Contribution of the Twin Arginine Translocation System to the Virulence of Enterohemorrhagic Escherichia coli O157:H7

Abstract

Shiga toxin-producing Escherichia coli O157:H7 is a major food-borne infectious pathogen. In order to analyze the contribution of the twin arginine translocation (TAT) system to the virulence of E. coli O157:H7, we deleted the tatABC genes of the O157:H7 EDL933 reference strain. The mutant displayed attenuated toxicity on Vero cells and completely lost motility on soft agar plates. Further analyses revealed that the ΔtatABC mutation impaired the secretion of the Shiga toxin 1 (Stx1) and abolished the synthesis of H7 flagellin, which are two major known virulence factors of enterohemorrhagic E. coli O157:H7. Expression of the EDL933 stxAB₁ genes in E. coli K-12 conferred verotoxicity on this nonpathogenic strain. Remarkably, cytotoxicity assay and immunoblot analysis showed, for the first time, an accumulation of the holotoxin complex in the periplasm of the wild-type strain and that a much smaller amount of StxA₁ and reduced verotoxicity were detected in the ΔtatC mutant cells. Together, these results establish that the TAT system of E. coli O157:H7 is an important virulence determinant of this enterohemorrhagic pathogen.

Bacteria export numerous proteins across the cytoplasmic membrane via either the Sec machinery (26) or the twin arginine translocation (TAT) (also called membrane targeting and translocation [MTT]) system (41). The TAT system is different from the Sec pathway because of its unusual ability to transport folded proteins and even enzyme complexes into the periplasm. Most of the TAT substrates are cofactor-containing enzymes taking part in oxidation-reduction systems involved in energy conservation under anaerobic conditions. The essential tat genes are found in most bacterial genomes, including the pathogens Helicobacter pylori, Yersinia pestis, Vibrio cholerae, Salmonella enterica subsp. enterica serovar Typhi, Haemophilus influenzae, Mycobacterium tuberculosis, Staphylococcus aureus, and Pseudomonas aeruginosa (41). The functionality of the TAT pathway has been reported for Escherichia coli (32, 40), Zymomonas mobilis (12), Ralstonia eutropha (5), Bacillus subtilis (17), Pseudomonas stutzeri (14), and Streptomyces lividans (33). It has been demonstrated that the P. aeruginosa TAT system operates in parallel with the Sec machinery in the secretion of virulence factors via the type II secretion pathway (38). Further investigation has shown that the TAT system of P. aeruginosa contributes to virulence through the secretion of various factors associated with either pathogenesis or stress response (23). Recently, Ding and Christie reported that the TAT system is an important virulence determinant of the phytopathogen Agrobacterium tumefaciens (10).

Widespread E. coli is a major component of the normal intestinal flora of humans and other mammals and is usually harmless to the host. However, some specific E. coli strains represent primary pathogens with an enhanced potential to cause disease. Shiga toxin- or verotoxin-producing E. coli (STEC or VTEC, respectively) causes not only a broad range of symptoms in humans, including uncomplicated diarrhea, but also more severe diseases like hemorrhagic colitis (HC) and the often deadly hemolytic-uremic syndrome (HUS) (22). Among STEC, enterohemorrhagic E. coli (EHEC) serotype O157:H7 is the most important cause for HC and HUS. The number of O157:H7-related infections is increasing steadily worldwide, and this strain is considered to be both an emerging pathogen and a major threat to public health (19). Genomes of two EHEC O157:H7 isolates (EDL933 and Sakai) have been sequenced (13, 24). The genome of the EDL933 isolate contains about 177 so-called O-specific islands, which are DNA segments of more than 50 bp present only in EDL933 but are absent from the genome of the harmless MG1655 isolate of the K-12 strain (24). Some of the O-specific islands and the pO157 plasmid of O157:H7 encode virulence factors, including hemolysin, intimin, Shiga toxins 1 (Stx1) and 2 (Stx2), the specific H7 type of flagella, and other factors required for the virulence, such as the type II and type III secretion systems. Moreover, the tat genes of the K-12 strain, which encode the best-characterized prototype of bacterial TAT system, are conserved in the two EHEC genomes. However, it is unknown whether the TAT system contributes to the virulence unique to the EHEC.

In this study, we knocked out the tatABC genes from the genome of the E. coli O157:H7 EDL933. Remarkably, the toxicity of the mutant was attenuated compared to the parental strain in the Vero cell assay. In addition, immunoblot analysis showed that the amount of Stx1 secreted by the mutant was consistently reduced. Expression of EDL933 stxAB₁ genes in the nonpathogenic E. coli K-12 strain confirmed the impairment of Stx1 translocation in the tat mutant. In addition, the TAT system was required for flagellin biosynthesis and the motility of EDL933. Together, these results showed that the TAT system is important for the synthesis of two major known virulence factors specific to EHEC O157:H7.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The EHEC reference strain used in this study was the EDL933 serotype O157:H7 (ATCC 43895) provided by A. O'Brien. The E. coli O157 stx mutant is a laboratory stock strain at the University of Auvergne Clermont-1, which carries deletions in both the stx1A and stx2A genes, as revealed by PCR. The nonpathogenic E. coli K-12 strains used in this study are MC4100A (F⁻ Δ(argF-lac)U169 araD139 rpsL150 (strR) thi flhD5301 deoC1 ptsF25 relA1 fruA25 rbsR22 e14⁻ LAM⁻ ara⁺) and B1LK0A (same as MC4100A, ΔtatC) (30). Plasmid pKOBEG is a thermosensitive replicon that carries the λ phage redγβα operon expressed under the control of the arabinose-inducible P_bad promoter (9). The p8737 (tatABCD⁺) plasmid is described in reference 1.

The bacteria were routinely grown in Luria-Bertani (LB) broth, on LB plates, or in the minimal M9 media (20). Anaerobic growth was achieved normally in stoppered bottles or tubes filled to the top. As required, ampicillin (100 μg/ml), chloramphenicol (50 μg/ml), kanamycin (50 μg/ml), l-arabinose (10 mM), trimethylamine N-oxide (TMAO; 1 mg/ml), glycerol (0.5%), ammonium molybdate (1 μM), or potassium selenite (1 μM) was added. Precultures were grown from single colonies and used at 100-fold dilutions for inoculation of experimental cultures.

Construction of the EDL933 ΔtatABC mutant.

Replacement of the tatABC genes on the E. coli O157:H7 EDL933 chromosome by the aphA-3 gene was performed according to the method of Ghigo et al. (http://www.pasteur.fr/recherche/unites/Ggb/Pcr3T.ppt) and Datsenko and Wanner (9). pKOBEG was introduced into the target strain EDL933 by selection for Cm^r in a special low-salt, low-Mg²⁺ broth (10 g of tryptone, 2.5 g of yeast extract, 0.25 g of NaCl, 2.5 ml of 1 M KCl [pH 7.4]) at 30°C overnight. A fresh culture in the same medium was inoculated from the overnight preculture by 100-fold dilution. To induce the recombinase expression from the λ phage redγβα operon on pKOBEG, 10 mM l-arabinose was added in this culture. The incubation was performed at 30°C with shaking until an optical density at 620 nm (OD₆₂₀) of 0.45 to 0.55 was reached, and then the cultures were placed in a water bath at 42°C with shaking for 15 min to eliminate pKOBEG containing a thermosensitive replicon. Competent cells were prepared from these cultures. The aphA-3 gene was amplified by PCR with the primers Kngbtataup (5′-AACGTATAATGCGGCTTTGTTTAATCATCATCTACCACAGAGGAACATGTAAAGCCACGTTGTGTCTCAA-3′) and Kngbtatcdn (5′-ATATCAAACATCCTGTACTCCATATGACAACCGCCCTGACGGGCGGTTGATTAGAAAAACTCATCGAGCA-3′) and the Taq DNA polymerase according to the manufacturer's instructions (Appligène-oncor, Illkrich, France). The amplified fragment carried, at the two ends, 50 bp (underlined) identical to the sequences from 1 to 50 bp upstream of the tatA gene and from 51 to 100 bp downstream of the tatC gene of E. coli EDL933. The PCR product was purified by using the QIAquick PCR purification kit in accordance with the supplier's instructions (QIAGEN S.A., Courtaboeuf, France) and transformed, by electroporation, into EDL933 competent cells prepared as above and selected on LB-kanamycin plates. The Km^r Cm^s phenotype of the recombinants was verified by the disk diffusion method on Mueller-Hinton agar (BioMérieux, Marcy-L'Etoile, France), with disks purchased from Sanofi Diagnostic Pasteur (Marnes La Coquette, France). The replacement of the tatABC genes by aphA-3 was verified by PCR with the oligonucleotides tat1 (5′-AACGTATAATGCGGCTTTGTTTAATCATCATCTACCACAGAGGAACATGT-3′), tat2 (5′-ATATCAAACATCCTGTACTCCATATGACAACCGCCCTGACGGGCGGTTGA-3′), GBnp-3 (5′-TTAGAAAAACTCATCGAGCA-3′), and GB-5 (5′-AGCCACGTTGTGTCTCAAAATC-3′). Further verification was performed by digestion of the PCR products with the SacII restriction enzyme. The general properties of the wild-type strain EDL933 and the mutant EDLtat were biochemically confirmed by using an API 20E test (BioMérieux).

Cloning of the stxAB₁ genes.

The stxAB₁ genes were amplified from the chromosome of E. coli O157:H7 isolate EDL933 by PCR with the primers stx1-1 (5′-ATGAATTCATATGAAAATAATTATTTTTAGAGT-3′) and stx1-4 (5′TCACTAAGCTTGTCGACGTCAACGAAAAATAACTTCGCT-3′). The amplified fragment was digested with EcoRI and HindIII and cloned into the corresponding sites of the pBAD24. The structure of the resulting plasmid, pSTX1AB, was verified by endonuclease digestion. Expression of stxAB₁ genes was under the tight control of the P_BAD promoter.

Enzyme assays.

To detect the TMAO and dimethyl sulfoxide (DMSO) reductase activities, protein fractions were separated on nondenaturing polyacrylamide (12.5%) gels. DMSO and TMAO reductase activities were visualized by an activity staining method which is based on a methyl viologen-linked TMAO reduction (31).

Cellular fractionation, electrophoresis, immunoblot, and mass spectrometry.

To prepare cellular fractions, bacteria were grown anaerobically in 200 ml of LB medium and the cells were harvested by centrifugation at 17,000 × g for 10 min. Periplasm and spheroplasts were prepared by the lysozyme-EDTA-cold osmoshock method (31). Spheroplasts were washed once and disrupted by sonication in 2 ml of 40 mM Tris-HCl (pH 7.6), in a Branson Sonifier 450 in the continuous mode and with an output setting of 2 for 30 s. The supernatant filtrates prepared from the bacterial cells grown in brain heart infusion broth were precipitated with 10% trichloroacetic acid, and washed with acetone. Protein samples were separated by polyacrylamide gel electrophoresis in the presence (denaturing) or in the absence (nondenaturing) of sodium dodecyl sulfate (SDS) on 12.5% or 15% acrylamide gels. Protein samples were electrotransferred, after the electrophoresis, onto a polyvinylidene difluoride membrane and analyzed by immunoblotting, by using the enhanced chemiluminescence method according to the manufacturer's instructions (Amersham Biosciences). The images were digitized and quantified with Kodak Image analysis software. Polyclonal antibodies against Stx were kindly provided by A. O'Brien. Anti-H7 flagellin sera were purchased from Difco (Detroit, Mich.) and used at 1:10,000-fold dilution for immunoblot analysis.

The two-dimensional gel electrophoresis (2DE) was performed according to Amersham Biosciences' instructions. For mass spectrometry, protein samples were separated by 2DE and stained by Coomassie blue or silver staining and the specific protein spots were excised. After crushing and washing of the excised gel, the proteinaceous material was reduced with dithiothreitol and alkylated with iodoacetamide in 100 mM NH₄HCO₃. Proteolytic digestion by trypsin was then performed overnight at 37°C. The supernatant was collected, the salts were removed by flow through an R2 Poros column, and the sample was analyzed by mass spectrometry. The protein was identified by using the ProFound software (http://129.85.19.192/profound_bin/). The identification significance was evaluated by the Est'dZ score, as described at http://129.85.19.192/profound/help.html.

Vero cell assay.

The production of Shiga toxin was checked by a cytotoxicity assay on Vero cells (18). Vero cells (African green monkey kidney cells; ATCC CRL 1587) were grown at 37°C in Eagle basal medium (Seromed, Berlin, Germany) supplemented with 10% fetal calf serum (Seromed), 1% l-glutamine (Life Technologies, Paisley, Scotland), 2% penicillin-streptomycin-amphotericin B, and 1% minimal essential medium vitamin solution (Life Technologies) in an atmosphere of 5% CO₂. Vero cells were cultivated as monolayers in 75-cm² tissue culture flasks. The bacterial strains were inoculated into 10 ml of brain heart infusion broth (Biokar Diagnostics) and incubated at 37°C to an OD₆₂₀ of 1. After centrifugation of 1.5 ml at 12,000 × g for 5 min, supernatant filtrates were obtained with a 0.45-μm-pore-size filter (PolyLabo, Molsheim, France) and screened for verotoxicity. Twofold serial dilutions of bacterial filtrates were done in 96-well flat-bottom microtiter plates (Nunc, Roskilde, Denmark) (50 μl per well). A total of 50 μl of Eagle basal medium containing 10⁵ Vero cells in suspension was added to each well. The culture plates were incubated for 24 h at 37°C in a 5% CO₂ atmosphere. After 24 h, the cell monolayers were washed with phosphate-buffered saline (PBS) (pH 7.4) (Seromed) and stained with a crystal violet solution (1.3% crystal violet-5% ethanol in PBS). The verotoxin titer was expressed as the reciprocal of the highest sample dilution of culture filtrate which caused 50% cell detachment after 24 h of incubation, as judged by the dye intensity and by microscopic observation. The stx mutant was used as a negative control. Alternatively, 50 μl of each dilution of cellular fractions prepared from K-12 strains expressing cloned stxAB₁ genes were used in place of filtrates.

Motility assay.

Motility assays were performed in 0.3% LB soft agar medium. Cell cultures were normalized to an OD₆₀₀ of 0.9, and 10 μl of each strain was inoculated onto the motility plates. Motility was examined after 24 h of incubation at 37°C. At least three independent motility assays were carried out for each strain.

Extraction of flagella.

Bacteria were recovered from 5 LB plates incubated overnight at 37°C, centrifuged for 10 min at 5,000 × g in PBS buffer (200 mM NaCl, 3 mM KCl, 1.5 mM KH₂PO₄, and 16 mM Na₂HPO₄ [pH 7.4]). The pellet was resuspended in 10 ml of PBS and vortexed for 5 min. The suspension was centrifuged for 15 min at 16,000 × g. After recovery of the supernatant, SDS was added (0.1% wt/vol). The solution was centrifuged at 11,000 × g for 3 h at 4°C. The pellet was recovered in 400 μl of Tris buffer (40 mM, pH 8).

RESULTS AND DISCUSSION

TAT system of EHEC O157:H7 and construction of the ΔtatABC mutant of the EDL933 isolate.

The genetic structure and composition of the tat genes can vary from genome to genome (41). The TAT apparatus of the nonpathogenic E. coli K-12 strain is encoded by the tatABCD operon and the tatE gene (32). TatA, TatB, and TatE share sequence homology at their N termini, including one transmembrane segment and an adjacent amphipathic domain, whereas their C termini vary in sequence and length (8). TatC is an integral membrane protein with apparently four transmembrane helices (11). The deletion of tatC leads to mislocation of all substrates analyzed and displays pleiotropic phenotypes (34). Therefore, TatC is essential for TAT function. In contrast, TatD is a soluble protein with no discernible role in Tat-dependent protein export (4). We analyzed the genomes of the two E. coli O157:H7 isolates, EDL933 and Sakai, and found that they contained the same tatABCD operon and the tatE gene as the E. coli K-12 strain. In addition, the sequences of the TatA, TatB, and TatC proteins were 100% identical in all three genomes. Therefore, the basic physiological function and the operation mechanism of the TAT system must be conserved in EHEC and nonpathogenic E. coli. Remarkably, the tatE gene and the particular genetic structure of the tat coding genes, i.e., the tatABCD operon plus the tatE gene, is observed only for enteric bacteria, including E. coli, S. enterica serovar Typhi, V. cholerae, and Y. pestis (41). Therefore, the TAT system might provide an important mechanism for adaptation of the environment of the gut.

In order to analyze the potential contribution of the TAT system to the virulence of the EHEC, we constructed a tat mutant of the E. coli O157:H7 reference strain, EDL933, by replacing the tatABC genes with the aphA-3 gene encoding kanamycin resistance. The deletion of the tatABC genes in the mutant (EDLtat) was confirmed by PCR amplification and by endonuclease digestion. Using the tat1-tat2 oligonucleotide pair, the tatABC region of the wild-type parental strain was amplified as an ∼1.7-kb fragment. In contrast, the replacement of the tatABC genes by aphA-3 resulted in a reduction of the size of the corresponding PCR product to ∼1 kb in the recombinant colonies analyzed. A SacII cleavage site is present only in the tatB gene. When subjected to the SacII digestion, the tatABC-carrying 1.7-kb fragment was cleaved into an ∼1.1-kb tat′BC fragment and an ∼0.6-kb tatAB′ fragment, whereas the 1-kb aphA-3 fragment remained intact, thus confirming the replacement of the tatABC genes by aphA-3 in the chromosome of the EDLtat mutant.

Three standard means for assaying TAT function in the nonpathogenic E. coli K-12 strain are to analyze the translocation of known TAT substrates, to test for anaerobic growth on TMAO-glycerol minimal media, and to inspect the morphology of the cells. SufI is one of the most used substrates in the study of the TAT system (1, 35, 42). As shown in Fig. 1A, it was translocated into the periplasm of only the wild-type strain but was absent from the periplasm of the ΔtatABC mutant. As a control, the translocation of the maltose binding protein, MalE, was also analyzed. MalE is a typical substrate of the Sec pathway. Since MalE was normally translocated into the periplasm of the tat mutant (Fig. 1A, lanes 3 and 4), the ΔtatABC mutation affected only the translocation of TAT substrates but had no effect on the Sec-dependent protein export.

FIG. 1.

Phenotype of the EDL933-derived ΔtatABC mutant. Periplasmic fractions (3 μg of protein each, all lanes in panel A, lanes 1 to 3 in panels B and C) and spheroplasts (30 μg of protein each, lanes 4 to 6 in panels B and C) of the E. coli O157:H 7 EDL933 (wild type [WT]), its derivative EDLtat mutant (ΔtatABC), and the E. coli K-12 strain (K-12) were resolved on a native polyacrylamide (12.5%) gel. SufI and MalE were resolved on SDS denaturing gels and detected by immunoblotting with polyclonal antisera against these two proteins, respectively (A). The DMSO reductase (DMSOR) and TMAO reductase (TMAOR) activities were visualized by activity staining (B). The gel was then analyzed by immunoblotting with polyclonal antisera against the TMAO reductase (C) (TorA). Light microscope phase-contrast images of EDL933 (WT) and the EDLtat mutant (ΔtatABC) are presented in panel D.

Both membrane-bound DMSO reductase and periplasmic TMAO reductase support anaerobic growth on TMAO (40). The interruption of the tat genes of the harmless E. coli K-12 strain abolishes the translocation of these two enzymes but has only a slight effect on the enzymatic activity. Because these two enzymes are mislocated in the cytoplasm and cannot form the anaerobic respiratory chains, the K-12 tatB and tatC mutants are incapable of anaerobic growth with TMAO as the sole electron acceptor (6, 40). Like the E. coli K-12 tat mutants, the E. coli O157:H7 ΔtatABC mutant could not grow on the minimal glycerol-TMAO medium, but the parental wild-type strain grew normally under the same conditions (data not shown). Surprisingly, neither TMAO reductase activity nor the TMAO reductase protein (TorA) was detected in the E. coli O157:H7 strain EDL933 (Fig. 1B and C, lanes 1 and 4). The genomes of the E. coli O157:H7 strains EDL933 and Sakai contain the fully conserved tor operon. In addition, we confirmed, by PCR, the presence of the torA gene in the EDL933 stock strain used in this study (data not shown). Therefore, the depletion of the TMAO reductase might result from a mutation in the regulatory genes or in the torA structure gene and be specific to the EDL933 stock strain used in this study. This result is consistent with the high genomic diversity of EHEC O157:H7, as revealed by analysis of 1,102 isolates in Japan (36). The growth of the wild-type EDL933 strain used in this study on the minimal medium is thus supported only by the DMSO reductase. Interestingly, DMSO reductase activities were absent from both the periplasm and the spheroplasts of the ΔtatABC mutant (Fig. 1B, lanes 2 and 5). These results showed that the ΔtatABC mutation impairs both the translocation and the activity of the DMSO reductase of E. coli O157:H7 EDL933. Therefore, the deletion of the tatABC genes seems to have much more severe consequence in EDL933 than in the K-12 strain.

In addition to the failure of growth on minimal medium, tat mutations in the K-12 strain seem to affect the late stage of cellular division. Thus, the mutant cells form chains with various lengths depending on the tat mutants (34). Similarly, we found that the ΔtatABC mutant exhibited an elongated shape and formed chains, in contrast to the single-cell short bacilli of wild-type strains (Fig. 1D). Together, these results confirmed the phenotype expected for the ΔtatABC mutation.

Requirement of the TAT system for virulence of E. coli O157:H7 and secretion of Shiga toxin.

E. coli O157:H7 EDL933 belongs to the group of STEC, which is distinguished from other types of diarrheagenic E. coli by the production of Shiga toxins. The Shiga toxins account for the severe clinical manifestations of STEC infections, including HC and HUS (22). We wondered if the TAT system of E. coli O157:H7 could be involved in the pathogenesis specific to EHEC. Shiga toxins (Stx) display a specific cytotoxicity for Vero cells in tissue culture, hence STEC are also known as VTEC. One standard means to study VTEC virulence is to analyze the specific cytotoxic effect of bacterial culture filtrates on the Vero cells. We prepared broth culture supernatants from the wild-type strain and the mutant and tested for cytotoxic activity in a tissue culture assay system with a Vero cell line. The verotoxin titer of the stx mutant was ≤2-fold dilution, whereas that of the E. coli O157:H7 EDL933 was ≥1/4,092 (1/2¹²) dilution (Fig. 2A). Importantly, the culture filtrates prepared from two independent ΔtatABC colonies both gave a verotoxin titer of 1/512 (1/2⁹) (Fig. 2A), which is about eightfold lower than that of the wild-type strain. These findings strongly suggest the requirement of the TAT system for the specific cytotoxic effect of EHEC O157:H7.

FIG. 2.

Vero cell toxicity and secretion of Stx. (A) Twofold serial dilutions (indicated on top) were prepared from bacterial filtrates of EDL933 (wild type [WT]) or the EDLtat mutant (ΔtatABC) and the stx mutant. The verotoxin titer was expressed as the reciprocal of the highest sample dilution of culture filtrate which caused 50% cell detachment after 24 h of incubation, as judged by the dye intensity (see Materials and Methods). Two independent sets of data regarding EDL933 and the EDLtat mutant are presented. (B) Proteins in culture filtrates of EDL933 (WT, lane 1), the EDLtat mutant (ΔtatABC, lane 2) and the stx mutant (lane 3) were precipitated with 10% trichloroacetic acid and resolved on an SDS-denaturing polyacrylamide (12.5%) gel. Stx1A was visualized by immunoblotting with polyclonal antisera against Stx1 and quantified by Kodak Image Analysis software. Molecular mass markers (MW) are indicated on the right.

The genome of E. coli O157:H7 EDL933 contains both stxAB₁ and stxAB₂ genes, encoding the immunologically distinct Stx1 and Stx2, respectively (13, 24). Each of them consists of a catalytic monomeric A subunit and a specific cell-binding part composed of a homopentameric B subunit (29). We confirmed, by PCR, the presence of both stxA₁ and stxA₂ in EDL933 and their absence from the O157 stx stock strain used in this study (data not shown). Secretion of Stx in the culture filtrates of these strains was analyzed by immunoblotting. The polyclonal antibodies raised against Stx1 recognize a polypeptide with a molecular mass of 30 kDa, which is expected for the StxA₁ subunit (Fig. 2B, lane 1). It was absent from the filtrates of the E. coli O157 stx mutant (Fig. 2B, lane 3). Therefore, it should be the StxA₁ subunit of E. coli EDL933. The 30-kDa StxA₁ band was also detected in culture filtrates of the E. coli O157:H7 ΔtatABC mutant (Fig. 2B, lane 2). However, the intensity of this specific band, quantified by using Kodak image analysis software, was fivefold weaker than that of the corresponding band found in the wild-type strain (Fig. 2B, lane 2 versus lane 1). This result is fully consistent with the reduction of verotoxicity of the mutant. Intriguingly, StxA₁ was absent from the spheroplasts of the ΔtatABC mutant (data not shown). Therefore, as observed for the DMSO reductase, the ΔtatABC mutation seems to affect Stx1 synthesis or stability. Notably, because a strong nonspecific signal covering the area with a molecular mass less than 10-kDa was developed by the same antisera, we could not analyze the Stx1B subunit (7.7 kDa) in these samples.

We attempted to analyze Stx2 secretion in the wild-type strain and the ΔtatABC mutant by immunoblotting. The anti-Stx2 polyclonal antisera did not reveal any specific band in the culture supernatants of these strains (data not shown). It might be due to poor production or release of Stx2 under the conditions used. The production and release of both Stx1 and Stx2 are dependent on bacteriophage induction (21, 39). However, only the stxAB₁ is expressed at the constitutive level independent of phage induction (39). Under the growth conditions used in this study, the stxAB₁ genes might be expressed at uninduced basic levels, whereas the Stx2 synthesized might be too low to be detected.

Synthesis and export of Stx1 in the nonpathogenic E. coli K-12 strain.

The stx genes are located in the late-phase region of the Stx-converting phages, upstream and in the same transcriptional orientation as the S gene that encodes the holin necessary for host lysis (37). It has been demonstrated recently that phage-mediated lysis regulates the quantity of Stx1 produced and provides a mechanism for Stx1 release (39). However, unlike Stx2, growth in low-iron media stimulates Stx1 production from the iron-regulated promoter but does not lead to host lysis. Therefore, Wagner et al. proposed the contribution of unknown phage-independent mechanisms, for example, the type II secretion system, to Stx1 release in vivo under these circumstances (39). Interestingly, in this study we detected the StxA₁ subunit in the culture supernatant of E. coli O157:H7 grown under noninduced conditions, thus confirming the phage-independent release of Stx1. Most importantly, the quantity of StxA₁ detected in the culture supernatant of the ΔtatABC mutant was reduced fivefold compared to the wild-type strain, which was paralleled by the attenuation of virulence of the mutant in a Shiga toxin-specific in vitro assay. Since StxA₁ was absent from the spheroplasts of the ΔtatABC mutant, it is impossible to distinguish an effect of the ΔtatABC mutation on Shiga toxin secretion from that on Shiga toxin synthesis or stability in the EDLtat mutant.

To better understand the effect of the ΔtatABC mutation on Stx1 synthesis and export, we cloned the stxAB₁ genes of EDL933 and studied Stx1 export in the nonpathogenic E. coli K-12 strain. The wild-type K-12 MC4100A strain does not contain lambda phage; release of Stx1 via phage-mediated lysis was thus impossible. In addition, the expression of cloned stxAB₁ genes was under the tight control of the P_BAD promoter; we could dissociate Stx1 export from its synthesis. StxA₁ was undetectable by immunoblot in the culture filtrates of both the wild-type strain and the B1LK0A (ΔtatC) mutant (Fig. 3A, lanes 8 and 9). The anti-Stx1 antisera recognized two bands in the periplasm of the wild-type strain harboring the plasmid pStx1AB. The bigger one was found also in the control strain carrying the vector pBAD24 (Fig. 3A, lanes 1 and 2). In contrast, the smaller one was specific for MC4100A/pSTX1AB and had a molecular mass (30 kDa) expected for StxA₁. Importantly, this specific band was barely detectable in the periplasm of the ΔtatC mutant expressing the cloned stxAB₁ genes (Fig. 3A, lane 3). A nonspecific band was detected at the migration position expected for StxA₁ in the spheroplasts of the three strains analyzed (Fig. 3A, lanes 4 to 6). Because of the poor resolution on the gels, it is unclear whether StxA₁ was present in the spheroplasts.

FIG. 3.

Synthesis and translocation of Stx1 in nonpathogenic E. coli K-12 strains. Culture filtrates and cellular fractions were prepared from the wild-type K-12 MC4100A strain (WT) and its B1LK0A ΔtatC derivative containing pStx1AB grown with arabinose. Samples were either subjected to electrophoresis on SDS denaturing gels and analyzed by immunoblotting with antibodies against Stx1 (A) or assayed for verotoxicity (B) as described in the legend to Fig. 2 and in Materials and Methods. Two sets of independent verotoxicity assays gave the same results. Only one set is presented here. NS, nonspecific band.

The location of Stx1 expressed in nonpathogenic E. coli K-12 strains was further analyzed by verotoxicity assay. As a control, neither the wild type nor the ΔtatC mutant transformed with vector pBAD24 exhibited cytotoxicity (Fig. 3B, lines 7 and 8). In contrast, cytotoxicity was detected in the 2¹⁸-fold dilution of the periplasm of the wild-type strain expressing Stx1 (Fig. 3B, line 3). Therefore, expression of the stxAB₁ genes is sufficient to allow the synthesis of active Shiga toxin in nonpathogenic E. coli. In addition, it shows, for the first time, the periplasmic location of Stx1. Similarly, Shiga toxin was also detected in the periplasm of the ΔtatC mutant, but it was about 16-fold less active than that of the wild-type strain (Fig. 3B, line 4 versus line 3). The Vero cell cytotoxicity was found in 2¹⁵- and 2¹¹-fold dilutions of the extracts prepared from the spheroplasts of the wild type and the ΔtatC mutant, respectively (Fig. 3B, lines 5 and 6). A weak cytotoxicity was detected in the eightfold dilution of the filtrates of the wild type and the ΔtatC mutant expressing the stxAB₁ genes (Fig. 3B, lines 1 and 2). The detected cytotoxicity in the filtrates probably resulted from phage-independent cell lysis.

The Stx1A protein and verotoxicity were significantly reduced in both the O157:H7 and K-12 tat mutants. In the K-12 strain, the Stx1A synthesis was induced by arabinose from the P_BAD promoter of the pBAD24 plasmid. This promoter has been widely used for expressing endogenous or heterologous genes in K-12 tat mutants, and no effect of the tat mutations on the transcription from this promoter has been observed (7, 30). Therefore, the reduction of Stx1A in the tat mutants must be due to a posttranscriptional event, probably proteolysis. It has been observed that various TAT substrates are degraded when the TAT system is nonfunctional or the translocation is mediated by defective signal peptides (7). It is necessary to remove these polypeptides, since accumulation of folded proteins in the cytoplasm may have severe consequences.

The Shiga toxin consists of an enzymatically active A subunit and five identical B subunits (29). The B subunits are responsible for the binding of toxin to glycolipid receptor of specific target eucaryotic cells. After endocytosis, the A subunit inhibits protein synthesis by releasing an adenine base from the 28S rRNA of the 60S ribosomal subunit. Since the spheroplasts of MC4100A/pStx1AB and B1LK0A/pStx1AB displayed Vero cell toxicity, the assembly of subunits A and B to form active holotoxin may occur in the cytoplasm. Because it is capable of exporting an enzyme complex of up to 100 kDa (28), the Tat system would be the most suitable known apparatus to export the ∼70-kDa holotoxin into the periplasm. At present, we could not exclude an in vitro assembly of the holotoxin after breaking the cells. Nevertheless, much less Shiga toxin was detected in the periplasm of the tat mutant than in the wild-type strain. The holotoxin accumulated mainly in the periplasm, which is consistent with the fact that the K-12 strains used here are not lysogens of lambda phage and lack the plasmid-borne type II secretion system. The residual verotoxicity found in the filtrates might result from phage-independent lysis. Together, these results showed that the TAT system plays a very important role in secretion or stability of one major known virulence factor of EHEC, although a direct effect of the tat mutation on Stx1 secretion could not be established.

Effect of the ΔtatABC mutation on protein export into the periplasm of the O157:H7 strain.

The proteins exported by the TAT system are synthesized with twin arginine signal peptides. The twin arginine signal peptides resemble Sec-dependent signal peptides considering their overall structures but possess a twin arginine motif in the N region, a weakly hydrophobic H region, and a Sec-avoidance signal composed of basic residues around the H region. Increasing evidence suggests that signal peptides without the conserved RR motifs can also mediate TAT pathway translocation (15, 16, 25). Therefore, the criterion of the presence of the twin arginine motif in the signal peptide seems not to be sufficient for identifying the proteins that are exported by the TAT system. Interestingly, Stx1B possesses a double lysine motif in the signal peptide, which is compatible with a TAT-dependent translocation under certain contexts (16, 25). It is possible that the TAT pathway might be involved, directly or indirectly, in the translocation of virulence factors synthesized with the signal peptide without the twin arginine motif in E. coli O157:H7. To assess this hypothesis, we compared the periplasmic protein profile of the wild-type EDL933 strain with that of the ΔtatABC derivative by 2DE. The results showed that at least 8 proteins were absent from, or present in a significantly reduced amount, in the periplasm of the mutant (Fig. 4). Six of them were identified by mass spectrometry. They included four periplasmic substrate-binding proteins of ABC transporters, the hyperosmotically inducible protein (OsmY) and the H7 flagellin.

FIG. 4.

Comparison of periplasmic protein profiles obtained by 2DE. Periplasmic fractions (100 μg of protein each) of EDL933 (wild type [WT]) or its ΔtatABC derivative were separated first on 7-cm immobilized pH gradient strips (pH 3 to 10 in panel A, pH 3 to 6 in panel B) and then on SDS denaturing polyacrylamide (12.5%) gels, and proteins were visualized by Coomassie blue staining. Molecular masses (MW, in kilodaltons) and pH ranges are indicated. The spots of which the counterparts are absent or present with significantly reduced quantity in the mutant periplasmic fractions are indicated by white circles. The spots identified by mass spectroscopy are as follows: OppA (oligopeptide binding protein, Est'dZ score of 2.30, 47% covering), Asg2 (asparaginase type II, Est'dZ score of 2.26, 48% covering), RKObp (arginine-lysine-ornithine-binding protein, Est'dZ score of 2.27, 68% covering [pH 3 to 10], Est'dZ score of 2.37, 70% covering [pH 3 to 6]), 3Rbp (arginine third transport system periplasmic binding protein, Est'dZ score of 2.25, 26% covering), OsmY (hyperosmotically inducible protein, Est'dZ score of 2.31, 41% covering), FliC (flagellin, Est'dZ score of 2.14, 67% covering), and GGbp (galactose-glucose binding protein, Est'dZ score of 1.06, 25% covering).

The periplasmic substrate-binding proteins are oligopeptide binding protein (OppA), lysine-arginine-ornithine binding protein (KRObp), galactose-glucose binding protein (GGbp), and the putative third arginine binding protein (3Rbp). The predicated signal peptide of OppA contains a KR motif, and the three others contain a KK motif. These motifs are compatible with TAT-dependent translocation (16, 25). The absence of these proteins from the periplasm of the ΔtatABC mutant would suggest a TAT-dependent translocation, but it needs to be confirmed. Although periplasmic substrate binding proteins are required for chemotactic responses, contributions of the proteins identified in this study to the virulence of E. coli O157:H7 are unknown. osmY gene expression is induced by hyperosmotic stress, and interruption of osmY increases bacterial sensitivity to hyperosmotic stress (43). Adaptation to high osmolarity plays an important role for the survival of pathogens in hosts. The TAT system might contribute indirectly to the adaptation to high osmolarity.

The ΔtatABC mutation affects flagellar biogenesis.

In contrast to the substrate binding proteins and OsmY, the H7 flagellin is an established virulence factor. In addition to the requirement for motility of the pathogen in the hosts, increasing evidence suggests that flagella also enhance pathogenicity either by promoting adherence to host tissues or by directly activating host inflammatory signaling pathways (27). Flagellin, encoded by fliC, is the major structural protein of the flagella of gram-negative bacteria, and it specifies the H antigen of E. coli (27). Sequence analysis of FliC from pathogenic E. coli strains with distinct H antigens has revealed that their N- and C-terminal regions are largely conserved, whereas the central region is highly diverse and distinguishes EHEC O157:H7 or O55:H7 from EHEC O55:H6, O127:H6, and O142:H6 (27). In this study, the mass spectrometry analysis identified not only peptides located in the N- and C-terminal regions but also those specific to H7 flagellin in the central region. Recently, it has been reported that the H7 flagellin of E. coli O157:H7 activates epithelial cell mitogen-activated protein kinase and NF-κB pathways leading to interleukin-8 secretion (3). Therefore, the H7 flagellin is one of the major virulence factors specific to EHEC O157:H7.

Flagellum assembly is a sequential process including a successive secretion of proteins through the flagellar assembly apparatus by a type III mechanism to the distal end, thus adding protein subunits from the most proximal cell to the most distal cell. Intriguingly, FliC was detected in the periplasm of the wild-type EDL933 strain but was absent from the ΔtatABC derivative (Fig. 4B). The periplasmic fractions were prepared by the lysozyme-EDTA-cold osmoshock method. Lysozyme is a muramidase and is used to destroy the peptidoglycan in this protocol. Since flagella are held by P ring and L ring located, respectively, in the peptidoglycan and outer membrane layers, this protocol might destabilize the flagella and thus release the flagellin into the periplasmic fraction. To test this hypothesis and to study the impact of the ΔtatABC mutation on flagellin synthesis, flagella were extracted from the wild-type strain and the ΔtatABC mutant. Cellular fractionation was performed after the extraction. These fractions were resolved on SDS-denaturing gels, proteins were visualized by Coomassie blue staining, and flagellin was detected by immunoblotting. As shown in Fig. 5A1, a strong protein band with a molecular mass expected for flagellin (∼66 kDa) was detected in the flagellar fraction of the wild-type strain (Fig. 5A1, lane 5), indicating the high efficiency of flagellum extraction. In contrast, no obvious flagellin band was revealed in the flagellar fraction of the mutant (Fig. 5A1, lane 6). Moreover, more proteins were detected in the mutant fraction than in the corresponding fraction of the wild-type strain (Fig. 5A1, lane 6 versus lane 5), which must result from the leakage of periplasmic proteins of the EDL933 ΔtatABC mutant, as observed for the nonpathogenic E. coli K-12 tat mutants (34). Remarkably, immunoblot analysis showed that flagellin was detected only in the flagellar fraction of the wild-type strain, but it was absent from the periplasm and the spheroplasts of the wild-type strain and all fractions of the mutant (Fig. 5A2). These results showed that lysozyme treatment could release flagella into the periplasmic fractions and, most importantly, suggested that the ΔtatABC mutation may affect flagellin synthesis or stability. Consistent with the absence of flagellin, the ΔtatABC mutant completely lost motility on soft agar plates (Fig. 5B). In addition, the introduction of the p8757 plasmid carrying the K-12 tatABC operon restored the motility of the EDLtat mutant. Therefore, the absence of flagellin and the loss of the motility of the mutant must be the consequence of the ΔtatABC mutation.

FIG. 5.

Impact of the ΔtatABC mutation on FliC synthesis and cellular motility. Periplasmic fractions, spheroplasts, and flagellar fractions of the E. coli O157:H7 EDL933 (wild type [WT]) or its derivative EDLtat mutant (ΔtatABC) were resolved on SDS denaturing polyacrylamide (12.5%) gels and visualized either by Coomassie blue straining (A1) or immunoblotting with polyclonal antisera against the H7 flagellin by an enhanced chemiluminescent procedure (A2). Flagellar fractions were prepared as described in Materials and Methods. The periplasmic fractions and the spheroplasts were prepared from the cells after flagellar extraction. The quantities of fractions loaded on gels correspond to those prepared from 15 mg (flagella) or 0.5 mg (periplasmic fractions and the spheroplasts) (wet weight) of cells. FliC represents flagellin. (B) Motility assay plates of EDL933 (WT), EDLtat (ΔtatABC), and EDLtat complemented by the K-12 tatABC operon (ΔtatABC/p8737).

The expression of over 50 genes that encode structural subunits of flagella, regulatory proteins, motor force generators, and the chemosensory machinery is hierarchically controlled based on three groups of promoters corresponding to their temporal requirement during the flagellum assembly process (2). Secretion and polymerization of FliC to form the flagellar filament are among the latest events, and the fliC gene is expressed from a class III promoter depending on an alternative sigma factor, σ²⁸ (2). Flagellin was not detected in the ΔtatABC mutant. From this result, it is unclear whether FliC was synthesized at an extremely lower level in the mutant or if it was degraded due to the absence of a functioning TAT system. Analysis of the flagellar protein sequences revealed that the processed FliP signal peptide contains a twin arginine motif. FliP is an integral membrane protein located in the basal body of the flagellar assembly apparatus and is essential for the flagellar protein secretion (2). It is tempting to speculate that the ΔtatABC mutation might affect the FliP membrane insertion. FlgM is an anti-σ²⁸ factor, and its secretion through the flagellar assembly apparatus outside of the cells triggers transcription from the class III promoters, including that of fliC (2). The impairment of FliP membrane insertion could result in the inhibition of the fliC gene expression via the accumulation of FliM in the cytoplasm. At present, we have assessed this hypothesis.

Concluding remarks.

Recently, it has been reported that the TAT system of P. aeruginosa is essential for the export of phospholipases, proteins involved in pyoverdine-dedicated iron uptake, anaerobic respiration, osmotic stress defense, motility, and biofilm formation (23, 38). In addition, the corresponding tat mutant cells are attenuated for virulence. Similarly, the TAT system of the phytopathogen A. tumefaciens contributes to flagellar biogenesis, chemotactic responses, and virulence (10). In this study, we revealed for the first time that the TAT system is required for the secretion of Shiga toxin and synthesis of the H7 flagellin, which are two of the major virulence factors specific to EHEC O157:H7. Together, these findings indicate that the TAT system is an important virulence determinant of both human pathogens and phytopathogens. Since the tat coding genes are absent from animal genomes (41), it could represent a potential target for novel antimicrobial compounds.