Type III secretion systems, designed to deliver effector proteins across the bacterial cell envelope and the plasma membrane of the target eukaryotic cell, are involved in subversion of eukaryotic cell functions in a variety of human, animal and plant pathogens. In enteropathogenic Escherichia coli (EPEC), several protein substrates for the secretion apparatus were identified, including EspA, EspB and EspD. EspA is a structural protein and the major component of a large transiently expressed filamentous surface organelle that forms a direct link between the bacterium and the host cell, whereas EspD and EspB seem to form the mature translocation pore. Recent studies of the type III secretion systems of Shigella and Salmonella pathogenicity island (SPI)-1 revealed the existence of a macromolecular complex that spans both bacterial membranes and consists of a basal structure with two upper and two lower rings and a needle-like projection that extends outwards from the bacterial surface. MxiH (Shigella) and PrgI (Salmonella) are the main components of the needle of the type III secretion complex. A needle-like complex has not yet been reported in EPEC. In this study, we investigated EscF, a protein sharing sequence similarity with MxiH and PrgI. We report that EscF is required for type III protein secretion and EspA filament assembly. Moreover, we show that EscF binds EspA, suggesting that EspA filaments are an extension of the type III secretion needle complexes in EPEC.
Induction of an ‘attaching and effacing’ (A/E) lesion on the intestinal mucosa is a pathogenic mechanism shared by a number of enteric human and animal pathogens (reviewed by Frankel et al., 1998; Kaper et al., 1998). The A/E lesion is characterized by localized destruction (effacement) of brush border microvilli, intimate attachment of the bacillus to the host cell plasma membrane and the formation of an underlying pedestal-like structure in the host cell. A/E lesions were first described for strains of enteropathogenic Escherichia coli (EPEC) (Ulshen and Rollo, 1980; Moon et al., 1983), an established aetiological agent of human infantile diarrhoea in the developing world (Nataro and Kaper, 1998). Similar intestinal lesions were later associated with other enteric mucosal pathogens including enterohaemorrhagic E. coli (EHEC) (Tzipori et al., 1986), a food-borne pathogen of industrialized countries that is often associated with a life-threatening complication, the haemolytic uraemic syndrome (Nataro and Kaper, 1998).
The genes encoding the A/E phenotype are encoded on a pathogenicity island termed the locus of enterocyte effacement (LEE; McDaniel et al., 1995; Perna et al., 1998). Both EPEC and EHEC LEE contain 41 open reading frames (ORFs). The majority of the genes are organized in five polycistronic operons (LEE1, LEE2, LEE3, tir and LEE4;Mellies et al., 1999). LEE1, LEE2 and LEE3 encode components of a type III secretion system (TTSS), a macromolecular complex spanning both bacterial membranes (Elliott et al., 1998; Frankel et al., 1998). The tir operon encodes the outer membrane adhesion molecule, intimin, the translocated intimin receptor (Tir) and CesT (the Tir chaperone) (Abe et al., 1999; Elliott et al., 1999), and the LEE4 operon encodes proteins secreted by the TTSS termed E. coli secreted proteins (Esps) (reviewed by Frankel et al., 1998), including EspA, EspD, EspB and EspF.
Using EspA antiserum to visualize the EspA protein on EPEC during growth and A/E lesion formation, we demonstrated that EspA is a structural protein and the major component of a long (12 nm diameter) transiently expressed filamentous surface organelle that forms a direct link between the bacterium and the host cell (Knutton et al., 1998). After bacterial attachment, EspB is translocated into the host cell, where it is localized to both membrane and cytosolic cell fractions (Taylor et al., 1998; Wolff et al., 1998). EspB exhibits weak homology with YopD (from Yersinia; 19% identity), and the putative structural organization of EspB is reminiscent of the YopD protein. Both proteins have only one putative transmembrane region and one predicted trimeric coiled-coil region (Pallen et al., 1997). YopD, like EspB, is required for the translocation of effector proteins, but is itself also translocated (Francis and Wolf-Watz, 1998; Wolff et al., 1998). The fact that EspA and EspB are required for the translocation of effector proteins to the host cell suggests that they may both be components of the translocation apparatus. Indeed, we showed recently that EspB can bind and be co-purified with EspA (Hartland et al., 2000). Significantly, we also showed that intact EspA filaments can be observed on the surface of an EPEC espB mutant and that binding of EspA filaments to the target host cell occurred even in the absence of EspB (Hartland et al., 2000). This suggests that EspA–EspB protein interaction is a subsequent event that allows protein translocation, and that translocation of EspB modulates EspA filament activity and signals the transition from an adhesive to a translocation function.
Previous studies have shown that proteins belonging to the EspD family (YopB from Yersinia, IpaB from Shigella) are part of the type III secretion apparatus involved in the formation of a translocation pore in the host cell membrane (Blocker et al., 1999; Neyt and Cornelis, 1999). We have reported previously that, although EspD does not appear to be a structural component of the EspA filament, an EPEC espD mutant secretes only low levels of EspA and produces barely detectable EspA filaments (Knutton et al., 1998). EspD is translocated into the host cell membrane and is required for cell attachment (Wachter et al., 1999) and EPEC-induced haemolysis (Warawa et al., 1999). Recently, we showed EspD–EspD protein interaction and demonstrated that a radical mutation in the C-terminus coiled-coil domain of EspD resulted in significantly reduced EspA filament-mediated cell attachment and significantly reduced EPEC-induced haemolysis without any detectable effect on EspA filament biosynthesis (Daniell et al., 2001). These observations suggest that EspD may have multiple activities: to influence the length of the EspA filament; to promote adhesion during early stages of infection; and to generate the translocation pore once contact has been established. Integrating these data into a model representing our current understanding of the EPEC/EHEC type III protein translocation apparatus predicts that EspA filaments are hollow cylindrical structures connecting a bacterial membrane-associated TTSS with a host cell EspD/EspB translocation pore, through which the effector proteins [Tir (Kenny et al., 1997) and EspF (McNamara and Donnenberg, 1998)] are translocated.
Recent electron microscopic studies of the Shigella (Blocker et al., 1999; 2001; Tamano et al., 2000) and Salmonella pathogenicity island (SPI)-1 TTSSs revealed the existence of a macromolecular complex that spans both bacterial membranes and consists of a basal structure with two upper and two lower rings, which are connected by a cylindrical structure, and a needle-like projection (composed of MxiH and PrgI respectively) that extends outwards from the bacterial cell surface. Although a needle-like complex has not yet been reported in EPEC or EHEC, we have recently observed, using our newly developed red blood cell (RBC) infection model (Shaw et al., 2001), a short ≈ 50 nm section of the EspA filament structures involved in cell attachment, adjacent to the bacterial surface, that did not label with the EspA antiserum (Shaw et al., 2001). We hypothesize that the part of the EspA filaments not stained with the EspA antiserum is homologous to the needle-like structures seen in Shigella and Salmonella. Based on its similarity with PrgI and MxiH, we hypothesised that EscF, the sixth gene in the LEE4 operon (Elliott et al., 1998), is the main component of this substructure in the EPEC/EHEC TTSS. The aim of this study was to investigate the role of EscF in TTSS protein secretion and translocation and in A/E lesion formation.
EscF is required for A/E lesion formation and protein secretion
Sequence analysis revealed that EscF is a 73-amino-acid-long polypeptide with a molecular weight of 8.127 kDa and a pI of 4.32. Based on sequence similarity to the needle complex proteins MxiH and PrgI of Shigella and Salmonella, respectively (Fig. 1), we investigated the effect of an insertion inactivation mutation in escF on the function of the EPEC (strain E2348/69) TTSS and the outcome of the interaction between EPEC and mammalian cells. A non-polar aphT kanamycin cassette was introduced as described in Experimental procedures, producing strain ICC171. The ability of ICC171 to induce A/E lesions and actin polymerization in cultured epithelial cells was investigated using the FAS test. After a 3 h infection of HEL cells, we observed that ICC171 was deficient in the reorganization of the host cell cytoskeletal proteins (Fig. 2). In this assay, the phenotype of ICC171 was similar to mutations in a number of other LEE4 genes, including espA(Fig. 2). However, the wild-type A/E phenotype was restored when escF was introduced on a multicopy number plasmid, pICC196 (Fig. 2). These results indicate that EscF is either an effector protein essential for A/E lesion formation or an essential component in the putative hollow conduit of the TTSS involved in effector protein translocation.
In order to determine the effect of the escF mutation on the secretion of the Esp proteins, supernatants of bacterial cultures, grown under conditions that favour the secretion of type III proteins in EPEC, were analysed by Western blots with EspA, EspB and EspD polyclonal sera. Unlike the parent E2348/69, we found that none of these proteins could be detected in the culture supernatant (Fig. 3). Secretion was regained once pICC196 was introduced into ICC171 (Fig. 3). These results show that EscF is directly involved in the secretion process of the Esps.
EscF is required for EspA filament biosynthesis and Tir translocation
We have reported recently on a new infection model for EPEC using monolayers of RBCs (Shaw et al., 2001). Moreover, we have shown that binding of EPEC to RBCs is mediated exclusively by EspA filaments (Shaw et al., 2001). Significantly, by gold labelling electron microscopy (EM), there appeared to be a short ≈ 50 nm section of the EspA filament structure adjacent to the bacterial surface that did not label with the EspA antiserum (Fig. 4). We hypothesized that this could be the equivalent external needle structure of the EPEC TTSS onto which EspA filaments are attached.
In order to investigate the effect of the escF mutation on the formation and function of EspA filaments, infected HEL cells were immunostained with polyclonal EspA antisera. Unlike the parent strain, no EspA filaments were observed on the surface of ICC171 (Fig. 2). This phenotype was consistent with the fact that ICC171 did not adhere to RBC monolayers (Fig. 2) or translocate Tir to the membrane of RBCs (Fig. 2). However, EspA filament biosynthesis, attachment to the RBC monolayers and translocation of Tir were all reconstituted in strain ICC171pICC196, although EspA filaments were now only observed on a relatively small fraction of bacterial cells, and the filaments were longer than in the wild-type control (Fig. 2). These results suggest that EscF is a structural protein involved in the assembly of EspA filaments.
In order to visualize EscF, escF was cloned into pET28a (generating plasmid pICC197), and expression of the recombinant His-tagged protein was induced by IPTG. As the recombinant protein was insoluble, 6 M guanidine was added to sonicate bacterial extracts before loading onto a nickel column. The eluted material was dialysed, and the precipitated powder was used as an antigen for antiserum production in both mice and a rabbit. Unfortunately, using the immune sera, we could not detect EscF in culture supernatants, whole bacterial extracts and enriched bacterial membranes (data not shown). Moreover, in immunostaining, the antiserum also failed to react with EscF on the bacterial cell surface of wild-type and espA− (UMD872) backgrounds (data not shown). In order to exclude the possibility that the antisera were unable to recognize the EscF protein itself, we cloned escF into pMAL-c2 (pICC198), generating a transcription fusion (maltose binding protein–EscF; MBP–EscF). After purification and Western blotting, we observed strong reactivity of the anti-EscF antiserum against the MBP–EscF fusion, whereas no reactivity was observed against MBP only (Fig. 5). These results show that the EscF antiserum was specific and indicated that our inability to detect EscF might result from the low level of EscF in EPEC or the immunization performed with denatured EscF polypeptide incapable of binding to the native protein.
EscF binds EspA
Despite the fact that we were unable to visualize EscF at the bacterial surface, we investigated the ability of EscF to bind EspA. We have already reported a number of homo- and heteroprotein interactions between different Esps (including EspA–EspA, EspA–EspB and EspD–EspD) using the yeast two-hybrid system and biochemical protein-binding assays. In this study, we used column pull-down experiments to determine EscF–Esp interaction. For this, an MBP–EscF fusion or MBP alone was immobilized on agarose columns and assessed for their ability to interact specifically with, and retain, any of the Esps in EPEC culture supernatants subsequently washed through. The eluted material was subjected to Western blots using anti-EspA, -EspB and -EspD antisera. Only EspA was retained and co-eluted with MBP–EscF, whereas none of the Esps co-eluted with MBP alone (Fig. 6A). The validity of this result was further demonstrated by showing that an MBP–EscF column also retained His–EspA (Fig. 6B). These results demonstrate EspA–EscF interaction and support the hypothesis that EspA filaments are, in fact, an extension to the EPEC needle complex.
Over the past several years, it has become apparent that TTSSs are involved in the subversion of eukaryotic cell functions in a variety of human, animal and plant pathogens, including species of Bordetella, Burkholderia, Chlamydia, Erwinia, E. coli, Pseudomonas, Ralstonia, Rhizobia, Salmonella, Shigella, Xanthomonas and Yersinia. TTSSs are designed to deliver effector proteins across the bacterial cell envelope and the plasma membrane of the target eukaryotic cell (Lee, 1997; Hueck, 1998). The components responsible for the secretion of proteins across the bacterial cell envelope are broadly conserved in all TTSSs, so that one TTSS can often export proteins usually secreted by another type III system.
A needle-like complex has not yet been reported in EPEC or EHEC. However, EscJ and EscC (encoded on the LEE1 operon) show sequence homology with the inner and upper ring proteins, respectively, of the basal TTSS complex of Shigella and Salmonella, and EscF (encoded on the LEE4 operon) is homologous to the needle proteins MxiH and PrgI. Accordingly, it is likely that EPEC assembles a similar secretion apparatus. However, there is a special interest in studying the basal TTSS complex in EPEC because, in this enteric pathogen, there appears to be an extension to the TTSS (EspA filaments), which is absent in Shigella and Salmonella. This and the fact that, in EPEC, direct association of TTSS with the plasma membrane can be seen and studied makes the EPEC TTSS novel and a worthwhile object for detailed molecular investigation. In this report, we studied the role of EscF in EspA filament assembly, Esp protein secretion, protein translocation and interaction of EPEC with mammalian cells.
Here, we confirmed by immunogold labelling that there appeared to be a short ≈ 50 nm section of the EspA filament structure adjacent to the bacterial surface that did not label with the EspA antiserum, which we suggest could be the equivalent external needle structure of the EPEC type III secreton onto which EspA filaments are attached. Unfortunately, our EscF antibodies did not stain any needle-like structures at the surface of wild-type EPEC, espA− (strain UMD872) or strain ICC171 (pICC196), nor were we able to detect EscF in Western blots. As we have shown that the antisera reacted with MBP–EscF, it is possible that the negative results are due to the fact that antibodies against the recombinant, solubilized EscF are not cross-reactive with the native protein or to the low quantity of EscF in EPEC. Indeed, if the number of TTSSs correlates with the number of EspA filaments (12 per cell), then the number of needle complexes expressed by EPEC is much lower than the numbers expressed by Salmonella and Shigella.
Mutation in escF abolished Esp protein secretion. This is in contrast to an espA mutant that secretes wild-type levels of EspB but reduced levels of EspD, implying that, in the absence of EscF, the putative hollow conduit in the basal body is either blocked or inactive. Importantly, mutation in mxiH in Shigella (Tamano et al., 2000; Blocker et al., 2001) and in prgI in Salmonella (Kimbrough and Miller, 2000; Kubori et al., 2000) resulted in the abolition of secretion of IpaB, IpaC and IpaD and SipB and SipC respectively.
Consistent with the fact that EscF is required for EspA filament assembly, the escF mutant was FAS test negative and failed to translocate Tir into RBC membranes. These phenotypes were restored when escF was introduced in trans on a multicopy number plasmid. Importantly, only a small proportion of the complemented strain [ICC171(pICC196)] expressed EspA filaments, which, although considerably longer than wild-type EspA filaments, retained the biological function of wild-type structures. The reason why overexpression of EscF leads to the production of elongated EspA filaments is not clear but, interestingly, overexpression of EspD leads to a similar phenotype (Daniell et al., 2001). Furthermore, a recent report by Tamano et al. (2000) showed that overexpression of MxiH leads to the production of elongated needles on the surface of Shigella that were capable of protein translocation.
Our working model of the EPEC TTSS predicts that EspA filaments are hollow cylindrical structures connecting a membrane-associated needle-like complex with an EspD/EspB translocation pore, through which effector proteins (Tir, EspF) are translocated. On this basis, we predicted that EspA would bind directly or indirectly to the EscF needle structures. This study has shown that both recombinant and native EspA can bind directly to recombinant EscF, a result that provides an additional layer of evidence supporting our model; a relevant part of this is presented in Fig. 7. A great challenge for future work is to isolate the needle-like structure–EspA filament complexes from EPEC and identify its components.
Growth and maintenance of bacterial strains and plasmids
The strains used in this study included EPEC strains E2348/69 (wild type), UMD872 (espA−) (Kenny et al., 1996) and the laboratory E. coli strains TG1, BL21 and JM109. The strains were grown in Luria–Bertani (LB) broth supplemented with ampicillin (50 µg ml−1) as required. The plasmids used in the study are listed in Table 1.
pICC193 with 120 bp in frame deletion of escF gene
pICC194 with insertion of aphT cassette at basepair 37 of escF gene
ptrc99a expressing wild-type escF
pET28a expressing escF
pMAL-c2 expressing escF
Construction of escF mutant strain ICC171
Primers used for polymerase chain reaction (PCR) are listed in Table 2. escF (222 bp) and a flanking 1 kb sequence were amplified using an Expand high-fidelity PCR kit (Roche) and primers escF BW-1 and escF BW-2 (designed on the basis of the LEE region sequence accession number AFO2223). The PCR product was cloned into pGEM-Teasy (Promega), generating plasmid pICC193, and sequenced. Primers were designed for inverse PCR (escF BW-3 and escF BW-4), which resulted in an in frame deletion of a 120 bp central region of the escF gene (from basepair 37 to 157) with the incorporation of a SnaB1restriction site creating plasmid pICC194. To facilitate the identification of E. coli containing the mutated gene, a non-polar aphT cassette, which conferred kanamycin resistance, was incorporated into the SnaB1site of pICC194 creating plasmid pICC195.
Table 2. Primers used in this study (5′−3′).
In order to enhance allelic exchange, plasmid pKD46, an easily curable low-copy-number plasmid that expressed lambda red recombinase, described by Datsenko and Wanner (2000), was transformed into E2348/69 by electroporation generating strain E2348/69(pKD46). pKD46 has been shown to aid chromosomal recombination of foreign DNA in E. coli K-12 (Datsenko and Wanner, 2000). pICC195 was used as the template for PCR using primers escF BW-1 and escF BW-2, resulting in a linear product containing the escF gene with an in frame 120 bp deletion, aphT cassette and the flanking 1 kb DNA sequences. The product was transformed by electroporation into strain E2348/69(pKD46). Clones were grown on kanamycin plates to select for kanamycin resistance. pKD46 was cured by growth at 37°C. The mutation in escF, strain ICC171, was verified by PCR and DNA sequencing using combinations of primers escF BW-1 to escF BW-7 and aphT-R.
Cloning of wild-type escF and complementation of strain ICC171
escF was amplified by PCR using primer pair escF BW-5 and escF BW-6 and cloned into ptrc99a (Pharmacia), a low-copy-number expression vector with an IPTG-inducible promoter, generating plasmid pICC196. The plasmid was transformed into the escF mutant strain ICC171 using electroporation creating strain ICC171(pICC196). Complementation of the non-polar mutant was verified by restoration of a wild-type phenotype, shown by FAS test, immunofluorescence and Esp protein secretion profiles.
Purification of His-tagged EscF
escF was amplified by PCR using primers escF BW-5 and escF BW-6 and cloned into pET28a for expression of an in frame 6-His-tagged fusion protein generating plasmid pICC197. Log phase cultures were induced by the addition of 0.5 mM IPTG and incubated for a further 4 h at 30°C. After sonication, 6 M guanidine was added, and the His-tagged EscF was purified on a nickel column (Novagen). Fractions (1 ml) were collected, pooled and the guanidine removed by dialysis. Solid EscF protein precipitate was collected and stored in aliquots.
Raising EscF polyclonal antisera in mice and rabbits
Solid His–EscF antigen was sent to Abcam for commercial antibody production in mice. For rabbit antisera, female sandy half-lops were immunized subcutaneously with 80 µg of solid His–EscF protein resuspended in an equal ration of PBS and complete Freund's adjuvant, followed by two boosts at 2 week intervals. Antisera were tested by enzyme-linked immunosorbent assay (ELISA) and Western blots.
HEL cell adhesion
Adhesion to human embryonic lung (HEL) cells was carried out according to the method of Cravioto et al. (1979). Subconfluent cell cultures on glass coverslips were washed and incubated with bacteria (10 µl of bacterial broth culture ml−1 DMEM−2% FCS) for 3 h at 37°C. After thorough washing to remove non-adhering bacteria, coverslips were fixed in 4% formalin.
Haemolysis of human RBCs was carried out as described previously (Shaw et al., 2001). RBCs were obtained from human type O blood by centrifugation, washed three times in PBS, and a 3% suspension was added to polylysine-coated 30-mm-diameter tissue culture dishes for 20 min. Unattached RBCs were removed by washing with PBS, and the resulting RBC monolayer was covered with 2 ml of HEPES-buffered DMEM without phenol red. Forty microlitres of an overnight Luria broth culture of EPEC was added to each RBC monolayer, and the dishes were incubated for 4 h at 37°C. After sedimentation of bacteria from the culture medium, supernatants were monitored for release by measuring the optical density at 543 nm. For microscopy, RBC monolayers were washed three times with PBS and fixed with 4% formaldehyde.
A/E lesion formation. A/E lesion formation was assessed using the fluorescence actin staining (FAS) test (Knutton et al., 1989). Fixed HEL cell preparations were washed, permeabilized with 0.1% Triton X-100 for 4 min and cytoskeletal actin stained with a 5 µg ml−1 solution of fluorescein isothiocyanate (FITC)-conjugated phalloidin (Sigma) for 20 min. Coverslips were mounted and examined by incident light fluorescence and phase contrast; A/E lesion formation was indicated by intense actin fluorescence at the site of bacterial adhesion.
EspA, EscF and Tir immunofluorescence For microscopy, RBC monolayers were prepared on glass coverslips and, for Tir staining, RBC monolayers were permeabilized with 0.1% Triton X-100 for 4 min. All antibody dilutions and immune reactions were carried out in PBS containing 0.2% bovine serum albumin (PBS–BSA). Formalin-fixed and washed HEL cell or RBC monolayers were incubated with EspA (Knutton et al., 1998), EscF or Tir-N (Hartland et al., 1999) antisera (1:50–1:100) in PBS–BSA for 45 min at room temperature. After three 5 min washes in PBS, samples were stained with either FITC-conjugated goat anti-rabbit IgG or goat anti-mouse IgG (Sigma) diluted 1:20 in PBS–BSA for 45 min; red cell preparations were simultaneously stained with wheat germ agglutinin (WGA) conjugated to Texas red (Molecular Probes) in order to visualize the red cell membrane. Preparations were washed a further three times in PBS, mounted in glycerol–PBS and examined by incident light fluorescence using a Leitz DMR microscope.
Immunogold labelling of EspA filaments and examination by transmission and scanning electron microscopy was carried out as described previously (Knutton et al., 1998; Hartland et al., 1999).
Preparation of secreted proteins for Western blots
Maltose binding protein (MBP)–EscF affinity columns
To determine possible protein interactions between EscF and secreted EPEC proteins, we constructed a maltose-binding protein (MBP) fusion with EscF. The constructs were made by PCR from E2348/69. EcoRI–SalIended PCR products were obtained using the primer combination escFBW5 and escFBW6. The PCR products were cloned into pMal-c2 (NEB) generating plasmid pICC198. Column overlay experiments were carried out as described previously (Daniell et al., 2001). Briefly, the recombinant plasmids were transformed into E. coli TG1, and log phase cultures were induced with 1 mM IPTG, followed by incubation at 37°C for 3 h with shaking. For affinity columns, MBP or MBP–EscF was bound to amylose resin according to the standard purification procedure (Hartland et al., 2000). The columns were then overlaid with 25 ml of filtered culture supernatant from EPEC strain E2348/69 grown overnight in DMEM. After washes with 10 volumes of column buffer (50 mM Tris-Cl, pH 7.4, 200 mM NaCl, 1 mM EDTA), MBP or MBP–EscF and associated proteins were eluted with 10 mM maltose dissolved in column buffer. Fractions were collected in 1 ml volumes, and 15 µl of each fraction was subjected to SDS–PAGE and immunoblotting with anti-MBP or anti-EspA, -EspB and -EspD sera and alkaline phosphatase-conjugated anti-rabbit antibodies.
We thank Dr B. L. Wanner, Purdue University, for the λ red system, and Dr Ariel Blocker for useful discussions. This project is supported by the Wellcome Trust.