Enteropathogenic Escherichia Coli Activates the RhoA Signaling Pathway Via the Stimulation of GEF‐H1
Takeshi Matsuzawa,Asaomi Kuwae,Sei Yoshida,Chihiro Sasakawa,Akio Abe
DOI: https://doi.org/10.1038/sj.emboj.7600359
2004-01-01
Abstract:Article19 August 2004free access Enteropathogenic Escherichia coli activates the RhoA signaling pathway via the stimulation of GEF-H1 Takeshi Matsuzawa Takeshi Matsuzawa Department of Infection Control and Immunology, Laboratory of Bacterial Infection, Kitasato Institute for Life Sciences, Kitasato University, Shirokane, Minato-ku, Tokyo, Japan The Kitasato Institute, Shirokane, Minato-ku, Tokyo, Japan Search for more papers by this author Asaomi Kuwae Asaomi Kuwae Department of Infection Control and Immunology, Laboratory of Bacterial Infection, Kitasato Institute for Life Sciences, Kitasato University, Shirokane, Minato-ku, Tokyo, Japan The Kitasato Institute, Shirokane, Minato-ku, Tokyo, Japan Search for more papers by this author Sei Yoshida Sei Yoshida Department of Microbiology and Immunology, Institute of Medical Science, University of Tokyo, Shirokanedai, Minato-ku, Tokyo, Japan Search for more papers by this author Chihiro Sasakawa Chihiro Sasakawa Department of Microbiology and Immunology, Institute of Medical Science, University of Tokyo, Shirokanedai, Minato-ku, Tokyo, Japan Search for more papers by this author Akio Abe Corresponding Author Akio Abe Department of Infection Control and Immunology, Laboratory of Bacterial Infection, Kitasato Institute for Life Sciences, Kitasato University, Shirokane, Minato-ku, Tokyo, Japan The Kitasato Institute, Shirokane, Minato-ku, Tokyo, Japan Search for more papers by this author Takeshi Matsuzawa Takeshi Matsuzawa Department of Infection Control and Immunology, Laboratory of Bacterial Infection, Kitasato Institute for Life Sciences, Kitasato University, Shirokane, Minato-ku, Tokyo, Japan The Kitasato Institute, Shirokane, Minato-ku, Tokyo, Japan Search for more papers by this author Asaomi Kuwae Asaomi Kuwae Department of Infection Control and Immunology, Laboratory of Bacterial Infection, Kitasato Institute for Life Sciences, Kitasato University, Shirokane, Minato-ku, Tokyo, Japan The Kitasato Institute, Shirokane, Minato-ku, Tokyo, Japan Search for more papers by this author Sei Yoshida Sei Yoshida Department of Microbiology and Immunology, Institute of Medical Science, University of Tokyo, Shirokanedai, Minato-ku, Tokyo, Japan Search for more papers by this author Chihiro Sasakawa Chihiro Sasakawa Department of Microbiology and Immunology, Institute of Medical Science, University of Tokyo, Shirokanedai, Minato-ku, Tokyo, Japan Search for more papers by this author Akio Abe Corresponding Author Akio Abe Department of Infection Control and Immunology, Laboratory of Bacterial Infection, Kitasato Institute for Life Sciences, Kitasato University, Shirokane, Minato-ku, Tokyo, Japan The Kitasato Institute, Shirokane, Minato-ku, Tokyo, Japan Search for more papers by this author Author Information Takeshi Matsuzawa1,2, Asaomi Kuwae1,2, Sei Yoshida3, Chihiro Sasakawa3 and Akio Abe 1,2 1Department of Infection Control and Immunology, Laboratory of Bacterial Infection, Kitasato Institute for Life Sciences, Kitasato University, Shirokane, Minato-ku, Tokyo, Japan 2The Kitasato Institute, Shirokane, Minato-ku, Tokyo, Japan 3Department of Microbiology and Immunology, Institute of Medical Science, University of Tokyo, Shirokanedai, Minato-ku, Tokyo, Japan *Corresponding author. Department of Infection Control and Immunology, Laboratory of Bacterial Infection, Kitasato Institute for Life Sciences, Kitasato University, 5-9-1 Shirokane, Minato-ku, Tokyo 108-8641, Japan. Tel.: +81 3 5791 6123; Fax: +81 3 5791 6125; E-mail: [email protected] The EMBO Journal (2004)23:3570-3582https://doi.org/10.1038/sj.emboj.7600359 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Enteropathogenic Escherichia coli delivers a subset of effectors into host cells via a type III secretion system, and this step is required for the progression of disease. Here, we show that the type III effectors, EspG and its homolog Orf3, trigger actin stress fiber formation and the destruction of the microtubule networks beneath adherent bacteria. Both effectors were shown to possess the ability to interact with tubulins, and to stimulate microtubule destabilization in vitro. A recent study showed that microtubule-bound GEF-H1, a RhoA-specific guanine nucleotide exchange factor, was converted to its active form by microtubule destabilization, and this sequence of events resulted in RhoA stimulation. Indeed, EspG- and Orf3-induced stress fiber formation was inhibited by the expression of dominant-negative forms of GEF-H1 and RhoA, but not of Rac1 and Cdc42, and by treatment with a ROCK inhibitor. These results indicate that the impact of EspG/Orf3 on microtubule networks triggers the activation of the RhoA–ROCK signaling pathway via GEF-H1 activity. This report reveals for the first time that a pathogen can exploit the host factor GEF-H1. Introduction Enteropathogenic Escherichia coli (EPEC) remains a major cause of infantile diarrhea worldwide, mostly in the developing world (Donnenberg and Kaper, 1992). Related pathogens, which cause disease by similar mechanisms (DeVinney et al, 1999), include enterohemorrhagic E. coli O157:H7, rabbit EPEC, and Citrobacter rodentium, which is found in mice. These pathogens induce a characteristic histopathological lesion, referred to as an attaching/effacing (A/E) lesion (Moon et al, 1983), which is defined by the intimate attachment of bacteria to the intestinal epithelial surface and the effacement of host cell microvilli, which is involved in the development of diarrhea (Abe et al, 1998). Factors responsible for A/E lesion formation are encoded by a 35-kilobase pair region referred to as the locus of enterocyte effacement (LEE) (McDaniel et al, 1995), which includes the type III secretion system, the translocated intimin receptor (Tir) (Kenny et al, 1997), intimin (Donnenberg et al, 1993b), and EPEC-secreted proteins (Esps). We demonstrated the supermolecular structure of the type III secretion system (Sekiya et al, 2001), which includes an EscF-needle portion and an EspA-sheath-like structure at the tip of the needle. EscN, a cytoplasmic ATPase, is essential for the secretion of effectors (Jarvis et al, 1995), and the presumed pore-forming factors EspB and EspD are delivered to the host plasma membrane, and these factors are a prerequisite for the translocation of effectors into host cells (Ide et al, 2001). To date, five effectors, which are translocated into host cells via the type III secretion system, have been identified and characterized in EPEC; these effectors include Tir (Kenny et al, 1997), EspF (McNamara et al, 2001), EspG (Elliott et al, 2001), EspH (Tu et al, 2003), and Map (Kenny et al, 2002). In addition, the EspG homolog Orf3, which is encoded on the EspC pathogenicity islet within the EPEC genome (distinct from the LEE), is thought to be an effector (Elliott et al, 2001). Both EspG and Orf3 effectors share homology with the Shigella VirA effector. Although it has been demonstrated that Shigella VirA promotes the destabilization of host microtubules and Rac1 activation (Yoshida et al, 2002), the contributions of EspG/Orf3 effectors to EPEC virulence remain unclear. Map induces the transient formation of Cdc42-dependent filopodia, while EspH represses filopodium formation and enhances the actin-rich pedestal formation beneath adherent bacteria. A/E lesion formation is mediated by a Tir–intimin interaction (Kenny et al, 1997), which is required for the initiation of the disease process (Abe et al, 1998). However, an in vivo study with adult human volunteers suggested that additional EPEC factor(s) trigger diarrhea, since diarrhea still developed in four of 11 volunteers who ingested an intimin mutant strain that was incapable of A/E lesion formation (Donnenberg et al, 1993a). It has been reported that EPEC-induced diarrhea may be involved in a loss of intestinal barrier function, with ion and fluid movement from the intestinal submucosa into the lumen (Hecht, 2001). When polarized epithelial cells were infected with EPEC, redistribution of the tight junction-associated protein occludin was observed, and this redistribution was correlated with the disruption of the tight junction barrier (Yuhan et al, 1997; Simonovic et al, 2000). Although the precise function of EspF in host cells is not yet clear, EspF is known to be required for the disruption of epithelial cell junctions during EPEC infection, but EspF is not involved in the formation of the A/E lesion (McNamara et al, 2001). Thus, EPEC effectors have multiple functions in host cells, and their synergistic effects are thought to contribute to EPEC-mediated diarrhea. In this study, we demonstrated that actin stress fiber formation was induced by EPEC infection in Swiss 3T3 cells. This effect was dependent on the type III secretion system, but not on Tir–intimin interactions. It has been reported that the formation of actin stress fibers is regulated by the RhoA–ROCK signaling pathway (Kimura et al, 1996; Leung et al, 1996; Hall, 1998). Although the Tir-induced accumulation of actin filaments has been observed during A/E lesion formation, this phenotype was not found to be affected by the addition of the RhoA inhibitor (Ben-Ami et al, 1998); moreover, this phenotype was required for the activity of N-WASP (Kalman et al, 1999), the Arp2/3 complex, and Nck (Gruenheid et al, 2001), indicating that EPEC may exploit different signal transduction pathways in the formation of A/E lesions and actin stress fibers. To determine whether or not the type III effectors are involved in actin stress fiber formation, various effector mutants were constructed, and an infection assay was carried out. We found that EspG and Orf3 effectors are involved in the induction of actin stress fiber formation in a Tir–intimin interaction-independent manner. We further demonstrated that both effectors induce the disruption of microtubule networks beneath adherent bacteria by the direct association of EspG and Orf3 with tubulins. One of the aims of the present study was to determine how EspG/Orf3 governs the crosstalk between the actin cytoskeleton and microtubules. We revealed here that a newly identified guanine nucleotide exchange factor, GEF-H1 (Ren et al, 1998; Krendel et al, 2002), is exploited by EspG/Orf3 effectors to induce actin stress fiber formation during EPEC infection. A recent study also showed that endogenous GEF-H1 is localized in the microtubules in its inactive form and it switches to an active form when it is released from microtubules. Indeed, both EspG/Orf3 effectors were revealed to possess the ability to release GEF-H1 from the cytoskeleton to the cytosol as a result of the microtubule destabilization. We report here that the GEF-H1-mediated signaling pathway is triggered by EPEC infection via the activity of EspG/Orf3 effectors. Results EPEC infection of mammalian cells induces stress fiber formation To investigate whether EPEC infection influences the host actin cytoskeleton, serum-starved Swiss 3T3 cells were infected with wild-type (WT) EPEC and the architecture of the actin networks was visualized by F-actin staining with rhodamine-labeled phalloidin. Although EPEC is a human pathogen, we used mouse Swiss 3T3 cells for this experiment, since the dynamics of the actin cytoskeleton are easily observed in these fibroblastic cells. EPEC has the ability to adhere to and form tight microcolonies on tissue culture cells and human intestinal cells (Rosenshine et al, 1992); these activities are mediated by a bundle-forming pilus (Giron et al, 1991). As expected, the accumulation of F-actin beneath microcolonies of adherent bacteria was also observed in the mouse fibroblastic cell line, Swiss 3T3 (Figure 1A, arrows). In addition to this accumulation, actin stress fibers with parallel bundles extending into projections of the cell periphery were prominently induced in the cytoplasm under conditions of EPEC infection (Figure 1A, arrowhead). Although the formation of a pedestal structure, represented by the accumulation of F-actin beneath the adherent bacteria, was not detected in all of the infected cells with the eae mutant (defective in intimin production) or the tir mutant (defective in Tir production), these cells still developed extensive actin stress fibers (Figure 1B and C). It was of note that stress fiber formation, observed here in EPEC WT-infected cells, was not detected with either the espA mutant (defective in the formation of the EspA-sheath in type III machinery) or the escF EPEC mutant (defective in the formation of the EscF-needle in type III machinery) (Figure 1B and C). The same was true for the espB and espD mutants (defective in pore formation on the host plasma membrane) (Figure 1B and C) and the escN mutant (defective in secretion of type III-secreted proteins) (Figure 2A and B). Therefore, the results of the present series of experiments strongly suggested that certain unknown type III-secreted effector(s) are involved in EPEC-induced actin stress fiber formation in infected cells. Figure 1.The host cytoskeletal rearrangement as a result of EPEC infection. (A) Swiss 3T3 cells were infected with EPEC WT for 3 h, and cells were fixed and stained with rhodamine-labeled phalloidin. The formation of pedestal structures beneath EPEC microcolonies (arrows) and actin stress fibers (arrowhead) was observed in the cells infected with EPEC. (B) Swiss 3T3 cells were infected with EPEC WT or its mutants for 3 h, and the cells were fixed and visualized as in (A). All fluorescence images were taken with the same exposure time using a CCD camera. (C) The percentage of Swiss 3T3 cells showing marked formation of actin stress fibers is indicated. Swiss 3T3 cells were infected with EPEC WT or its mutants for 3 h, and the cells were fixed and visualized as in (A). Percentages were based on counts of at least 150 cells, and the values are the means±standard deviation from three independent experiments. Download figure Download PowerPoint Figure 2.Determination of the EPEC type III effectors required for stress fiber formation. (A) Swiss 3T3 cells were infected with EPEC WT, various type III effector mutants, or an espG/orf3 mutant complemented with cloned espG (p99-EspG), orf3 (p99-Orf3), or vector alone for 3 h. Cells were fixed and stained with rhodamine-labeled phalloidin in order to detect F-actin. All fluorescence images were taken with the same exposure time. (B) The percentage of Swiss 3T3 cells showing marked formation of actin stress fibers is indicated. Swiss 3T3 cells were infected with EPEC WT or its mutants for 3 h, and the cells were fixed and visualized as in (A). Percentages were based on counts of at least 150 cells, and the values are the means±standard deviation from three independent experiments. (C) Swiss 3T3 cells were transfected with the plasmids encoding EGFP-tagged EspG, EGFP-tagged Orf3, and vector alone. Cells were fixed and stained as in (A), and the transfected cells were visualized by EGFP fluorescence. (D) EGFP fluorescence-positive Swiss 3T3 cells were randomly obtained from panel C, and the percentage of cells showing actin stress fiber formation is indicated. Percentages were based on a count of at least 100 cells, and the values are the means±standard deviation from three independent experiments. Download figure Download PowerPoint EspG/Orf3 is involved in stress fiber formation in host cells infected with EPEC To identify the secreted bacterial protein(s) involved in the formation of stress fibers in host cells, we constructed a series of nonpolar mutants with a defect in the genes espF, orf19 (defective in Map production), espG, espH, or orf3, and we also considered an espG/orf3 double-knockout mutant (due to the fact that EspG shares 43.5% identity with Orf3; Elliott et al, 2001). Swiss 3T3 cells were infected with each of the mutants in order to investigate their ability to elicit actin stress fiber formation. The results demonstrated that the ability of the espG/orf3 double-knockout mutant to induce stress fiber formation was greatly reduced, in comparison with that of the other type III effector mutants, such as the espF, orf19, espG, espH, and orf3 mutants (Figure 2A and B). Complementation of the espG/orf3 double-knockout mutant with a plasmid encoding EspG (p99-EspG) or Orf3 (p99-Orf3) restored the ability to form actin stress fibers (Figure 2A and B), indicating that both proteins participated in the formation of stress fibers. To confirm this possibility, pEGFP-EspG or pEGFP-Orf3 (enhanced green fluorescent protein (EGFP)-tagged EspG or Orf3, respectively) was transfected into Swiss 3T3 cells and the cells were observed for the formation of stress fibers. The Swiss 3T3 cells overexpressing EGFP-EspG or EGFP-Orf3, but not EGFP alone, were able to induce actin stress fibers (Figure 2C and D), thereby suggesting the following conclusions: (i) EspG and Orf3 are responsible for the formation of stress fibers, and they possess functional redundancy, and (ii) other effectors are not involved in this series of events. Although EspG is secreted via the type III machinery (Elliott et al, 2001), it remains unclear as to how Orf3 is secreted. To address this issue, the secretion profile of Orf3 was analyzed by immunoblotting using anti-Orf3 antibodies. We detected Orf3 in the culture supernatant prepared from the EPEC WT, but not from the type III machinery mutant, indicating that Orf3 is also secreted via the type III machinery (data not shown). EspG/Orf3 interaction with tubulins and stimulation of microtubule destruction EspG and Orf3 shared a striking homology not only with each other but also with the Shigella VirA effector (Elliott et al, 2001). VirA is secreted via the type III machinery, and is required for promoting the uptake of bacteria by epithelial cells. Since VirA has the ability to interact with tubulins and trigger microtubule destabilization, this finally leads to Rac1 activation (Yoshida0 et al, 2002). Interestingly, cloned EPEC espG and orf3 were able to individually rescue the ability of the Shigella virA mutant to invade host cells, implying that both EspG and Orf3 shared functions similar to those of VirA (Elliott et al, 2001). To investigate the effects of both EspG and Orf3 on host microtubule networks, Swiss 3T3 cells were infected for 3 h with EPEC WT or the espG/orf3 double mutant, and the reorganization of both F-actin and the microtubule structures was observed by immunofluorescence microscopy. As shown in Figure 3A, the local microtubule structures beneath the EPEC microcolonies were destabilized by infection with EPEC WT (Figure 3A, arrows), but not with the espG/orf3 mutant. These events were reproducibly detected in EPEC WT-infected HeLa and Cos-7 cells, and also with the introduction of cloned espG or orf3 (pEGFP-EspG, pEGFP-Orf3) into Swiss 3T3 cells (data not shown). In addition, the time course of microtubule destabilization after EPEC WT or espG/orf3 mutant infection was measured, and the percentage of destabilization beneath EPEC microcolonies was calculated (Figure 3B). Destabilization of the local microtubule structures was greatly increased as a result of WT infection. In contrast, no such effect was detected in infection with the espG/orf3 mutant. Like EPEC WT, the espG or orf3 single mutant still possessed the ability to destabilize the microtubule networks, and complementation of the espG/orf3 double-knockout mutant with a cloned espG or orf3 restored the ability to destabilize microtubule networks due to functional redundancy (Figure 3C and D). Collectively, both EPEC effectors, EspG and Orf3, are involved in the induction of actin stress fiber formation and the destabilization of microtubule networks in vivo. Figure 3.Effect of EPEC infection on microtubule dynamics. (A) Swiss 3T3 cells were infected with EPEC WT and the espG/orf3 mutant for 3 h, and the cells were fixed and stained with rhodamine-labeled phalloidin and fluorescently labeled anti-tubulin antibodies in order to detect the actin cytoskeleton and microtubule networks, respectively. The arrows indicate disruptions of the microtubule network beneath the adherent EPEC microcolonies. (B) Time course of the local destabilization of microtubule networks after EPEC infection. The percentages of microtubule destabilization beneath the EPEC microcolonies were based on a count of at least 150 microcolonies, and the values given are the means±standard deviation from three independent experiments. (C) Swiss 3T3 cells were infected with EPEC WT and its mutants for 3 h and the cells were fixed and visualized as in (A). (D) The percentage of local destabilization of microtubule networks beneath the EPEC microcolonies using various EPEC derivatives. After 3 h infection, percentages were calculated as in (B). Download figure Download PowerPoint Next, using a GST pull-down assay, we investigated GST-EspG and GST-Orf3 for their ability to interact with tubulins contained in Swiss 3T3 or HeLa cell lysates. Analysis of the precipitated proteins with GST-EspG or GST-Orf3 in an immunoblot analysis with anti-tubulin antibodies revealed that the tubulins were co-precipitated with GST-EspG or GST-Orf3, but not with GST alone (Figure 4A). The direct binding of EspG/Orf3 to tubulins was further confirmed by a ligand overlay assay (see Materials and methods; data not shown) and by a pull-down assay with purified twice-cycled microtubule proteins (Yoshida et al, 2002) (Figure 4B). Moreover, we considered whether or not EspG/Orf3 would be able to affect either microtubule stabilization or destabilization using twice-cycled microtubule proteins (Figure 4C and D). Twice-cycled microtubule proteins were polymerized at 37°C for 1 h, and the resulting polymerized microtubules (15 μM) were incubated with an equal volume of GST alone, GST-EspG, or GST-Orf3 (15 μM each), and the microtubule dynamics were monitored by spectrophotometry (absorbance at 340 nm) (Figure 4C). In the absence of GST-EspG or GST-Orf3, the destabilization of the microtubules was slightly induced after 30 min (Figure 4C, circles), and similar results were observed in an incubation with GST alone (Figure 4C, triangles). In contrast, depolymerization of the microtubules was triggered by the addition of GST-EspG (Figure 4C, squares) or GST-Orf3 (Figure 4C, crosses). Furthermore, microtubule assembly was inhibited by the addition of GST-EspG and GST-Orf3, but not GST alone (Figure 4D). These results clearly demonstrate that EspG/Orf3 has the ability to bind directly to tubulins, and that it promotes the destabilization of microtubules in vitro. Figure 4.EspG/Orf3 bound to tubulins and induced the destabilization of microtubules in vitro. (A) GST fusion proteins bound to glutathione–Sepharose 4B beads were incubated with Swiss 3T3 and HeLa cell lysates for 2 h at 4°C. The beads were washed, and then bound proteins were analyzed by immunoblotting using anti-tubulin antibodies. (B) GST fusion proteins bound to glutathione–Sepharose 4B beads were incubated with purified twice-cycled microtubule proteins for 2 h at 4°C. The beads were washed, and then each sample was subjected to immunoblotting as in (A). (C) EspG/Orf3 induces microtubule destabilization. Purified twice-cycled microtubule proteins (15 μM) were polymerized at 37°C for 1 h, and then equal volumes of PM-1.3 M buffer alone (circles) or PM-1.3 M buffer each containing 15 μM of GST alone (triangles), GST-EspG (squares), or GST-Orf3 (crosses) were added. Destabilization of microtubules was monitored spectrophotometrically (absorbance at 340 nm) at 37°C for 30 min. (D) EspG/Orf3 inhibited microtubule polymerization. Purified twice-cycled microtubule proteins (7.5 μM) were polymerized with PM-1.3 M buffer alone (circles) or PM-1.3 M buffer each containing 7.5 μM of GST alone (triangles), GST-EspG (squares), and GST-Orf3 (crosses). Download figure Download PowerPoint EspG/Orf3-induced rearrangement of the actin cytoskeleton is dependent on GEF-H1 Regulation of the actin cytoskeleton by microtubules relies on the activity of the Rho family GTPases (Wittmann and Waterman-Storer, 2001). Microtubule disassembly results in the activation of RhoA, which enhances myosin contractility and actin stress fiber formation (Enomoto, 1996; Liu et al, 1998; Ren et al, 1999). The newly identified GEF family members, including murine Lfc (Glaven et al, 1999), murine p190RhoGEF (van Horck et al, 2001), and human GEF-H1 (Lfc homolog) (Krendel et al, 2002), are associated with microtubules and function as RhoA/Rac activators. Since the destruction of local microtubule networks and the formation of actin stress fibers were the most pronounced host responses to EPEC infection, we investigated the involvement of GEF-H1 (human) in EspG- or Orf3-induced stress fiber formation in human epithelial HeLa cells, instead of in mouse Swiss 3T3 cells. When HeLa cells were infected with EPEC WT, extensive actin accumulation was detected beneath a substantial number of adherent bacteria. However, actin stress fiber formation was very weak in HeLa cells (data not shown). (In fact, this has been one of the major reasons as to why EPEC-induced stress fiber formation has not been characterized to date.) It is possible that actin may be preferentially supplied for the formation of a pedestal structure, rather than for stress fiber formation in EPEC-infected HeLa cells. To facilitate the detection of pathogen-induced stress fiber formation, including GEF-H1 activation, we created tir/espG/orf3 triple mutants carrying either p99-EspG or p99-Orf3. Our preliminary data indicated that mutation of the tir gene turned off the formation of F-actin aggregation beneath the EPEC adherent; this finding allowed us to detect the F-actin architecture in epithelial cells such as HeLa cells. As shown in Figure 5A, upon infection of HeLa cells with the tir/espG/orf3 triple mutant overexpressing EspG, actin stress fibers appeared, a result that was not observed in HeLa cells infected with the triple mutant carrying the vector alone. Importantly, EspG-induced actin cytoskeletal rearrangement was almost completely abolished by the expression of a dominant-negative form of GEF-H1 (GEF-H1 DHmut) in HeLa cells. Indeed, the percentage of cells showing a marked formation of actin stress fibers increased greatly among HeLa cells infected with the tir/espG/orf3 triple mutant carrying p99-EspG or p99-Orf3 (Figure 5B). To further analyze the direct involvement of GEF-H1 in the EspG/Orf3-mediated actin stress fiber formation, a dominant-negative form of p115-RhoGEF (p115-RGS) (Johnson et al, 2003) was also used in this study, since p115-RhoGEF is well characterized as a RhoA-specific activator. RhoA is activated by the treatment of lysophosphatidic acid (LPA) via p115-RhoGEF activity (Hart et al, 1998; Kozasa et al, 1998). Indeed, LPA-induced stress fiber formation was inhibited by the expression of p115-RGS (Figure 5B). As expected, the formation of EspG-induced stress fibers was not affected by the expression of p115-RGS (Figure 5A and B). In contrast, GEF-H1 was found to be modulated by the addition of nocodazole, which induces the depolymerization of microtubules. Moreover, the nocodazole-, but not LPA-, induced stress fiber formation was inhibited by the expression of GEF-H1 DHmut. The impact of actin stress fiber formation elicited by EspG or Orf3 in HeLa cells was similar to that elicited via the destruction of microtubule structures by nocodazole treatment, but not by LPA treatment (Figure 5B). The EspG/Orf3–GEF-H1 signaling pathway was further confirmed by a small interfering RNA (siRNA) assay (Figure 6). Two siRNAs homologous to genes coding for GEF-H1 in different regions (GEF-H1-siRNA-I and GEF-H1-siRNA-II; Benais-Pont et al, 2003) were used; nonsilencing siRNA was used as a negative control. These siRNAs were introduced into HeLa cells, and the reduction in the transcription of GEF-H1 mRNA was confirmed by RT–PCR (Figure 6A). EspG- and Orf3-induced actin cytoskeletal rearrangement was almost completely abolished by the introduction of GEF-H1-siRNA-I or GEF-H1-siRNA-II into HeLa cells (Figure 6B and C). Collectively, these data clearly demonstrate that GEF-H1 is absolutely required for EspG- and Orf3-induced actin cytoskeletal rearrangement. Figure 5.Effect of GEF-H1 on EPEC-induced actin stress fiber formation. (A) HeLa cells were transfected with a plasmid encoding the dominant-negative form of GEF-H1 (GEF-H1 DHmut), containing a Tyr-to-Ala substitution at residue 393 in the DH domain of GEF-H1, or RGS domain of p115-RhoGEF (p115-RGS). Transfected cells were infected with a tir/espG/orf3 mutant complemented with a cloned espG (p99-EspG) or with vector alone for 3 h, at a final concentration of 1 mM of IPTG in order to induce EspG expression. The cells were fixed and stained with rhodamine-labeled phalloidin in order to detect F-actin, and the transfected cells were visualized by EGFP fluorescence. The arrows indicate GEF-H1 DHmut or p115-RGS. (B) HeLa cells were transfected with GEF-H1 DHmut or p115-RGS, and then transfected cells were infected with the EPEC mutant complemented with a cloned espG (p99-EspG), orf3 (p99-Orf3), or vector alone (vector). Otherwise, transfected cells were treated with 1 μg/ml of LPA for 3 min or 10 μM nocodazole for 1 h. The percentage of HeLa cells showing marked formation of actin stress fibers is indicated. Percentages were based on counts of at least 75 cells, and the values given are the means±standard deviation from three independent experiments. Download figure Dow