RhoG regulates endothelial apical cup assembly downstream from ICAM1 engagement and is involved in leukocyte trans-endothelial migration.

Abstract

During trans-endothelial migration (TEM), leukocytes use adhesion receptors such as intercellular adhesion molecule-1 (ICAM1) to adhere to the endothelium. In response to this interaction, the endothelium throws up dynamic membrane protrusions, forming a cup that partially surrounds the adherent leukocyte. Little is known about the signaling pathways that regulate cup formation. In this study, we show that RhoG is activated downstream from ICAM1 engagement. This activation requires the intracellular domain of ICAM1. ICAM1 colocalizes with RhoG and binds to the RhoG-specific SH3-containing guanine-nucleotide exchange factor (SGEF). The SH3 domain of SGEF mediates this interaction. Depletion of endothelial RhoG by small interfering RNA does not affect leukocyte adhesion but decreases cup formation and inhibits leukocyte TEM. Silencing SGEF also results in a substantial reduction in RhoG activity, cup formation, and TEM. Together, these results identify a new signaling pathway involving RhoG and its exchange factor SGEF downstream from ICAM1 that is critical for leukocyte TEM.

Full text

Page 1

T H E J O U R N A L O F C E L L B I O L O G Y
JCB: ARTICLE
© The Rockefeller University Press $30.00
The Journal of Cell Biology, Vol. 178, No. 7, September 24, 2007 1279–1293
http://www.jcb.org/cgi/doi/10.1083/jcb.200612053
JCB
1279
Introduction
Leukocyte trans-endothelial migration (TEM) is a key event in
host defense. The passage of leukocytes across the vascular wall
into the underlying tissues can be divided into distinct phases,
including fi rm adhesion of leukocytes to the endothelium and
subsequent diapedesis (Vestweber, 2002; Johnson-Leger and
Imhof, 2003; van Buul and Hordijk, 2004; for review see Muller,
2003). Leukocyte adhesion to the endothelium initiates the for-
mation of dynamic dorsal membrane protrusions, assembling
a cuplike structure, which surrounds adherent leukocytes and con-
tains the cell adhesion molecules intercellular adhesion molecule-1
(ICAM1) and VCAM1 (Barreiro et al., 2002; Carman et al.,
2003; Carman and Springer, 2004). They have been referred to
as docking structures (Barreiro et al., 2002) or trans-migratory
cups (Carman and Springer, 2004). Little is known about the
mechanisms that regulate their assembly, and their role in TEM
remains uncertain.
During TEM, leukocytes adhere to ICAM1 on the endo-
thelial cell surface, and this triggers diverse intracellular signals
(Vestweber, 2002; Kluger, 2004). Engagement of ICAM1 can
be mimicked by cross-linking ICAM1 with ICAM1-specifi c
antibodies (Wojciak-Stothard et al., 1999; Etienne-Manneville
et al., 2000; Thompson et al., 2002) or by beads coated with anti-
bodies against ICAM1 (Tilghman and Hoover, 2002). Actin
dynamics in endothelial cells are important for leukocyte TEM,
which is prevented by inhibiting endothelial actin polymeriza-
tion by cytochalasin D (Adamson et al. 1999; Carman and
Springer, 2004). Cross-linking of ICAM1 stimulates the assem-
bly of actin stress fi bers (Wojciak-Stothard et al., 1999; Van
Buul et al., 2002). In addition, actin polymerization is involved
in assembly of the cups (Carman and Springer, 2004).
Actin membrane dynamics are controlled by small Rho-
like GTPases. These proteins function as molecular switches
and cycle between an inactive GDP-bound state and an active
GTP-bound state. Blocking RhoA activity using Clostridium
botulinum C3 transferase prevents the adhesion or migration
RhoG regulates endothelial apical cup assembly
downstream from ICAM1 engagement and is
involved in leukocyte trans-endothelial migration
Jaap D. van Buul,1,2 Michael J. Allingham,1,2 Thomas Samson,1,2 Julia Meller,3,4 Etienne Boulter,1,2
Rafael García-Mata,1,2 and Keith Burridge1,2
1Department of Cell and Developmental Biology and 2Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC 27599
3Cardiovascular Research Center and 4Department of Microbiology, Mellon Prostate Cancer Institute, University of Virginia, Charlottesville, VA 22908
D
thelium. In response to this interaction, the endothelium
throws up dynamic membrane protrusions, forming a cup
that partially surrounds the adherent leukocyte. Little is
known about the signaling pathways that regulate cup
formation. In this study, we show that RhoG is activated
downstream from ICAM1 engagement. This activation re-
quires the intracellular domain of ICAM1. ICAM1 colocalizes
uring trans-endothelial migration (TEM), leukocytes
use adhesion receptors such as intercellular adhe-
sion molecule-1 (ICAM1) to adhere to the endo-
with RhoG and binds to the RhoG-specifi c SH3-containing
guanine-nucleotide exchange factor (SGEF). The SH3
domain of SGEF mediates this interaction. Depletion of
endothelial RhoG by small interfering RNA does not affect
leukocyte adhesion but decreases cup formation and in-
hibits leukocyte TEM. Silencing SGEF also results in a sub-
stantial reduction in RhoG activity, cup formation, and TEM.
Together, these results identify a new signaling pathway
involving RhoG and its exchange factor SGEF downstream
from ICAM1 that is critical for leukocyte TEM.
Correspondence to Jaap D. van Buul: j.vanbuul@sanquin.nl; or Keith Burridge:
keith_burridge@med.unc.edu
J.D. van Buul’s present address is Dept. of Molecular Cell Biology, Sanquin
Research and Landsteiner Laboratory, Academic Medical Center, University of
Amsterdam, 1012 ZA Amsterdam, Netherlands.
Abbreviations used in this paper: GEF, guanine-nucleotide exchange factor;
HUVEC, human umbilical vein endothelial cell; ICAM1, intercellular adhesion
molecule-1; MHC, major histocompatibility complex; miRNA, micro-RNA; PBD,
p21-activated kinase–binding domain; ROCK, Rho-associated coil-containing
protein kinase; SDF-1, stromal cell–derived factor-1; SGEF, SH3-containing GEF;
TEM, trans-endothelial migration; VE, vascular endothelial; wt, wild type.
The online version of this article contains supplemental material.

Page 2

JCB • VOLUME 178 • NUMBER 7 • 2007 1280
of leukocytes across endothelial cell monolayers (Adamson
et al., 1999; Wojciak-Stothard et al., 1999). However, the role
of RhoA in the assembly of the cups is unclear. Barreiro et al.
(2002) reported that assembly of these structures induced by
VCAM1 is inhibited by Y27632, an inhibitor of Rho- associated
coil-containing protein kinase (ROCK)/Rho kinase, which
is a downstream effector of RhoA. In contrast, Carman and
Springer (2004) found that treatment with Y27632 or C3 was
Figure 1. Endothelial cells protrude ICAM1-expressing
membrane ruffl es around an adhered HL60 cell.
(A) TNF-α–treated endothelial cells were incubated with
HL60 cells for 30 min, processed, and stained for ICAM1
in green and for VE-cadherin in red. Confocal imaging
shows that ICAM1 is recruited to sites of leukocyte
adhesion at the baso-lateral focal plane (a) and as a
ring structure surrounding the leukocyte at the apical
focal plane (b). (B) Z-stack imaging shows ICAM1 staining
in green (a) surrounding a leukocyte, which is stained for
F-actin in red (asterisks; b). Image c shows the merge.
Reconstruction of the Z-stack imaging shows ICAM1
surrounding a leukocyte in a cup-like structure (green, d).
Vertical bar at the right shows the height of the pro-
trusions (6 μm). (C) TNF-α–treated endothelial cells
transiently transfected with GFP-actin were incubated
with HL60 cells for 30 min, processed, and imaged for
GFP-actin in green (a and e), ICAM1 in red (b and f),
merge of GFP-actin and ICAM1 in yellow (c and g),
and F-actin using phalloidin in white to visualize the
adhered leukocyte (d and h). Confocal imaging re-
vealed that actin and ICAM1 are recruited to sites of
leukocyte adhesion, surrounding the leukocyte at the
apical plane (e–h). (D) Scanning EM images show pro-
truding endothelial membrane sheets (arrowheads) around
adhered HL60 cells (asterisks). Bars (A), 10 μm; (B and C)
5 μm; (D) 1 μm.

Page 3

ICAM1-INDUCED RHOG ACTIVATION • VAN BUUL ET AL.1281
unable to prevent cup formation downstream from ICAM1
engagement. The similarity of these apical cups to phago-
cytic cups (Barreiro et al., 2002; Carman et al., 2003) together
with the role of RhoG in the phagocytosis of apoptotic cells
(deBakker et al., 2004) has led us to examine whether RhoG
may contribute to the formation of endothelial cups and par-
ticipate in TEM.
In this study, we demonstrate that RhoG is a critical medi-
ator of leukocyte TEM. RhoG and a guanine-nucleotide ex-
change factor (GEF) for RhoG, SH3-containing GEF (SGEF),
are recruited to sites of ICAM1 engagement, where RhoG be-
comes activated. We fi nd that ICAM1 interacts with SGEF
through its SH3 domain. Finally, reduction of RhoG or SGEF
expression in endothelial cells using siRNA decreases the as-
sembly of the cups as well as the migration of leukocytes across
endothelial cell monolayers.
Results
Endothelial cells form apical cups
around leukocytes
Adhesion of myeloid leukemia HL60 cells to TNF-α–activated
endothelial cells induced not only the recruitment of ICAM1 to
sites of adhesion (Fig. 1 A) but also ICAM1-positive membrane
protrusions that surrounded the adhered leukocyte (Fig. 1 B),
which is consistent with previously reported fi ndings (Barreiro
et al., 2002; Carman et al., 2003). Also, GFP-actin, which is
transiently expressed in endothelial cells, distributed to sites
of leukocyte binding and colocalized with ICAM1 (Fig. 1 C).
Of note, the endothelial cell–cell junctional marker vascular
endothelial (VE) cadherin did not localize to these membrane
protrusions (Fig. 1 A). Three-dimensional projections showed
that ICAM1-positive protrusions arose from the apical plane of
the endothelial cells but did not fully cover the leukocyte (Fig.
1 B). These protrusions resembled cuplike structures that extended
?6–7 μm above the baso-lateral membrane (Fig. 1 B, d). To
determine whether these ICAM1-rich cups formed around cells
that were transmigrating, HL60 cells were plated on endothelial
monolayers growing on transwell fi lters. Confocal analysis of
fi xed and stained preparations revealed rings of ICAM1 stain-
ing at the apical surface (i.e., cups) surrounding cells that were
traversing the monolayer (Fig. S1, available at http://www.jcb
.org/cgi/content/full/jcb.200612053/DC1). Scanning EM con-
fi rmed the presence of endothelial cuplike protrusions surroun-
ding but not fully covering leukocytes 30 min after leukocyte
adhesion (Fig. 1 D).
RhoG and SGEF are enriched in dorsal
membrane ruffl es
The small GTPase RhoG and its specifi c GEF, SGEF, are
known to induce dorsal ruffl es (Ellerbroek et al., 2004).
RhoG and SGEF are endogenously expressed in endothelial
cells as well as in COS7 and HeLa cells (Fig. 2 A). Over-
expression of the constitutively active mutant RhoG-Q61L or
Figure 2. RhoG and SGEF are expressed endogenously in
endothelial cells and are localized to dorsal endothelial mem-
brane protrusions. (A) Western blot analysis of tissue lysates
of mouse brain (a positive control for SGEF), HUVECs, HeLa,
and COS7 cells show the endogenous expression of SGEF
(100 kD; top blot) and RhoG (18 kD; bottom blot). (B) Endo-
thelial cells were transiently transfected with GFP–RhoG-Q61L (a)
or GFP-SGEF (b) and stained for F-actin in red. Images re present
the merge. Arrowheads show membrane ruffl es.

Page 4

JCB • VOLUME 178 • NUMBER 7 • 2007 1282
SGEF in endothelial cells induced ruffl es on the apical sur-
face (Fig. 2 B).
To study the involvement of ICAM1 in the regulation of dor-
sal ruffl es, COS7 cells that lack endogenous ICAM1 were used.
The expression of ICAM1 tagged with GFP or the V5 epitope in
COS7 cells showed distributions similar to ICAM1 in endothe-
lial cells (Fig. 3 A). Interestingly, cotransfection of RhoG-Q61L
or SGEF not only induced dorsal ruffl es but also induced a
redistribu tion of ICAM1 to these ruffl es (Fig. 3, A and B; and
Videos 1 and 2, available at http://www.jcb.org/cgi/content/full/
jcb.200612053/DC1). ICAM1 colocalized with RhoG-Q61L or
SGEF (Fig. 3, A and B). The localization of ICAM1 to ruffl es re-
quired active RhoG because neither wild-type (wt) RhoG nor a
dominant-negative mutant, T17N, colocalized with ICAM1 (un-
published data). As a control, transmembrane protein PECAM-1
was expressed together with RhoG-Q61L or SGEF and showed
no colocalization (unpublished data). These data suggested a role
for RhoG and SGEF in the formation of endothelial apical cup
structures; therefore, we next tested the involvement of RhoG
and SGEF in ICAM1 signaling and cup formation.
Figure 3. SGEF and RhoG-Q61L colocalized
with ICAM1. (A) COS7 cells were transiently
cotransfected with ICAM1-GFP (a and g) or
ICAM1-V5 (d) and with GFP–RhoG-Q61L (e) or
myc-SGEF (h). Image b shows F-actin. Images c,
f, and i represent the merge. ICAM1 colocalizes
with RhoG-Q61L and SGEF. Moreover, RhoG-
Q61L and SGEF induce a change in ICAM1
distribution from spikes (a) to ruffl es (arrowheads;
d and g). (B) COS7 cells were transiently co-
transfected with ICAM1-GFP (a and d), myc-
SGEF (b), or myc-RhoG-Q61L (e). Panels c and
f show merged images. Confocal x-z section
images show colocalization between ICAM1
and SGEF (a–c) and ICAM1 and RhoG-Q61L
(d–f) in dorsal membrane ruffl es (arrowheads).
Bars (A), 20 μm; (B) 10 μm.

Page 5

ICAM1-INDUCED RHOG ACTIVATION • VAN BUUL ET AL.1283
Recruitment of ICAM1-GFP to sites
of adhesion
COS7 cells lacking endogenous ICAM1 were used to express
ICAM1-GFP. Incubation of these COS7 cells with HL60 cells
resulted in the majority of HL60 cells adhering to the ICAM1-
GFP–transfected cells (Fig. 4 A). Three-dimensional projections
showed that ICAM1-positive protrusions surrounded the ad-
hered HL60 cells (Fig. 4 A, d), similar to those observed with
endothelial cells (Fig. 1 B). To specifi cally study ICAM1 en-
gagement and downstream signaling that would mimic leuko-
cyte binding to ICAM1, beads coated with antibodies against
ICAM1 were used as described in Materials and methods
Figure 4. HL60 cells and beads coated with
ICAM1 antibodies recruited ICAM1-GFP. (A)
COS7 cells were transiently transfected with
ICAM1-GFP (a). HL60 cells were allowed to
adhere for 30 min. Samples were fi xed, per-
meabilized, and stained for F-actin (b). Image c
shows the merge of images a and b. Note that
the majority of the HL60 cells adhere to the
ICAM1-GFP–expressing cells. Image d shows an
x-z projection of ICAM1-GFP in green (arrow-
heads) that surrounds adhered HL60 cells
(asterisks) stained with F-actin in red. (B) αICAM1
beads added for 30 min to ICAM1-GFP–
expressing COS7 cells. Beads were stained with
secondary AlexaFluor594 anti–mouse anti-
bodies in red (b). Image c shows merge of
images a and b, with additional F-actin distri-
bution in white. Note that the majority of the
αICAM1 beads adhere to the ICAM1-GFP–
expressing cells. Image d shows x-z projection of
ICAM1-GFP in green (arrowheads) that surrounds
adhered αICAM1 beads (asterisks). (C) Scan-
ning EM image shows endothelial membrane
sheets (arrowheads) that surround an αICAM1
bead (asterisk). Bars, (A and B, panel c), 20 μm;
(A and B, panel d) 5 μm; (C) 1 μm.

Page 6

JCB • VOLUME 178 • NUMBER 7 • 2007 1284
(see Bead adhesion assay section; Tilghman and Hoover, 2002).
These beads, which are hereafter referred to as αICAM1 beads,
specifi cally adhered to ICAM1 and recruited ICAM1-GFP within
30 min (Fig. 4 B and Video 3, available at http://www.jcb.org/
cgi/content/full/jcb.200612053/DC1). X-Z projections showed
that ICAM1-GFP protruded around adhered αICAM1 beads
(Fig. 4 B, d). Additionally, scanning EM images revealed that
adhe sion of αICAM1 beads induced dorsal ruffl es comparable
with those induced by leukocytes (Fig. 4 C). The αICAM1 beads
did not bind to VCAM1-GFP–transfected cells or to nontransfected
Figure 5. Beads coated with ICAM1 antibodies recruited SGEF and active RhoG to sites of adhesion. (A) As a control, cells expressing GFP alone and
ICAM1-V5 were incubated with αICAM1 beads. ICAM1 mAbs are used to visualize ICAM1-V5. αICAM1 beads are stained with secondary AlexaFluor anti-
bodies that also stained ICAM1 in red (a). Images show that GFP alone (green; b) is not recruited around αICAM1 beads that adhered to ICAM1-V5 (arrow-
heads; a–c). Image c shows the merge. Image d shows a magnifi cation of image c. GFP-SGEF (f) and GFP–RhoG-Q61L (j) are recruited (arrowheads) to
αICAM1 bead adhesion sites on ICAM1-V5–expressing COS7 cells (e and i). Merged images are shown in g and k. Images h and l show magnifi cations
of g and k, respectively. (B) Quantifi cation of GFP-expressing proteins that are recruited to adhesion sites induced by αICAM1 beads. All cells were trans-
fected with ICAM1-V5 and subsequently cotransfected with GFP-tagged proteins except for single-transfected ICAM1-GFP. αICAM1 beads recruit ICAM1-
GFP, GFP-SGEF, and GFP–RhoG-Q61L in 55–80% of the cases to sites of adhesion. In contrast, GFP, β-catenin–GFP, and VE-cadherin–GFP are not recruited.
Data are means ± SEM (error bars) of at least four independent experiments. Bars (A, panel c), 10 μm; (A, panels d, h, and l) 5 μm; (A, g and k) 20 μm.

Page 7

ICAM1-INDUCED RHOG ACTIVATION • VAN BUUL ET AL.1285
cells (unpublished data). In addition, blocking antibodies to
ICAM1 completely inhibited binding of the αICAM1 beads to
ICAM1 (unpublished data).
RhoG and SGEF are recruited to sites
of ICAM1 engagement
To show that ICAM1-GFP was recruited specifi cally to the
beads, cotransfections with ICAM1-V5 and GFP as a control
were performed and revealed that GFP alone was not recruited
to sites of adhesion (Fig. 5 A). Also, neither β-catenin–GFP nor
VE-cadherin–GFP was recruited to sites of adhesion (Fig. 5 B).
In contrast, GFP-SGEF and GFP–RhoG-Q61L were recruited
to sites of ICAM1 engagement (Fig. 5, A and B). Additionally,
as a control, beads coated with major histocompatibility complex
(MHC) antibodies were incubated on human umbilical vein endo-
thelial cells (HUVECs), and z-stack analysis was performed to
measure actin-rich protrusions around adhered beads. The re-
sults revealed that αICAM1 beads induced substantially more
F-actin–rich protrusions than the αMHC class I beads, whereas
the total number of beads that adhered to the endothelium was
equivalent (Fig. S2 A, available at http://www.jcb.org/cgi/
content/full/jcb.200612053/DC1). Expression of GFP–RhoG-
Q61L in HUVECs showed that RhoG is recruited by αICAM1
beads but not by αMHC class I beads (Fig. S2 B). Previous work
has indicated that actin is a major component of the ICAM1-
positive cup structures (Barreiro et al., 2002; Carman et al., 2003;
Carman and Springer, 2004). Using GFP-actin, which is tran-
siently expressed in endothelial cells, we confi rmed that αICAM1
beads effi ciently recruited actin to sites of adhesion (Fig. S2 C).
These data indicate that ICAM1 specifi cally induces these pro-
trusions and recruits RhoG to sites of adhesion.
ICAM1 engagement activates RhoG
We next performed RhoG activation assays to determine RhoG
activity downstream from ICAM1 engagement. We made use of
the RhoG downstream effector ELMO (engulfment and cell
motility), which specifi cally binds GTP-bound RhoG (Katoh and
Negishi, 2003; Ellerbroek et al., 2004). In our initial experiments,
Figure 6. RhoG is activated downstream
from ICAM1. (A) HUVECs were transiently
transduced with myc-RhoG-wt using adenovirus.
αICAM1 beads were added as described in
Materials and methods. Using GST-ELMO,
activated RhoG was pulled down from the lysate
and detected by Western blot analysis using
anti-myc antibodies. The middle blot shows
protein expression levels of myc-RhoG-wt in
endothelial cell lysates. The graph on the right
shows quantifi cation of two independent ex-
periments. (B) αICAM1 beads were added to
TNF-α–stimulated endothelial cells as described
in Materials and methods. Using GST-ELMO,
activated endogenous RhoG was isolated and
detected by Western blot analysis using anti-
RhoG mAbs. (A and B) αICAM1 beads increased
RhoG activity in endothelial cells within 30 min
(top). The bottom panel shows endogenous
ICAM1 expression in endothelial cell lysates.
(B) The middle panel shows expression levels of
endogenous RhoG in endothelial cell lysates.
(C) COS7 cells were transiently transfected with
myc-RhoG-wt and ICAM1-GFP. αICAM1 beads
were added as described in Materials and
methods. Using GST-ELMO, activated RhoG was
isolated and detected by Western blot analysis
using anti-myc antibodies. αICAM1 beads in-
creased RhoG activity up to 30 min (top). The
middle panel shows expression levels of myc-
RhoG-wt in cell lysates. The bottom panel shows
ICAM1-GFP expression detected by anti-GFP
antibodies in cell lysates. (D) Experiments
were performed as in B, but with HL60 cells
(5 × 105 cells per six wells). Single-transfected
myc-RhoG-Q61L-COS7 cells were used as a
positive control. The top panel shows that HL60
cell adhesion increased RhoG-GTP levels. The
middle panel shows levels of transfected myc-
RhoG-wt, and the bottom panel shows levels
of ICAM1-GFP. (B–D) The graph on the right
shows quantifi cation of three independent exper-
iments. Data are means ± SEM (error bars).
*, P < 0.05; **, P < 0.01.

Page 8

JCB • VOLUME 178 • NUMBER 7 • 2007 1286
we used an adenoviral vector to deliver myc-tagged RhoG to
HUVECs and found that engagement of ICAM1 with αICAM1
beads induced RhoG activation (Fig. 6 A). Examining the acti-
vation of endogenous RhoG using a monoclonal antibody re-
vealed that ICAM1 engagement showed a similar response (Fig.
6 B). It should be noted that TNF-α pretreatment did not change
the activity of RhoG in endothelial cells, although overnight
treatment slightly diminished RhoG expression (unpublished data).
To delineate the pathway downstream from ICAM1, myc-tagged
RhoG-wt together with ICAM1-GFP were expressed in COS7
cells as described in Materials and methods (see RhoG, RhoA,
and Rac1 activation assay section). Treatment with αICAM1
beads induced RhoG activation after 10 and 30 min (Fig. 6 C).
This activation was transient because the induced activity of
RhoG declined after 60 min (Fig. 6 B and Fig. S3 A, available
at http://www.jcb.org/cgi/content/full/jcb.200612053/DC1).
Beads coated with MHC class I antibodies did not induce any
RhoG activation (Fig. S3 B). To examine whether leukocytes
could activate RhoG through ICAM1, we added HL60 cells
to myc–RhoG-wt and ICAM1-GFP–expressing COS7 cells.
RhoG activation was stimulated by the adhesion of HL60 cells
(Fig. 6 D). To study whether closely related GTPases Rac1 and
Cdc42 are activated downstream from ICAM1 engagement, pull-
down assays using the p21-activated kinase–binding domain (PBD)
as bait were performed. Interestingly, Rac1 and Cdc42 were tran-
siently activated downstream from ICAM1 engagement as well,
although Rac1 activation peaked at 10 min (Fig. S3 C). RhoA
activity measurements confi rmed that RhoA became activated
after ICAM1 engagement (Adamson et al. 1999; Wojciak-Stothard
et al., 1999), and this was maximal after 10 min (Fig. S3 E).
ICAM1–intracellular domain is required
for RhoG activation
Previously, it has been shown that the intracellular domain of
ICAM1 is required for leukocyte passage across the endothe-
lium but is dispensable for the initial adhesion (Lyck et al., 2003;
Sans et al., 2001). To investigate whether the intracellular domain
of ICAM1 is required to transmit the signal that triggers RhoG
activation, a C-terminal deletion mutant of ICAM1 lacking the
intracellular domain and tagged to a V5 epitope (ICAM1-∆C-V5)
Figure 7. ICAM1 intracellular domain was
required for RhoG localization and activation.
(A) COS7 cells were transiently cotransfected
with full-length ICAM1 tagged with V5 (b) or a
C-terminal deletion mutant of ICAM1-V5 (ICAM1-
∆C-V5; e) and with GFP–RhoG-Q61L (a and d).
Panels c and f show the corresponding merged
images. ICAM1-∆C-V5 does not colocalize with
RhoG-Q61L. (B) Experiments were performed
as described in Fig. 6 C except ICAM1-GFP
has been replaced with V5-tagged ICAM1-wt
(ICAM1-wt) or the V5-tagged C-terminal trun-
cated ICAM1 mutant (ICAM1-∆C). The top panel
shows that ICAM1 engagement increases
RhoG-GTP levels in cells expressing ICAM1-wt
but not in cells expressing ICAM1-∆C. The
bottom panel shows levels of myc-RhoG-wt in
total cell lysates. The graph on the right shows
quantifi cation of three independent experiments.
*, P < 0.05. (C) ICAM1 intracellular tail was
required for effi cient recruitment around an
adhered leukocyte. Confocal imaging was
used to visualize the apical and the baso-
lateral plane of the COS7 cells that were
expressing either ICAM1-wt or ICAM1-∆C.
Quantifi cation of ICAM1-rich rings around HL60
cells that were allowed to adhere for 30 min
showed a requirement of the ICAM1 tail for
proper cup formation. The experiment was re-
peated three times. *, P < 0.01. (B and C) Data
are means ± SEM (error bars). Bars, 20 μm.

Page 9

ICAM1-INDUCED RHOG ACTIVATION • VAN BUUL ET AL.1287
was generated and expressed in COS7 cells. The overexpression
of ICAM1-∆C-V5 together with GFP–RhoG-Q61L showed
that ICAM1 required its intracellular domain to localize to
RhoG-induced dorsal ruffl es (Fig. 7 A). No difference in the
adhesion of αICAM1 beads to either full-length or ICAM1-∆C
was observed (unpublished data). However, the αICAM1 beads
were unable to activate RhoG in cells expressing ICAM1-∆C
(Fig. 7 B). Additionally, cells that expressed ICAM1-∆C in-
duced substantially less ICAM1-positive protrusions around ad-
hered leukocytes than ICAM1-wt (Fig. 7 C). Together, these data
show that ICAM1 engagement induces RhoG activation and
subsequent membrane protrusions in a pathway that is depen-
dent on its intracellular domain.
ICAM1 associates with SGEF through
its SH3 domain
The fi nding that RhoG is activated downstream from ICAM1
engagement coupled with the observation that SGEF and RhoG
colocalized with ICAM1 led us to investigate whether ICAM1
and SGEF physically interact. Immunoprecipitation experiments
showed that endogenous ICAM1 was precipitated with endo-
genous SGEF from TNF-α–treated endothelial cells (Fig. 8 A).
To study this interaction in more detail, pull-down experi-
ments were performed using biotinylated peptides. A peptide
corresponding to the cytoplasmic domain of ICAM1 bound
myc-tagged SGEF as well as endogenous SGEF (Fig. 8, B and C,
respectively). Interestingly, the intracellular domain of ICAM1
comprises only 28 amino acids, and its C terminus contains four
prolines in close proximity. We examined whether the SH3 do-
main of SGEF could directly associate with the cytoplasmic
domain of ICAM1. Biotinylated ICAM1–intracellular domain
peptide sedimented the SH3 domain of SGEF, which was fused
to GST (GST-SH3SGEF) in vitro (Fig. 8 D, a). To further explore
the interaction of SGEF with ICAM1, we used a myc-tagged
mutant of SGEF lacking the SH3 domain (SGEF-∆SH3). This
mutant SGEF failed to coimmunoprecipitate with ICAM1-GFP
(Fig. 8 D, b). Interestingly, the association between SGEF
and ICAM1 did not depend on the GEF activity of SGEF;
ICAM1 still associated with a catalytically dead mutant of
SGEF (SGEF-∆DH) that contained the SH3 domain (Fig. 8 D, b).
An inactivating point mutant in the SH3 domain of SGEF (myc–
SGEF-W826R) was previously generated in which the catalytic
activity of SGEF remained intact (Ellerbroek et al., 2004).
This construct and SGEF-wt were overexpressed in COS7 cells
together with ICAM1-GFP. Immunoprecipitation assays con-
fi rmed that SGEF-wt interacted with ICAM1, but SGEF-W826R
revealed decreased binding (Fig. 8 E). These data indicated that
the ICAM1–SGEF interaction requires an intact SGEF-SH3
domain. To test whether ICAM1 associates through its proline-
rich sequence to SGEF, we deleted this proline-rich sequence
from the cytoplasmic domain of ICAM1. Immunoprecipitation
studies revealed that ICAM1 lacking the proline-rich sequence
failed to bind to SGEF (Fig. 8 F).
RhoG is required for leukocyte TEM
To study RhoG involvement in TEM, siRNA was used to reduce
RhoG expression in primary endothelial cells. Western blot
analysis revealed that the relevant siRNA reduced RhoG protein
expression in endothelial cells but did not affect other proteins
known to be present in cup structures or involved in transmigra-
tion, such as moesin and ICAM1 (Barreiro et al., 2002; Carman
et al., 2003; Millán et al., 2006). Also, the expression levels of
other closely related small GTPases such as Rac1, Cdc42, and
RhoA were unaffected (Fig. 9 A). Adhesion of leukocytes to endo-
thelial monolayers that showed reduced RhoG expression was
not affected. Similarly, expression of dominant-negative RhoG
did not affect leukocyte adhesion (unpublished data). However,
the formation of cup structures, which was quantifi ed as ICAM1-
positive ringlike structures that surrounded adhered leukocytes,
was decreased compared with control cells (Fig. 9 B). Trans-
migration of HL60 cells across endothelial cell monolayers
was also substantially attenuated by the knockdown of RhoG
expression (Fig. 9 C).
Several previous studies have addressed the role of RhoA
in endothelial cells during leukocyte TEM, demonstrating that
it is required for TEM (Adamson et al., 1999; Wojciak-Stothard
et al., 1999) and showing that it becomes activated downstream
from ICAM1 cross-linking (Wojciak-Stothard et al., 1999; Etienne-
Manneville et al., 2000; Thompson et al., 2002). We were in-
terested to relate our RhoG results to this previous body of
work on RhoA. Reducing RhoG expression by siRNA did not
affect RhoA activation downstream of ICAM1 engagement
(Fig. S4 A, available at http://www.jcb.org/cgi/content/full/
jcb.200612053/DC1), which is consistent with the activation
of RhoA occurring faster than the activation of RhoG (Fig. S4,
A and E). Interestingly, reducing RhoA expression by siRNA
depressed ICAM1-induced RhoG activation (Fig. S4 B). This
suggested that RhoA acts upstream of RhoG activation in the
pathway from ICAM1 engagement. Whether RhoA has a role
in cup formation has been controversial. Barreiro et al. (2002)
found that inhibiting the RhoA effector ROCK/Rho kinase
with Y27632 diminished cup formation. However, this was not
found by Carman and Springer (2004), who also were unable
to block cup formation by treating endothelial cells with C3
or Y27632 (Carman et al., 2003). Our fi nding that RhoA is
required upstream of RhoG activation suggested that RhoA may
be necessary for cup formation. Consequently, we investigated
this directly using micro-RNA (miRNA) of RhoA to depress its
expression. We have found that the depletion of RhoA reduced
the formation of cups induced by αICAM1 beads (Fig. S4 C).
SGEF and leukocyte TEM
We wished to explore whether SGEF has a role in leukocyte
TEM and, thus, have used siRNA to knockdown SGEF expres-
sion in endothelial cells. We confi rmed that the siRNA decreased
SGEF expression and that it did not affect the expression of
RhoG, Rac1, or other proteins involved in cups, such as ICAM1
or moesin (Fig. 10 A). Importantly, SGEF knockdown did im-
pair the activation of RhoG downstream from ICAM1 engage-
ment (Fig. 10 B), and, consistent with this, it also resulted in
decreased cup formation, as judged by the number of ICAM1-
positive rings surrounding adherent leukocytes (Fig. 10 C).
Together, these data indicate a pathway from ICAM1 clustering
to SGEF to RhoG activation resulting in the formation of cups.

Page 10

JCB • VOLUME 178 • NUMBER 7 • 2007 1288
Figure 8. SGEF interacts through its SH3 domain with the
C-terminal domain of ICAM1. (A) ICAM1 was immunoprecipi-
tated (IP) using anti-ICAM1 antibodies, and IgG was used as
control. Immunoblotting with anti-SGEF antibodies showed
that endogenous ICAM1 and SGEF interacted in endothelial
cells, whereas IgG did not show any interaction with SGEF
(top). The second panel shows that ICAM1 was effi ciently
immunoprecipitated by anti-ICAM1 antibodies (left lane) but not
by IgG (right lane). The two bottom panels show levels
of endogenous SGEF and ICAM1 in endothelial cell lysates.
(B) COS7 cells were transfected with myc-SGEF-wt, lysed, and
subsequently incubated with biotinylated peptides that corre-
spond to the intracellular domain of ICAM1 (ICAM1-C-term.)
or to the intracellular domain of αv-integrin, which was used
as control (αv-C-term). Streptavidin-based pull-downs show
that the intracellular domain of ICAM1 binds myc-SGEF. The
right lane shows myc-SGEF expression in one tenth of the total
cell lysate using anti-myc antibodies. (C) Endothelial cells were
lysed and subsequently incubated with biotinylated peptides as
described in B. The top panel shows that ICAM1-peptide
bound endogenous SGEF, whereas the αv-C-term-peptide
did not. (D, a) GST-SH3SGEF (amino acids 789–850) was puri-
fi ed and incubated with the peptides described in B. Pull-
down experiments using streptavidin-coated Sepharose beads
were performed, and GST-SH3SGEF was detected using anti-
GST antibodies. SH3SGEF interacted with the ICAM1 tail but
not with the tail of αv-integrin (αv-C-term). (D, b) COS7 cells
were transfected with ICAM1-GFP and myc-SGEF-∆DH or
myc-SGEF-∆SH3 and were processed for immunoprecipitation
using anti-GFP antibodies. Western blot analysis revealed
that ICAM1-GFP binds to myc-SGEF that contains the SH3 do-
main independent of the DH domain. (E) COS7 cells were
transiently cotransfected with ICAM1-GFP and myc-SGEF-wt
or myc-SGEF-W826R. ICAM1-GFP was immunoprecipitated
using anti-GFP antibodies. Immunoblotting with anti-myc anti-
bodies showed the binding of SGEF-wt but reduced binding
of SGEF-W826R to ICAM1-GFP (top). The bottom blots show
levels of immunoprecipitated ICAM1-GFP and myc constructs
in total cell lysates as indicated. (F) COS7 cells were tran-
siently cotransfected with myc-SGEF-wt and ICAM1-wt-GFP or
ICAM1-∆Proline (Pro). GFP was immunoprecipitated using
anti-GFP antibodies. Immunoblotting with anti-myc antibodies
showed the binding of SGEF-wt but reduced binding of SGEF-wt
to ICAM1-∆Pro-GFP (top). The bottom blots show levels of
immunoprecipitated ICAM1-wt-GFP or ICAM1-∆Pro-GFP and
GFP and myc constructs in total cell lysates as indicated. Note
that ICAM1-∆Pro-GFP runs a little bit lower than wt.

Page 11

ICAM1-INDUCED RHOG ACTIVATION • VAN BUUL ET AL.1289
Finally, we examined the effect of SGEF knockdown on TEM
and found that it caused a decrease in the migration of HL60
cells across endothelial monolayers by up to 50% (Fig. 10 D).
Discussion
During the last decade, it has become increasingly clear that
endothelial cells, rather than being a passive barrier, actively par-
ticipate in the process of leukocyte TEM. This study focuses on a
recently discovered phenomenon that occurs during TEM in which
the endothelial cell extends sheets of membrane to form a cuplike
structure that surrounds adherent leukocytes (Barreiro et al.,
2002; Carman et al., 2003; Carman and Springer, 2004; Doulet
et al., 2006). Although their precise function is unclear, evidence
has been presented that these structures assist leukocytes on
their way through the endothelium (Carman and Springer, 2004).
Our data reveal a new signaling pathway downstream from
leukocyte adhesion that involves the small GTPase RhoG. We
show here that RhoG activation is triggered through the engage-
ment of ICAM1 and is critical for formation of the apical cups.
Additionally, RhoG expression is needed for optimal leukocyte
passage across the endothelium. Our data show a strong correla-
tion between formation of the cups and TEM. The endothelial
apical cups resemble phagocytic cups, and it is notable that RhoG
has been implicated previously in the phagocytosis of apoptotic
cells in Caenorhabditis elegans (deBakker et al., 2004). Recent
work has also implicated RhoG as well as its exchange factor
SGEF in the uptake of Salmonella by epithelial cells (Patel and
Galan, 2006). Engulfment of Salmonella is promoted by several
bacterial proteins that function to activate multiple Rho family
GTPases. Interestingly, the Salmonella protein SopB was found
to activate SGEF and RhoG, thereby stimulating the formation
of phagocytic cups on the surfaces of epithelial cells (Patel and
Galan, 2006). Together, these results suggest that SGEF and
RhoG may function in a variety of physiological and pathological
situations in which phagocytosis or the uptake of particulate
material is involved.
The route by which leukocytes pass through the endo-
thelium, whether it is paracellular or transcellular, has generated
considerable debate for many years. In tissue culture models,
it has been estimated that only 10–25% of all leukocytes use the
transcellular route, with the majority migrating through cell–
cell junctions (Carman and Springer, 2004). Millán et al. (2006)
have shown that the redistribution of ICAM1 to caveolin-rich
membrane domains in response to engagement is followed by
transcytosis to the baso-lateral side of the endothelium. The in-
duction of apical cups by RhoG as well as the similarity of these
structures to phagocytic cups might lead to the idea that RhoG
would function primarily in transcellular rather than paracellular
migration. However, our data show that silencing RhoG re-
sults in >70% inhibition of leukocyte TEM. Although our work
does not discriminate between the para- and trans-cellular mi-
gration routes, this decrease in TEM cannot be explained by
blocking the trans-cellular pathway only. Consistent with this,
the work of Carman and Springer (2004) suggests that trans-
migratory cups are not restricted to the trans-cellular route but
may function to facilitate and guide leukocyte TEM in general.
Alternatively, RhoG may have additional functions in TEM
besides mediating cup formation.
The role of Rho family GTPases in formation of the cup
structures has begun to be investigated. Barreiro et al. (2002)
found that Y27632, which inhibits ROCK/Rho kinase down-
stream of RhoA, decreased the assembly of these structures in-
duced by VCAM1 engagement. However, Carman and Springer
reported that neither C3 nor Y27632 inhibited the assembly of
the cups induced by ICAM1 cross-linking (Carman et al., 2003;
Figure 9. Transmigration of HL60 cells and formation of cups depended
on RhoG expression levels. (A) siRNA against RhoG effi ciently reduced the
expression of endogenous RhoG in endothelial cells. Longer exposure did
show some expression of RhoG. The siRNA did not affect the expression
levels of RhoA, Rac1, Cdc42, tubulin, ICAM1, or moesin. (B) Quantifi cation
of ICAM1-positive rings around adhered HL60 cells, which was measured
as described in Materials and methods, showed that cup formation was
signifi cantly decreased when RhoG expression was reduced. *, P < 0.01.
(C) Endothelial cells were cultured on transwell fi lters and transfected with
RhoG siRNA as described in Materials and methods. 48 h later, differenti-
ated HL60 cells were allowed to transmigrate for 4 h under spontaneous
conditions (black bars) or to 50 ng/ml SDF-1 (white bars). Reduced endo-
thelial RhoG expression resulted in decreased spontaneous and SDF-1–
induced transmigration. *, P < 0.001; **, P < 0.01. (B and C) The
experiment was repeated four times. Data are means ± SEM (error bars).

Page 12

JCB • VOLUME 178 • NUMBER 7 • 2007 1290
Carman and Springer, 2004). In our hands, we have observed
the partial inhibition of cup formation by Y27632 (our unpub-
lished data) and have found that knockdown of RhoA also in-
hibits cup formation (Fig. S4 C). The depression of cup formation
may, in part, be caused by the inhibition of RhoG activation in
cells in which RhoA has been knocked down (Fig. S4 B). How
RhoA regulates RhoG activation remains to be determined. In
addition, RhoA may play other roles in the assembly of endo-
thelial apical cups.
In this study, we have focused on RhoG, a close relative of
Rac1 (Wennerberg et al., 2002), because it induces dorsal mem-
brane ruffl es and has been implicated in phagocytosis (deBakker
et al., 2004). However, we have observed that ICAM1 engage-
ment leads to the activation of not only RhoG and RhoA but also
Rac1 and Cdc42 (Figs. 6 and S3). It is notable that RhoG can
activate Rac1 through the DOCK180-binding protein ELMO
(Katoh and Negishi, 2003), raising the possibility that the acti-
vation of RhoG we observe stimulates Rac1 activation. How-
ever, the time course of the activation of Rac1 and RhoG is not
consistent with this idea. In future work, it will be interesting to
identify the pathways leading to the activation of these other
Rho family members.
The intracellular domain of ICAM1 is a prerequisite for
optimal TEM of leukocytes (Sans et al., 2001; Lyck et al., 2003).
ICAM1 lacking its intracellular domain (ICAM1-∆C) fails
to promote leukocyte TEM, although leukocyte adhesion to
ICAM1-∆C is unaffected. Engagement of ICAM1-∆C by
αICAM1 beads also fails to activate RhoG. The fact that ICAM1-
∆C cannot activate RhoG is likely the result of its inability to
bind SGEF. We found that the proline-rich region of the intracel-
lular domain of ICAM1 binds the SH3 domain of SGEF. This
interaction is independent of SGEF activation because catalyti-
cally inactive mutants of SGEF that express the SH3 domain still
bind ICAM1. Engagement of ICAM1 does not promote the as-
sociation between SGEF and ICAM1 but does increase the
activation of SGEF, as judged by the increased binding of SGEF to
nucleotide-free RhoG (unpublished data). Thus, SGEF and
ICAM1 likely form a stable interacting pair.
Additional signals such as tyrosine phosphorylation may
be necessary to trigger SGEF activation, as has been shown for
other GEFs (Rossman et al., 2005). One such signal may de-
pend on Src-kinase activity. Src-kinase is rapidly activated after
ICAM1 engagement and is required for optimal leukocyte TEM
but also does not affect leukocyte adhesion (Etienne-Manneville
et al., 2000; Tilghman and Hoover, 2002; Wang et al., 2003;
Yang et al., 2006a). Our preliminary results show that inhibiting
Src family kinases using PP2 prevented RhoG activation down-
stream from ICAM1 engagement (unpublished data). These
data support the idea that additional signals such as tyrosine
phosphorylation are needed to activate SGEF. It is likely that
there are multiple targets for Src downstream from ICAM1.
One Src substrate that has been implicated in TEM is cortactin
Figure 10. Reducing SGEF expression in
HUVECs blocked RhoG activation and TEM.
(A) siRNA against SGEF reduces the expression
of endogenous SGEF in endothelial cells. The
siRNA did not affect the expression levels of
endogenous RhoG, Rac1, ICAM1, or moesin.
(B) Knockdown of SGEF expression by siRNA
inhibits the activation of RhoG downstream
from ICAM1. HUVECs were transiently trans-
duced with myc-RhoG-wt and treated with TNF-α.
SGEF expression was reduced by siRNA in
HUVECs, and RhoG activity, which was induced
by αICAM1 beads, was measured using GST-
ELMO. The top panel shows RhoG activity
after 30 min, which was depressed when SGEF
expression was reduced (bottom panel). The
second panel shows equal expression for myc-
RhoG-wt in HUVECs. The third panel shows
equal levels of ICAM1 in cell lysates. The
bottom panel shows reduced SGEF expression
after siRNA treatment in HUVECs. The experi-
ment was performed two times. (C) Knock-
down of SGEF expression decreased ICAM1
cup formation. Endothelial cells transfected
with SGEF siRNA were incubated with HL60
cells. Quantifi cation of ICAM1-positive rings
around adhered leukocytes, which was mea-
sured as described in Materials and methods,
shows that cup formation was signifi cantly
decreased when SGEF levels were reduced.
*, P < 0.05. (D) Knockdown of SGEF expres-
sion inhibited transmigration. Endothelial cells
were cultured on transwell fi lters and trans-
fected with the appropriate siRNA as de-
scribed in Materials and methods. 48 h later,
differentiated HL60 cells were allowed to
transmigrate for 4 h under spontaneous conditions (black bars) or toward 50 ng/ml SDF-1 in the lower chamber (white bars). Reduced SGEF levels
diminish SDF-1–induced transmigration signifi cantly. *, P < 0.001; **, P < 0.05. (C and D) The experiment was repeated nine times. Data are means ± SEM
(error bars).

Page 13

ICAM1-INDUCED RHOG ACTIVATION • VAN BUUL ET AL.1291
(Yang et al., 2006b). Cortactin is a regulator of the actin cyto-
skeleton that is notably prominent in structures like membrane
ruffl es and phagocytic cups (Weed and Parsons, 2001).
The passage of leukocytes across the endothelium is a
critical event in immune surveillance and in infl ammation.
Although infl ammation is physiologically important, it also
underlies many pathological conditions. Consequently, there is
considerable interest in understanding the pathways by which
leukocytes cross the endothelial barrier so that inappropriate in-
fl ammation can be controlled. Much remains to be learned about
TEM, including the role of the cups that are formed in response
to ICAM1 engagement. Different leukocyte types may induce
different effects on the kinetics of ICAM1 signaling and sub-
sequent apical cup formation. In this study, we have identifi ed a
pathway downstream from ICAM1 involving RhoG and its ex-
change factor SGEF that leads to endothelial apical cup forma-
tion. Inhibition of either RhoG or SGEF not only inhibits apical
cup formation but also depresses TEM, which is consistent with,
although does not prove, a role for the cups in TEM.
Materials and methods
Reagents and antibodies
pAbs against ICAM1 (for Western blotting) and mAb against RhoA were
obtained from Santa Cruz Biotechnology, Inc. mAbs against Rac1, Cdc42,
and MHC class I (MHC-A, -B, and -C) were purchased from BD Bio-
sciences. Recombinant TNF-α and a mAb against ICAM1 were purchased
from R&D Systems. The GFP and myc (clone 9E10) mAbs were purchased
from Invitrogen. Polyclonal rabbit antibody against VE-cadherin was pur-
chased from Cayman Chemical. The SGEF rabbit pAb was generated
in our laboratory as described previously (Ellerbroek et al., 2004). The
mAb against RhoG (clone IF-3-B3-E5) was raised in the laboratory of
M.A. Schwartz (Robert M. Berne Cardiovascular Research Center, University
of Virginia, Charlottesville, VA) against the C-terminal RhoG peptide (AA162-
180) of the sequence Q Q D G V K E V F A E A V R A V L N P T . Dot blots showed that
the mAb did not cross react with bacterially expressed Rac1, Cdc42, and
RhoA. Western blotting analysis showed that the RhoG antibody did recog-
nize GFP–RhoG-wt but not GFP–Rac1-wt expressed in COS7 cells.
Expression vectors
SGEF cDNA was subcloned using BamHI–EcoRI restriction sites into pCMV6M,
an N-terminal myc epitope–tagged eukaryotic expression vector, as described
previously (Ellerbroek et al., 2004). SGEF deletion mutants were gener-
ated using the QuikChange Site-Directed Mutagenesis kit (Stratagene)
and were subcloned into pCMV6M. pGEX-4T2-ELMO was a gift from
K. Ravichandran (University of Virginia, Charlottesville, VA). Generation
of eukaryotic expression vectors pCMV-myc-Rac(Q61L), pCMV-myc–Rac-wt,
pCMV-myc-Rac(T17N), pCMV-myc-RhoG(Q61L), pCMV-myc–RhoG-wt, and
pCMV-myc-RhoG(T17N) was described previously by our laboratory
(Wennerberg et al., 2002). wt and mutant Rac1 and RhoG constructs were
subsequently subcloned into pEGFP-C3 (CLONTECH Laboratories, Inc.) as
described previously (Wennerberg et al., 2002). SGEF was subcloned into
pEGFP-C2. ICAM1-GFP was a gift from F. Sanchez-Madrid (Hospital de la
Princesa, Universidad Autónoma de Madrid, Madrid, Spain). For ICAM1-
∆Pro-GFP, the last 11 amino acids of the intracellular tail of ICAM1 were
deleted. ICAM1-wt and C-terminal deletion mutant (lacking the last 28
amino acids) cDNA was subcloned into the pAdCMV-V5-DEST vector using
the Gateway expression system (Invitrogen).
Cell cultures, treatments, and transfections
HUVECs were obtained from Cambrex and cultured as described previously
(Worthylake et al., 2001). Endothelial cells were activated with 10 ng/ml
TNF-α overnight as indicated to mimic infl ammation. All cell lines were cul-
tured or incubated at 37°C at 10% CO2. The HL60 promyelocytic cell line
was obtained from the University of North Carolina’s Lineberger Compre-
hensive Cancer Center Tissue Culture Facility and grown in Optimem plus
5% FBS. In all experiments described, differentiated HL60 cells were used.
Differentiation to a neutrophil-like lineage was achieved by adding 1.3%
DMSO for 3–5 d (Back et al., 1992). COS7 cells were maintained in growth
medium (Iscove’s modifi ed Dulbecco’s medium with 10% FCS; Sigma-
Aldrich). Cells were transiently transfected with the expression vectors indi-
cated in each experiment according to the manufacturer’s protocol using
LipofectAMINE PLUS (Invitrogen) or Fugene 6 (Roche). Myc–RhoG-wt cDNA
was transferred to an AdV expression vector and transfected into 293 cells,
and high titer virus stocks were produced. Subsequently, myc–RhoG-wt was
transiently delivered into HUVECs by adenovirus transduction.
Immunofl uorescence
Cells were cultured on glass coverslips, fi xed, and immunostained with the
indicated primary antibodies for 60 min at RT as described previously (van
Buul et al., 2002). Subsequent visualization was performed with Alexa-
Fluor-conjugated secondary antibodies for 30 min (Invitrogen). F-actin was
visualized with fl uorescently labeled phalloidin (Invitrogen). Glass cover-
slips were mounted in MOWIOL at RT. Images were collected with a confo-
cal microscope (LSM510; Carl Zeiss MicroImaging, Inc.) equipped with a
microscope (Axiovert 100M; Carl Zeiss MicroImaging, Inc.) and an oil im-
mersion plan-Neofl uar 63× NA 1.3 oil lens (Carl Zeiss MicroImaging, Inc.).
Cross talk between the different channels was avoided by the use of se-
quential scanning. Images were processed using imaging examiner soft-
ware (Carl Zeiss MicroImaging, Inc.) and Photoshop CS (Adobe).
Scanning EM
Transfected cells were grown on glass coverslips, fi xed in 2.5% glutaral-
dehyde/PBS for 30 min at room temperature, and processed for scanning
EM as described previously (Ellerbroek et al., 2004). In brief, samples
were incubated with 2% aqueous osmium tetroxide for 45 min, dehydrated
in a graded ethanol series, and critical point dried in liquid CO2 using a
drying apparatus (CPD 010; Balzers Instruments). Samples were mounted
on aluminum stubs (Ted Pella, Inc.) and sputter coated with gold/palladium
using Polaron scanning EM. Cells were examined on a scanning electron
microscope (model 820; JEOL) at 15 kV.
TEM assay
Migration assays were performed in transwell plates (Corning) of 6.5-mm
diameter with 8-μm pore fi lters. Approximately 20,000 endothelial cells
were plated on matrigel-coated transwell fi lters, which were treated the
next day with siRNA as indicated. The following day, endothelial cells
were treated with siRNA again and with 10 ng/ml TNF-α overnight at
37°C and 10% CO2. 100,000 differentiated HL60 cells were added to the
upper compartment, and HL60 cells were allowed to migrate to 50 ng/ml
stromal cell–derived factor-1 (SDF-1; placed in the lower chamber to gener-
ate a chemotactic gradient; R&D Systems) for 4 h at 37°C and 10% CO2.
An input control (i.e., 100,000 HL60 cells) was set as 100%. After collect-
ing the migrated HL60 cells, fi lters were inspected by confocal laser-
scanning microscopy using fl uorescently labeled phalloidin to stain F-actin;
coating of matrigel on the transwell fi lter did not affect the formation of a
confl uent endothelial monolayer. Migrated HL60 cells were counted and
compared with 100% input, and the percent migration of HL60 cells was
calculated. To confi rm effi cient knockdown of the protein by siRNA, cells
were simultaneously grown in six-well plates and equally treated with
siRNA constructs and were analyzed by Western blotting.
Immunoprecipitation and Western blotting
Cells were grown to confl uency, washed twice gently with ice-cold
Ca2+- and Mg2+-containing PBS, and lysed in 300 μl lysis buffer (25 mM
Tris, 150 mM NaCl, 10 mM MgCl2, and 1% Triton X-100 with the addi-
tion of fresh protease inhibitors, pH 7.4). Immunoprecipitation was per-
formed as previously described (Barreiro et al., 2002) and analyzed by
Western blotting using an enhanced ECL detection system (GE Healthcare).
The intensity of the bands was quantifi ed by using ImageJ version 1.36
(National Institutes of Health, Bethesda, MD).
Apical cup quantifi cation
Using confocal laser-scanning microscopy, z-stacks were taken to confi rm the
formation of a cup around an adhered leukocyte. The length of the protru-
sion was ?6–7μm above the baso-lateral plane of the substrate (Fig. 1 B, d).
The apical plane was set to 4 μm from the baso-lateral plane (Fig. 1 A).
ICAM1-positive rings in the apical plane were counted as positive cups.
RhoG, RhoA, and Rac1 activation assay
For RhoG activation assays, a transient coexpression of myc-tagged RhoG
was used because of the lack of a high affi nity antibody that is appropriate
for these assays (according to Katoh and Negishi [2003]). Transfected cells
were lysed in 300 μl of 50 mM Tris, pH 7.4, 10 mM MgCl2, 150 mM NaCl,

Page 14

JCB • VOLUME 178 • NUMBER 7 • 2007 1292
1% Triton X-100, 1 mM PMSF, and 10 μg/ml each of aprotinin and leupeptin.
Lysates were cleared at 14,000 g for 10 min. Supernatants were rotated
for 30 min with 60–90 μg GST-ELMO (GST fusion protein containing
the full-length RhoG effector ELMO) conjugated to glutathione–Sepharose
beads (GE Healthcare). Beads were washed in 50 mM Tris, pH 7.4, 10 mM
MgCl2, 150 mM NaCl, 1% Triton X-100, and protease inhibitors. Pull-downs
and lysates were then immunoblotted for the myc epitope tag. For RhoA and
Rac1, GST-Rhotekin and GST-PBD were used as baits, respectively, and
used as described for GST-ELMO.
Fusion proteins
GST-ELMO, GST-SH3SGEF (SGEF789–850), GST-Rhotekin, and GST-PBD fusion
proteins were purifi ed from BL21 Escherichia coli cells (Stratagene) using
glutathione–Sepharose 4B as previously described (Ellerbroek et al., 2004).
GST fusion proteins were eluted with free, reduced glutathione in TBS medium
(50 mM Tris, 150 mM NaCl, 5 mM MgCl2, pH 7.4, and 1 mM DTT) and
stored in 30% glycerol at −80°C.
Antibody-coated beads
3 μm polystyrene beads (Polysciences, Inc.) were pretreated with 8% glutar-
aldehyde overnight, washed fi ve times with PBS, and were incubated with
300 μg/ml ICAM1/MHC mAb according to the manufacturer’s protocol.
Bead adhesion assay
For immunofl uorescence or scanning EM, 1 μg/ml of antibody-containing
beads was washed and resuspended in culture medium. 1 μg/ml of antibody-
coated beads was incubated in wells of 24-well dishes containing glass
coverslips, on which TNF-α–pretreated HUVECs or COS7 cells were
cultured. After the appropriate time, unbound beads were removed, and
coverslips were put on ice, gently washed three times with ice-cold PBS
containing 1 mM Ca2+/Mg2+, and subsequently processed for immuno-
fl uorescence. For biochemistry, 10 μg/ml of antibody-coated beads were
incubated on the cells, after which cells were washed as described above
(see Bead adhesion assay section) and subsequently lysed and processed
as described (see Immunoprecipitation and Western blotting and RhoG,
RhoA, and Rac1 activation assay sections).
Knockdown using siRNA
siRNA duplexes against human RhoG (sense, G C A A C A G G A U G G U G U-
C A A G U U ; antisense, 5′-P-U C G U C C A A G A U C G A C A U C C UU) and SGEF
mRNA (sense, C A A A U G G C C U U G C C G C U A A U U ; antisense, 5′-P-U U A G-
C G G C A A G G C C A U U U G U U ) and siControl nontargeting siRNA were ob-
tained from the Dharmacon siRNA collection. HUVECs were transfected
twice with 50 nmol/l siRNA using RNAifect transfection reagent (QIAGEN).
After 48 h, cells were processed as described in the previous paragraph.
Knockdown using miRNA adenovirus
miRNA adenoviral constructs were engineered according to the manufac-
turer’s protocol (Invitrogen). In brief, two sets of DNA oligonucleotides were
designed to target human RhoA mRNA and were named RhoA#1 and
RhoA#2: T G C T G A A G A C T A T G A G C A A G C A T G T C G T T T T G G C C A C T G A C T-
G A C G A C A T G C T C T C A T A G T C T T and C C T G A A G A C T A T G A G A G C A T G T C-
G T C A G T C A G T G G C C A A A A C G A C A T G C T T G C T C A T A G T C T T C (RhoA#1)
and G C T G T T T C C A T C C A C C T C G A T A T C T G T T T T G G C C A C T G A C T G A C A-
G A T A T C G G T G G A T G G A A A and C C T G T T T C C A T C C A C C G A T A T C T G T C A-
G T C A G T G G C C A A A A C A G A T A T C G A G G T G A T G G A A A C (RhoA#2). The
oligonucleotides were annealed and ligated into pcDNA6 EmGFP. The
EmGFP MiR RNA cassette was subsequently transferred to pDONR221
and fi nally to pAd by two sequential Gateway BP and LR recombinations.
Each construct was sequence verifi ed, and viral particles were produced
by transfection in 293A cells.
Biotinylated peptides
Peptides were synthesized with the following sequence: ICAM1–intra-
cellular tail peptide; biotin-G R Q R K I K K Y R L Q Q A Q K G T P M K P N T Q A T P P -OH;
αv peptide; and biotin-G H E N G E G N S E T -OH.
Live cell imaging
COS7 cells were cultured on glass coverslips and transfected with cDNA
as indicated. After 24 h, cells were placed in a heating chamber at 37°C
and recorded with a confocal microscope (LSM510; Carl Zeiss MicroImaging,
Inc.) equipped with a microscope (Axiovert 100M; Carl Zeiss MicroImaging,
Inc.) and an oil immersion plan-Neofl uar 63× NA 1.3 oil lens (Carl Zeiss
MicroImaging, Inc.).
Online supplemental material
Fig. S1 shows the recruitment of endogenous ICAM1 around a migrat-
ing HL60 cell. Fig. S2 shows the recruitment of GFP–RhoG-Q61L and
F-actin to αICAM1 beads but not to αMHC class I beads. Fig. S3 shows
activation of the small GTPases RhoG, Rac1, Cdc42, and RhoA down-
stream of ICAM1 engagement. Fig. S4 shows that reduced RhoG ex-
pression does not affect ICAM1-mediated RhoA activation but that the
reduced expression of RhoA does infl uence RhoG activity downstream
from ICAM1 engagement, which is induced by αICAM1 beads. Video 1
shows a real-time recording of 10 min of GFP–RhoG-Q61L expression
in COS7 cells. Video 2 shows a real-time recording of 10 min of GFP-
SGEF expression in COS7 cells. Video 3 shows a real-time recording of
ICAM1-GFP expressed in COS7 cells and incubated with αICAM1 beads
for 30 min. Online supplemental material is available at http://www.jcb
.org/cgi/content/full/jcb.200612053/DC1.
We thank Wendy Salmon and Dr. Michael Chua (Michael Hooker Microscopy
Facility, University of North Carolina, Chapel Hill, NC) as well as Hal Mekeel
(University of North Carolina, Chapel Hill, NC) for their technical assistance
with confocal and electron microscopy. We thank Lisa Sharek for outstanding tech-
nical assistance and members of the Burridge laboratory for their encouragement.
We thank Dr. Peter Hordijk for critically reading the manuscript. We also thank
Jos van Rijssel and Floris van Alphen for their technical assistance.
This work was supported, in part, by National Institutes of Health
grants HL45100 and HL080166 and a Kenan Distinguished Professorship
to K. Burridge. J.D. van Buul was supported by the Ter Meulen Fund, Royal
Netherlands Academy of Arts and Sciences, and an E. Dekker stipendium
from the Netherlands Heart Foundation. R. Garcia-Mata was supported by
a postdoctoral fellowship from the Susan Komen Foundation. T. Samsom
was supported by a postdoctoral fellowship from the Deutsche Forschungs-
gemeinschaft (grant Sa 1636/1-1).
Submitted: 15 December 2006
Accepted: 23 August 2007
References
Adamson, P., S. Etienne, P.O. Couraud, V. Calder, and J. Greenwood. 1999.
Lymphocyte migration through brain endothelial cell monolayers involves
signaling through endothelial ICAM-1 via a Rho-dependent pathway.
J. Immunol. 162:2964–2973.
Back, A.L., K.A. Gollahon, and D.D. Hickstein. 1992. Regulation of expression of the
leukocyte integrin CD11a (LFA-1) molecule during differentiation of HL-60
cells along the monocyte/macrophage pathway. J. Immunol. 148:710–714.
Barreiro, O., M. Yanez-Mo, J.M. Serrador, M.C. Montoya, M. Vicente-Manzanares,
R. Tejedor, H. Furthmayr, and F. Sanchez-Madrid. 2002. Dynamic interac-
tion of VCAM-1 and ICAM-1 with moesin and ezrin in a novel endothelial
docking structure for adherent leukocytes. J. Cell Biol. 157:1233–1245.
Carman, C.V., and T.A. Springer. 2004. The transcellular railway: insights into
leukocyte diapedesis. J. Cell Biol. 167:377–388.
Carman, C.V., C.D. Jun, A. Salas, and T.A. Springer. 2003. Endothelial cells
proactively form microvilli-like membrane projections upon intercellular
adhesion molecule 1 engagement of leukocyte LFA-1. J. Immunol.
171:6135–6144.
deBakker, C.D., L.B. Haney, J.M. Kinchen, C. Grimsley, M. Lu, D. Klingele,
P.K. Hsu, B.K. Chou, L.C. Cheng, A. Blangy, et al. 2004. Phagocytosis
of apoptotic cells is regulated by a UNC-73/TRIO-MIG-2/RhoG sig-
naling module and armadillo repeats of CED-12/ELMO. Curr. Biol.
14:2208–2216.
Doulet, N., E. Donnadieu, M.P. Laran-Chich, F. Niedergang, X. Nassif, P.O.
Couraud, and S. Bourdoulous. 2006. Neisseria meningitidis infection
of human endothelial cells interferes with leukocyte transmigration by
preventing the formation of endothelial docking structures. J. Cell Biol.
173:627–637.
Ellerbroek, S.M., K. Wennerberg, W.T. Arthur, J.M. Dunty, D.R. Bowman, K.A.
DeMali, C.J. Der, and K. Burridge. 2004. SGEF, a RhoG guanine nucleo-
tide exchange factor that stimulates macropinocytosis. Mol. Biol. Cell.
15:3309–3319.
Etienne-Manneville, S., J.B. Manneville, P. Adamson, B. Wilbourn, J. Greenwood,
and P.O. Couraud. 2000. ICAM-1-coupled cytoskeletal rearrangements
and transendothelial lymphocyte migration involve intracellular calcium
signaling in brain endothelial cell lines. J. Immunol. 165:3375–3383.
Johnson-Leger, C., and B.A. Imhof. 2003. Forging the endothelium during
infl ammation: pushing at a half-open door? Cell Tissue Res. 314:93–105.
Katoh, H., and M. Negishi. 2003. RhoG activates Rac1 by direct interaction with
the Dock180-binding protein Elmo. Nature. 424:461–464.

Page 15

ICAM1-INDUCED RHOG ACTIVATION • VAN BUUL ET AL.1293
Kluger, M.S. 2004. Vascular endothelial cell adhesion and signaling during leuko-
cyte recruitment. Adv. Dermatol. 20:163–201.
Lyck, R., Y. Reiss, N. Gerwin, J. Greenwood, P. Adamson, and B. Engelhardt.
2003. T-cell interaction with ICAM-1/ICAM-2 double-defi cient brain endo-
thelium in vitro: the cytoplasmic tail of endothelial ICAM-1 is necessary
for transendothelial migration of T cells. Blood. 102:3675–3683.
Millán, J., L. Hewlett, M. Glyn, D. Toomre, P. Clark, and A.J. Ridley. 2006.
Lymphocyte transcellular migration occurs through recruitment of endo-
thelial ICAM-1 to caveola- and F-actin-rich domains. Nat. Cell Biol.
8:113–123.
Muller, W.A. 2003. Leukocyte-endothelial-cell interactions in leukocyte trans-
migration and the infl ammatory response. Trends Immunol. 24:327–334.
Patel, J.C., and J.E. Galan. 2006. Differential activation and function of Rho GTPases
during Salmonella-host cell interactions. J. Cell Biol. 175:453–463.
Rossman, K.L., C.J. Der, and J. Sondek. 2005. GEF means go: turning on RHO
GTPases with guanine nucleotide-exchange factors. Nat. Rev. Mol. Cell
Biol. 6:167–180.
Sans, E., E. Delachanal, and A. Duperray. 2001. Analysis of the roles of ICAM1
in neutrophil transmigration using a reconstituted mammalian cell
expression model: implication of ICAM1 cytoplasmic domain and Rho-
dependent signaling pathway. J. Immunol. 166:544–551.
Thompson, P.W., A.M. Randi, and A.J. Ridley. 2002. Intercellular adhesion mol-
ecule (ICAM)-1, but not ICAM-2, activates RhoA and stimulates c-fos
and rhoA transcription in endothelial cells. J. Immunol. 169:1007–1013.
Tilghman, R.W., and R.L. Hoover. 2002. The Src-cortactin pathway is required
for clustering of E-selectin and ICAM-1 in endothelial cells. FASEB J.
16:1257–1259.
van Buul, J.D., and P.L. Hordijk. 2004. Signaling in leukocyte transendothelial
migration. Arterioscler. Thromb. Vasc. Biol. 24:824–833.
van Buul, J.D., C. Voermans, V. van den Berg, E.C. Anthony, F.P. Mul, S. van
Wetering, C.E. van der Schoot, and P.L. Hordijk. 2002. Migration of
human hematopoietic progenitor cells across bone marrow endothelium
is regulated by vascular endothelial cadherin. J. Immunol. 168:588–596.
Vestweber, D. 2002. Regulation of endothelial cell contacts during leukocyte
extravasation. Curr. Opin. Cell Biol. 14:587–593.
Wang, Q., G.R. Pfeiffer II, and W.A. Gaarde. 2003. Activation of SRC tyro-
sine kinases in response to ICAM-1 ligation in pulmonary microvascular
endothelial cells. J. Biol. Chem. 278:47731–47743.
Weed, S.A., and J.T. Parsons. 2001. Cortactin: coupling membrane dynamics to
cortical actin assembly. Oncogene. 20:6418–6434.
Wennerberg, K., S.M. Ellerbroek, R.Y. Liu, A.E. Karnoub, K. Burridge, and C.J.
Der. 2002. RhoG signals in parallel with Rac1 and Cdc42. J. Biol. Chem.
277:47810–47817.
Wojciak-Stothard, B., L. Williams, and A.J. Ridley. 1999. Monocyte adhesion
and spreading on human endothelial cells is dependent on Rho-regulated
receptor clustering. J. Cell Biol. 145:1293–1307.
Worthylake, R.A., S. Lemoine, J.M. Watson, and K. Burridge. 2001. RhoA is
required for monocyte tail retraction during transendothelial migration.
J. Cell Biol. 154:147–160.
Yang, L., J.R. Kowalski, X. Zhan, S.M. Thomas, and F.W. Luscinskas. 2006a.
Endothelial cell cortactin phosphorylation by Src contributes to poly-
morphonuclear leukocyte transmigration in vitro. Circ. Res. 98:394–402.
Yang, L., J.R. Kowalski, P. Yacono, M. Bajmoczi, S.K. Shaw, R.M. Froio, D.E.
Golan, S.M. Thomas, and F.W. Luscinskas. 2006b. Endothelial cell cor-
tactin coordinates intercellular adhesion molecule-1 clustering and actin
cytoskeleton remodeling during polymorphonuclear leukocyte adhesion
and transmigration. J. Immunol. 177:6440–6449.