Type III secretion (T3S) from enteric bacteria is a co-ordinated process with a hierarchy of secreted proteins. In enteropathogenic and enterohaemorrhagic Escherichia coli, SepL and SepD are essential for translocator but not effector protein export, but how they function to control this differential secretion is not known. This study has focused on the different activities of SepL including membrane localization, SepD binding, EspD export and Tir secretion regulation. Analyses of SepL truncates demonstrated that the different functions associated with SepL can be separated. In particular, SepL with a deletion of 11 amino acids from the C-terminus was able to localize to the bacterial membrane, export translocon proteins but not regulate Tir or other effector protein secretion. From the repertoire of effector proteins only Tir was shown to bind directly to full-length SepL and the C-terminal 48 amino acids of SepL was sufficient to interact with Tir. By synchronizing induction of T3S, it was evident that the Tir-binding capacity of SepL is important to delay the release of effector proteins while the EspADB translocon is secreted. The interaction between Tir and SepL is therefore a critical step that controls the timing of T3S in attaching and effacing pathogens.
Enterohemorrhagic Escherichia coli (EHEC) O157:H7 is the main EHEC serotype associated with outbreaks of gastrointestinal disease in North America, parts of Europe and Japan. It is an important pathogen that can be life-threatening particularly in the young and the elderly. EHEC strains express a type III secretion (T3S) system encoded on the locus of enterocyte effacement (LEE) pathogenicity island (Jerse et al., 1990; McDaniel et al., 1995). The main phenotype associated with EHEC O157 T3S is the formation of attaching and effacing (A/E) lesions (Knutton et al., 1989; 1998). A/E lesions require the injection of the translocated intimin receptor (Tir) into the enterocyte (Kenny et al., 1997; Deibel et al., 1998) and the cytoskeleton rearrangements associated with A/E lesions require Tir and EspFu/TccP (Campellone et al., 2004; Garmendia et al., 2004) that act through N-WASP activation of the Arp2/3 complex leading to alpha-actinin polymerization (Garmendia et al., 2005).
The T3S system is evident as a needle-like projection on the surface of the bacterium. Its predicted structure is considered to have the proteins EscD (Pas), R, S, T, U and V in the inner membrane joined to an outer membrane ring of EscC, with EscF forming a needle-like structure at the base of the EspA filament (Daniell et al., 2001; Yip et al., 2005). The export of Tir and other effector proteins occurs through the hollow filament, with the timing and regulation of translocation potentially driven by chaperone proteins (Wainwright and Kaper, 1998; Elliott et al., 1999; Neves et al., 2003). The filament is considered to link to a pore in the epithelial cell membrane that is composed of the bacterial protein EspD, and possibly EspB (Wachter et al., 1999). Export of effector proteins into the host cell is therefore dependent on the initial export and assembly of EspADB. The conduit that these three proteins form has been referred to as a ‘translocon’ and the proteins as ‘translocators’ (Deng et al., 2005).
The LEE pathogenicity island contains at least 41 genes in five main operons. espADB are expressed from the LEE4 operon and are preceded in this operon by sepL. sepL encodes a protein composed of 351 amino acids (aa) with a predicted molecular weight of 39.95 kDa (Kresse et al., 2000). Like many genes in the LEE it is highly conserved between EHEC and enteropathogenic E. coli (EPEC) strains (93.7–94.3% identity). SepL is central to a switch that occurs between the export of EspADB and the export of effector proteins. Original research by Kresse et al. (2000) indicated that SepL was associated predominately with the bacterial membrane and in particular the outer membrane. Research using EPEC O127 has demonstrated that the majority of SepL is either in the cytoplasm (O'Connell et al., 2004) or in both the cytoplasm and the inner membrane (Deng et al., 2005).
How SepL works to govern the switch between translocon and effector protein secretion is not known, but both yeast two-hybrid and in vitro studies have demonstrated that SepL interacts directly with SepD, a protein encoded on LEE2 (Creasey et al., 2003; O'Connell et al., 2004). A deletion of sepD has a similar phenotype to a sepL mutant, in that the export of translocators but not effectors is prevented (Deng et al., 2004). In fact, deletion of either sepL or sepD leads to an increase in the levels of secreted Tir and other effector proteins and this increased effector protein secretion is not considered to be controlled at the transcriptional level (Deng et al., 2004; 2005). A recent hypothesis is that SepL and SepD act as a gate to allow translocator export, but that this gate dissociates or changes in response to a drop in calcium levels following the opening of a conduit to the host cell, allowing effectors to be exported (Deng et al., 2005). However, there is no evidence that SepL or SepD interacts directly with the translocon proteins EspA, D or B. The majority of SepL (267 aa) shares some homology with YopN and its carboxy terminus (83 aa) homology with TyeA, both from Yersinia spp. (Pallen et al., 2005). As TyeA controls the secretion of specific effector proteins in Yersinia spp. (Cheng and Schneewind, 2000; Day et al., 2003; Sundberg and Forsberg, 2003) it was decided to investigate full-length and carboxy-terminal truncates of SepL in terms of the known activities of SepL including localization, SepD binding, translocon export and Tir secretion control.
SepL is homologous to a combination of YopN and TyeA from Yersinia spp. (Pallen et al., 2005; Fig. 8). The carboxy terminus of SepL (final 83 aa) is homologous with TyeA and so carboxy-terminal truncates of SepL fused to eGFP were tested to determine if these deletions can separate the different functions of SepL. The functions analysed were the capacity to: (i) localize to the bacterial membrane, (ii) bind to SepD, (iii) restore EspD secretion in sepL mutants and (iv) reduce Tir secretion in sepL mutants. Five truncated proteins were initially constructed as illustrated in Fig. 1A.
Membrane localization and the role of SepD
Previous research has demonstrated that SepL in EHEC O157 can be membrane-associated and can also bind to SepD (Kresse et al., 2000; Creasey et al., 2003; O'Connell et al., 2004). The different-length fusion constructs were examined for their presence in bacterial membrane-containing fractions. The full-length SepL–eGFP construct was present in membrane-containing preparations of bacteria prepared from T3S-permissive conditions while this was not the case for eGFP alone, which was only detectable in the cytoplasmic fraction (Fig. 1B). Deletion of either 11 aa or 61 aa from the carboxyl end of SepL (creating 340 aa and 290 aa proteins fused to eGFP) still allowed detection in the membrane-containing fractions, although this distribution was prevented by any further truncation with detection only in the cytoplasmic fraction. As controls, the distributions of OmpA and GroEL in the same samples were as expected (OmpA > 90% in the membrane-containing fraction; GroEL > 99% in the cytoplasmic fraction) (Fig. 1B). Direct imaging of individual bacteria containing the full-length SepL–eGFP clearly showed a higher concentration of fluorescence localized to the periphery of the bacteria (Fig. 1C). This was not the case for bacteria expressing just eGFP (Fig. 1C). The biochemical and imaging data indicated that SepL localizes to the bacterial membrane and this association does not require the carboxy-terminal 61 aa of SepL.
We next wanted to determine what effect sepD deletion would have on the localization of SepL. sepD was deleted by allelic exchange and the construct confirmed by complementation with sepD on a plasmid (pDW20, data not shown). Localization of SepL–eGFP was examined using a combination of Western blotting and single cell imaging (Figs 2A and B and 1C). The proportion of the full-length SepL–eGFP hybrid protein that was in the membrane-containing fraction was reduced in a sepD background (Fig. 2A). In addition, in this background, the localization of SepL–eGFP was clearly different in the bacteria. The distribution now appeared asymmetric (Fig. 2B), unlike the distribution of eGFP alone, indicating that in the absence of SepD, SepL may associate with another cellular protein that exhibits this asymmetric distribution. The different SepL truncates fused to eGFP were then assessed for their capacity to bind to purified GST–SepD attached to a glutathione-sepharose 4B column. Full-length SepL as well as the 340 aa and 290 aa fusions bound to SepD but further truncation from the carboxy terminus prevented the interaction (Fig. 3). The SepL truncates that localized to the membrane are the same as those that bound to SepD. Given that deletion of sepD also reduced membrane association of SepL and altered its cellular distribution, it is evident that SepD is probably required for SepL localization to the bacterial membrane.
Complementation of translocon (EspD) export in a sepL mutant
SepL is required for translocon but not effector protein export and translocon export can be complemented in a sepL mutant by supplying sepL in trans (Kresse et al., 2000). This was also the case in this study in which the full-length SepL–eGFP fusion could complement a complete sepL deletion mutant for EspD secretion (Fig. 4A). In our hands this complementation never achieved the EspD secretion levels of the wild-type strain and this was also the case for complementation with sepL alone, pDW24 (data not shown). Analysis of the SepL truncates in the sepL deletion strain indicated that only the full-length protein and the 11 aa carboxy-terminal deletion were able to export EspD. Deletion of 11 aa from SepL partially complemented the sepL deletion for EspD secretion (Fig. 4A). It was interesting to note that a deletion of 61 aa (leaving a 290 aa SepL derivative) failed to export EspD despite membrane localization and SepD binding activity.
Our ongoing work indicates that the failure to completely complement a full sepL deletion is due to changes in the LEE4 transcript (data not shown). To verify the phenotypes of the SepL truncates in a sepL-mutated background, a frameshift mutation was constructed in sepL (ZAP1211) that will have less impact on the structure of the LEE4 transcript. In this background, EspD secretion could be complemented completely by sepL in trans (Fig. 4D). The 11 aa carboxy-terminal deletion in this background was still able to restore some EspD secretion, although at markedly reduced levels compared with the full-length complement (Fig. 4D).
Regulation of effector protein secretion
A sepL mutant is characterized by high levels of Tir secretion (Kresse et al., 2000 and Fig. 4B) and of other effector proteins including NleA (Deng et al., 2004). To investigate the function of the different SepL truncate fusions in this capacity, the levels of Tir secreted were determined by Western blotting in the sepL mutants transformed with the different SepL constructs. Of note was that only the full-length SepL construct was able to lower Tir secretion levels to those demonstrated for the wild-type strain (Fig. 4B). To confirm this result and to rule out any impact of eGFP on the phenotypes, four His-tagged SepL constructs were also tested. These were: (i) full-length SepL, (ii) a protein with the first 340 aa of SepL but deleted for the carboxy-terminal 11 aa, (iii) the first 290 aa of SepL but deleted for the carboxy-terminal 61 aa and (iv) the first 267 aa of SepL but deleted for the carboxy-terminal 84 aa. These variants had exactly the same phenotypes as the respective eGFP fusions in the sepL deletion background. For example, analysis of the general secretion profiles indicated that the 11 aa carboxy-terminal SepL truncate failed to control secretion of Tir, whereas complementation with full-length SepL could (Fig. 4B and C). Moreover, it was apparent that this regulation also applied to other T3S effectors with sizes equivalent to those of NleA and EspZ, as complementation with full-length SepL restored normal secretion levels in a sepL mutant but de-regulated effector protein secretion was still apparent when the 11 aa carboxy-terminal SepL truncates (or any of the other truncates) were transformed into this background (Fig. 4C). This result was also confirmed in the sepL frameshift mutant background (Fig. 4D).
Tir binds to the carboxy terminus of SepL
To determine if any of the effector proteins directly interact with SepL, the secreted supernatant effector proteins from a sepL mutant were separated by SDS-PAGE and a Far-Western was carried out to examine their interaction with 6× His-tagged SepL. Using this approach, SepL only interacted clearly with one protein in the bacterial supernatant and this protein was of a molecular weight equivalent to Tir (Fig. 5A). This interaction was confirmed using a GST–SepL construct that bound to 6× His-tagged Tir (Fig. 5B). As a control, His-tagged Tir bound to GST–CesT (and GST–SepL) but not to GST alone. As an 11 aa deletion at the carboxy terminus of SepL failed to limit Tir secretion, we hypothesized that the SepL interaction with Tir that limits effector protein secretion requires the carboxy terminus of SepL. This was confirmed as 6× His-tagged Tir was shown to bind to full-length SepL–eGFP but not to any of the SepL truncates (Fig. 5C). We next wanted to determine if the carboxy terminus of SepL alone could bind to Tir. The final 48 aa of SepL was fused to GST and immobilized on a glutathione-sepharose 4B column. This region was selected based on a comparative domain analysis with YopN/TyeA (Fig. 8). Tir-His was able to bind to this hybrid protein but not GST alone (Fig. 5D).
The Far-Western analysis indicated that only Tir was interacting with SepL from the different secreted effector proteins. However, this may be a result of conformation or levels of the proteins present in this type of analysis. To determine if SepL could interact with another hyper-secreted effector protein, NleA was expressed with either an amino- or carboxy-terminal 6× His-tag but neither construct demonstrated any interaction with immobilized GST–SepL (Fig. S1). Therefore, despite the fact that SepL regulates the secretion of a combination of effector proteins, the data presented here indicate that the interaction of SepL with effector proteins may be limited to Tir and this interaction is prevented by removal of the final 11 aa of SepL and that the final 48 aa of SepL are sufficient to bind to Tir.
Analysis of Tir domains that interact with SepL and CesT
Tir is stabilized by CesT that is also required for efficient secretion of Tir (Abe et al., 1999; Elliott et al., 1999). CesT is also the main chaperone for the other effector proteins that are hyper-secreted in a sepL mutant (Thomas et al., 2005). Prevention of Tir secretion may simply require SepL binding to Tir but the impact of CesT in this interaction is unknown. To determine where SepL and CesT bind to Tir, different Tir constructs were expressed as His-tagged fusions and interactions with immobilized GST–SepL, GST–CesT and GST alone were analysed (Fig. 6). Previous research has shown that the N-terminal 233 aa of Tir contains a CesT binding domain (Abe et al., 1999; Elliott et al., 1999). Our results confirmed that the first 200 aa of Tir bound to CesT (Fig. 6). However, another CesT binding domain was mapped in Tir deleted for its first 200 aa. Even deletion of the first 382 aa of Tir still produced a polypeptide that could bind to CesT. By contrast, the first 200 aa of Tir did not interact with SepL but the remainder of the protein did bind to SepL as did the 382 amino-terminal truncate. The data indicate that there are at least two regions in Tir that can interact with CesT and that one of these could compete with SepL–Tir binding.
The interaction of Tir with SepL controls the timing of secretion
As an 11 aa deletion of SepL retains the capacity to export translocon proteins but is unable to limit effector protein secretion, it raised the possibility that effector protein export was now occurring at the same time as translocon export as the capacity of SepL to bind Tir potentially sequesters Tir export and somehow limits the secretion of other effector proteins during translocon assembly. To test this, the timing of EspD and Tir secretion was analysed in the wild-type strain and a sepL mutant complemented with either full-length sepL or the 11 aa carboxy-terminal deletion. Bacteria were cultured initially in a medium (Luria–Bertani, LB) that is not permissive for T3S and then transferred to a medium (MEM-HEPES) that induces T3S. Following the transition, samples were taken at regular intervals and the levels of secreted Tir and EspD determined as described in Experimental procedures. For the full-length SepL complement in the sepL deletion, EspD secretion was detectable but not Tir at early time points (Fig. 7A). This pattern was similar in the wild-type strain (Fig. 7A) By contrast, in the sepL mutant complemented with the C-terminal 11 aa deletion of SepL, Tir secretion was detectable along with EspD secretion at early time points (Fig. 7A) following the induction of T3S. While it is appreciated that Tir secretion levels are higher in the truncate-complemented background it is clear from analysis of the EspD/Tir secretion ratios (Fig. 7B) that Tir secretion is no longer delayed in the sepL mutant complemented with the 11 aa truncate by comparison with full-length sepL complementation or the wild type. Consequently, secretion hierarchy is disrupted when the capacity of SepL to bind Tir is removed.
SepL and SepD are critical proteins controlling the switch between translocon and effector protein secretion in A/E E. coli. From the current study, we propose that SepL directly contributes to the T3S hierarchy by binding to Tir and through this sequestration prevents the secretion of Tir and other effector proteins while the translocon components are exported. This activity requires the carboxy terminus of SepL and can be separated from other phenotypes associated with SepL, including its membrane localization, SepD binding and translocon export.
The C-terminus of SepL shares some homology with TyeA and the remainder of SepL some homology with YopN, both from Yersinia spp. (Pallen et al., 2005) (Fig. 8). TyeA controls the export of specific effector proteins so it was logical to investigate C-terminus deletions of SepL to see if these can separate the known activities of SepL and this did prove to be the case. Membrane localization was investigated by both imaging and biochemical analysis of SepL–eGFP constructs. While some cleavage of eGFP was detected from these heterologous proteins it was evident that only the full-length SepL and SepL constructs with deletions of either 11 aa or 61 aa from the C-terminus were able to localize to membrane-containing fractions. The same three SepL constructs were also able to bind to SepD in vitro. In addition, a sepD mutation reduced the levels of SepL–eGFP associated with the membrane-containing fractions and led to its asymmetric distribution in the bacterial cell. SepD is expressed from LEE2 along with T3S basal apparatus proteins and it is likely that the SepL interaction with SepD is responsible for the membrane localization of SepL, possibly to the T3S apparatus, although higher-resolution imaging is required to investigate this. This work confirms previous research that indicated that SepL can be membrane-associated (Kresse et al., 2000; Deng et al., 2005), although one report for EPEC concluded that it was only cytoplasmic (O'Connell et al., 2004). This same study also concluded that effector proteins were not hyper-secreted in a sepL mutant, although this does not agree with the current study and other published research (Kresse et al., 2000; Deng et al., 2004; 2005). In our work the ratios obtained for the SepL–eGFP fusion in the different fractions were variable, potentially indicating only a weak association of SepL with the membrane, potentially via SepD. Each membrane preparation may disassociate this interaction to different levels and account for the variation seen.
SepL-dependent translocator secretion was abolished with a C-terminus deletion of 61 aa, even though this construct could still bind SepD and localize to the membrane. By contrast, only the full-length SepL protein was able to restore normal levels of Tir and other effector protein secretion in a sepL mutant. Therefore, the 11 aa deletion still functioned to some extent to export translocon proteins but had lost effector protein secretion control. As the final 11 aa of SepL were essential for controlling effector protein secretion, we examined whether any of these hyper-secreted proteins could actually bind to SepL. Tir was shown to interact with SepL and this interaction required the carboxy-terminal 11 aa of SepL. In addition, the C-terminal 48 aa of SepL, when combined with GST, was capable of binding to Tir (Fig. 5D). When the structure of the YopN76-293–TyeA complex (Schubot et al., 2005) is overlaid with SepL, it is evident that the final 48 aa of SepL map to the final two alpha helices of TyeA and this domain is sufficient to interact with Tir. The sequence divergence at the C-terminal 12 aa between TyeA and SepL may relate to the recognition of different proteins in the two organisms. Our work also demonstrates that the region of SepL required for SepD binding must lie within the YopN homologous region (Fig. 8). There was no evidence from our work that any of the other secreted effector proteins could bind directly to SepL, although this cannot be discounted as they have not all been tested individually. The Far-Western analysis only identified one clear binding partner, Tir. While NleA is known to be secreted at higher levels in a sepL mutant, it did not bind to SepL in vitro using similar approaches that were successful with Tir. Therefore, it appears likely that the Tir–SepL interaction is critical in limiting the secretion of effector proteins in general. This finding fits well with recently published research (Thomas et al., 2007) that demonstrated that Tir is required for hyper-secretion of other effector proteins in a sepD mutant background.
Our previous work has demonstrated that LEE4 and LEE5 are co-ordinately expressed, indicating that Tir will be produced in individual bacteria while the translocon is being assembled (Roe et al., 2004). Therefore, we investigated the hypothesis that the binding of Tir by SepL actually sequesters it and prevents its early release while the translocon is being assembled. The timing of release of Tir and the translocon protein EspD were investigated using a shift in culture conditions from a non-permissive to a permissive medium for T3S. Under these conditions, Tir secretion was demonstrated to be delayed in the wild type and a sepL deletion strain complemented by full-length SepL. However, Tir secretion occurred at the same time as EspD secretion when the strain was complemented with SepL deleted for the C-terminal 11 aa. The data support the proposition that the timing of Tir and effector protein secretion is directly controlled by SepL binding to Tir. The altered timing may also account for the higher levels of secreted effector proteins found in bacterial supernatants of sepL mutants as these can be exported over a longer period by each cell. However, as the deletion of the C-terminus of SepL will have other effects on SepL function, we cannot rule out that another mechanism may be responsible for limiting effector protein export.
How Tir binding to SepL could prevent secretion of other effector proteins is not understood but it must presumably stall a series of T3S apparatus interactions with effector proteins prior to EscN/ATPase-driven export. Another key question is how such a mechanism is then switched once translocon export has finished, allowing effector protein export. It has been suggested that opening a conduit to the host cell via the translocon could induce a change in local ion concentrations in particular calcium, which may disassociate or alter the SepL/SepD complex (Deng et al., 2005). Our current work has indicated that the SepL–Tir interaction could also be a target for such a trigger. Alternatively, SepL and/or SepD may have limited stability so their activity is only for a defined period. Another possibility is that SepL (like YopN) may be secreted to initiate effector secretion (Pallen et al., 2005). We have tested whether His-tagged fusions to SepL or the SepL region homologous to YopN (the first 267 aa of SepL) are exported into the bacterial supernatant. There was no evidence for this even though both were expressed inside the bacterial cell (data not shown). An indication of a potential mechanism to release Tir from SepL comes from previous research that has shown a direct interaction between Tir and EscD (Pas) (Kresse et al., 1998). EscD is considered to be a protein in the inner membrane complex of the T3S (Ogino et al., 2006) and when deleted prevents both translocon and effector protein secretion. However, plasmid complementation of escD led to high levels of Tir secretion (Ogino et al., 2006), indicating that overexpression of EscD also leads to loss of Tir secretion control. We have shown that SepL binds directly to EscD (D. Wang and D. Gally, unpubl. data) and this binding requires the same final 11 aa of SepL that are required for SepL binding to Tir. It may be that two different organizations of the SepL–SepD complex are required to permit translocator and effector protein export with different binding partners at the C-terminus of SepL.
Normal levels of Tir secretion also require presentation by its chaperone, CesT, and will require the interaction of complexed Tir with the ATPase, EscN, which then energizes the export of Tir (Gauthier and Finlay, 2003). Many other effector proteins utilize CesT, especially those known to be hypersecreted in a sepL mutant (Thomas et al., 2005). Our in vitro data indicate that the CesT binding domain shown in the amino-terminal third of Tir is not the only region of Tir that interacts with CesT. It remains possible that while the amino-terminal domain is important for stability a further domain is necessary for its export. Multiple chaperone binding sites in Tir have been suggested by others (Elliott et al., 1999). In the context of the current work, SepL and CesT were shown to bind in the C-terminal half of Tir and preliminary data support competitive interactions between SepL and CesT for Tir binding. CesT and Tir are both expressed from the LEE5 transcript (Elliot et al., 1999) and LEE4 and LEE5 are co-ordinately expressed (Roe et al., 2004). As a consequence, Tir stabilized by CesT will be present in the cell while EspA filaments are assembled. CesT–Tir must interact with the T3S system to allow Tir release from CesT and its subsequent export. Therefore, we propose that the CesT–Tir complex interacts with a SepD–SepL ‘filter’ but the interaction of Tir with SepL becomes the stalling point for effector protein secretion. Once Tir export is triggered by disassociation of this interaction, other effectors can then be exported but these also have to pass through the SepD–SepL 'filter', perhaps requiring an interaction of CesT and/or effectors with SepD. This combination of possible interactions is the focus of ongoing research.
Logically, secretion of Tir and effector proteins should be restricted until the translocon is assembled so they can be secreted directly into the host cell. If Tir is secreted prior to this it may interfere with translocon assembly and perhaps, more importantly, it may bind to surface-expressed intimin and therefore block the subsequent interaction of intimin with host membrane-inserted Tir. A/E E. coli along with Citrobacter rodentium are unique in having SepL and SepD proteins and are the only bacterial pathogens to date known to inject their own receptor. This may not be a coincidence as tight control over the release of this receptor is required.
Bacterial strains, plasmids, oligonucleotides, media and antibodies
The bacterial strains, media, antibodies and plasmids used in the study are described in Tables 1 and 2. Table 3 lists the oligonucleotide primers used. MEM-HEPES is minimal essential medium with HEPES buffer (Sigma), containing additional glucose to a final concentration of 0.2%. LB broth was also used (Oxoid). Antibiotics were included when required at the following concentrations: chloramphenicol 12.5 μg ml−1, kanamycin 25 μg ml−1 and ampicillin 50 μg ml−1.
Table 1. Plasmids used in the study.
Low-copy-number cloning vector
Plasmid contained GST gene fusion system from Amersham Biosciences
Plasmid contained 6× His gene fusion system from Novagen
pET21d digested with XbaI/XhoI; fragment contained full-length tir gene amplified from ZAP193 and inserted
pET21d digested with XbaI/XhoI; fragment contained the first 100 aa tir gene amplified from ZAP193 and inserted
pET21d digested with XbaI/XhoI; fragment contained the first 200 aa tir gene amplified from ZAP193 and inserted
pET28a digested with NdeI/XhoI; fragment contained the 201–558 aa tir gene amplified from ZAP193 and inserted
pET28a digested with NdeI/XhoI; fragment contained the 382–558 aa tir gene amplified from ZAP193 and inserted
Preparation of secreted proteins and bacterial fractions for protein analyses
Bacteria were cultured in 50 ml of MEM-HEPES at 37°C (200 r.p.m.) to an OD600 of 0.8 unless specifically stated. The bacterial cells were pelleted by centrifugation at 4000 g for 20 min, and supernatants were passed through filters (0.45 μm). The proteins were precipitated overnight with 10% TCA, and separated by centrifugation at 4000 g for 30 min (4°C); the proteins were suspended in 150 μl of 1.5 M Tris (pH 8.8). The bacterial pellet was initially suspended in 150 μl of sonication buffer [10 mM Tris-HCl (pH 7.5), 0.5 mM PMSF, aprotinin (0.5 μg ml−1)] and sonicated on ice. Cell envelopes and unbroken bacteria were removed by two rounds of centrifugation (5000 g for 10 min at 4°C). The supernatant (whole-cell fraction) was removed and the membranes pelleted by ultra-centrifugation of the samples for 1 h at 500 000 g at 4°C. The supernatant containing cytoplasmic proteins was collected. The membrane preparation was washed twice with sonication buffer and re-suspended in 150 μl of SDS sample buffer. Proteins were separated by SDS-PAGE using standard methods and Western blotting performed as described previously (Roe et al., 2003; Naylor et al., 2005) using the relevant antibodies listed in Table 2. Tir and EspD secretion levels were measured following enhanced chemi-luminescence detection from Western blots using Multi-analyst (Bio-Rad) software.
Full-length SepL and different carboxy-terminal truncates of SepL (Fig. 1) were fused to eGFP in pAJR70 (Roe et al., 2003) using the primers described in Table 3. A 6× His-tag was introduced at the carboxy end of SepL and three of the truncated SepL proteins (267, 290 and 340 aa) by PCR before cloning into pACYC184 (Table 1). All constructs were confirmed by sequencing.
Construction of GST and 6× His-tagged proteins and binding assays
For the GST–SepL construct, sepL was amplified from EHEC O157 ZAP193 by PCR using the primers sepL 5′g and sepL 3′g. The resulting PCR product was digested with BamHI and SmaI, and cloned into the BamHI and SmaI sites of pGEX-4T-2. This creates a GST–SepL hybrid protein fusion (in pDW9) used in GST pull-down assays. A similar stategy was used to clone the 48 aa carboxy terminus of SepL, SepD and CesT using the primers described in Table 3. For His-tagged proteins, tir and sepD open reading frames were amplified by the primers listed in Table 3, digested with XbaI and XhoI and cloned into the XbaI and XhoI sites of pET21d. For sepL and tir domain analyses, the amplified fragments were cloned into pET28a via NdeI and XhoI to create N-terminal 6× His-tags. All constructs were expressed in E. coli BL21 (Table 2) following IPTG induction (0.1 mM). The GST fusions were expressed in AAEC 185 (an E. coli K-12 derivative, Table 2) and the His-tag fusions in E. coli BL21, both following IPTG induction in LB (0.1 mM) at OD600 = 0.5. For protein preparations, the bacteria were harvested at 4000 r.p.m. (4°C) for 30 min. 2 h post IPTG inoculation. The bacterial pellet was suspended in PBS and sonicated. The supernatant was collected by centrifugation at 12 000 g for 10 min at 4°C. For the GST fusions, the supernatant was mixed with PBS-balanced glutathione-sepharose 4B beads for 30 min at room temperature. The beads were separated by centrifugation at 500 g for 5 min. An aliquot of the supernatant was saved for analysis and the rest of the supernatant discarded. The beads were washed three times using 10 vols of PBS and separation by centrifugation at 500 g for 5 min. The beads were mixed gently in the same volume of Glutathione Elution Buffer (0.154 g of reduced glutathione dissolved in 50 ml of 50 mM Tris-HCl, pH 8.0) and incubated at room temperature for 10 min. Supernatant was collected by centrifugation at 500 g for 5 min. The elution and centrifugation step was repeated and the two eluates pooled. Equal volumes of washes and eluates were loaded onto the protein gels. To check initial loading of columns, some eluates were stained with colloidal blue and/or Western blotted to confirm the presence of the expected GST fusion protein and His- or GFP-tagged binding partners.
For His-tagged proteins, these were expressed and purified as above except a Ni-NTA column was used and elution was with the supplied Qiagen buffer. For the in vitro binding assays, these were carried out on either glutathione or Ni-NTA columns on which the bait protein was first retained and then the protein being investigated was run through the column in a lysate prepared as above. Following washes, elution was carried out as described above. Eluted samples were analysed by SDS-PAGE followed by colloidal blue staining and Western Blotting.
Construction of sepL and sepD mutants
The experiments were carried out essentially as described previously (Roe et al., 2003; Emmerson et al., 2006) using allelic exchange methodology. The respective primer sets used to amplify the sepL, sepD are described in Table 3, the ler deletion was published previously (Low et al., 2006). To generate plasmids for the sepL frameshift mutation, a fragment containing sepL gene and flanking regions (upstream 1 kb + downstream 1 kb) was amplified from ZAP193 by PCR using sepL 5′allf and sepL 3′allf. It was digested with by KpnI and XbaI, and cloned into the KpnI and XbaI sites of pIB307. Following the methodology in the Stratagene ‘Site-Directed Mutagensis’ kit, a single base pair was inserted at 345 bp of sepL to generate a plasmid (pDW52) for allelic exchange. Final plasmid constructs (Table 1) were sequenced prior to the deletion exchange and each deletion confirmed by PCR analysis. The sepL and sepD mutants could be functionally complemented by pDW24 (sepL), pDW48 (sepL::6xhis), pDW6 (sepL::egfp) and pDW20 (sepD::egfp) (Table 1) to restore translocon (EspD) secretion as determined by Western blotting (Fig. 4A, C and D and data not shown).
Fluorescence imaging was carried out using a Leica DM LB2 microscope and a 100× objective lens. Narrow-bandwidth filters to excite and detect eGFP/FITC were used (41017 Endow GFP, CHROMA). Images were captured using a Hamamatsu ORCA-ER black and white CCD digital camera. Images were analysed using OpenLab software (Improvision). To measure levels of fluorescence in individual cells, transects were marked on bacteria and the fluorescence levels determined using QFluor software (Leica).
Supernatant proteins were prepared from ZAP1143 (ΔsepL) as described in the relevant section above. The secreted proteins were separated by SDS-PAGE using standard methods and then transfered to an enhanced chemi-luminescence Nitrocellulose membrane. The membrane was first blocked with 8% milk PBS overnight at 4°C and washed three times with PBS-Tween (0.5% v/v) before being incubated overnight with SepL-His in an E. coli BL21 lysate at 4°C. After the incubation, the nitrocellulose membrane was washed three times with PBS-Tween and the standard Western procedure for detecting the 6× His-tags was carried out.
D.W. is supported by a studentship from the College of Medicine and Veterinary Medicine at the University of Edinburgh. This research was also supported by funding from DEFRA with a research Fellowship (VF0304) to D.L.G./A.J.R. and then with funding from DEFRA under the Veterinary Training and Research Initiative (VT0102). We are indebted to Prof. Trinad Chakraborty at the University of Giessen, Germany and Prof. John Leong at the University of Massachusetts Medical School, USA for supplying antibodies used in the study. We would like to thank Dr. Mads Gabrielsen for help with the comparative analysis of SepL and YopN/TyeA structures. We also thank Andreas Kresse for initial helpful discussions and reagents.