Molecular Microbiology (2000) 35(6), 1483±1492

The type III protein translocation system ofenteropathogenic Escherichia coli involvesEspA±EspB protein interactions

Elizabeth L. Hartland,1 Sarah J. Daniell,1

Robin M. Delahay,1 Bianca C. Neves,1 Tim Wallis,2

Robert K. Shaw,3 Christine Hale,1 Stuart Knutton3

and Gad Frankel1*1Department of Biochemistry, Imperial College of Science,

Technology and Medicine, London SW7 2AZ, UK.2Institute of Animal Health, Compton, Berkshire, UK.3Institute of Child Health, University of Birmingham,

Birmingham, UK.

Summary

Enteropathogenic Escherichia coli (EPEC), like many

bacterial pathogens, use a type III secretion system to

deliver effector proteins across the bacterial cell wall.

In EPEC, four proteins, EspA, EspB, EspD and Tir are

known to be exported by a type III secretion system

and to be essential for `attaching and effacing' (A/E)

lesion formation, the hallmark of EPEC pathogenicity.

EspA was recently shown to be a structural protein and

a major component of a large, transiently expressed,

filamentous surface organelle which forms a direct link

between the bacterium and the host cell. In contrast,

EspB is translocated into the host cell where it is

localized to both membrane and cytosolic cell frac-

tions. EspA and EspB are required for translocation of

Tir to the host cell membrane suggesting that they may

both be components of the translocation apparatus. In

this study, we show that EspB co-immunoprecipitates

with the EspA filaments and that, during EPEC infec-

tion of HEp-2 cells, EspB localizes closely with EspA.

Using a number of binding assays, we also show that

EspB can bind and be copurified with EspA. Never-

theless, binding of EspA filaments to the host cell

membranes occurred even in the absence of EspB.

These results suggest that following initial attachment

of the EspA filaments to the target cells, EspB is

delivered into the host cell membrane and that the

interaction between EspA and EspB may be important

for protein translocation.

Introduction

Most bacterial virulence determinants are surface located

or secreted outside the bacterium. Many pathogens of

animals (Yersinia, Salmonella, Shigella and Bordetella sp.)

and plants (Pseudomonas syringae, Ralstonia solanacearum,

Xanthomonas campestris) utilize a specialized mechanism of

secretion, termed type III secretion, to transport virulence

effector proteins across the bacterial cell envelope into the

host cell membrane or the cytosol of the target cell (Hueck,

1998). Generally the proteins secreted by type III secretion

systems are classified either as effector proteins, which act

to subvert host cell processes, or translocases, which

deliver the effector proteins into the eukaryotic cell. The

structural basis for protein translocation has not been fully

elucidated for any type III secretion system. Nevertheless,

data from studies of different type III secretion systems

suggest the presence of channel-forming proteins of bac-

terial origin, in both the bacterial outer membrane and in

the plasma membrane of the infected host cells (Hueck,

1998).

The components responsible for protein secretion are

broadly conserved in all type III secretion systems so that

one system can often export the proteins usually secreted

by another type III system (Rosqvist et al., 1995; Elliott

et al., 1999). Many of these type III components also

share homology with proteins involved in the export of

flagellar subunits (Hueck, 1998) and electron microscopy

studies of the Salmonella SPI1 and Shigella flexneri type III

secretion systems have shown the existence of a macro-

molecular complex which spans both bacterial membranes.

This complex consists of a basal-body-like structure, a

neck domain and an external needle (Kubori et al., 1998;

Blocker et al., 1999).

The enteric pathogens, enteropathogenic (EPEC) and

enterohaemorrhagic (EHEC) Escherichia coli, constitute a

significant risk to human health worldwide (reviewed by

Nataro and Kaper, 1998) and both depend on a type III

secretion system to colonize the intestinal mucosa (Jarvis

et al., 1995; Jarvis and Kaper, 1996). A hallmark of infec-

tion with either pathogen is the development of character-

istic `attaching and effacing' (A/E) lesions on the intestinal

mucosa (reviewed by Frankel et al., 1998). The genes

required for A/E lesion formation map to a chromosomal

pathogenicity island termed the LEE (Locus of Enterocyte

Q 2000 Blackwell Science Ltd

Received 25 August, 1999; revised 13 December, 1999; accepted 16December, 1999. *For correspondence. E-mail g.frankel@ic.ac.uk;Tel. (144) 171 594 5253; Fax (144) 171 594 5255.

1484 E. L. Hartland et al.

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1483±1492

Effacement) (McDaniel et al., 1995). The LEE-encoded

bacterial adhesion molecule intimin (Jerse et al., 1990;

Kelly et al., 1999), is essential for A/E lesion formation and

mediates intimate bacterium±host cell interaction through

binding to an intimin receptor (Tir), another LEE-encoded

protein that is delivered by the type III secretion system

into the host cell membrane (Kenny et al., 1997).

Three major E. coli secreted proteins (Esps) are known

to be exported by the LEE-encoded type III secretion

system (Elliott et al., 1998; Jarvis et al., 1995), EspA, EspB

and EspD (reviewed by Frankel et al., 1998). All three Esp

proteins are required for signal transduction and for the

formation of the A/E lesion, although until recently very

little was known about their biological function. EspA is

now known to be a structural protein and a major com-

ponent of a large (< 50 nM diameter) transiently

expressed filamentous surface organelle which, prior to

intimin-mediated intimate attachment, forms a direct link

between the bacterium and the host cell (Ebel et al., 1998;

Knutton et al., 1998). However, it is not known if EspA

binds host cells directly or via other proteins. EspB is not

thought to be a structural component of the filament

because antibodies to EspB do not stain EspA filaments

and intact EspA filaments can be observed on the surface

of an espB mutant of EPEC (Knutton et al., 1998). An

espD mutant, on the other hand, secretes only low levels of

EspA and produces barely detectable EspA filaments

(Knutton et al., 1998). At present, the reason for this is not

clear.

Following bacterial attachment, EspB is translocated

into the host cell where it is distributed in both membrane

and cytosol. This process requires the presence of EspA

filaments and is strongly enhanced by intimate bacterial

attachment (Wolff et al., 1998). EspB has weak homology

(21% identity) to YopB of Yersinia (Pallen et al., 1997), a

protein believed to form the translocation channel in the

host cell membrane through which other effectors are

delivered (Hakansson et al., 1996). While this similarity

suggests that EspB plays a similar role in EPEC/EHEC

pathogenesis, EspB also exhibits weak homology with

YopD (19% identity) and the structural organization of

EspB is more reminiscent of the YopD protein. Both pro-

teins 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 also itself translocated (Francis

and Wolf-Watz, 1998). Recently, EspB was expressed in

transfected HeLa cells where it induced changes in cell

morphology and actin stress fibres (Taylor et al., 1999). It is

possible therefore that EspB has a dual role in virulence.

We propose that in EPEC and EHEC, the EspA fila-

ments form a structural link between a bacterial mem-

brane pore (suggested by homology to be EscC, a putative

outer membrane component of the EPEC type III secretion

system) and the host cell translocation channel (EspD/

EspB), a model which implies interaction between EspA,

and the predicted pore forming proteins, EspB and EspD

(Frankel et al., 1998). Here we investigated protein±

protein interactions between EspA and EspB. The results

show that although EspA filaments can interact with host

cells when expressed in an espB deficient strain, EspB

binds EspA, is copurified with EspA filaments and local-

izes closely with EspA in tissue culture cells infected with

EPEC.

Results

Co-immunoprecipitation of EspB and EspA with EspA

antibodies

EspA filaments are a crucial factor in EPEC pathogenesis

and are believed to act in concert with other Esps to make

up the type III protein translocation apparatus (trans-

locon). In this study, we investigated whether other Esps

associate directly with the EspA filaments during bacterial

growth by immunoprecipitating the filaments with mono-

clonal antibodies to EspA. The prototype wild-type EPEC

strain (E2348/69) was grown under conditions that favour

EspA filament production. An espA mutant derivative of

E2348/69 (UMD872) (Kenny et al., 1996) was used as a

negative control. Following a 3 h incubation, the bacteria

were washed, incubated with the mouse monoclonal

EspA antibodies overnight and then immunoprecipitated

with protein-G agarose beads. The immunoprecipitated

material and bacterial pellets were subjected to Western

blots using rabbit polyclonal EspA and mouse polyclonal

EspB antisera. EspA was specifically detected in both the

immunoprecipitated material and bacterial cell pellet from

E2348/69. EspA could not be detected in either sample

obtained from UMD872. As expected, EspB was detected

in the bacterial cell pellets of both E2348/69 and UMD872

whereas EspB was detected in the immuno-precipitated

material from E2348/69 only (Fig. 1). EspB was not

detected in the bacterial cell pellet or immunoprecipitated

material of an espB mutant, strain UMD864 (Donnenberg

et al., 1993) (data not shown). These results suggest that

EspB is associated with EspA filaments during bacterial

growth and in the absence of host cells.

Binding of EspB and EspA on solid phase

To verify the specificity of the association, EspA and EspB

were subjected to a number of in vitro binding assays to

examine protein±protein interactions. Purified His±EspA

and MBP±EspB fusion proteins were used in gel

overlays to probe the ability of EspB to bind EspA.

Immobilized His±Tir±M (a 107-amino-acid Tir fragment

involved in intimin binding) (Hartland et al., 1999) was

EspA±EspB protein interactions 1485

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1483±1492

used as a negative control. In parallel, Western blots

using monoclonal anti-EspA (Fig. 2, lane 1) and poly-

clonal anti-Tir±M (Fig. 2, lane 2) were used to visualize

the immobilized EspA and Tir±M polypeptides. In solution,

EspA is present in several polymerized forms (due, in part,

to coiled±coil interactions between monomeric EspA

polypeptides) (Delahay et al., 1999). Indeed the EspA

monoclonal antibodies reacted with several EspA bands

(Fig. 2, lane 1). Overlaying MBP±EspB (5 mg ml21) on

immobilized His±EspA, followed by incubation with anti-

MBP and AP conjugated anti-rabbit antiserum, revealed

interactions with similar bands (Fig. 2, lane 3), although

an extra, faint band was recognized by MBP±EspB

(Fig. 2 lane 3) but not the anti-EspA monoclonal

antibodies (Fig. 2 lane 1). Because this band is missing

from the control it is unlikely to represent a non-specific

interaction, although this possibility can not be excluded at

present. MBP alone did not bind any of the EspA bands

(data not shown) and MBP±EspB did not bind Tir±M

(Fig. 2, lane 4).

To investigate EspA±EspB protein interactions further,

an ELISA-based overlay-binding assay was employed.

For this, 0.5 mg purified His-tagged EspA was used to coat

ELISA plate wells. Following washes, increasing dilutions

of an MBP±EspB fusion were added to the wells and

incubated for 2 h. Equivalent amounts of MBP alone were

used as control. Binding of MBP±EspB was detected by

incubating the wells with anti-MBP followed by HRP-

conjugated anti-rabbit antiserum. This revealed a dose-

dependent interaction between MBP±EspB and EspA

(Fig. 3A). MBP±EspB binding to immobilized His±EspA

Fig. 2. Detection of EspA±EspB interactions using gel overlays.Lanes 1 and 3, His±EspA; Lanes 2 and 4, His±Tir±M. Lanes 1 and2 represent Western blots using anti-EspA monoclonal antibodyand anti±Tir±M polyclonal antiserum respectively. Lanes 3 and 4show an overlay with MBP±EspB (5 mg ml21), detected with anti-MBP antiserum and AP-conjugated anti-rabbit antibodies.

Fig. 1. Co-immunoprecipitation of EspA and EspB. Lane 1, EPECE2348/69 (wild type) cell pellet; Lane 2, EPEC UMD872 (EspA-)cell pellet; Lane 3. EPEC E2348/69 (wild type) immunoprecipitate;Lane 4, EPEC UMD872 (EspA-) immunoprecipitate. Blots weredetected with anti-EspA and anti-EspB polyclonal antiserasimultaneously.

Fig. 3. A. ELISA-based overlay showing binding between EspA andEspB. His±EspA (0.5 mg well21) was used to coat ELISA plateswhich were then incubated with serial dilutions of MBP±EspB (A)or MBP (V).B. Competitive inhibition of binding between EspA and MBP±EspBusing His±EspB. His±EspA was used to coat ELISA plates asabove and wells were incubated with serial dilutions of MBP±EspB(A) and MBP±EspB in the presence of 100 mg ml21 His±EspB (V).Binding was detected with anti-MBP antiserum and anti-rabbit HRPantibodies. Error bars represent standard deviation from sixindividual tests.

1486 E. L. Hartland et al.

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1483±1492

was partially inhibited by performing the binding assay in

the presence of excess soluble His±EspB (100 mg ml21)

as a competitor (Fig. 3B). No binding was detected with

the MBP-only negative control (Fig. 3A). Taken together,

these results demonstrate that EspB binds directly to EspA

in solid phase.

Copurification of EspB with a GST±EspA fusion protein

In addition, we utilized an independent bioassay to demon-

strate EspA±EspB interactions. For this purpose, we con-

structed a synthetic operon consisting of a GST±EspA

fusion and EspB (pICC35) and an operon consisting of

GST only and EspB (pICC36) as a control (Fig. 4A). E. coli

XL1 Blue was transformed with pICC35 and pICC36, the

cultures induced with IPTG, sonicated and the supernatant

subjected to affinity purification with glutathione sepharose

resin. Copurification of EspB with the EspA fusion was

examined by Western blot using GST and EspB antisera.

Figure 4(B) shows that both GST±EspA and GST were

eluted following affinity purification. In contrast, EspB was

eluted only when it was loaded together with GST±EspA.

No EspB copurified with GST only. This result provides

further support for a specific EspB±EspA interaction.

Binding of secreted EspB to a MBP±EspA column

To confirm that recombinant EspA binds the native

secreted form of EspB from EPEC, columns containing

bound MBP±EspA or MBP only were overlaid with 25 ml

of filtered culture supernatant derived from E2348/69 grown

in DMEM. MBP±EspA was generated from plasmid

pICC38 following amplification and insertion of espA into

pMALc-2. Following elution of MBP or MBP±EspA with

10 mM maltose, 1 ml fractions were collected and sub-

jected to Western blots with anti-MBP and anti-EspB

antiserum. The results showed that EspB elutes specifi-

cally with MBP±EspA but not MBP (Fig. 5), demonstrat-

ing the ability of EspA to bind the secreted form of EspB

from EPEC.

Interaction of EPEC espB mutant with host cells

We previously showed by immunofluorescence and

immunogold-labelling electron microscopy that EspA fila-

ments of wild-type EPEC strain E2348/69 interact with the

surface of host cells during early stages of A/E lesion

formation (Knutton et al., 1998); similar-sized filamentous

structures seen linking bacteria and the host cell surface

Fig. 4. A. Schematic diagram showing the artificial operonsconstructed from pGEX-2T between a GST±EspA fusion and EspB(pICC35), and GST and EspB (pICC36). RBS represents the espBribosome binding site.B. Copurification of GST±EspA and EspB. Lanes 1 and 3, E. coliXL1-Blue pICC35; Lanes 2 and 4, E. coli XL1-Blue (pICC36).Exponential cultures of E. coli XL1-Blue carrying pICC35 or pICC36were induced with IPTG for 3 h and GST derivatives were purifiedwith glutathione sepharose. Samples were analysed by Westernblot. Lanes 1 and 2 show an immunoblot with anti-GST antibodies.Lanes 3 and 4 show an immunoblot with mouse anti-EspBantiserum. Arrows point to purified GST±EspA and GST andcopurified EspB.

Fig. 5. Co-elution of MBP±EspA and secreted EspB. Lane 1,Supernatant from EPEC strain E2348/69; Lane 2, MBP±EspA;Lane 3, MBP±EspA 1 supernatant (fraction 1); Lane 4, MBP±EspA 1 supernatant (fraction 2); Lane 5, MBP±EspA 1 supernatant (fraction 3); Lane 6, MBP 1 supernatant(fraction 1); Lane 7, MBP 1 supernatant (fraction 2); Lane 8,MBP 1 supernatant (fraction 3).A. Immunoblot with anti-MBP antiserum.B. Immunoblot with anti-EspB antiserum. Arrow shows EspB.

EspA±EspB protein interactions 1487

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1483±1492

by scanning electron microscopy were interpreted as being

EspA filaments (Knutton et al., 1998). In this study, we

have demonstrated by immunofluorescence production of

typical EspA filaments by the espB mutant strain UMD864

and show that EspA filaments similarly interact directly

with the host cell surface (Fig. 6). Filamentous structures

seen to link strain UMD864 and the host cell were con-

firmed as EspA filaments by immunogold labelling and

scanning electron microscopy (SEM) (Fig. 6). These obser-

vations show that EspB is not essential for EspA filament

interaction with host cells.

Detection of EspB in infected host cells

Previous reports have shown that following infection of

epithelial cells in culture, EspB fractionates with host cell

cytosolic and membranous fractions; immunofluorescence

and confocal microscopy revealed zones of EspB accumu-

lation immediately beneath adherent bacteria (Knutton

et al., 1998; Wolff et al., 1998). Here, we used EspB serum

and fluorescence microscopy at higher power to visualize

EspB in infected host cells. First, we noted that EspB

staining was only associated with a small proportion of

adherent bacterial microcolonies and second we noted, in

Fig. 7A, these foci were localized adjacent to individual

adherent bacteria and had a diameter similar to that of

EspA filaments. Double immunostaining using EspB

and EspA antisera, revealed close localization of EspA

filaments expressed on the surface of adherent bacteria

and these more intense accumulations of EspB (Fig. 7).

These observations suggest that concentrated EspB

might, like EspA filaments, have a transient function during

the operation of the type III secretion system.

Discussion

Like many other Gram-negative enteric pathogens, EPEC

and EHEC appear to rely on a complex specialized type

Fig. 6. EspB is not required for interaction of EspA filaments with host cells. HEL cells were infected for 3 h with wild type and espB-EPEC,washed, fixed and processed for microscopy. EspA filaments of espB±EPEC, like wild-type EPEC, were seen by immunofluorescence (A,arrows) and scanning electron microscopy (B, arrows) to interact with the HEL cell surface; filaments seen by scanning electron microscopywere confirmed as EspA filaments by immunogold labelling (C, arrows). Magnification bars, A, 5-mm; B and C, 0.5 mm.

Fig. 7. EspB localizes closely with EspA filaments in infected HEp-2 cells. HEp-2 cells were infected for 3 h with wild-type EPEC, fixed,permeabilized and double stained with anti-EspA and anti-EspB antibody. EspB (A) and EspA (B) immunofluorescence and a correspondingphase contrast (C) image of a microcolony of adherent bacteria are shown. Discrete foci of EspB staining (A, arrow) seen within zones ofmore diffuse EspB staining (A) localize closely with EspA filaments expressed on the surface of adherent bacteria (B, arrow). Magnificationbar, 5 mm.

1488 E. L. Hartland et al.

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1483±1492

III secretion system for their exploitation of host cell

machinery (Hueck, 1998). In EPEC, three major Esp

proteins are secreted by the type III secretion system and

their precise function is the subject of intense investiga-

tion. Because Esps are known to be essential for signal

transduction and A/E lesion formation, elucidating their

role in virulence is crucial for the understanding of EPEC

pathogenesis. Using a molecular approach to clone, purify

and raise antiserum to EspA, we and others showed that

EspA is a structural protein and a major component of a

large filamentous surface organelle (Ebel et al., 1998;

Knutton et al., 1998), the precise structure and composi-

tion of which we are now investigating. Recently, we

demonstrated that the EspA filaments are made up of

multimeric EspA and that intermolecular EspA coiled±coil

interactions are essential for the structural integrity of the

filaments (Delahay et al., 1999). In this study, we have

investigated protein±protein interactions between the

EspA filaments and EspB. We have shown that EspB is

copurified with the EspA filaments in the absence of host

cells and that EspB binds EspA in solid and liquid phase.

We also demonstrate that EspB associates closely with

EspA filaments on infected host cells but that EspB is not

required for attachment of the filaments to the host cell

surface.

Given that EspB copurifies with EspA filaments in the

absence of host cells, it is possible that EspB is located

inside the filament prior to infection where it is inacces-

sible to anti-EspB antibodies or folded into a form that is

not recognized and is then released only upon contact

with the host cell surface. So far we have been unable to

demonstrate co-immunoprecipitation of EspA filaments with

EspD or Tir, suggesting that these Esps are not located in

the filament prior to host cell attachment (unpublished

data).

EspB is a translocated protein that requires both the

EspA filaments and functional EspB to be transported into

the host cell where it is found in both the cell membrane

and the cytosol (Knutton et al., 1998; Wolff et al., 1998).

Initial studies failed to show staining of the EspA filaments

with polyclonal EspB antiserum, suggesting that EspB is

not a structural component of the filament. Here we show

that EspA filaments can interact with host cells in the

absence of EspB, demonstrating that EspB is not the

initial receptor for the filament. EspB [and EspD, which

was also shown to be translocated to the host cell mem-

brane (Wachter et al., 1999)] has weak structural similarity

to YopB and YopD, two proteins secreted by Yersinia that

form a translocation channel through which cytotoxins are

delivered to the host cell (Hakansson et al., 1996; Tardy

et al., 1999). However, to date no EspA filament-like struc-

tures have been observed in Yersinia. Importantly, the

results presented here show that attachment of EspA

filaments to host cells can be distinguished from protein

translocation. Based on similarity of EspB/D with YopD/B

(Neyt and Cornelis, 1999a) and IpB/C (Menard et al., 1996;

Blocker et al., 1999), if a pore is produced by a complex of

these Esp proteins, then EspA filaments do not need a

functional pore to bind to the host cell surface; the associ-

ation of EspB/D in the membrane and their presumed

interaction with EspA filaments must be a subsequent

event.

Evidence supporting the formation of membrane pores

was demonstrated using contact haemolytic activity (Hakans-

son et al., 1996) and we (unpublished observations) and

others (Warawa et al., 1999) have recently shown that

EPEC induce haemolysis of red blood cells. In support of

EspB generated pores in the red blood cell membrane is

the observation that, unlike wild-type EPEC, strain UMD864

(espB±) was unable to lyse red blood cells. In this paper,

high power fluorescence microscopy of translocated EspB

revealed discrete foci of fluorescent staining closely

adjacent to adherent bacteria and with a diameter similar

to that of the EspA filaments. Although, at the present

time, we are unable to precisely define the localization of

such EspB staining, the discrete nature of the fluores-

cence suggests that localized EspB could be concen-

trated in the host cell membrane. Previous studies have

shown that in the presence of host cells EspB is protease

resistant (Kenny and Finlay, 1995), suggesting this pro-

tein does not have a surface location. However, it is likely

that EspB is protected from proteolytic degradation by its

interaction with EspA and insertion into the host membrane.

As mentioned above, EspB also has similarity with

YopD, which is translocated into the host cell cytosol and

membrane by Yersinia (Francis and Wolf-Watz, 1998).

Although the precise mechanism of action of YopD remains

unclear, this protein appears to have direct roles in viru-

lence gene regulation and translocation, as well as pos-

sible functions in the host cell. Thus, although EspB

appears to form part of the type III protein translocation

apparatus with EspA and possibly with EspD, it may also

have a cytosolic function.

EscC (formerly SepC) is a major component of the

EPEC type III secretion system and is predicted to be

exposed at the bacterial surface (Elliott et al., 1998). YscC,

a Yersinia homologue of EscC, has recently been shown to

form a ring-shaped oligomeric complex in the outer mem-

brane with an < 20 nM diameter central pore (Koster et al.,

1997). Thus, EscC may also represent this type of novel

channel-forming protein in the bacterial outer membrane.

Our working hypothesis is that proteins are carried from

pathogenic bacteria to the host cell through a pore in

hollow EspA-like filaments which bridge the bacterial (EscC)

and host cell membrane (EspB and EspD) channels.

Although it remains to be shown if the separate com-

ponents of the proposed EPEC translocon are function-

ally integrated as suggested, such an apparatus could

EspA±EspB protein interactions 1489

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1483±1492

function as a `molecular syringe' which, powered by the

type III secretion system ATPase, EscN, injects proteins

directly from the bacterium into the host cell cytosol

(Frankel et al., 1999).

In conclusion, our observations now suggest that the

EspA filaments might have a dual function: they may func-

tion initially to sense host target cells and subsequently,

following pore formation, function as an integral part of the

type III translocon. The data presented in this study show,

for the first time, protein±protein interactions between

different Esps. We believe that this provides an important

step forward in our understanding of EPEC and EHEC

pathogenesis and the function of type III secretion

systems in general.

Experimental procedures

Bacterial strains and plasmids

Bacterial strains used in this study include EPEC strainsE2348/69 (wild type), UMD872 (espA-) (Kenny et al., 1996)and UMD864 (espB-) (Donnenberg et al., 1993), E. coli XL-1Blue and BL21. Strains were grown in Luria broth or on Luriaagar supplemented with kanamycin (100 mg ml21) or ampi-cillin (100 mg ml21) where appropriate. The plasmids usedare listed in Table 1.

Preparation of monoclonal antibodies to EspA

For the production of monoclonal antibodies, BALB/c micewere immunized by two subcutaneous injections with 90 mgof purified His±EspA (Knutton et al., 1998) in incompleteFreund's adjuvant. A final intraperitoneal injection of antigen(90 mg) without adjuvant was performed 4 days before themouse was killed and the spleen removed. Splenocytes wereprepared and fused with SP2/0 mouse myeloma cells usingpolyethylene glycol 1500 (Boehringer) following standardprocedures (Harlow and Lane, 1988). The fusion mix wasdiluted in RPMI/HAT medium with 10% foetal bovine serumand plated into 96-well plates seeded 48 h previously withmouse peritoneal macrophages. The plates were incubatedat 378C in a 5% CO2 atmosphere for 10 days when samples

of the supernatant were taken to test for antibody. Positivewells were subcultured and the hybridoma cells cloned twiceby limiting dilution. Four lines, HF7, HA9, AC2 and EG4 wereconsidered to be monoclonal. All four lines were isotyped asIgG1, and were able to detect secreted EspA by ELISA andWestern blot. Culture supernatant (50 ml) containing anti-body secreted from clone HA9 was selected for purificationusing a Protein G Sepharose column (Pharmacia) followingstandard techniques.

Preparation of mouse polyclonal antiserum to EspB

His±EspB was purified as described previously (Frankelet al., 1996). Six- to eight-week-old TO mice were immunizedsubcutaneously with < 20 mg purified His-EspB and wereboosted twice with the same antigen in incomplete His±EspBin complete Freund's adjuvant. Animals Freund's adjuvantwith 2 week intervals before exsanguination. Antiserum wastested against purified EspB by Western blotting.

Co-immunoprecipitation of EspA and EspB

EPEC strains E2348/69 and UMD872 (EspA-) were grownin 5 ml Luria±Bertani (LB) standing at 378C overnight. Theovernight cultures were used to inoculate 5 mls of DMEMcontaining 10% fetal calf serum at a 1:100 dilution. Thecultures were incubated at 378C without agitation for 3 h andcentrifuged at 4000 r.p.m. for 10 min at 48C to pellet the cells.The cell pellets were resuspended gently in 500 ml PBS andre-centrifuged. This was repeated twice. Cell pellets were theneither used for immunoprecipitation, or resuspended in100 ml of Laemmli buffer for SDS±PAGE and Westernblotting. For immunoprecipitation, the cell pellets were resus-pended in 50 ml of PBS and 5 ml of monoclonal anti-EspAantibody was added. The samples were incubated at 48Covernight. EspA filaments were then gently dislodged fromthe bacteria and the samples centrifuged at 15 000 r.p.m. ina microfuge for 5 min to pellet the bacterial cells. The super-natant was transferred to a fresh tube and the filamentsprecipitated with 10 ml of a 50% protein G-agarose beadslurry. The samples were centrifuged (as above) to pellet theprotein G-beads. The beads were washed in 50 ml of PBSand resuspended in 100 ml of Laemmli buffer for analysis bySDS±PAGE and Western blotting.

Table 1. Plasmids used in this study.

Plasmid Relevant features Reference

pETespA pET28a expressing His-tagged EspA Knutton et al. (1998)pETespB pET28a expressing His-tagged EspB Frankel et al. (1996)pMALc-2 Cloning vector generating

translational fusions with MBPNEB

pGEX-2T Cloning vector generatingtranslational fusions with GST

Pharmacia

pICC35 pGEX-2T expressing GST±EspAfusion protein and EspB

This study

pICC36 pGEX-2T expressing GST and EspB This studypICC37 pMALc-2 expressing MBP±EspB

fusion proteinThis study

pICC38 pMALc-2 expressing MBP±EspAfusion protein

This study

1490 E. L. Hartland et al.

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1483±1492

EspA/EspB gel overlays

His±EspA was purified as described previously (Knuttonet al., 1998). EspB was amplified by PCR as describedpreviously (Frankel et al., 1996) and cloned into pMALc-2(NEB) to generate a translational fusion with maltose bindingprotein (MBP) (pICC37). The recombinant plasmids weretransformed into E. coli TG1 and log phase cultures wereinduced with 1 mM IPTG followed by incubation at 378C for3 h with shaking. MBP±EspB was purified with amylose resinas described (Frankel et al., 1994).

His±EspA and His±Tir±M were separated by SDS±PAGE,blotted onto nitrocellulose and blocked with 10% skimmedmilk in PBS, 0.1% Tween-20 for 1 h. Membranes were incu-bated with MBP±EspB or MBP (5 mg ml21) overnight at 48C,washed and detected with anti-MBP antiserum and AP-conjugated anti-rabbit antibodies as described previously(Hartland et al., 1999).

EspA/EspB ELISA overlays

Purified His±EspA (5 mg ml21) in carbonate/bicarbonate bufferwas used to coat an ELISA plate (100 ml well21). Wells wereincubated with serial dilutions of MBP±EspB or MBP alone(100 ml well21) for 2 h and then detected with anti-MBPantiserum (1/2000 dilution, 1.5 h) followed by HRP-conjugatedanti-rabbit antiserum (1/2000, 1.5 h). The colour reaction wasdeveloped with o-phenylenediamine(OPD)/H2O2 (Sigma) andstopped with 3 M sulphuric acid. The final OD was measuredat 490 nM.

Construction of gst±espA, espB synthetic operons

espA was amplified by PCR from EPEC strain 2348/69 usingthe primer pair, 5 0-CGGGATCCATGGATACATCAACTAC-3 0

and 5 0-CGGGATCCTTATTTACCAAGGGATATTCC-3 0. ThePCR product was cloned into the BamHI site of pGEX-2T(Pharmacia) to generate a translational fusion with gluta-thione S-transferase (GST) (Neyt and Cornelis, 1999b). espBwas amplified by PCR using the primers 5 0-CGGAATTCT-TATTAAAGAGGCGTTTATTATG-3 0 and 5 0-CGGAATTCTT-ACCCAGCTAAGCGAGC-3 0, which were designed to includethe native ribosome binding site of espB. This fragment wascloned downstream of the GST±EspA fusion into the EcoRIsite of pGEX-2T to generate a transcriptional fusion with gst±espA. The resulting plasmid was called pICC35. A negativecontrol plasmid (gst, espB) was constructed by redigestingpICC35 with BamHI to remove espA. 5 0 overhangs were thenrepaired with Klenow enzyme to generate blunt ends andintroduce a stop codon after gst. This plasmid was religatedand called pICC36. PICC35 and pICC36 were transformedinto E. coli XL1-Blue.

Copurification of GST±EspA and EspB

Production of GST or GST±EspA was induced after theaddition of 1 mM IPTG to log phase cultures and incubationat 378C for 3 h with shaking. For purification of GST proteins,induced cultures were sonicated and 10 ml clarified sonicatewas mixed with 100 ml (50% slurry) glutathione sepharose 4Bbeads (Pharmacia). After 30 min, beads were washed three

times with 1 ml PBS and resuspended directly into 50 ml ofLaemmli buffer for analysis by SDS±PAGE and Westernblotting.

MBP±EspA/EspB affinity column

espA was amplified by PCR from EPEC strain 2348/69 usingthe primer pair, 5 0-GGAATTCATGGATACATCAACTACAG-CATCA-3 0 and 5 0-TATCTGCAGTTATTTACCAAGGGATATTCCTG-3 0. The PCR product was cloned into the site of pMALc-2(NEB) to generate a translational fusion with MBP. Therecombinant plasmid was transformed into E. coli TG1 andlog phase cultures were induced with 1 mM IPTG followed byincubation at 378C for 3 h with shaking. MBP±EspA waspurified with amylose resin as described (Frankel et al.,1994). For affinity columns, MBP or MBP±EspA was boundto amylose resin according to the standard purificationprocedure (Frankel et al., 1994). The column was thenoverlaid with 25 ml of filtered culture supernatant from EPECstrain 2348/69 grown overnight in DMEM. Following washeswith 10 volumes of column buffer (50 mM Tris-Cl pH 7.4,200 mM NaCl, 1 mM EDTA), MBP or MBP±EspA andassociated proteins were eluted with 10 mM maltosedissolved in column buffer. Fractions were collected in 1 mlvolumes and 15 ml of each fraction was subjected to SDS±PAGE and immunoblotting with anti-MBP and anti-EspB andAP-conjugated anti-rabbit antibodies.

Immunofluorescence and scanning electron microscopy

Epithelial cells [HEp-2, HEL (human embryonic lung)] wereseeded and grown overnight on glass coverslips. After a 3 hEPEC infection, cells were washed with PBS and fixed ineither 4% formalin (immunofluorescence) or 0.1% glutaralde-hyde (SEM).

For immunofluorescence, fixed cells were washed threetimes with PBS, permeabilized with 0.1% Triton X-100 in PBSfor 4 min and washed a further three times with PBS. Cellswere incubated with 20 ml of primary (anti-EspA, anti-EspB)antibodies for 45 min, washed and incubated with 20 ml ofsecondary Alexa488 conjugated anti-rabbit IgG antibody(Molecular Probes). For double staining of EspA and EspB, aprimary mouse anti-EspA and secondary anti-mouse Alexa594conjugated IgG antibody was used. Samples were examinedand photographed with a Leitz Dialux microscope.

For scanning electron microscopy, glutaraldehyde fixedHEL cell monolayers were washed, incubated with EspAantiserum for 2 h, washed again and fixed in 3% bufferedglutaraldehyde. For immunogold labelling, samples wereincubated for 2 h with 30 nM gold conjugated anti-rabbit IgG(British BioCell) prior to fixation in 3% glutaraldehyde. Mono-layers were post-fixed in 1% osmium tetroxide, dehydratedthrough graded acetone solutions and critical point dried.Mounted specimens were sputter coated with platinum(Polaron) and examined in a Jeol 1200EX scanning trans-mission electron microscope operated at 40 kV.

Acknowledgements

E.L.H. is the recipient of a Royal Society/NHMRC Howard Florey

Fellowship. This work was supported by the BBSRC and the

Leverhulme and Wellcome Trusts.

EspA±EspB protein interactions 1491

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1483±1492

References

Blocker, A., Gounon, P., Larquet, E., Niebuhr, K., Cabiaux, V.,

Parsot, C., et al. (1999) The tripartite type III secreton of

Shigella flexneri inserts IpaB and IpaC into host membranes. JCell Biol 147: 683±693.

Delahay, R.M., Knutton, S., Shaw, R.K., Hartland, E.L., Pallen,

M.J., and Frankel, G. (1999) The coiled-coil domain of EspA is

essential for the assembly of the type III secretion transloconon the surface of enteropathogenic E coli. J Biol Chem 274:

35969±35974.

Donnenberg, M.S., Yu, J., and Kaper, J.B. (1993) A secondchromosomal gene necessary for intimate attachment of

enteropathogenic Escherichia coli to epithelial cells. J Bacteriol

75: 4670±4680.

Ebel, F., Podzadel, T., Rohde, M., Kresse, A.U., Kramer, S.,Deibel, C., et al. (1998) Initial binding of Shiga toxin-producing

Escherichia coli to host cells and subsequent induction of actin

rearrangements depend on filamentous EspA-containing sur-face appendages. Mol Microbiol 30: 147±161.

Elliott, S.J., Dubois, M.S., Hutcheson, S.W., Wainwright, L.A.,

Batchelor, M., Frankel, G., et al. (1999) Identification of CesT, a

chaperone for the type III secretion of Tir in EnteropathogenicEscherichia coli. Mol Microbiol 33: 1176±1189.

Elliott, S.J., Wainwright, L.A., McDaniel, T.K., Jarvis, K.G., Deng,

Y.K., Lai, L.C., et al. (1998) The complete sequence of thelocus of enterocyte effacement (LEE) from enteropathogenic

Escherichia coli E2348/69. Mol Microbiol 28: 1±4.

Francis, M.S., and Wolf-Watz, H. (1998) YopD of Yersinia

pseudotuberculosis is translocated into the cytosol of HeLaepithelial cells: evidence of a structural domain necessary for

translocation. Mol Microbiol 29: 799±813.

Frankel, G., Candy, D.C., Everest, P., and Dougan, G. (1994)Characterization of the C-terminal domains of intimin-like pro-

teins of enteropathogenic and enterohemorrhagic Escherichia

coli, Citrobacter freundii, and Hafnia alvei. Infect Immun 62:

1835±1842.

Frankel, G., Phillips, A.D., Novakova, M., Field, H., Candy, D.C.,

Schauer, D.B., et al. (1996) Intimin from enteropathogenic

Escherichia coli restores murine virulence to a Citrobacterrodentium eaeA mutant: induction of an immunoglobulin A

response to intimin and EspB. Infect Immun 64: 5315±5325.

Frankel, G., Phillips, A.D., Rosenshine, I., Dougan, G., Kaper,J.B., and Knutton, S. (1998) Enteropathogenic and entero-

haemorrhagic Escherichia coli: More Subversive Elements.

Mol Microbiol 30: 911±921.

Frankel, G., Wren, B., Dougan, G., Delahay, R.M., Batchelor, M.,Hale, C., et al. (1999) Host/pathogen interactions during

infection by enteropathogenic Escherichia coli: a role for

signaling. In Microbial Signaling and Communication. England,

R., Hobbs, G., Bainton, N., and Roberts, D. McL. (eds). Societyfor General Microbiology Symposium 57. Cambridge, UK:

Cambridge University Press, 197±213.

Hakansson, S., Schesser, K., Persson, C., Galyov, E.E.,Rosqvist, R., Homble, F., et al. (1996) The YopB protein of

Yersinia pseudotuberculosis is essential for the translocation of

Yop effector proteins across the target cell plasma membrane

and displays a contact-dependent membrane disruptingactivity. EMBO J 15: 5812±5823.

Harlow, E., and Lane, D., (eds)(1988) Antibodies, A Laboratory

Manual. Cold Spring Harbour, NY: Cold Spring HarbourLaboratory Press, p. 211.

Hartland, E.L., Batchelor, M., Delahay, R.M., Hale, C., Matthews,

S., Dougan, G., et al. (1999) Binding of intimin from

enteropathogenic Escherichia coli to Tir and to host cells.Mol Microbiol 32: 151±158.

Hueck, C.J. (1998) Type III protein secretion systems in bacterial

pathogens of animals and plants. Microbiol Mol Biol Rev 62:

379±433.

Jarvis, K.G., Giron, J.A., Jerse, A.E., McDaniel, T.K., Donnenberg,M.S., and Kaper, J.B. (1995) Enteropathogenic Escherichia coli

contains a putative type III secretion system necessary for the

export of proteins involved in attaching and effacing lesion

formation. Proc Natl Acad Sci USA 92: 7996±8000.

Jarvis, K.G., and Kaper, J.B. (1996) Secretion of extracellularproteins by enterohemorrhagic Escherichia coli via a putative

type III secretion system. Infect Immun 64: 4826±4829.

Jerse, A.E., Yu, J., Tall, B.D., and Kaper, J.B. (1990) A genetic

locus of enteropathogenic Escherichia coli necessary for theproduction of attaching and effacing lesions on tissue culture

cells. Proc Natl Acad Sci USA 87: 7839±7843.

Kelly, G., Prasannan, S., Daniell, S., Fleming, K., Frankel, G.,

Dougan, G., et al. (1999) Structure of the cell-adhesion

fragment of intimin from enteropathogenic Escherichia coli.Nature Struct Biol 6: 313±318.

Kenny, B., DeVinney, R., Stein, M., Reinscheid, D.J., Frey, E.A.,

and Finlay, B.B. (1997) Enteropathogenic E. coli (EPEC)

transfers its receptor for intimate adherence into mammalian

cells. Cell 91: 511±520.

Kenny, B., and Finlay, B.B. (1995) Protein secretion by entero-

pathogenic Escherichia coli is essential for transducing signals

to epithelial cells. Proc Natl Acad Sci USA 92: 7991±7995.

Kenny, B., Lai, L.C., Finlay, B.B., and Donnenberg, M.S. (1996)

EspA, a protein secreted by enteropathogenic Escherichia coli,is required to induce signals in epithelial cells. Mol Microbiol 20:

313±323.

Knutton, S., Rosenshine, I., Pallen, M.J., Nisan, I., Neves, B.C.,

Bain, C., et al. (1998) A novel EspA-associated surface organ-elle of enteropathogenic Escherichia coli involved in protein

translocation into epithelial cells. EMBO J 17: 2166±2176.

Koster, M., Bitter, W., de Cock, H., Allaoui, A., Cornelis, G.R.,

and Tommassen, J. (1997) The outer membrane component,

YscC, of the Yop secretion machinery of Yersinia enterocoliticaforms a ring-shaped multimeric complex. Mol Microbiol 26:

789±797.

Kubori, T., Matsushima, Y., Nakamura, D., Uralil, J., Lara-Tejero,

M., Sukhan, A., et al. (1998) Supramolecular structure of the

Salmonella typhimurium type III protein secretion system.Science 280: 602±605.

McDaniel, T.K., Jarvis, K.G., Donnenberg, M.S., and Kaper, J.B.

(1995) A genetic locus of enterocyte effacement conserved

among diverse enterobacterial pathogens. Proc Natl Acad SciUSA 92: 1664±1668.

Menard, R., Prevost, M.C., Gounon, P., Sansonetti, P., and

Dehio, C. (1996) The secreted Ipa complex of Shigella flexneri

promotes entry into mammalian cells. Proc Natl Acad Sci USA

93: 1254±1258.

Nataro, J.P., and Kaper, J.B. (1998) Diarrheagenic Escherichia

coli. Clin Microbiol Rev 11: 142±201.

Neyt, C., and Cornelis, G.R. (1999a) Insertion of a Yop

translocation pore into the macrophage plasma membrane by

Yersinia enterocolitica: requirement for translocators YopB andYopD, but not LcrG. Mol Microbiol 33: 971±981.

Neyt, C., and Cornelis, G.R. (1999b) Role of SycD, the chaper-

one of the Yersinia Yop translocators YopB and YopD. Mol

Microbiol 31: 143±156.

Pallen, M.J., Dougan, G., and Frankel, G. (1997) Coiled-coil

1492 E. L. Hartland et al.

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1483±1492

domains in proteins secreted by type III secretion systems. MolMicrobiol 25: 423±425.

Rosqvist, R., Hakansson, S., Forsberg, A., and Wolf-Watz, H.(1995) Functional conservation of the secretion and transloca-

tion machinery for virulence proteins of yersiniae, salmonellae

and shigellae. EMBO J 14: 4187±4195.

Tardy, F., HombleÂ, F., Neyt, C., Wattiez, R., Cornelis, G.R.,

Ruysschaert, J.-M., et al. (1999) Yersinia enterocolitica type IIIsecretion-translocation system: channel formation by secreted

Yops. EMBO J 18: 6793±6799.

Taylor, K.A., Luther, P.W., and Donnenberg, M.S. (1999) Expres-

sion of the EspB protein of enteropathogenic Escherichia coli

within HeLa cells affects stress fibers and cellular morphology.Infect Immun 67: 120±125.

Wachter, C., Beinke, C., Mattes, M., and Schmidt, M.A. (1999)

Insertion of EspD into epithelial target cell membranes by

infecting enteropathogenic Escherichia coli. Mol Microbiol 31:1695±1707.

Warawa, J., Finlay, B.B., and Kenny, B. (1999) Type III secretion-

dependent hemolytic activity of enteropathogenic Escherichiacoli. Infect Immun 67: 5538±5540.

Wolff, C., Nisan, I., Hanski, E., Frankel, G., and Rosenshine, I.

(1998) Protein translocation into HeLa cells by infecting

enteropathogenic Escherichia coli. Mol Microbiol 28: 143±155.