M and the Michaelis-Menten parameters (k cat ϭ 5 Ϯ 1 s Ϫ1 and K m ϭ 6 Ϯ 2 M; Table  II) were in the same order of magnitude as those reported for SopE 78 -240 -mediated nucleotide exchange on Cdc42 V12 ⅐mGDP (k cat ϭ 0.95 Ϯ 0.06 s Ϫ1 and K m ϭ 4.5 Ϯ 0.9 M; Ref. 32). It is unclear whether the slight differences might be attributable to effects of the G12V mutation of Cdc42 used in the earlier study (32). SopE2 69 -240 is an even more efficient GEF for Cdc42-Hs 1-192 than SopE 78 -240 (Table II).

Multiple Turnover Kinetics of SopE-and SopE2-mediated
Neither with SopE 78 -240 nor with SopE2 69 -240 did the observed nucleotide dissociation rate constants (v) of Rac1 1-191 ⅐mGDP reach a plateau at concentrations up to 50 M Rac1 1-191 ⅐mGDP; Fig. 3, a and b). Therefore, we could only estimate the catalytic efficiency of SopE 78 -240 and SopE2 69 -240 from the slopes of the linear plots shown in Fig. 3 (Table II). This indicates that the catalytic efficiency of SopE2 69 -240 (k obs / [Rac1  ⅐mGDP]) is about 6-fold lower than the catalytic efficiency of SopE 78 -240 .
Affinity Precipitation Assays to Determine Substrate Specificities of SopE and SopE2 in Vivo-The biochemical analyses presented above show that SopE is an efficient GEF for Rac1 and Cdc42 whereas SopE2 is an efficient GEF for Cdc42 but not for Rac1. It was of interest to also analyze this specificity in vivo. The levels of GTP-bound Rac1 and Cdc42 in tissue culture cells can be analyzed directly in an affinity precipitation assay (35,40). This assay is based on the ability of the Cdc42/Rac1binding domain (CD; aa 56 -272) of PAK-1 to specifically bind to activated Cdc42⅐GTP and Rac1⅐GTP, but not to inactive Cdc42⅐GDP or Rac1⅐GDP (35).
For the analysis of RhoGTPase activation by SopE and SopE2 during the course of an infection, we have employed the S. typhimurium SL1344 mutant M516 (sopE::aphT; sopE2::pM218; ⌬sopB; Ref. 26), which lacks the three major effector proteins necessary for host cell invasion (26). This strain has a fully functional SPI1 type III secretion system (26) but it is unable to activate Cdc42-or Rac1-signaling (see Fig. 4). To analyze the in vivo specificity of SopE and SopE2, we infected COS7 tissue culture cells for 40 min with the plasmidless control S. typhimurium strain M516 or with M516 complemented with expression vectors for epitope-tagged SopE 1-240-M45 (pM136) or SopE2 1-240-M45 (pM226; Ref. 22). COS7 cell lysates were subsequently analyzed using the GST-PAK-CD affinity precipitation assay (see "Materials and Methods"). M516 complemented with pM136 (SopE 1-240-M45 ) was able to efficiently activate Cdc42 and Rac1 (Fig. 4, lanes 2a and  2b). In contrast, M516 complemented with pM226 (SopE2 1-240-M45) only activated Cdc42 but not Rac1 (Fig. 4, lanes 3a and 3b). Control experiments verified that the observed differences were not attributable to different amounts of Rac1 or Sop-proteins present in the lysates (Fig. 4, lanes 2c, 3c, 1d, 2d, and  3d). In conclusion, these are in line with the results from the biochemical analyses and show that SopE2 has the capacity to specifically activate Cdc42 signaling in vivo, whereas SopE activates both Rac1 and Cdc42.
SopE and SopE2 Have Different Effects on the Actin Cytoskeleton of HUVECs-In mammalian cells specific activation of Rho, Rac, and Cdc42 leads to characteristic rearrangements of the actin cytoskeleton. Usually, activation of Cdc42 is associated with the formation of filopodia and activation of Rac1 with formation of lamellipodia ("ruffles"; Refs. 9 -11). Therefore, the differential signaling capacity of SopE and SopE2 might lead to different cytoskeletal rearrangements in infected tissue culture cells. In HUVECs, activation of Cdc42 and Rac1 induces formation of filopodia and lamellipodia, respectively (37,41). Therefore, we have infected HUVECs for 40 min with S. typhimurium strain M516 or with M516 complemented with pM136 (SopE 1-240-M45 ) or with pM226 (SopE2 1-240-M45 ). M516 did not induce actin cytoskeletal rearrangements (Fig. 5a). In contrast, M516 complemented with pM136 (SopE 1-240-M45 ) induced the formation of lamellipodia. Infection with M516 complemented with pM226 (SopE2 1-240-M45 ) induced formation of filopodia, whereas only a minority of infected cells formed lamellipodia (Fig. 5, a and b). The small number of cells forming lamellipodia ( Fig. 5b; M516 ϩ pM226) might be attributable to indirect activation of Rac1 by activated Cdc42 (11). In conclusion, these data are in line with our observation that SopE can efficiently activate Rac1 (and Cdc42) signaling, whereas SopE2 activates Cdc42 but not Rac1 in vivo.

DISCUSSION
It is well established that the translocated S. typhimurium protein SopE acts as an efficient GEF for host cellular Rho-   (24,32). This was of special interest, since SopE does not share any recognizable sequence similarity to eukaryotic GEFs or any other known proteins. Recently, it was discovered that S. typhimurium translocates an additional, 69% identical effector protein named SopE2 into host cells (21,22). SopE2 is also capable of activating Rho-GTPase signaling cascades leading to cytoskeletal rearrangements and bacterial entry (22). Why does S. typhimurium translocate two structurally and functionally similar effector proteins into host cells? We hypothesized that differences in the specificity of SopE and SopE2 for certain RhoGTPases might be one possible explanation. As activation of different RhoGTPases leads to specific changes in key cellular functions (i.e. actin cytoskeletal rearrangements, activation of transcription factors; Ref. 4), this might enable S. typhimurium to precisely manipulate host cell physiology. Therefore, we have analyzed in the present study the specificity of SopE and SopE2 for Cdc42 and Rac1. Indeed, we found SopE is a potent GEF for Cdc42 and for Rac1 both in vitro and in vivo, whereas SopE2 is much more active on Cdc42 than on Rac1. 1) In the absence of free guanine nucleotide, the equilibrium binding of SopE2 to Cdc42 is ϳ40-fold stronger than binding of SopE2 to Rac1. 2) SopE2-mediated in vitro guanine nucleotide exchange is ϳ10fold more efficient for Cdc42 than for Rac1. 3) Affinity precipitation assays revealed that, upon translocation into COS7 tissue culture cells, SopE2 activates Cdc42, but essentially no Rac1. 4) Translocation of SopE2 into HUVEC tissue culture cells induces actin cytoskeletal rearrangements characteristic for specific activation of Cdc42. In contrast, SopE interacted efficiently with both Cdc42 and Rac1 in vitro and in vivo. Analysis of a SopE-SopE2 chimeric protein (promoter region and aa 1-95 of SopE (ϭ translocation signal) fused to aa 96 -240 of SopE2 (ϭ catalytic domain)) verified that the observed specificities in vivo are really attributable to the catalytic Cterminal domains (data not shown). Altogether, SopE and SopE2 provide S. typhimurium with a means to specifically activate either Cdc42 and Rac1 or Cdc42 but not Rac1.

GTPases both in vitro and in vivo
Eukaryotic GEFs share a common functional unit (DH plus PH domain), which facilitates GTPase binding and catalysis. Yet, the eukaryotic GEFs for RhoGTPases display different specificities for different subsets of RhoGTPases (5,42). The three-dimensional structure of Rac1 complexed with the DH and PH domains of the eukaryotic GEF Tiam1 has identified the amino acid residues determining the binding specificity (43). Unfortunately, as SopE and SopE2 do not share any recognizable sequence similarity with eukaryotic GEFs, it is impossible to predict the amino acid residues responsible for a Determined from measurements at low RhoGTPase concentrations corresponding to linear ranges of the curves shown in Fig. 3. Cells were fixed, f-actin was stained with rhodamine-phalloidin (red), and bacteria were stained with a polyclonal ␣-Salmonella antiserum and a secondary ␣-rabbit fluorescein isothiocyanate antibody (green; see "Materials and Methods"). b, quantitative analysis of the SopE/SopE2-induced cytoskeletal rearrangements. Cells with altered actin cytoskeletal morphology (ϳ35% of all cells) were classified based on their morphological features: profound membrane ruffling (i.e. a, panel 2), weak filopodia formation (Ͻ20 filopodia/cell), pronounced filopodia formation (Ͼ20 filopodia/cell; i.e. a, panel 3). For each S. typhimurium strain, at least 100 cells with altered actin cytoskeletal morphology were evaluated in three independent experiments (experimental error: Ϯ10%). *, no cytoskeletal rearrangements observed. the different substrate preferences of SopE and SopE2 from the data of Worthylake et al. (43).
Do all Salmonella strains address Cdc42 and Rac1? Phylogenetic analyses have shown that sopE2 is present in all contemporary Salmonella lineages (21,22,26). Therefore, the capacity to directly activate Cdc42 but not Rac1 (via SopE2) inside cells of the animal host is common to all Salmonellae. In contrast, sopE is encoded in the genome of a bacteriophage, which is only present in very few Salmonella strains, including the S. typhimurium strain SL1344 used in this study (26,27,44,45). A second S. typhimurium strain (ATCC 14028) that is commonly used to study virulence mechanisms does not carry SopE⌽ and does not express SopE. 2 Therefore, the capacity to directly activate Rac1 signaling (via SopE) is not strictly required for Salmonella virulence per se. However, it is certainly possible that SopE2 can activate Rac1 via an indirect mechanism, as it is known that specific activation of Cdc42 will finally result in activation Rac1 in Swiss 3T3 fibroblasts (11).
Nonetheless, expression of SopE improves virulence. Interestingly, SopE-expressing S. typhimurium strains are associated with severe epidemics (44). It has been speculated that the improved epidemic virulence of these strains might simply be attributable to a higher "sopE" gene dosage and higher total amounts of SopE-like proteins delivered into host cells (44). However, the data presented here suggest that the improved virulence of sopE-positive S. typhimurium strains is much rather linked to the capacity of these strains to directly activate Rac1 and Cdc42 (via SopE) inside host cells.
Indeed, there are several lines of evidence from tissue culture experiments suggesting that direct activation of Cdc42 and Rac1 are needed to optimize host cell invasion. 1) Disruption of the sopE gene in S. typhimurium strain SL1344 leads to a 2-fold decreased invasiveness into COS7 tissue culture cells (27), whereas a sopE2 mutant is equally invasive as the wildtype SL1344 strain (22). This argues that the presence of SopE can fully compensate for the loss of SopE2, whereas SopE2 (possibly due to its inability to directly activate Rac1) cannot completely compensate for the loss of SopE. 2) Complementation of a non-invasive S. typhimurium strain (M516 ϭ SL1344, sopE Ϫ , sopE2 Ϫ , sopB Ϫ ), which lacks all three translocated effector proteins triggering bacterial entry with a SopE expression vector, is 2-fold more efficient than complementation with a SopE2 expression vector (26). 3) Disruption of Cdc42 signaling inside COS7 cells by transfection with Cdc42 N17 expression vectors is 2-fold more efficient at blocking S. typhimurium SL1344 invasion than disruption of Rac1 signaling via Rac1 N17 (23). In conclusion, these observations suggest that the capacity to directly activate Rac1 (in addition to Cdc42) via SopE improves S. typhimurium virulence.
Taken together, our results show for the first time that S. typhimurium can specifically activate different RhoGTPases of the host cell via the translocated effector proteins SopE and SopE2. This allows the bacteria to fine tune host cellular responses very precisely. Future work must address how SopE/ SopE2-triggered signaling may be further modulated by the other translocated effector proteins like the actin-binding protein SipA (19), the phosphatidyl inositol phosphatase SopB (31), or the GTPase-activating protein SptP (46). This will further advance our current knowledge of the intricate network of responses triggered by the translocated effector proteins of S. typhimurium in order to alter host cell signaling in a very precise manner.