Insulin receptor tyrosine kinase substrate links
the E. coli O157:H7 actin assembly effectors
Tir and EspFUduring pedestal formation
Didier Vingadassaloma, Arunas Kazlauskasb, Brian Skehana, Hui-Chun Chengc, Loranne Magouna, Douglas Robbinsa,
Michael K. Rosenc, Kalle Sakselab, and John M. Leonga,1
aDepartment of Molecular Genetics and Microbiology, University of Massachusetts Medical School, Worcester, MA 01655;bDepartment of Virology,
Haartman Institute, University of Helsinki and HUSLAB, Helsinki University Central Hospital, FIN-00014, Helsinki, Finland; andcDepartment of
Biochemistry and Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, Dallas, TX 75390
Edited by R. John Collier, Harvard Medical School, Boston, MA, and approved March 2, 2009 (received for review September 12, 2008)
Enterohemorrhagic Escherichia coli O157:H7 translocates 2 effec-
tors to trigger localized actin assembly in mammalian cells, result-
intimin receptor (Tir), is localized in the plasma membrane and
clustered upon binding the bacterial outer membrane protein
intimin. The second, the proline-rich effector EspFU (aka TccP)
activates the actin nucleation-promoting factor WASP/N-WASP,
and is recruited to sites of bacterial attachment by a mechanism
dependent on an Asn-Pro-Tyr (NPY458) sequence in the Tir C-
terminal cytoplasmic domain. Tir, EspFU, and N-WASP form a
complex, but neither EspFUnor N-WASP bind Tir directly, suggest-
ing involvement of another protein in complex formation. Screen-
ing of the mammalian SH3 proteome for the ability to bind EspFU
identified the SH3 domain of insulin receptor tyrosine kinase
substrate (IRTKS), a factor known to regulate the cytoskeleton.
Derivatives of WASP, EspFU, and the IRTKS SH3 domain were
capable of forming a ternary complex in vitro, and replacement of
protein competent for actin assembly in vivo. A second domain of
IRTKS, the IRSp53/MIM homology domain (IMD), bound to Tir in a
manner dependent on the C-terminal NPY458 sequence, thereby
of either the IRTKS SH3 domain or the IMD, or genetic depletion of
translocates 2 effectors that bind to distinct domains of a common
host factor to promote the formation of a complex that triggers
robust actin assembly at the plasma membrane.
enterohemorrhagic Escherichia coli ? IRSp53/MIM homology domain ?
IRTKS ? N-WASP ? SH3 domain
diarrheal and systemic disease (1). Along with the closely related
pathogen, enteropathogenic E. coli (EPEC), it is a member of
the attaching and effacing (AE) family of Gram-negative enteric
pathogens, so named because they generate striking histopatho-
logical lesions on intestinal epithelia, characterized by a loss of
microvilli, intimate attachment of the bacteria to the host cell,
and the formation of filamentous (F)-actin-rich pedestal struc-
tures beneath the host cell membrane at sites of bacterial
attachment (1). The ability to form AE lesions correlates with
the ability to colonize the intestine and cause disease in animal
models (2, 3). In addition, the ability to stimulate the localized
assembly of F-actin in the host cell has been a model for
understanding the control and modification of the mammalian
Actin pedestal formation by EHEC and EPEC depends on the
delivery of bacterial effector proteins into host cells via a type III
secretion system (4, 5). One effector required for pedestal forma-
tion is the translocated intimin receptor (Tir) (6, 7). After trans-
nterohemorrhagic Escherichia coli (EHEC) O157:H7 is a
food-borne pathogen that is an important agent of both
host cell plasma membrane with N- and C-terminal intracellular
domains and a central extracellular domain that binds to the
bacterial outer membrane protein intimin. Clustering of Tir in the
host cell membrane upon intimin binding initiates a signaling
cascade, ultimately leading to actin pedestal formation.
For the canonical EPEC strain, serotype O127:H6, Tir is the
only translocated effector required for pedestal formation, and
after becoming phosphorylated on tyrosine residue 474 (Y474)
by mammalian kinases, recruits the SH2 domain-containing
mammalian adapter protein Nck (8, 9). Nck promotes recruit-
ment of the neuronal Wiskott-Aldrich syndrome protein (N-
WASP), which in turn activates actin assembly by stimulating the
actin nucleating complex Arp2/3 (10).
In contrast, EHEC O157:H7 Tir generates pedestals indepen-
dent of Nck (11). The C-terminal cytoplasmic domain of EHEC
Tir harbors an Asn-Pro-Tyr458(NPY458) sequence that is essen-
tial for actin signaling (12–14). In addition, EHEC translocates
into host cells a second effector, EspFU(aka TccP) that acts in
concert with Tir to promote pedestal formation (15, 16). An
EHEC?espFUmutant generates pedestals at approximately one
tenth the efficiency of WT on cultured monolayers (15) and is
impaired at the expansion of an initial infectious niche during
infection of infant rabbits (17). EspFUcontains multiple 47-aa
proline-rich repeats, and a 20-residue sequence of the repeat is
capable of binding and activating WASP/N-WASP (15, 18–20).
EspFUis recruited to sites of bacterial attachment in a manner
dependent on the Tir NPY458sequence (13), and Tir and EspFU
form a co-immunoprecipitable complex with N-WASP in in-
fected cells (15).
Although N-WASP and EspFU are in complex with Tir,
neither protein appears to directly bind this protein (15, 16),
suggesting that another factor (or factors) binds Tir and pro-
motes complex formation. No other bacterial effectors besides
Tir and EspFUare required for pedestal formation (21), so this
pedestal formation occurs, albeit at low levels, in the absence of
EspFU, the putative host factor may itself stimulate actin assem-
bly. In the current study, we report that the insulin receptor
tyrosine kinase substrate (IRTKS), a homologue of insulin
receptor substrate protein of 53 kDa (IRSp53) and thus a
Author contributions: D.V., A.K., B.S., H.-C.C., M.K.R., K.S., and J.M.L. designed research;
D.V., A.K., B.S., H.-C.C., L.M., and D.R. performed research; M.K.R., K.S., and J.M.L. contrib-
uted new reagents/analytic tools; D.V., A.K., B.S., H.-C.C., M.K.R., K.S., and J.M.L. analyzed
data; and D.V. and J.M.L. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
See Commentary on page 6431.
1To whom correspondence should be addressed. E-mail: email@example.com.
This article contains supporting information online at www.pnas.org/cgi/content/full/
April 21, 2009 ?
vol. 106 ?
member of a protein family that is capable of transducing actin
assembly signals in mammalian cells, is targeted by both Tir and
EspFUand is thus essential to the formation of a potent actin
assembly complex during EHEC pedestal formation.
Formation When Artificially Clustered as Tir Fusion Proteins. The
C-terminal 47-residue repeats of EspFU each contain an
amphipathic helix that interacts with WASP/N-WASP (18, 20),
as well as a region that harbors up to 3 copies of the sequence
PxxP, a motif associated with recognition by SH3 domain-
containing proteins (22). To identify possible SH3 domain-
containing host proteins that could link EspFU and Tir, we
screened an essentially complete collection of human SH3
domains expressed on phage surface (23) for the ability to bind
to GST-EspFUC, a GST fusion protein containing 6 C-
terminal proline-rich repeats of EspFU[supporting informa-
tion (SI) Fig. S1]. Affinity panning of the phage display library
revealed that GST-EspFUC bound avidly to SH3 clones, as
indicated by more than 100-fold higher recruitment of phages
than observed with a GST protein that was used as a negative
control (not shown). Sequencing of resultant phagemids re-
vealed that the SH3 domains of IRTKS (24) or its close
homologue IRSp53 were the only clones consistently enriched,
and constituted 70% of the selected phages isolated from these
enrichments. IRSp53, via its SH3 domain, interacts with
known regulators of actin assembly, such as Scar2/WAVE2 and
N-WASP (25). In addition to the SH3, IRSp53 and IRTKS
contain an N-terminal IRSp53/MIM-homology domain (IMD)
that may bundle actin, bind and deform membranes, and
interact with small G proteins (26).
To better define the region of EspFUrecognized by the SH3
domains of IRSp53 and IRTKS, derivatives of EspFUC were
tested in yeast 2-hybrid assays for their ability to interact with
the IRSp53 or IRTKS SH3 domains. A single 47-residue repeat
(‘‘R47,’’ Fig. S2) of EspFUwas capable of SH3 binding because
co-expression of SH3IRTKSor SH3IRSp53fusions with an R47
fusion activated the ?–galactosidase reporter between 35- and
120-fold (Fig. S2). This signal was specific to SH3 domains, as
neither IMDIRTKS nor IMDIRSp53 interacted with R47, and
required the proline-rich sequence of an EspFU repeat, be-
cause R33, a 33-residue fragment of EspFUthat lacks most of
the proline-rich sequence, did not bind either SH3 domains.
To determine if the interaction of IRSp53 and IRTKS with
EspFU detected in vitro is reflected by recruitment to actin
pedestals, we examined the distribution of IRSp53 and IRTKS
in infected cells by immunofluorescence microscopy. Upon
EHEC infection of HeLa cells, both IRSp53 and IRTKS were
recruited to the tip of phalloidin-stained actin pedestals (Fig.
1A), similar to the localization of EspFU(15, 16).
As pedestal formation involves direct interaction of EspFU
with the GTPase binding domain (GBD) of WASP/N-WASP
(15, 16, 18–21), we tested whether binding of the IRTKS SH3
domain to EspFUwas compatible with simultaneous binding to
GBDWASP. GBDWASP, fluorescently labeled with FITC, was
added to EspFU-5R, a 5-repeat derivative of EspFU(26), or to
both EspFU-5R and GST-SH3IRTKS(Fig. S1), all at equivalent
molar concentrations (taking into account the 5 repeats of
EspFU-5R). The relative size of GBDWASP-containing com-
plexes, detected by absorbance at 494 nm, was determined by
gel filtration chromatography. As expected, GBDWASPbound
to EspFU-5R, as indicated by an increase in the apparent size
(i.e., earlier elution) of FITC-GBDWASP(Fig. 2, blue vs. purple
traces). The addition of GST-SH3IRTKScaused a further shift
of the GBD to a more rapidly eluting peak (Fig. 2, green trace).
The GBD did not shift upon addition of GST-SH3IRTKSalone
(Fig. 2, orange trace). Thus, the earliest eluting peak repre-
sents a ternary complex of GBD, SH3IRTKS, and EspFU-5R
(confirmed by SDS/PAGE; not shown). These data indicate
that SH3IRTKSand the WASP GBD domain can simultaneously
The localization of IRSp53 and IRTKS at the tips of
pedestals and the ability of the IRTKS SH3 domain to bind an
EspFU-5R/GBD complex in vitro raised the possibility that
IRSp53 and/or IRTKS might promote pedestal formation by
recruiting EspFU/N-WASP. To test whether the requirement
for the C-terminal domain of Tir, which is normally essential
for EspFU recruitment and pedestal formation, can be by-
passed by direct fusion of Tir to the IRSp53 and IRTKS SH3
domains, we replaced the Tir C terminus with SH3IRTKS or
IRTKS F-actinEHECΔ ΔespFU
were infected with EHEC?dam, which generates actin pedestals more effi-
ciently on cultured mammalian cells than does WT EHEC (thereby facilitating
evaluation of recruitment; ref. 38), and examined after staining with anti-
IRSp53 or anti-IRTKS antibody (green), DAPI to localize attached bacteria
(blue), and Alexa568-phalloidin (red). (B) HeLa cells were infected with
EHEC?dam?espFUand examined after staining as in A.
IRTKS and IRSp53 are recruited to actin pedestals but only IRTKS
Absorbance 494 (mAU
Elution volume (ml)
form a tripartite complex in vitro. Interactions between GST-SH3IRTKS(50 ?M)
and EspFU-5R (10 ?M) in complex with FITC-labeled GBD (50 ?M) were exam-
ined by gel filtration chromatography. The A494profile, which detects FITC-
GBD, is shown.
The IRTKS SH3 domain, EspFUproline-rich domain, and WASP GBD
Vingadassalom et al.PNAS ?
April 21, 2009 ?
vol. 106 ?
no. 16 ?
SH3IRSp53, and infected HeLa cells ectopically expressing these
fusions with KC14/pEspFU, an EPEC strain engineered to
translocate EspFUbut that does not normally generate ped-
estals because it lacks Tir (8). In fact, infection of transfected
HeLa cells expressing either Tir?C-SH3IRTKS or Tir?C-
SH3IRSp53resulted in the formation of phalloidin-stained actin
pedestals beneath bound bacteria, and in a manner dependent
on EspFU (Fig. S3). Thus, the C terminus of Tir can be
functionally replaced by the IRSp53 or IRTKS SH3 domains,
indicating that the interactions of these domains with EspFU
are sufficient to trigger EspFU-mediated pedestal formation in
IRTKS Binds to Tir and Localizes at Sites of Bacterial Attachment
Independently of EspFU. Given ability of IRSp53 and IRTKS to
bind EspFU, their localization at the pedestal tip could simply
reflect the interaction of these proteins with EspFU. To test
this hypothesis, we assayed recruitment of IRSp53 and IRTKS
upon infection of HeLa cells with an espFUmutant of EHEC.
As expected given the absence of EspFU, actin pedestals were
not readily observed under adherent bacteria (Fig. 1B).
IRSp53 was not associated with bound bacteria, indicating that
this protein requires EspFUfor localization to these sites. In
contrast, IRTKS was readily recruited to sites of bacterial
attachment in the absence of EspFU(Fig. 1B). Thus, whereas
localization of IRSp53 at the pedestal tip is likely secondary to
binding to EspFU, IRTKS might be actively involved in re-
cruiting EspFUto these sites.
The EspFU-independent localization of IRTKS at the sites
of bacterial attachment raised the possibility that IRTKS could
bind to the Tir C-terminal cytoplasmic domain. To determine
whether IRTKS or IRSp53 interacts with TirC, we used the
yeast 2-hybrid assay and analyzed the IMD and SH3 domains
separately. Neither the SH3 nor the IMD of IRSp53 bound to
TirC (Fig. S4), an observation consistent with the lack of
recruitment of IRSp53 to sites of bacterial attachment in the
absence of EspFU. In contrast, co-expression of IMDIRTKSand
TirC derivatives indicated an interaction, resulting in an 8-fold
induction of ?–galactosidase reporter activity (Fig. S4). These
data suggest that the IMD of IRTKS mediates its recruitment
to sites of bacterial attachment by binding to the C-terminal
cytoplasmic domain of translocated Tir.
The EHEC Tir NPY458Sequence Is Required for Binding of IRTKS to Tir
and Its Recruitment to Sites of Bacterial Attachment. The Tir
tripeptide NPY458within the C-terminal cytoplasmic domain
of EHEC Tir is critical for Tir function and alanine substitu-
tion of any of these residues resulted in severe defects in both
EspFU recruitment and pedestal formation (13). To test
whether the ability of IRTKS to bind Tir requires the NPY458
sequence, we assessed IRTKS-Tir interaction in GST pull-
down assays using purified derivatives of these proteins (Fig.
S1). GST-IMDIRTKS bound to WT Tir (‘‘Tir NPY’’) in this
assay, but was not capable of binding to a mutant Tir harboring
substitutions of the NPY458 motif to alanine residues (‘‘Tir
AAA’’; Fig. 3A Upper). Full-length GST-IRTKS also inter-
acted with WT Tir, and the efficiency of binding was signifi-
cantly diminished by mutation of the NPY458sequence (Fig. 3A
Single alanine substitutions of the Tir NPY458 sequence
abrogate both EspFUrecruitment and pedestal formation (13).
To determine if these mutants are also incapable of recruiting
IRTKS, HeLa cells that ectopically express GFP-IRTKS were
infected with KC12?tir/pTirEHEC, an EPEC strain engineered
to express EHEC Tir (8), or isogenic strains that express
alanine-substituted Tir NPY458 mutants. When HeLa cells
ectopically expressing GFP-IRTKS (Fig. S5) were infected
with KC12?tir expressing WT Tir, IRTKS was recruited to
sites of bacterial attachment (Fig. 3B), consistent with our
previous finding (Fig. 2). As expected because of the lack of
EspFU, no actin pedestals were formed. In contrast, no
recruitment of GFP-IRTKS to sites of bacterial attachment
was detected when the transfected HeLa cells were infected
with bacteria expressing Tir derivatives that carry alanine
substitutions in N456, P457, or Y458 (Fig. 3B). Thus, IRTKS
directly binds to Tir via the IMD and is recruited to sites of
bacterial attachment in an NPY458-dependent manner.
Ectopic Expression of the IRTKS SH3 or IMD Domain Inhibits EspFU-
Dependent Pedestal Formation. To examine the functional role of
IRTKS in actin signaling by EHEC, we assessed pedestal
formation after ectopic expression of its IMD or SH3 domain
in HeLa cells. HeLa cells were transfected with plasmids
producing a variety of GFP derivatives, including GFP-
SH3IRTKS and GFP-IMDIRTKS. Immunoblotting confirmed
that all GFP derivatives were efficiently produced (Fig. S5).
Expression of the GFP control had no effect on pedestal
formation, as virtually all transfected cells displayed actin
pedestals (Fig. 4 Top). In contrast, expression of GFP-
SH3IRTKSstrongly inhibited pedestal formation: only 15% of
cells expressing high levels of GFP-SH3IRTKSexhibited ped-
estals (Fig. 4, row 3). Consistent with the hypothesis that this
inhibition was caused specifically by the ability of GFP-
SH3IRTKSto bind EspFU, expression of a GFP fusion contain-
L S PD L S PD
OSP O S
incubated with TirNPY458or TirAAA458and pulled down using glutathione magnetic beads. Proteins present in the original incubation (O), the supernatant (S)
or the pull-down (P) were visualized by Coomassie staining after 10% SDS/PAGE. (B) HeLa cells transfected with GFP-IRTKS were infected with KC12?tir (8)
harboring plasmids encoding EHEC Tir carrying the WT NPY458sequence (Tir NPY) or alanine substitutions of this sequence, as indicated (Left). Monolayers were
examined after staining with DAPI to localize bacteria (blue), Alexa568-phalloidin (red), and anti-myc antibody to detect GFP-IRTKS-myc (green).
The EHEC Tir NPY458sequence is required for binding of IRTKS to Tir and its recruitment to sites of bacterial attachment. (A) GST-IRTKS derivatives were
www.pnas.org?cgi?doi?10.1073?pnas.0809131106Vingadassalom et al.
ing the SH3 domain of IRSp53, which also binds EspFU,
blocked pedestal formation (Fig. 4, row 2), whereas expression
of a fusion containing an SH3 domain of Nck, which was not
enriched from the SH3 phage display library by affinity
panning on EspFU, had no discernible effect (Fig. 4, row 4).
Importantly, inhibition by GFP-SH3IRTKSand GFP-SH3IRSp53
was specific to EspFU-mediated pedestals and not caused by
non-specific inhibition of translocation or the actin assembly
machinery, because expression of these fusions did not inhibit
pedestal formation by EPEC (Fig. S6), which generates ped-
estals independent of EspFU(8, 9, 15, 16).
As shown in Fig. 4, expression of GFP-IMDIRTKS also
efficiently inhibited actin pedestal formation, because only
12.6% of cells that expressed GFP-IMDIRTKS and bound
bacteria demonstrated pedestals (Fig. 4, row 6). This inhibition
was specific because expression of GFP-IMDIRTKS had no
effect on actin pedestal formation by EPEC (Fig. S6). In
contrast to the strong inhibitory activity of GFP-IMDIRTKS, the
pedestal index for cells expressing GFP-IMDIRSp53was 82.6%
(Fig. 4, row 5), which, although somewhat lower than for cells
expressing GFP alone (i.e., 95.6%), is consistent with our
inability to discern recruitment of IRSp53 to sites of bacterial
attachment in the absence of EspFU.
Genetic Depletion of IRTKS Inhibits EspFU-Dependent Pedestal For-
mation. To further examine whether IRTKS function is required
for EHEC actin assembly, we used an RNAi approach based on
previously published siRNA sequences that efficiently and spe-
cifically silence expression of IRTKS or IRSp53 (27). RT-PCR
analysis of cells transfected with a combination of 2 IRTKS
siRNAs showed an approximately 90% depletion of IRTKS
mRNA compared with control siRNA (Fig. S7A). Similarly, a
combination of 2 IRSp53 siRNAs knocked down IRSp53 mRNA
more than 90% (Fig. S7A).
To assess the role of IRTKS in pedestal formation, IRTKS-
depleted and control cells were infected with KC12/pEspFU.
As expected, pedestals formed with high efficiency on control
siRNA-treated cells—visual quantitation revealed that virtu-
ally all infected cells displayed pedestals. In contrast, pedestal
formation was diminished more than 5-fold on cells treated
with a combination of the 2 IRTKS siRNAs (Fig. 5 and Fig.
S7B). IRTKS depletion with only 1 siRNA resulted in partial
(?50%) but significant inhibition (Fig. S7B). The decrease in
pedestal formation was specific for IRTKS, as cells depleted
for IRSp53 generated pedestals with undiminished efficiency
(Fig. 5 and Fig. S7B), and the cells depleted for both IRTKS
and IRSp53 generated pedestals at a frequency indistinguish-
able from cells depleted only for IRTKS (Fig. 5 and Fig. S7B).
Importantly, EPEC formed pedestals with high efficiency on
IRTKS-depleted cells (Fig. S7C). Thus, in agreement with the
data on ectopic expression of IRTKS SH3 or IMD domains,
these RNAi studies indicate that IRTKS is specifically re-
quired for EspFU-mediated actin assembly.
EspFU binds and activates WASP-family actin nucleation-
promoting factors (15, 16) and artificial fusion of EspFUto Tir
clustered at the plasma membrane is sufficient to trigger actin
assembly (18, 20, 21). However, although Tir, EspFU, and
N-WASP are associated in host cells, neither EspFU nor
N-WASP directly interact with Tir (15, 16), indicating that a
(host-encoded) factor is required for formation of this actin
assembly complex. Because EspFU contains multiple PxxP
sequences, we screened an essentially complete collection of
human SH3 domains (23) and identified IRTKS as an avid
binding partner of EspFU. Detection of a ternary complex of
SH3IRTKS, EspFU, and GBDWASPin vitro supports the model
that IRTKS is part of an EspFU/N-WASP-containing complex
that potently stimulates Arp2/3. Consistent with this hypoth-
esis, a Tir-SH3IRTKSfusion lacking the Tir C terminus, which
is normally required for function, generated robust EspFU-
mediated pedestals upon clustering by intimin.
IRTKS was localized to the tip of actin pedestals, raising the
possibility that it mediated Tir-EspFU interaction. Indeed,
whereas the SH3 domain of IRTKS bound to EspFU, its IMD
15.0 +/- 3.0
82.6 +/- 3.2
12.6 +/- 2.4
dependent pedestal formation. Transfected HeLa cells expressing GFP or GFP
fusion proteins were infected with KC12/pEspFU(15). Transfected cells were
identified by GFP fluorescence (Merge), and monolayers were stained with
DAPI (blue) and Alexa568-phalloidin (red). (The lack of localization of GFP-
of Tir foci.) The percentage of cells competent for actin pedestal formation
after infection is shown (Right). Shown is the mean ? SD of at least 3
experiments; *P ? 0.0001;?P ? 0.01.
Ectopic expression of the IRTKS SH3 or IMD domain inhibits EspFU-
% of cells
99.0 +/- 1.0
96.0 +/- 3.0
17.0 / 2.0
IRSp53 siRNA1 + 2
IRTKS siRNA1 + 2
15.0 +/- 2.0*
mation. HeLa cells transfected with pairs of control, IRTKS, or IRSp53 siRNAs,
or a pool of the pair of IRTKS and IRSp53 siRNAs, were infected with KC12/
pEspFU(15). Monolayers were examined after staining with DAPI (blue) and
Alexa568-phalloidin (red). The percentage of cells competent for actin ped-
estal formation after infection is shown (Right). Shown is the mean ? SD of at
least 3 experiments; *P ? 0.0001.
Genetic depletion of IRTKS inhibits EspFU-dependent pedestal for-
Vingadassalom et al.PNAS ?
April 21, 2009 ?
vol. 106 ?
no. 16 ?
bound to Tir in a manner dependent on the NPY458sequence,
which has previously been shown to be critical for EspFU
recruitment. Furthermore, IRTKS was recruited to sites of
bacterial attachment, dependent on Tir NPY458but indepen-
dent of EspFUand actin assembly. Finally, ectopic expression
of either the IMD or SH3 of IRTKS, or RNAi silencing of
IRTKS, inhibited EspFU-dependent pedestal formation with-
out affecting EspFU-independent pedestal formation by
EPEC. These results provide compelling evidence that IRTKS,
by interacting with Tir and EspFU, promotes the formation of
a complex of bacterial and host factors that trigger robust actin
assembly beneath bound bacteria. The SH3 domain of the
IRTKS homologue IRSp53 appears to be functionally equiv-
alent to that of IRTKS because it promotes EspFU-mediated
actin assembly when clustered at the plasma membrane.
However, in contrast to IRTKS, IRSp53 was not detectably
recruited to sites of bacterial attachment in the absence of
EspFU, and ectopic expression of the IRSp53 IMD, or siRNA-
mediated depletion of IRSp53, had no marked effect on
pedestal formation, correlating with our inability to detect
interaction of the IRSp53 IMD with the Tir C terminus in a
yeast 2-hybrid assay. It should be noted, however, that we have
been able to detect binding of recombinant IRSp53 to recom-
binant Tir in vitro (D.V., unpublished data), and this
activity might be reflected in the mild (?15%) inhibition of
pedestal formation by ectopic expression of GFP-IMDIRSp53
(see Fig. 4). In addition, Stradal and coworkers have impli-
cated IRSp53 in pedestal formation using murine cell lines
(39), and the relative roles of members of this family during
pedestal formation in different cell types remains to be fully
IRTKS, as a member of the IRSp53 family, is involved in signal
transduction pathways that link deformation of the plasma mem-
brane and remodeling of the actin cytoskeleton (26). IRTKS
promotes actin assembly and membrane protrusions when overex-
pressed in mammalian cells (24), so it is possible that its role in
pedestal formation may extend beyond simply recruiting EspFUto
sites of clustered Tir at the plasma membrane. In fact, an
EHEC?espFU mutant retains the ability to generate low-level
Tir-mediated localized actin assembly in vitro (15) and to trigger
some AE lesions during infection of the mammalian host (17). The
IRSp53 C-terminal SH3 domain has been shown to interact with
the formin mDia, as well as the Arp2/3 activators WAVE2 (see ref.
26 for review) and N-WASP (25). The IMD binds F-actin, the
of a Bin-amphiphysin-Rvs167 (BAR) domain, which binds and
deforms membranes, generating invaginations during endocytosis.
IMDs, also known as I-BAR (inverse BAR) domains because of
their opposite curvature, triggers protrusive membrane deforma-
tion (26, 28), and it is tempting to speculate that this activity of the
IRTKS IMD might contribute to the morphology of AE lesions.
Recent work has shown that, within a 47-residue C-terminal
EspFUrepeat, a segment consisting of approximately 20 residues
forms an amphipathic helix and an extended arm that binds and
activates WASP/N-WASP (18, 20). We show here that a different
segment of the repeat, one that is rich in prolines, is required for
binding to the IRTKS/IRSp53 SH3 domains, and that IRTKS and
N-WASP can bind EspFUsimultaneously. The division of a repeat
unit into 2 functional recognition elements parallels that of the
EspFU-related E. coli effector EspF. Like EspFU, EspF consists of
an N-terminal translocation domain and several 47-residue C-
terminal repeats, each of which contains an N-WASP binding
segment and a proline-rich sequence that is recognized by an SH3
domain-containing protein that binds and deforms membranes. In
the case of EspF, the SH3-containing protein is SNX9 (29, 30),
which contains a BAR domain and participates in membrane
remodeling during endocytosis (31). Although EspFUcan comple-
ment some functions of EspF (32), EspF plays no apparent role in
pedestal formation (15), presumably because its proline-rich se-
quences target a different SH3 domain. This difference notwith-
by acting as modular and repetitive adaptor proteins that link
N-WASP to a membrane-deforming protein.
With the identification of IRTKS as an essential link be-
tween Tir and EspFU, a striking feature of many components
of the actin pedestal signaling cascade is the ability to mul-
timerize. The membrane anchoring domain of intimin and the
extracellular domain of Tir each encode elements that pro-
mote homotypic dimerization (33, 34), leading to the hypoth-
esis that intimin-Tir interactions result in a reticular array-like
superstructure of Tir cytoplasmic domains beneath the clus-
tered receptor. This putative array of Tir cytoplasmic domains
is recognized by the IRTKS IMD, which, upon dimerization,
would be predicted to present physically linked pairs of IRTKS
SH3 domains to recruit EspFU. In this regard, it is notable that
the presence of at least 2 EspFU repeats is required for
recruitment to sites of bacterial attachment (19). Finally, the
repetitive nature of EspFU is also critical for downstream
signaling, because the tandem N-WASP-binding elements
synergistically activate the N-WASP/Arp2/3 pathway for actin
assembly (20, 21, 35). Thus, by targeting distinct domains of
IRTKS, Tir and EspFUpromote the formation of a multimeric
complex containing N-WASP-binding and activation elements
that triggers the robust actin assembly.
Materials and Methods
Strains, Plasmids, and DNA Manipulations. The bacterial strains and plasmids
As detailed in SI Methods, cDNA encoding IRTKS and IRSp53 derivatives were
amplified from the human cDNAs and cloned in the mammalian expression
plasmids pKC425 (21) and pKC689 (21) to generate GFP-fusion proteins and
Tir?C fusion proteins, respectively.
Assays for Protein-Protein Interaction. GST-EspFUC, His-tagged EHEC Tir deriv-
atives, and GST-tagged IRTKS derivatives were produced in E. coli strain
BL21(DE3) and purified by affinity chromatography according to manufac-
turers’ instructions. Screening of the phage-displayed SH3 domain library was
performed as described previously (23). Yeast 2-hybrid assays were used to
assess interaction among IRTKS, IRSp53, EspFU, and EHEC Tir as previously
described (36). Interaction between IRTKS and recombinant EHEC Tir was
assessed in GST pull-down assays. Formation of a tripartite complex between
GST-SH3IRTKS, EspFU-5R, and GBDWASPwas assessed by gel filtration chroma-
RNAi Experiments. siRNA experiments were performed using stealth RNAi
(Invitrogen). The sequences used were as described by Suetsugu et al. (26)
(see SI Methods). Transfections were performed using Lipofectamine 2000
(Invitrogen) according to the manufacturer’s instructions. To evaluate
knock-down efficiency, total mRNA from RNAi-treated HeLa cells was
isolated using TRIzol reagent (Invitrogen). A first-strand cDNA was synthe-
sized from the mRNA using the SuperScript first-strand cDNA synthesis
system for RT-PCR (Invitrogen).
described (14, 15). Transfection of mammalian cells for ectopic expression of
proteins, and infection of mammalian cells with bacteria were performed as
previously described (15, 37). Cells were treated with mouse anti-HA tag mAb
HA.11 (1:500; Covance), mouse anti-IRSp53 mAb (1:100; Novus Biologicals), or
mouse anti-IRTKS mAb (1:100; Novus Biologicals). Pedestal formation indices
were determined as detailed in SI Methods.
ACKNOWLEDGMENTS. We thank Dr. Nathalie Cohet for help with quantita-
tive PCR analyses, L. Soll for help with plasmid construction, T. Stradal and K.
K. Campellone and D. Tipper for critical reading of the manuscript. This work
was supported by National Institutes of Health Grant R01-AI46454 (to J.M.L.)
and Fondation pour la Recherche Medicale (Paris, France) post-doctoral fel-
lowship SPE20061208629 (to D.V.).
www.pnas.org?cgi?doi?10.1073?pnas.0809131106Vingadassalom et al.
1. Kaper JB, Nataro JP, Mobley HL (2004) Pathogenic Escherichia coli. Nat Rev Microbiol Download full-text
Escherichia coli infection. J Clin Invest 92:1412–1417.
infant rabbits Infect Immun 71:7129–7139.
4. Frankel G, Phillips AD (2008) Attaching effacing Escherichia coli and paradigms of Tir-
triggered actin polymerization: getting off the pedestal. Cell Microbiol 10:549–556.
5. Hayward RD, Leong JM, Koronakis V, Campellone KG (2006) Exploiting pathogenic
Escherichia coli to model transmembrane receptor signalling. Nat Rev Microbiol
6. Deibel C, Kramer S, Chakraborty T, Ebel F (1998) EspE, a novel secreted protein of
attaching and effacing bacteria, is directly translocated into infected host cells, where
it appears as a tyrosine-phosphorylated 90 kDa protein. Mol Microbiol 28:463–474.
adherence into mammalian cells. Cell 91:511–520.
8. Campellone KG, Giese A, Tipper DJ, Leong JM (2002) A tyrosine-phosphorylated
12-amino-acid sequence of enteropathogenic Escherichia coli Tir binds the host adap-
tor protein Nck and is required for Nck localization to actin pedestals. Mol Microbiol
9. Gruenheid S, et al. (2001) Enteropathogenic E. coli Tir binds Nck to initiate actin
pedestal formation in host cells. Nat Cell Biol 3:856–859.
10. Rohatgi R, et al. (2001) Nck and phosphatidylinositol 4,5-bisphosphate synergistically
activate actin polymerization through the N-WASP-Arp2/3 pathway. J Biol Chem
11. Campellone KG, Leong JM (2003) Tails of two Tirs: actin pedestal formation by
enteropathogenic E. coli and enterohemorrhagic E. coli O157:H7. Curr Opin Microbiol
12. Allen-Vercoe E, Waddell B, Toh MC, DeVinney R (2006) Amino acid residues within
enterohemorrhagic Escherichia coli O157:H7 Tir involved in phosphorylation, alpha-
13. Brady MJ, Campellone KG, Ghildiyal M, Leong JM (2007) Enterohaemorrhagic and
enteropathogenic Escherichia coli Tir proteins trigger a common Nck-independent
actin assembly pathway. Cell Microbiol 9:2242–2253.
14. Campellone KG, et al. (2006) Enterohaemorrhagic Escherichia coli Tir requires a C-
terminal 12-residue peptide to initiate EspFU-mediated actin assembly and harbours
N-terminal sequences that influence pedestal length. Cell Microbiol 8:1488–1503.
III effector protein that couples Tir to the actin-cytoskeleton. Cell Microbiol 6:1167–
epithelial association and late-stage intestinal colonization by E. coli O157:H7. Cell
18. Cheng HC, et al. (2008) Structural mechanism of WASP activation by the enterohae-
morrhagic E. coli effector EspF(U). Nature 454:1009–1013.
19. Garmendia J, et al. (2006) Characterization of TccP-mediated N-WASP activation
during enterohaemorrhagic Escherichia coli infection. Cell Microbiol 8:1444–1455.
20. Sallee NA, et al. (2008) The pathogen protein EspF(U) hijacks actin polymerization
using mimicry and multivalency. Nature 454:1005–1008.
21. Campellone KG, et al. (2008) Repetitive N-WASP-binding elements of the enterohe-
22. Mayer BJ, Saksela K (2004) SH3 domains. Structure and Function of Modular Protein
Domains, ed Cesareni G, Gimona M, Sudol M, Yaffe M (Wiley-VCH, Weinheim, Ger-
many), pp 37–58.
23. Karkkainen S, et al. (2006) Identification of preferred protein interactions by phage-
display of the human Src homology-3 proteome. EMBO Rep 7:186–191.
24. Millard TH, Dawson J, Machesky LM (2007) Characterisation of IRTKS, a novel IRSp53/
MIM family actin regulator with distinct filament bundling properties. J Cell Sci
25. Lim KB, et al. (2008) The Cdc42 effector IRSp53 generates filopodia by coupling
membrane protrusion with actin dynamics. J Biol Chem 283:20454–20472.
26. Scita G, Confalonieri S, Lappalainen P, Suetsugu S (2008) IRSp53: crossing the road of
membrane and actin dynamics in the formation of membrane protrusions. Trends Cell
27. Suetsugu S, et al. (2006) Optimization of WAVE2 complex-induced actin polymeriza-
tion by membrane-bound IRSp53, PIP(3), and Rac. J Cell Biol 173:571–585.
28. Cory GO, Cullen PJ (2007) Membrane curvature: the power of bananas, zeppelins and
boomerangs. Curr Biol 17:R455–457.
29. Alto NM, et al. (2007) The type III effector EspF coordinates membrane trafficking by
the spatiotemporal activation of two eukaryotic signaling pathways. J Cell Biol
30. Marches O, et al. (2006) EspF of enteropathogenic Escherichia coli binds sorting nexin
9. J Bacteriol 188:3110–3115.
31. Lundmark R, Carlsson SR (2003) Sorting nexin 9 participates in clathrin-mediated
endocytosis through interactions with the core components. J Biol Chem 278:46772–
32. Viswanathan VK, et al. (2004) Comparative analysis of EspF from enteropathogenic
and enterohemorrhagic Escherichia coli in alteration of epithelial barrier function.
Infect Immun 72:3218–3227.
33. Luo Y, et al. (2000) Crystal structure of enteropathogenic Escherichia coli intimin-
receptor complex. Nature 405:1073–1077.
34. Touze T, et al. (2004) Self-association of EPEC intimin mediated by the beta-barrel-
containing anchor domain: a role in clustering of the Tir receptor. Mol Microbiol
35. Padrick SB, et al. (2008) Hierarchical regulation of WASP/WAVE proteins. Mol Cell
pedestal formation highlight a putative Tir binding pocket. Mol Microbiol 45:1557–
37. Campellone KG, Leong JM (2005) Nck-independent actin assembly is mediated by two
phosphorylated tyrosines within enteropathogenic Escherichia coli Tir. Mol Microbiol
38. Campellone KG, et al. (2007) Increased adherence and actin pedestal formation by dam-
deficient enterohaemorrhagic Escherichia coli O157:H7. Mol Microbiol 63:1468–1481.
39. Weiss SM, et al. (2009) IRSp53 links the enterohemorrhagic E.Coli effectors Tir and
EspFufor actin pedestal formation. Cell Host Microbe 5:244–258.
coli necessary for the production of attaching and effacing lesions on tissue culture
cells. Proc Natl Acad Sci USA 87:7839–7843.
Vingadassalom et al.PNAS ?
April 21, 2009 ?
vol. 106 ?
no. 16 ?