Biochemical Journal

Research article

EscI: a crucial component of the type III secretion system forms the inner rod structure in enteropathogenic Escherichia coli

Neta Sal-Man, Wanyin Deng, B. Brett Finlay


The T3SS (type III secretion system) is a multi-protein complex that plays a central role in the virulence of many Gram-negative bacterial pathogens. This apparatus spans both bacterial membranes and transports virulence factors from the bacterial cytoplasm into eukaryotic host cells. The T3SS exports substrates in a hierarchical and temporal manner. The first secreted substrates are the rod/needle proteins which are incorporated into the T3SS apparatus and are required for the secretion of later substrates, the translocators and effectors. In the present study, we provide evidence that rOrf8/EscI, a poorly characterized locus of enterocyte effacement-encoded protein, functions as the inner rod protein of the T3SS of EPEC (enteropathogenic Escherichia coli). We demonstrate that EscI is essential for type III secretion and is also secreted as an early substrate of the T3SS. We found that EscI interacts with EscU, the integral membrane protein that is linked to substrate specificity switching, implicating EscI in the substrate-switching event. Furthermore, we showed that EscI self-associates and interacts with the outer membrane secretin EscC, further supporting its function as an inner rod protein. Overall, the results of the present study suggest that EscI is the YscI/PrgJ/MxiI homologue in the T3SS of attaching and effacing pathogens.

  • enteropathogenic Escherichia coli (EPEC)
  • inner rod
  • pathogenesis
  • secretin
  • substrate specificity switch
  • type III secretion system (T3SS)


EPEC (enteropathogenic Escherichia coli) is a Gram-negative bacterial pathogen, and is the major cause of infantile diarrhoea in developing countries, with outbreaks reporting a mortality rate of up to 30% [1,2]. This bacterium belongs to a family of pathogens called A/E (attaching and effacing) pathogens, which colonize the gut epithelium and induce a distinct histopathology characterized by intimate adherence to the epithelial surface and effacement of host enterocyte brush border microvilli [3].

Many important EPEC virulence factors are encoded within a 35-kb chromosomal pathogenicity island termed the LEE (locus of enterocyte effacement) [4]. The LEE contains 41 genes, most of which are organized in five operons that encode components of (i) the T3S (type III secretion) apparatus, (ii) E. coli secreted proteins and translocators, (iii) type-III specific chaperones and (iv) transcriptional regulators.

The T3SS (T3S system) is an export complex that allows injection of virulence factors (termed effectors) from the bacterial cytoplasm of many Gram-negative pathogens into host cells [57]. The main proteins that form the T3S export apparatus (also called nano-syringe or injectisome) are well conserved among the T3SS of other pathogens and share significant similarity with components of the flagellar system [5,810]. The T3S apparatus in EPEC is comprised of approximately 20 proteins that assemble into a multi-ring base structure, which transverses both inner and outer membranes as well as the periplasmic space [11,12], and a protruding needle. The needle, which is composed of a single polymerizing protein, EscF [13], is linked to a substructure termed the inner rod. This substructure forms the periplasmic portion of the channel that lies within the injectisome [14,15]. Besides the structural role of the inner rod proteins within the T3S apparatus and the flagellar system, it was previously suggested that the formation of the inner rod acts as the trigger for substrate specificity switching [14,16]. The switching process enables hierarchy in substrate export, with the first substrates to be secreted being proteins involved in the assembly of the inner rod and the extracellular needle (early substrates). On completion of the needle structure to its full length, components of the translocon are secreted (middle substrates), and finally effector proteins (late substrates) are secreted. Although the exact details of this switch are unknown, it was proposed to involve the cleavage and presumably a conformational change in the C-terminal domain of an integral membrane protein from the EscU/FlhB/SpaS/Spa40/YscU family [1723].

Although the inner rod proteins of Salmonella (PrgJ [14,24]), Shigella (MxiI [25,26]) and Yersinia (YscI [16]) are known, the EPEC homologue of these proteins has not yet been characterized. In a previous bioinformatics study by Pallen et al. [27], rOrf8, a small uncharacterized protein encoded by the LEE, was found to have low levels of homology with other inner rod proteins. Apart from this prediction, rOrf8's potential relationship with other T3SS components and its putative role in the assembly of the channel has never been thoroughly examined.

In the present study, we demonstrate that rOrf8 is essential for T3S and is secreted by the T3SS, as previously reported for inner rod proteins of other T3SSs [16,24,33,43]. Moreover, we found that rOrf8 secretion occurs regardless of EscU auto-cleavage, suggesting that it is an early substrate of the T3SS. In addition, we show that rOrf8 self-associates and interacts with EscC, the EPEC secretin, and with EscU, the integral membrane protein linked to the substrate specificity switch. Based on our experimental results, we conclude that rOrf8 forms the inner rod structure of the T3SS and therefore support the suggestion of renaming it as EscI [27], according to the standard Yersinia enterocolitica type III nomenclature. We use this terminology throughout the present paper.


Bacterial strains

WT (wild-type) EPEC O127:H6 strain E2348/69 [Smr (streptomycin-resistant)] was used in the present study. Strains were grown in LB (Luria–Bertani) broth supplemented with the appropriate antibiotics at 37°C. Antibiotics were used at the following concentrations: streptomycin (50 μg/ml), ampicillin (100 μg/ml), kanamycin (50 μg/ml) and chloramphenicol (30 μg/ml).

Construction of an escI non-polar mutant

A non-polar deletion mutant of the escI gene in the Smr (streptomycin-resistant) EPEC E2348/69 was generated using the sacB-based allelic exchange method [28]. Briefly, two PCR fragments were generated using primer pairs EscI-01F/EscI-01R and EscI-02F/EscI-02R (Supplementary Table S1 at, cloned into pCR2.1-TOPO (Invitrogen), verified by DNA sequencing, and then subcloned as a KpnI/NheI fragment and an NheI/SacI fragment into the suicide vector pRE112 [29] digested by KpnI/SacI. The resulting plasmid, containing the flanking regions of escI with 87% of escI deleted, was transformed into E. coli SM10λpir by electroporation, and introduced into EPEC by conjugation. After sucrose selection, EPEC colonies that were resistant to sucrose and susceptible to chloramphenicol were screened for the deletion of escI by PCR. Other deletion mutants that were used in the present study have been described elsewhere and are listed in Supplementary Table S2 (

Construction of plasmids expressing EscI and EscC

The escI gene was amplified using the appropriate primer pair EscI-F/EscI-2HA-R (Supplementary Table S1), cloned into pCR2.1-TOPO (Invitrogen), and then subcloned as a SacI/XhoI fragment into SacI/XhoI-digested pTOPO-2HA [30]. The resulting construct, pEscI-2HA, expressed a fusion protein of EscI with a double HA (haemagglutinin) tag at its C-terminus. Similarly, escI was amplified from EPEC by PCR using the primer pair EscI-F2/EscI-stop-R (Supplementary Table S1), which includes a stop codon at the end of the coding region. The PCR product was cloned into pCR2.1-TOPO (Invitrogen), and then subcloned as a BamHI/SalI fragment into BamHI/SalI-digested pACYC184 [31]. The resulting construct, pEscI, expressed an untagged version of EscI. Tagged versions of escI were cloned similarly into pACYC184 plasmid [31] with either HSV (herpes simplex virus) or V5 tags, using the primer pairs EscI-F2/EscI-HSV-R or EscI-F2/EscI-V5-R (Supplementary Table S1).

To generate an expression vector of EscC fused to both FLAG and double-HA tags, the entire coding region of escC from EPEC E2348/69 was amplified using the primer pair EscC-F/EscC-R (Supplementary Table S1). The PCR product was cloned as a HindIII/XhoI fragment into HindIII/XhoI-digested pTOPO-2HA. A FLAG tag was then added between the escC coding region and the double-HA tag using two complementary oligonucleotides FLAG-oligoF/FLAG-oligoR (Supplementary Table S1). All constructs were verified by DNA sequencing.

Secretion assay

EPEC strains were grown overnight in LB broth in a shaker at 37°C. The cultures were diluted 1:50 into DMEM (Dulbecco's modified Eagle's medium) that was pre-warmed in a CO2 tissue culture incubator overnight and grown 6 h to a D600 of 0.7 in a tissue culture incubator (with 5% CO2) with the tubes standing. The cultures were centrifuged at 16000 g for 5 min at 23°C to remove the bacteria; the supernatant was collected, and then filtered through a 0.22 μm filter (Millipore). The supernatant was then precipitated with 10% (v/v) trichloroacetic acid to concentrate proteins secreted into the culture medium. The secreted proteins precipitated were dissolved in SDS/PAGE sample buffer, and trichloroacetic acid was neutralized with saturated Tris. The volumes of buffer used to resuspend the secreted proteins were normalized relative to the D600 of the cultures to ensure equal loading of the samples. The proteins were analysed on SDS/12% PAGE gels and stained with Coomassie Blue.

Co-immunoprecipitation of protein complexes from EPEC lysates

EPEC strains were subcultured 1:50 into 50 ml of pre-warmed DMEM and grown for 6 h in a tissue culture incubator (with 5% CO2) with the flasks standing. The bacterial cultures were harvested by centrifugation at 5000 g for 30 min at 4°C. The pellet was washed with one culture volume of ice-cold PBS and pelleted by centrifugation at 16000 g for 5 min at 23°C. The washed pellet was resuspended in 1 ml of lysis buffer (20 mM Tris/HCl, pH 7.5, 150 mM NaCl, 3 mM MgCl2, 1 mM CaCl2 and 2 mM 2-mercaptoethanol with protease inhibitor cocktail) containing 2.5% glycerol. The cell suspension was subjected to sonication using a probe sonicator (3×15 s; Fisher Scientific). After sonication NP-40 (Nonidet P40) detergent was added to 0.1% and samples were incubated for 15 min on ice. Intact cells were removed by centrifugation at 5000 g for 15 min at 4°C. During cell lysate preparation, 100 μl of Protein G beads (GE Healthcare) per sample was washed three times with lysis buffer with 2.5% glycerol (centrifuged at 1000 g for 4 min at 4°C). To reduce non-specific binding, the lysates were pre-incubated with 60 μl of washed beads for 30 min at 4°C on a rotary wheel, then centrifuged (1000 g for 4 min at 4°C) and the pre-cleared lysates were collected. The pre-cleared lysates were then incubated with 2.5 μg of antibody for 30 min at 4°C on a rotary wheel. Finally, 40 μl of washed beads were added to each sample and incubated on a rotary wheel overnight at 4°C. The samples were then centrifuged at 1000 g for 4 min at 4°C to pellet the beads as described above and the supernatant removed. The beads were then washed three times with 1 ml of lysis buffer containing 2.5% glycerol. Interacting proteins were eluted by adding 50 μl of sample buffer and boiling the beads for 10 min. Equal amounts of the total lysate and eluted fractions were loaded on to an SDS/PAGE gel and immunoblotted (as described below). EPEC strains containing empty vectors were processed similar to the EPEC strains expressing both proteins, and were used as controls for non-specific binding.


Samples were subjected to SDS/PAGE and transferred on to nitrocellulose membranes (pure nitrocellulose; pore size, 0.45 μm; Bio-Rad). Blots were blocked for 1 h in 5% (w/v) non-fat dried skimmed milk powder/PBST (0.1% Tween in PBS), incubated first with the primary antibody diluted in 5% non-fat dried skimmed milk powder/PBST for 1 h at room temperature (23°C) and then with the secondary antibody diluted in 5% non-fat dried skimmed milk powder/PBST for 30 min at room temperature, and detected with the ECL® (enhanced chemiluminescence) reagents (GE Healthcare). Primary antibodies were mouse anti-FLAG-M2 (Sigma), diluted 1:1000; mouse anti-HA.11 (Covance), diluted 1:1000; rat anti-HA (Roche), diluted 1:1000; mouse anti-V5 (Invitrogen), diluted 1:5000; goat anti-HSV (Abcam), diluted 1:400; rabbit anti-EscI (raised against a peptide of EPEC EscI with the sequence SPEQVLIEEIKKRHLA; Pacific Immunology), diluted 1:500; mouse anti-EspB (laboratory stock), diluted 1:1000; mouse anti-DnaK (Stressgen), diluted 1:1000; and rabbit anti-β subunit of E. coli ATP synthase, diluted 1:15 000 (a gift from Dr Gabriele Deckers-Hebestreit, Arbeitsgruppe Mikrobiologie, Fachbereich Biologie/Chemie, Universität Osnabrück, Osnabrück, Germany). Secondary antibodies were HRP (horseradish peroxidase)-conjugated goat anti-mouse antibody (Jackson ImmunoResearch), diluted 1:5000; HRP-conjugated goat anti-rabbit antibody (Sigma), diluted 1:5000; HRP-conjugated goat anti-rat antibody (Jackson ImmunoResearch), diluted 1:5000; and HRP-conjugated donkey anti-goat antibody (Sigma), diluted 1:5000.

Analysis of T3S by epitope tagging

The EscI-2HA construct was introduced into EPEC WT, ΔescU, ΔescJ, ΔescC, ΔescN and ΔescF strains. Secreted proteins were processed as indicated above, separated by SDS/16% PAGE and analysed by Western blotting using a mouse monoclonal antibody against HA (Covance). ImageJ software was used to quantify band intensity.


EscI is essential for T3S

PrgJ constitutes the inner rod substructure of the Salmonella T3SS, which is positioned within the internal chamber of the T3S base extending from the socket-like structure to the needle [14,15]. A recent bioinformatics prediction proposed that the EPEC homologue of PrgJ is a small uncharacterized LEE-encoded protein, designated rOrf8, and suggested renaming it EscI [27]. To examine whether EscI is essential for T3S, as expected for structural proteins of the T3SS, we analysed the secretion of the T3SS translocators EspA, EspB and EspD in an EPEC escI-null strain (ΔescI) by SDS/PAGE and Coomassie Blue staining. The null strain did not secrete type III translocators, indicating that EscI is essential for T3S (Figure 1). A similar result was found in the related murine A/E pathogen, Citrobacter rodentium [30]. Complementation of the mutant strain with full-length escI, in trans, restored secretion of the translocators, thus confirming that the deletion of escI is non-polar. Overall, this result demonstrates that EscI is an essential component for T3S in EPEC, similar to what has been reported for the inner rod proteins of Yersinia and Shigella, YscI and MxiI respectively [16,25].

Figure 1 EscI is essential for T3S

Protein secretion profiles of EPEC WT, ΔescI and ΔescI complemented with escI in trans. Secreted proteins were concentrated from supernatants of bacterial cultures grown in DMEM and analysed by Coomassie Blue staining of an SDS/12% PAGE. The locations of the translocators EspA, EspB and EspD are indicated on the left-hand side of the gel. Also indicated is the location of EPEC EspC, which is not secreted via the LEE-encoded T3SS.

EscI self-associates

It has been suggested that inner rod proteins polymerize in order to form a hollow channel within the T3S apparatus [14]. To determine whether EscI shares this potential characteristic of inner rod proteins, we tagged it with two different tags (double-HA and HSV) and expressed both versions in EPEC WT and ΔescI strains, under T3S-inducing conditions. Analysis of whole-cell extracts confirmed the expression of both tagged proteins (Figure 2A). Immunoprecipitation was performed using an anti-HA antibody, while the eluted fractions were immunoblotted against the HSV epitope. EscI tagged with the HSV epitope co-eluted with EscI-2HA (Figure 2B) in both EPEC strains, indicating that EscI self-associates. This result is in agreement with current models that suggest that the inner rod protein polymerizes to form its substructure within the channel [5,14].

Figure 2 EscI can homo-dimerize

EscI-2HA is co-immunoprecipitated with EscI-HSV in WT and ΔescI strains. (A) Whole-cell extracts were immunoblotted for EscI using anti-HA or anti-HSV antibodies prior to immunoprecipitation. (B) Eluted samples of whole-cell extracts applied to Protein G beads linked to the anti-HA antibody were examined by immunoblotting using anti-HA and anti-HSV antibodies. pACYC is the vector control pACYC184. The molecular mass in kDa is indicated on the left-hand side.

EscI is secreted by the T3SS

EscI was previously identified as a secreted protein during an analysis of the global profile of secreted proteins of C. rodentium using quantitative proteomics [32]. To determine whether EscI is also secreted by EPEC, we assessed the ability of EPEC WT, ΔescI and ΔescI complemented with EscI to secrete the EscI protein by analysing their supernatants using immunoblotting with a rabbit anti-EscI antibody. EscI was detected in the secretion sample of ΔescI complemented with EscI, but not in the WT or ΔescI strains (Figure 3A). Mass spectrum analysis confirmed the presence of EscI in the WT secretion fraction (results not shown), suggesting that the detection ability of the anti-EscI antibody was poor and the anti-EscI antibody could only detect EscI when it was overexpressed. Using EPEC WT with a vector expressing EscI fused to a double-HA tag we confirmed that EscI was secreted under T3S-inducing conditions in WT strain (Figure 3B). To determine whether the secretion of EscI is T3SS-dependent, we transformed the vector pEscI-2HA into different T3S apparatus mutants and checked for EscI secretion. Mutations in any of the integral components of the T3S apparatus (EscU, EscN, EscJ, EscC and EscF) resulted in a complete loss of EscI secretion (Figure 3B). Based on these results, we concluded that EscI is a secreted substrate of the EPEC T3SS. Similar T3S was previously reported for other characterized inner rod proteins [16,24].

Figure 3 EscI is secreted via the T3SS

(A) EscI secretion was examined in WT, ΔescI and ΔescI complemented with pEscI (pTOPO). Secreted proteins were concentrated from supernatants of bacterial cultures grown under T3S- inducing conditions and analysed by SDS/16% PAGE and immunoblotting using an anti-EscI antibody. The molecular mass marker (kDa) is indicated on the left-hand side of the gel. (B) EscI-2HA secretion was examined in WT, ΔescU, ΔescJ, ΔescC, ΔescN and ΔescF strains. Secreted proteins concentrated from supernatants of bacterial cultures grown in DMEM (Sup) and whole-cell extracts (WCE) were analysed by SDS/16% PAGE and immunoblotting for the presence of EscI using an antibody against HA. (C) Quantitative analysis of the secreted EscI in WT EPEC. Samples containing the secreted fraction (Sup), the cell pellet fraction (Pellet) or both (Total) were collected and equal percentages of all samples were subjected to SDS/16% PAGE. Proteins from the total and secreted fraction were trichloroacetic acid precipitated. EscI was detected by Western blotting using an anti-HA antibody.

To quantitatively determine how much EscI is secreted against how much remains in the bacteria, WT EPEC harbouring EscI-2HA was grown under T3S-inducing conditions, and secreted supernatant and cell pellet fractions were collected, ran on an SDS/PAGE gel, analysed by Western blotting and the band intensity was assessed using ImageJ software. The results showed that EscI is more abundant in the pellet sample (~70%) than the supernatant (~30%) (Figure 3C), suggesting that, although partly secreted, much of the EscI protein remains associated with the bacterial membrane, as expected for an inner rod protein.

EscI is an early T3S substrate

Assembly of the distal portion of the T3S apparatus requires secretion of early T3SS substrates. This set of substrates includes the needle and the inner rod proteins [7,14,25,33,34]. According to the current dogma, regulated secretion is controlled by an inner membrane protein that belongs to the FlhB/YscU family [18,21,35,36]. EscU is the EPEC homologue of FlhB/YscU. Similar to other proteins in this family, it contains a conserved auto-catalytic cleavage site in its C-terminal domain. It was proposed that the cleavage of FlhB/YscU/EscU induces a conformational change of the protein and prompts a substrate-specificity switch [19,20,23,3742]. Deletion of the entire cleavage motif (NPTH amino acid sequence) of the Shigella homologue of EscU, Spa40, resulted in a non-functional T3SS demonstrating the critical role of this motif [43]. Point mutations in the cleavage motif were found to result in full arrest of the secretion of translocon and filament components (middle substrates), whereas the export of rod/needle/hook-type substrates (early substrates) remained intact [2123,3335].

In the present study, we examined the secretion of EscI–V5 in the ΔescU strain complemented with WT EscU (EscUWT) or with a non-cleavable EscU mutant (EscUN262A). EscI was detected (using an anti-V5 antibody) in the supernatants of ΔescU strains complemented with either EscUWT or the non-cleavable EscU mutant form (Figure 4A), indicating that EscI is an early substrate, and its secretion occurs independently of EscU cleavage. As a control, we examined the secretion of EspB, a known T3S translocator (middle substrate) and found that EspB was secreted by the ΔescU strain complemented with EscUWT, but not in the strain complemented with the non-cleavable form (Figure 4B). Expression of both EscI and EspB was observed in the bacterial lysates, indicating expression of both proteins (Figures 4A and 4B).

Figure 4 EscI is an early T3S substrate

EscI–V5 secretion was examined in the ΔescU strain complemented with EscUWT or the non-cleavable mutant EscUN262A grown under T3S-inducing conditions. Whole-cell lysates (pellet) and supernatant fractions (sup) were analysed by immunoblotting using anti-V5 (A) or anti-EspB (B) antibodies. The molecular mass in kDa is indicated on the left-hand side.

EscI interacts with EscU

As EscI is a T3SS substrate, it could be targeted to the secretion apparatus by interacting with integral components of the T3S apparatus. Since the FlhB/YscU family of proteins was previously suggested to serve as the gatekeepers of the T3SS [19,21,35], we examined whether EscI interacts with EscU. For this purpose, we expressed HA-tagged EscU together with V5-tagged EscI or in the presence of vector controls. The strains were induced for T3S and whole-cell lysates were used for co-immunoprecipitation experiments. Analysis of whole-cell extracts confirmed the expression of the tagged proteins (Figure 5). Immunoprecipitation was performed using an anti-V5 antibody, whereas the eluted fractions were immunoblotted against the HA epitope. EscUWT, tagged with a double-HA tag, co-eluted with EscI–V5 (Figure 5A). This finding is in agreement with a previous study that reported a weak interaction between EscU and EscI, using a yeast two-hybrid system [44]. Moreover, we observed that EscI can pull-down both the full-length EscU protein and the cleaved product (~12 kDa; Figure 5A). This result suggests that EscI interacts with the C-terminus of EscU, which is epitope-tagged. However, it was previously shown that the resulting subdomains produced by cleavage within the NPTH sequence of T3SS and flagellar EscU/FlhB proteins remain tightly associated with each other so that they are co-purified [21]. Thus we cannot conclude that EscI and EscU interact through the C-terminal part of EscU. The non-cleavable EscU mutant, EscUN262A, showed similar interactions with EscI (Figure 5B). Moreover, we detected interaction between the two proteins using an EPEC strain that lacks the entire LEE region (results not shown), suggesting that the interaction between EscI and EscU is not mediated through an additional T3SS protein.

Figure 5 EscI interacts with WT and a non-cleavable mutant form of EscU

(A) ΔescI EPEC expressing EscUWT–2HA was transformed with either EscI–V5 or a vector control (pACYC184) and grown under T3S-inducing conditions. The ΔescI strain expressing EscI–V5 together with pTOPO (pCR2.1-TOPO) served as a control. Whole-cell extracts were subjected to immunoprecipitation using an antibody directed against V5. Samples were analysed by Western blot probing with a rat anti-HA antibody. The heavy chain of the anti-V5 antibody is marked with an asterisk. (B) Similar experimental set-up as outlined in (A), but expressing the EscU non-cleavable mutant (EscUN262A–2HA) in the presence of EscI–V5 or a vector control (pACYC184). (C) Whole-cell extracts were examined for V5 expression. (D) Whole-cell extracts of EPEC ΔescI expressing EscUWT–2HA or EscUN262A–2HA were subjected to SDS/16% PAGE and immunoblotting analysis using an anti-HA antibody to demonstrate the different cleavage states of the EscU protein. ** indicates a cross-reactive band. The molecular mass in kDa is indicated on the left-hand side.

EscI interacts with EscC

The structural analysis of the needle complex of Salmonella T3SS identified a new substructure formed by InvG (secretin) and PrgJ (inner rod) [15]. Since T3SSs are highly conserved, we predicted that EscI would interact with the EPEC secretin EscC if it indeed serves as an inner rod protein. To examine this, we introduced EscC fused to a FLAG–2HA tag (which remains fully functional, Supplementary Figure S1 at and EscI fused to a V5 tag in WT EPEC, using two different vectors. This WT strain carrying the two vectors showed normal type III protein secretion, indicating that the overexpression of EscC and EscI proteins has no effect on the function of the T3SS (Supplementary Figure S2 at Vector controls were used to examine non-specific interactions. Analysis of whole-cell extracts confirmed the expression of the tagged proteins (Figure 6A). Immunoprecipitation was performed using an anti-FLAG antibody, whereas the eluted fractions were immunoblotted against the V5 epitope. EscI–V5 co-eluted with EscC–FLAG–2HA (Figure 6B). The interaction between EscI and EscC further supports our assertion that EscI is indeed the inner rod. EscI–V5 did not show non-specific binding to the beads in the absence of EscC–FLAG–2HA. Immunoprecipitation of EscC–FLAG–2HA expressed in the E. coli BL21strain, which does not have a T3SS, together with EscI–V5 did not co-elute EscI–V5 (Supplementary Figure S3 at These results suggest that EscI and EscC do not interact directly and additional T3SS component(s) is needed to mediate their interaction. Alternatively, EscI is not targeted to its correct location in the absence of a functional T3SS and therefore cannot interact with EscC in the BL21 strain.

Figure 6 EscI interacts with EscC when co-expressed in EPEC

Whole-cell extracts of EPEC strains expressing EscC–FLAG–2HA, EscI–V5 or both proteins were subjected to immunoprecipitation using an anti-FLAG antibody. Whole-cell extracts prior to immunoprecipitation (A) and eluted fractions (B) were analysed by Western blot probing for EscC (using an anti-FLAG antibody) and EscI (using an anti-V5 antibody).


T3SSs are complex structures that are essential for the virulence of many Gram-negative pathogens including Yersinia, Salmonella, Shigella, Pseudomonas aeruginosa and EPEC. Although the majority of the proteins that form the T3SS in EPEC are characterized, several components that are known to exist in the flagellar system and in other T3SSs are still missing or are poorly characterized in EPEC [45]. Here, we studied one of these poorly characterized proteins, EscI, and demonstrated that this protein shares many common features with the class of T3SS structural proteins designated as the inner rod proteins, suggesting that EscI is the inner rod protein of the EPEC T3SS. Based on the finding that EscI interacts with the YscU/FlhB protein and the recent suggestion that the formation of the inner rod is critical for substrate-specificity switching [14,16], we can speculate on how EscI is involved in the substrate-specificity switch: EscU has two substrate specificity states that alter its substrate affinities. Early substrates (like EscI) have higher binding affinity to full-length EscU. Binding of these early substrates slows down the auto-cleavage event of EscU. However, as the secretion of the early substrates proceeds, the local concentration of these substrates decreases, relieving their repressive effect on EscU cleavage. The auto-catalytic cleavage event of EscU induces a conformational change thereby enhancing the affinity of late substrates for binding cleaved EscU (Figure 7). A second switch that promotes secretion of effector proteins (late substrates) over translocon components is mediated by the SepD/L proteins [46]. Further studies will be required to clarify the functional consequences of the EscI and EscU interaction.

Figure 7 Schematic illustration of the interaction between EscI, EscU and EscC within the context of the T3SS

(A) Side view of the T3SS. The inner-rod protein, EscI (green circles labelled I), interacts with the inner-membrane protein EscU (coloured light blue and labelled U) regardless of the EscU cleavage state. This allows secretion of EscI together with the needle protein (grey blocks). (B) As secretion of the early substrates proceeds, the inner rod and the needle structure are assembled allowing EscI to interact with EscC (coloured dark blue and labelled C). We speculate that when the local concentration of the early substrates decreases it relieves their repressive effect on EscU auto-cleavage. The auto-catalytic cleavage event of EscU induces a conformational change that enhances the affinity of late substrates (grey shapes in the cytosol) for binding cleaved EscU.


Neta Sal-Man designed and performed the experiments, analysed the data and wrote the paper. Wanyin Deng cloned the EscI–2HA in the pTOPO plasmid, performed the preliminary secretion experiment and contributed to the writing of the paper. Brett Finlay supervised the work and the writing of the paper. All authors read and approved the final paper.


This work was supported by the Canadian Institutes of Health Research (CIHR). B.B.F. is the University of British Columbia Peter Wall Distinguished Professor. N.S. was supported by postdoctoral fellowships from the Michael Smith Foundation for Health Research (MSFHR), the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Rothschild Foundation.


We thank members of the Finlay laboratory for critical reading of the paper before submission, and for Nat F. Brown for helping with the model of Figure 7.

Abbreviations: A/E, attaching and effacing; DMEM, Dulbecco's modified Eagle's medium; EPEC, enteropathogenic Escherichia coli; HA, haemagglutinin; HRP, horseradish peroxidase; HSV, herpes simplex virus; LB, Luria–Bertani; LEE, locus of enterocyte effacement; T3S, type III secretion; T3SS, T3S system; WT, wild-type


View Abstract