THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 284, NO. 38, pp.

2595325961, September 18, 2009

2009 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

Intersectin-2L Regulates Caveola Endocytosis Secondary to

Cdc42-mediated Actin Polymerization*
Received for publication, June 18, 2009 Published, JBC Papers in Press, July 21, 2009, DOI 10.1074/jbc.M109.035071

Irene K. Klein, Dan N. Predescu, Tiffany Sharma, Ivana Knezevic, Asrar B. Malik, and Sanda Predescu1
From the Department of Pharmacology, Rush University Medical Center, and the Department of Pharmacology, University of
Illinois, Chicago, Illinois 60612
Here we addressed the role of intersectin-2L (ITSN-2L), a
guanine nucleotide exchange factor for the Rho GTPase Cdc42,
in the mechanism of caveola endocytosis in endothelial cells
(ECs). Immunoprecipitation and co-localization studies showed
that ITSN-2L associates with members of the Cdc42-WASpArp2/3 actin polymerization pathway. Expression of Dbl
homology-pleckstrin homology (DH-PH) region of ITSN-2L
(DH-PHITSN-2L) induced specific activation of Cdc42, resulting
in formation of extensive filopodia, enhanced cortical actin,
as well as a shift from G-actin to F-actin. The catalytically
dead DH-PH domain reversed these effects and induced significant stress fiber formation, without a detectable shift in
actin pools. A biotin assay for caveola internalization indicated a significant decrease in the uptake of biotinylated proteins in DH-PHITSN-2L-transfected cells compared with control and 1 M jasplakinolide-treated cells. ECs depleted of
ITSN-2L by small interfering RNA, however, showed
decreased Cdc42 activation and actin remodeling similar to
the defective DH-PH, resulting in 62% increase in caveolamediated uptake compared with controls. Thus, ITSN-2L, a
guanine nucleotide exchange factor for Cdc42, regulates different steps of caveola endocytosis in ECs by controlling the
temporal and spatial actin polymerization and remodeling
sub-adjacent to the plasma membrane.

The polymerization of actin has a central role in clathrin- and

caveola-mediated endocytosis (1). Studies have shown a number of protein-protein interactions that suggest a functional
relationship between the actin cytoskeleton and endocytic
machinery; however, the underlying mechanisms remain
unclear. ITSN-2L,2 a multifunctional domain protein with two
Epsin 15 homology domains, a central coiled-coil region followed by five consecutive Src homology 3 domains, a Dbl
homology (DH), a pleckstrin homology (PH), and finally a C2

* This work was supported, in whole or in part, by National Institutes of Health

Grants R01 HL089462 (to S. P.) and HL007829 (to A. B. M.). This work was
supported by American Heart Association Grant SDG 0635175N (to S. P.).
To whom correspondence should be addressed: 1750 West Harrison St.,
Chicago, IL 60612. Fax: 312-942-0339; E-mail: sanda_predescu@rush.edu.
2
The abbreviations used are: ITSN-2L, intersectin-2L; EC, endothelial cell;
ELISA, enzyme-linked immunosorbent assay; Jasp, Jasplakinolide; pAb,
polyclonal antibody; PM, plasma membrane; cav-1, caveolin-1; IEJs, interendothelial junctions; TCR, T cell antigen receptor; N-WASP, neural Wiskott-Aldrich syndrome protein; GEF, guanine nucleotide exchange factor;
DH, Dbl homology; PH, pleckstrin homology; siRNA, small interfering RNA;
PFA, paraformaldehyde; RT, reverse transcription; F, forward; R, reverse; Ab,
antibody; GST, glutathione S-transferase.
1

SEPTEMBER 18, 2009 VOLUME 284 NUMBER 38

domain, interacts via the Src homology 3 region with the ubiquitously expressed neural Wiskott-Aldrich syndrome protein
(N-WASP) that stimulates actin nucleation through Arp 2/3
complex activation (2). ITSN-2L interaction with N-WASP in
turn induces activation of N-WASP in a Cdc42-dependent
manner (2, 3). In this way, ITSN-2L on the basis of its DH
domain acts as a GEF for the small Rho GTPase Cdc42, similar
to its neuronal counterpart ITSN-1L (2, 4). The DH domain of
ITSN-2L shows high sequence homology with the corresponding region of ITSN-1L (5), and it possesses all the amino acid
residues required for its GEF enzymatic activity (6). Both long
ITSN isoforms display immediately distal to the DH domain a
PH domain, which may thereby modulate the intrinsic catalytic
activity of the DH region (6 8). It has been shown that the PH
domain enhances up to 100-fold the DH catalytic activity for
some Dbl proteins compared with that measured for DH alone
in vitro, whereas for other Dbl proteins the presence of the PH
domain negatively regulates GEF activity of the DH region (9).
This latter function is apparently mediated by interactions with
phosphoinositides (7, 9). However, the PH sequence was shown
to be dispensable for GEF activity of ITSN-2L in vitro, but it
enhanced the ability to activate Cdc42 in vivo (9). Despite high
sequence conservation among Rho GTPases, long ITSN isoforms apparently induce selective activation of Cdc42 due to
the overall increasing size of the specificity residues of the
GTPases (Cdc42 Rac1 RhoA) and the inability of ITSN to
accommodate in an analogous position the larger size amino
acid chains found in Rac1 and RhoA (10).
ITSN-2L, like its alternatively spliced short isoform, is widely
expressed in human tissue, and it shows subcellular distribution similar to components of the endocytic machinery (5).
In COS-7 cells overexpressing ITSN-2 isoforms, clathrinmediated transferrin uptake was blocked, consistent with
their involvement in the regulation of clathrin-mediated
endocytosis (5). By contrast, ITSN-2L overexpression in Jurkat cells stimulated T cell antigen receptor (TCR) internalization, whereas truncated ITSN-2L, deleted for the DH
domain, caused significant inhibition of TCR internalization
(2). The stimulatory effect of ITSN-2L on TCR endocytosis
may be secondary to the ability of ITSN-2L to bind through
its Src homology 3 domains the proline-rich domain of
N-WASP followed by Cdc42-mediated actin polymerization
(2). Although more work is needed to clarify these inconsistencies, both of these studies suggest that ITSN-2L may regulate endocytosis and function cooperatively with N-WASP
and Cdc42 to link WASP-mediated actin polymerization to
the endocytic machinery (2).
JOURNAL OF BIOLOGICAL CHEMISTRY

25953

ITSN-2L and Caveola Endocytosis

Live cultured fibroblast imaging showed that actin polymerization as regulated by the WASP-Arp2/3 complex participates
in the late stage of clathrin-mediated endocytosis (11). Therefore, we reasoned that ITSN-2L, as a specific activator of Cdc42,
may be essential for actin cytoskeleton polymerization and
caveola internalization in ECs. ECs are particularly rich in
caveola, and caveola-mediated endocytosis is a fundamental
step in mediating the transcytosis of proteins (12, 13), but the
mechanisms of caveola-mediated endocytosis and the essential
proteins involved remain enigmatic. In this study, we addressed
the role of ITSN-2L in the mechanism of caveola internalization in ECs. Our data employing morphological, biochemical,
and functional approaches show that ITSN-2L on the basis of
Cdc42-mediated spatial actin polymerization is required in the
mechanism of caveola-mediated endocytosis.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection
Human lung microvascular ECs (Clonetics) were cultured
with EGMTM-2 MV (Cambrex, Walkersville, MD). FuGENE
HD transfection reagent (Roche Applied Science) was used for
Myc-His-tagged DH-PHITSN-2L and catalytically dead ITSN2L-DH-PH1339 1347(DH-PH-(1339 1347)) transfection of ECs per the manufacturers instructions. Preliminary
experiments were carried out to establish optimized transfection and cell culture conditions. For specific silencing of
ITSN-2L isoform, custom-generated Dharmacon SMARTpool
siRNA reagents were used. The individual siGENOME duplex
most efficient in knocking down ITSN-2L protein expression,
GCACGGAUUCCUCUUCAAUUU (sense sequence) and
5-PAUUGAAGAGGAAUCCGGCUU (antisense), was delivered using Dharmafect2 transfection reagent. RNase-free conditions were used throughout the silencing experiments. All
controls for efficient transfection and to evaluate any off-target
effects caused by ITSN-2L siRNA in ECs were performed as in
Ref. 14. The Dharmacon siCONTROLTM functional, nontargeting siRNA sequence 1 (5-UAGCGACUAAACACAUCAAUU-3) and sequence 2 (5-UAAGGCUAUGAAGAGAUACUU-3) were used as controls for secondary effects caused
by ITSN-2L siRNA transfection. These two controls interact
with the RNA-induced silencing complex but lack sufficient
homology with any known human gene to effectively induce
mRNA knockdown. siGlo cyclophilin B siRNA, a silencer with a
fluorescent modification, was used to monitor transfection
efficiency.
Generation of DNA Constructs
Myc-His-ITSN-2L-DH-PH ConstructFull-length human
ITSN-2L cDNA was used as a template for PCR with DNA
polymerase Pfu (Stratagene, La Jolla, CA) to generate C-terminal Myc-His-tagged DH-PH domain (residues 12121534).
ITSN-2L cDNA was kindly provided by Dr. Suzana de la Luna
(Medical and Molecular Genetics Center, IRO, Barcelona,
Spain). Primer pair ITSN2L-DH-F, 5-AGTAGAATTCGCCACCATGGCATATATTCATGAGCTGATTCAGAC-3, and
ITSN2L-DH-R, 5-ACTTATCTAGACTCAGACGCCGCCTTGATCTTC-3, were used to generate the DH-PH domain of
ITSN-2L that lacks the stop codon. The PCR cycling conditions

25954 JOURNAL OF BIOLOGICAL CHEMISTRY

were as follows: initial denaturation 98 C for 30 s, 30 cycles of

denaturation, annealing, and extension; 94 C for 3 min, 25
cycles of denaturation, annealing, and extension; 94 C for 30 s,
58 C for 30 s, and 72 C for 1 min 15 s, followed by final extension 72 C for 7 min. The resulting PCR fragment was digested
with restriction enzymes EcoRI and XbaI (Invitrogen) and
subcloned into the EcoRI-XbaI restriction sites of
pcDNA6/Myc-His A (Invitrogen), resulting in the construct
pcDNA6A-ITSN2L-DH-PH, which was verified for sequence
integrity through DNA sequencing.
Myc-His-ITSN-2L-DH-PH-AA1339 1347 Construct (DHPH-(1339 1347))Full-length human ITSN-2 cDNA was
used as a template for PCR with High Fidelity DNA polymerase
Phusion (Finnzymes Oy, Espoo, Finland) to generate C-terminal Myc-His-tagged DH-PH domain (residues 12121534)
with residues 1339 1347 deleted. The deleted residues corresponded to the following amino acid sequence GMPLSSFLL.
To delete this region, a two-step PCR method was used. In the
first step, two separate reactions with primer pairs ITSN2LDH-F and ITSN2L-del AA1339 47-R or ITSN2L-del
AA1339 47-F and ITSN2L-DH-R were used to generate PCR
fragments that flanked the deleted region. The primers used
contained overlapping sequence that flanked the deleted residues. The resulting PCR fragments were analyzed by gel, purified, and used as overlapping DNA templates for the 2nd PCR
step with primers ITSN2L-DH-F and ITSN2L-DH-R. The
sequence of the primer pair that flanked the deleted region were
as follows: ITSN2L-del AA1339 47-F, 5-CGGTGTAAAAAACCCATGCAGAGGATCAC-3, and ITSN2L-del AA1339
47-R, 5-CATGGGTTTTTTACACCGCGGGTCAGATGCCAG-3 (the line between the base pairs in the primers
indicates the deleted region). The PCR cycling conditions were
as follows: initial denaturation 98 C for 30 s, 30 cycles of
denaturation, annealing, and extension; 98 C for 10 s, 67 C
for 30 s, and 72 C for 15 s, followed by a final extension 72 C
for 10 min. The final PCR product, which lacked the stop
codon, was digested with EcoRI and XbaI (New England Biolabs, Ipswich, MA) and ligated into the EcoRI and XbaI sites
of pcDNA6/Myc-His A (Invitrogen), resulting in the construct pcDNA6A-ITSN2L-DH-PH-AA1339 1347, which
was verified for sequence integrity through DNA sequencing. Cytomegalovirus-driven dominant active Cdc42 (V12)
was a gift from Dr. Tatyana Voyno-Yasenetskaya (University
of Illinois, Chicago).
RT-PCR
RNA was isolated from mouse lung or cultured ECs using
the RNeasy mini RNA isolation kit (Qiagen). RT-PCR was
performed using ThermoScript Two-step RT-PCR kit
(Invitrogen).
Immunoprecipitation
Cells were lysed in 20 mM Hepes-KOH, pH 7.4, 1.5% Triton
X-100, 100 mM KCl, 2 mM dithiothreitol, 2 mM EDTA, and
protease inhibitors (Roche Applied Science) for 1 h and 30 min
at room temperature. Protein content was determined using
MicroBCA assay (Pierce). 500 g of total protein were then
incubated with 2 g of a control nonspecific IgG, ITSN-2L or
VOLUME 284 NUMBER 38 SEPTEMBER 18, 2009

ITSN-2L and Caveola Endocytosis

FIGURE 1. ITSN-2L expressed in ECs interacts with protein members of N-WASP-Arp2/3 actin polymerization pathway. A, Western blotting using anti-ITSN pAb81177 shows expression of short and long ITSN isoforms in cultured lung ECs. Mr, molecular weight protein standards. B, RT-PCR shows relative amounts of mRNA
for ITSN isoforms in ECs. C, representative micrographs of ECs immunostained with anti-ITSN-2L-specific pAb
show the subcellular distribution of ITSN-2L (panel c1) and in detail the punctate staining pattern at the PM
level (panel c2, arrows) and in the cytosol (panel c3). D, immunoprecipitation with anti-ITSN pAb81177 revealed
ITSN (lane c) interactions with Cdc42 (lane d), Arp2/3 complex (lane e), and N-WASP (lane f). Two nonspecific
Abs, anti-rabbit IgG (lane a) and anti-rat IgG (lane b), did not immunoprecipitate these proteins. Immunoprecipitation using N-WASP pAb demonstrated similar interactions, ITSN isoforms (lane g), Cdc42 (lane h), Arp 2/3
(lane i) and N-WASP (lane j). Bars, 20 m (panel c1); 10 m (panels c2 and c3).

N-WASP pAb (Cytoskeleton, Denver, CO), for 1 h at 4 C. 40 l

of protein A/G bead slurry were then added, and the mixtures
were incubated end-over-end at 4 C overnight. The bound
proteins were resolved on 520% SDS-PAGE. For identifying
the immunoprecipitated proteins by Western blotting, primary
Abs were used in conjunction with the appropriate IgG TrueBlot horseradish peroxidase-conjugated reporters (EBioscience, San Diego), designed for detection of the immunoblotted target proteins without detection of light and heavy
chains of the immunoprecipitating immunoglobulin.

and collected into 1.5-ml Eppendorf

tubes. Cell pellets were then resuspended in 100 l of Lysis Buffer I (20
mM Hepes-NaOH, pH 7.2, 100 mM
NaCl, 1 mM sodium orthovanadate,
50 mM NaF, 1% Triton X-100, 1
protease inhibitors) for 1 h and then
centrifuged for 20 min at 10,000 g.
Supernatants were saved as the
G-actin fractions, and the pellets
were resuspended in Lysis Buffer II
(15 mM Hepes-NaOH, pH 7.5, 0.15
mM NaCl, 1% Triton X-100, 1%
sodium deoxycholate, 0.1% SDS, 10
mM EDTA, 1 mM dithiothreitol, 1
mM sodium orthovanadate, protease inhibitors) for 1 h. The mixtures
were then centrifuged at 45,000 rpm
in a Beckman OptimaTM TLX ultracentrifuge for 40 min. The ensuing
supernatants represented the F-actin fractions. Equal protein amounts
were loaded onto 4 12% SDSPAGE. Actin was detected by
immunoblotting with anti-actin Ab
(Cytoskeleton) and quantified by
densitometry using ImageJ1.37v
software.
Cdc42/Rac1/RhoA Activation Assay

Assays were performed as

instructed using Cdc42 activation
assay biochem kit (Cytoskeleton, Denver, CO). Briefly, cells
were scraped with ice-cold lysis buffer and centrifuged to
remove cell debris. Cell lysates were then incubated with 20 ml
of PAK-GST beads for Cdc42/Rac or 50 ml of Rhotekin-GST
beads for Rho for 90 min at 4 C. Beads were washed three times
with wash buffer. Beads were boiled for 5 min in 20 l of Laemmli buffer; samples were run in parallel with total cell lysate
and immunoblotted with Abs to the appropriate small GTPase.
Activation of Cdc42, Rac1, and RhoA in control and transfected
cells was determined by densitometry, with total lysate used to
normalize the total amount of small GTPase protein.

Immunofluorescence
ECs grown on glass coverslips or PM patches were immunostained as in Ref. 15. For phalloidin Alexa Fluor 488 staining,
control or transfected cells were fixed in 3.7% paraformaldehyde (PFA) in phosphate-buffered saline for 15 min at room
temperature, permeabilized in 0.1% Triton X-100 for 5 min on
ice, and stained with 25 g/ml phalloidin Alexa Fluor 488 for 30
min at room temperature. Cells were mounted with Prolong
Antifade reagent (Molecular Probes). Images were acquired on
a Zeiss Axioplan2 with Neofluor 100 objective (1.3 N.A.),
using the AxioCam with AxioVision 4.6 software.
Actin Polymerization Assay
Analysis of actin polymerization was performed as in Ref. 16.
Briefly, ECs were washed in phosphate-buffered saline, scraped,
SEPTEMBER 18, 2009 VOLUME 284 NUMBER 38

RESULTS
ITSN-2L Is Expressed in ECs and Interacts with Protein Components of the Cdc42/N-WASP/Arp2/3 Actin Polymerization
PathwayWe first investigated the expression and subcellular
localization of ITSN-2L in cultured ECs. Western blot of EC
lysates using a rabbit polyclonal Ab (anti-ITSN pAb81177) raised
against a GST fusion protein comprising the N-terminal 440
amino acids of human ITSN-1 (as in Ref. 14) showed two bands
corresponding to the short (140 kDa) and long (190 kDa) ITSN
isoforms (Fig. 1A). As immunoblot analysis did not distinguish
between splice variants of ITSN-1 and ITSN-2, we used RTPCR with specific primers for ITSN-1s, ITSN-1L, ITSN-2s, and
ITSN-2L to investigate the presence of mRNA for the two
ITSN-2 isoforms. Both ITSN-2s and ITSN-2L mRNA were
JOURNAL OF BIOLOGICAL CHEMISTRY

25955

ITSN-2L and Caveola Endocytosis

anti-ITSN pAb81177 for co-immunoprecipitation studies. Anti-ITSN
pAb co-immunoprecipitated endothelial ITSN isoforms (Fig. 1D, lane
c) and additional proteins were
identified by immunoblotting as
Cdc42 (lane d), Arp2/3 (lane e), and
N-WASP (lane f). Control nonspecific rabbit IgG and rat IgG did not
bring down these proteins (Fig. 1D,
lanes a and b). The same proteinprotein interactions were detected
when anti-N-WASP pAb was used
for immunoprecipitation (Fig. 1D).
Next, we analyzed by double immunofluorescence on PM patches the
co-localization of ITSN-2L with
cav-1, N-WASP, and Arp2/3. Endothelial PM patches bearing attached
caveolae allowed us to determine
with high resolution the distribution of these protein pairs on the
PM, the site of caveola endocytosis.
PM patches were generated by brief
sonication of endothelial monolayers grown on poly-lysine-coated
coverslips as described previously
(21). As shown in Fig. 2A, freshly
prepared PM sheets, fixed with 4%
PFA and immunostained with antiITSN-2L pAb, showed the nano-domain distribution similar to cav-1, a
FIGURE 2. ITSN-2L co-localizes with protein members of Cdc42/N-WASP-Arp2/3 complex actin polymer- reliable marker for the caveola
ization pathway. A, PM patches immunostained with ITSN-2L pAb show bright fluorescent puncta. Double
fluorescent staining with cav-1, N-WASP, and Arp2/3 Abs indicated partial co-localization. Higher magnifica- microdomains on the PM sheets
tion of the boxed areas shows a pool of PM-associated ITSN-2L co-localizing with cav-1 and significant co- (21). The overlapped images show
localization of ITSN-2L with WASP and Arp2/3 complex. B, representative PM patch triple fluorescently stained limited cav-1/ITSN-2L co-localizafor cav-1, N-WASP, and Arp2/3 complex were viewed sequentially through optical filter sets for the fluorophores used, and the resulting images were overlapped. Co-localization was scored as follows: for cav-1/Arp- tion, suggesting that only part of the
2/3 (yellow, panels b1 b3), for cav-1/N-WASP (purple, panels b.4 b6), and for cav-1/Arp2/3/N-WASP (white, PM-associated ITSN-2L is localized
panel b7). Co-localization of the three proteins was also considered when cav-1-positive puncta overlapped on
either side of the Arp2/3/N-WASP immunoreactive spots (panel b8). The extensive overlapping detected on the in caveolae. Similar punctate patedge of the PM patch was not used for quantification. All images used for quantification were acquired using tern and a greater degree of co-loidentical parameters per experiment. Bars, 5 m.
calization with ITSN-2L were
revealed by anti-N-WASP and antidetected in ECs (Fig. 1B). In addition, RT-PCR showed mRNA Cdc42 Abs (Fig. 2, merged panels). To determine the spatial
for ITSN-1s but not for the neuron-specific ITSN-1L (17). The coordination between cav-1 and protein members of the actin
specific primer for ITSN-1L detected this isoform in mouse polymerization pathway, we performed tri-color imaging (Fig.
brain lysate (data not shown). Immunofluorescence staining 2B) of PM patches as described previously (21). Morphometric
using an effective anti-ITSN-2L pAb (Santa Cruz Biotechnol- analysis of overlapped images indicated a similar degree of assoogy, Inc.) revealed a diffuse staining and strong numerous ciation between N-WASP/cav-1 (13 4%) and Arp3/cav-1
puncta throughout the cytosol (Fig. 1C, panels c1 and c3) and at (15 4%). In 22 7% (mean S.D., n 10 PM patches) of
the PM level (Fig. 1C, panel c2, arrows). This observation sug- co-localization events, all three proteins were found together.
gests that ITSN-2L, a protein without a transmembrane About 50% of cav-1 positive puncta did not co-localize with
domain, is present not only in the EC cytosol but also associated either N-WASP or Arp-2 or it was not possible to determine
their co-localization. Fig. 2B, panels b1 b8, shows representawith vesicular structures.
Since in other cell types ITSN-2L has been linked with tive images used to score for co-localization of cav-1 with
N-WASP regulators of actin cytoskeleton, we investigated if in Arp2/3 (panels b1 and b3), with N-WASP (panels b4 b6), and
ECs ITSN-2L interacts with essential proteins for actin polym- with both Arp2/3 and N-WASP (panels b7 and b8). Co-localerization and actin-mediated mobility, such as N-WASP, ization of the three proteins was also considered to have
Arp2/3, and Cdc42 (18 20). We used total ECs lysates and occurred when cav-1-positive puncta overlapped on either side

25956 JOURNAL OF BIOLOGICAL CHEMISTRY

VOLUME 284 NUMBER 38 SEPTEMBER 18, 2009

ITSN-2L and Caveola Endocytosis

activation and actin polymerization.
Although the PH domain does not
have an effect on the DH function, it
may play an important role in its
proper location (4). Ectopically
expressed
DH-PHITSN-2L
was
detected by fluorescence microscopy and in total ECs lysates with
anti-Myc mAb, at 48 h post-transfection (Fig. 3A). Morphometric
analysis performed on transfected
EC monolayers immunostained
with Myc mAb followed by antimouse IgG Alexa Fluor 488-conjugated indicated an average of 75
5% transfection efficiency. To determine the ability of DH-PHITSN-2L to
activate Cdc42, control and transfected ECs were subjected to GST
pulldowns using the Pak-PBD binding domain, which only interacts
with active Cdc42 and Rac1. Similarly, for RhoA activation, GST pulldown was carried out using rhotekin-GST beads. Expression of
DH-PHITSN-2L in ECs increased
Cdc42 activity 71%, compared with
controls, but did not affect Rac1 or
RhoA activity (Fig. 3, B and C).
Given that ITSN-2L activity
FIGURE 3. DH-PHITSN-2L expression selectively activates Cdc42 in ECs. A, expression of myc-DH-PHITSN-2L in
toward
Cdc42 via the DHECs was monitored by immunofluorescence and Western blot with anti-Myc mAb. B, DH-PHITSN-2L expression
stimulated formation of active Cdc42-GTP. No activation of Rac1 or RhoA was detected. Data are representative PHITSN-2L may spatially regulate
of three independent experiments. C, quantitative analysis of Cdc42, Rac1, and RhoA activation by densitom- actin polymerization, we next anaetry. D, phalloidin-488 staining of PFA-fixed control, DH-PHITSN-2L, and DA-Cdc42 transfected ECs to visualize
polymerized actin. Control cells exhibit polygonal shape, intact junctions, cortical actin band, and some stress lyzed the effects of DH-PHITSN-2L
fibers. DH-PHITSN-2L-transfected ECs demonstrate formation of filopodia (panels d2 and d4, arrows) and signif- on actin organization, using phalloiicant increase of cortical actin (panels d2 and d4, arrowheads), as well as inter-endothelial junctional gap for- din-Alexa 488 labeling. Control ECs
mation (panel d2, asterisk). The cytoskeletal changes seen in DH-PHITSN-2L-transfected cells are reminiscent of
cytoskeletal changes seen upon transfection with DA-Cdc42 (panel d3). E, biochemical and densitometric displayed a polygonal shape with
analysis of G- and F-actin in control and transfected ECs as well as 1 M Jasp-treated ECs. Data were collected actin microfilament bundles formfrom 3 to 4 different experiments. Bars, 10 m.
ing the dense peripheral band and
some stress fibers (Fig. 3D, panel
of the Arp2/3/N-WASP immunoreactive spots (Fig. 2B, panel d1), whereas transfected cells exhibited increased polymerized
b8). These observations indicate that a fraction of the PM-lo- actin seen as a wider cortical actin layer with accumulation of
cated ITSN-2L and proteins of the N-WASP-Arp2/3 complex F-actin (Fig. 3D, panel d2, arrowheads) and appearance of filopactin polymerization pathway are localized in close proximity odia (Fig. 3D, panels d2 and d4, arrows). We also observed
of the caveolar microdomains, suggesting that ITSN-2L-medi- appearance of gaps between ECs (Fig. 3D, panel d2, asterisk),
ated localized actin polymerization occurs in a spatially and suggesting that expression of DH-PHITSN-2L and resultant
temporally coordinated manner at the level of the endothelial Cdc42 activation and actin remodeling affect the integrity of
PM caveolae.
interendothelial junctions (IEJs). To substantiate that the
DH-PHITSN-2L Domain Specifically Activates Cdc42To recorded cytoskeletal changes are because of ITSN-2L-ininvestigate the role of ITSN-2L in caveola endocytosis, we duced Cdc42 activation, we transfected ECs with V12analyzed the morphological and functional effects of Cdc42, a dominant active form of Cdc42. Phalloidin-Alexa
DH-PHITSN-2L expression in ECs, a common approach used to 488 staining showed similar distribution of filamentous actin
assess functions of multimodular proteins (22, 23). This and disrupted IEJs (Fig. 3D, panel d3). Under the conditions
approach allowed us to investigate the effects exclusively of our experiment, the length of filopodia in ECs expressing
ascribed to the DH-PH domain, thus eliminating influences of the activated Cdc42 is significantly shorter and may be due
neighboring sequences. We transiently transfected subconflu- to the fact that Cdc42 activity induces activation of endogeent ECs with the Myc-tagged DH-PHITSN-2L, to assess the cat- nous Rac, which in turn activates Rho (24). These events do
alytic specificity of DH-PHITSN-2L in the mechanism of Cdc42 not occur in DH-PHITSN2L-transfected cells (Fig. 3C).
SEPTEMBER 18, 2009 VOLUME 284 NUMBER 38

JOURNAL OF BIOLOGICAL CHEMISTRY

25957

ITSN-2L and Caveola Endocytosis

FIGURE 4. DH-PH-(1339 1347) expression decreases Cdc42 activation

and causes enhanced stress fiber formation. A, DH-PH-(1339 1347)
expression inhibits formation of active Cdc42-GTP. B, quantitative analysis of
Cdc42 activation by densitometry. C, phalloidin-488 staining of PFA-fixed
DH-PH-(1339 1347)-transfected ECs to visualize polymerized actin (panel
c1). DH-PH-(1339 1347)-transfected ECs demonstrate enhanced formation
of stress fibers (panel c1 and c2), lack of cortical actin, as well as presence of
disrupted IEJs. D, morphometric analysis of the extent of stress fiber formation in DH-PH-(1339 1347)-transfected ECs. E and F, biochemical and densitometric analysis of G- and F-actin in control and transfected ECs as well as
DH-PH-(1339 1347)-transfected ECs. Data are representative of 3 4 independent experiments. Bar, 20 m (panel c1); 10 m (panel c2).

Biochemical analysis of changes in actin polymerization as

described previously (16) showed significant shift from G-actin
to F-actin in DH-PHITSN-2L-transfected cells by reference to
controls (Fig. 3E). As assessed by densitometry, in controls, the
amount of G-actin and F-actin was similar, whereas in
DH-PHITSN-2L-transfected cells the F-actin content was
increased (Fig. 3E). Together, these results support our hypothesis that ITSN-2L, by specifically activating Cdc42, stimulates
actin nucleation and polymerization through N-WASP, a substrate for activated Cdc42.
The specificity of DH-PH activity was also evaluated by
transfection of subconfluent EC monolayers with the
DH-PH-(1339 1347), an ITSN-2L construct comprising a
catalytically dead DH-PH region. The mutation involves several
amino acids essential for the catalytic activity (7, 10). Transfected ECs were then subjected to morphological and biochemical analyses to evaluate the status of cellular actin and activation of Cdc42. The defective DH-PH-(1339 1347) domain
suppressed Cdc42 activation (Fig. 4A and B) and filopodia formation (Fig. 4C). Infrequently, we still noticed short membrane
protrusions (Fig. 4C, panel c1). Transfected ECs became elon-

25958 JOURNAL OF BIOLOGICAL CHEMISTRY

gated (Fig. 4C, panel c1), and filamentous actin shifted from the
cortical pool to only stress fibers (Fig. 4C, panels c1 and c2).
Morphometric analysis indicated that about 70% of the ECs
immunoreactive to Myc Ab, and thus transfected with the catalytically dead DH-PH-(1339 1347) domain, show enhanced
stress fiber formation (Fig. 4D). Biochemical analysis of cellular
actin (Fig. 4E) and densitometric analysis (Fig. 4F) did not reveal
any significant change between F- and G-actin pools in
DH-PH-(1339 1347)-transfected cells by reference to controls. These results suggest that Cdc42 activity as regulated by
the DH-PH domain of ITSN-2L, a GEF for Cdc42, controls the
distribution of F-actin between cortical dense band and stress
fibers in ECs.
Increased Actin Polymerization Induced by the DH-PHITSN-2L
Domain Interferes with Caveola InternalizationThe functional effects of DH-PHITSN-2L expression on caveola internalization were studied in ECs transfected with the DH-PHITSN-2L,
at 48 h post-transfection. Caveola internalization, followed by
biochemical and morphological analyses were carried out as
described previously (15). Briefly, control and transfected cells
were subjected to biotinylation of cell surface proteins using a
cleavable biotin reagent. In ECs the internalization of biotinylated cell surface proteins is mediated primarily by caveolae
(15). After 30 min of internalization, biotinylated proteins still
on the cell surface were reduced with glutathione, a membraneimpermeant reducing reagent (25). For morphological analysis
by fluorescence microscopy, the internalized proteins were
detected by NeutrAvidin-Texas Red staining. Control cells
show a strong fine punctate staining throughout the cytoplasm
with significant accumulation of biotin in the perinuclear area
(Fig. 5A, a1 and a1.1). In contrast, DH-PHITSN-2L-transfected
ECs exhibited markedly limited staining with no perinuclear
accumulation (Fig. 5A, a2 and a2.1). Quantitative assessment of
internalized biotinylated proteins in control and transfected
ECs lysates by enzyme-linked immunosorbent assay (ELISA)
using streptavidin-horseradish peroxidase indicated that control cells internalized 19.2 1014 1.9 biotin molecules/mg of
total protein, whereas DH-PHITSN-2L-transfected ECs internalized 7.3 1014 0.9 biotin molecules/mg of total protein (Fig.
5B). The limited NeutrAvidin Texas Red staining and the 63%
decrease in the amount of internalized biotin in cells transfected with DH-PHITSN-2L by reference to controls indicate a
central role of actin nucleation and actin polymerization (subadjacent to the PM) induced by Cdc42 activation in regulating
caveola internalization. To substantiate this observation, we
used the cell-permeant cyclic peptide Jasp for inducing polymerization and stabilizing actin filaments (26). Although structural changes caused by 1 M Jasp treatment were distinct from
those induced by DH-PHITSN-2L (not shown) and the shift from
G-actin to F-actin was significantly greater (Fig. 3E), NeutrAvidin Texas Red staining was limited (Fig. 5A, a3 and a3.1) and
caveola internalization reduced by 20% (Fig. 5B). The smaller
reduction in caveola internalization following 1 M Jasp treatment by comparison with DH-PHITSN-2L-transfected cells suggests that overall augmented actin polymerization per se is not
responsible for inhibiting caveola internalization. Instead, temporal and spatial actin polymerization and remodeling sub-adVOLUME 284 NUMBER 38 SEPTEMBER 18, 2009

ITSN-2L and Caveola Endocytosis

FIGURE 5. Experimental manipulation of ITSN-2L protein expression causes actin cytoskeletal redistribution that interferes with caveola internalization. A, control (A, panels a1 and a1.1), DH-PHITSN-2L (A, panels a2 and a2.1), and 1 M Jasp-treated (A, panels a3 and a3.1) ECs subjected to internalization assay
followed by NeutrAvidin Texas Red staining show strong puncta throughout the cytoplasm, with prominent perinuclear accumulation (panels a1 and a1.1),
whereas DH-PHITSN-2L-transfected (panels a2 and a2.1) or 1 M Jasp-treated cells (panels a3 and a3.1) displayed limited staining and no perinuclear accumulation of biotin. B, number of biotin molecules in EC lysates from control, DH-PHITSN-2L-transfected cells, Jasp-treated ECs, and ITSN-2L siRNA was determined by
ELISA in 3 4 experiments for each experimental condition. Data are calculated as number of biotin molecules/mg of total protein/min and plotted as
percentage from control. Bar, S.D. C, Western blot analysis of ITSN isoforms protein expression using anti-ITSN pAb81177 in controls and ITSN-2L siRNA-treated
cells; MW, molecular weight protein standards. D, densitometric analysis of ITSN immunoreactivity in controls and siRNA-transfected ECs. E, RT-PCR analysis of
mRNA levels for ITSN isoforms in controls (lane a), control siRNA (lane b), and ITSN-2L siRNA-transfected ECs (lane c). F, control (panel f1) and siRNA ITSN-2Ltransfected ECs (panel f2) were subjected to internalization assay. ECs depleted of ITSN-2L (panel f2) show strong punctate staining similar to controls (panel f1).
Because of excessive perinuclear accumulation of biotin-labeled proteins in ECs deficient in ITSN-2L, both control and ITSN-2L siRNA-transfected ECs were
subjected to only 15 min of internalization. Bars, 10 m (panels a1, a2, and a3); 5 m (panels a1.1, a2.1, and a3.1); 10 m (panels f1 and f2).

jacent to the PM induced by DH-PHITSN-2L are crucial in inhibiting caveola endocytosis.

ITSN-2L-deficient ECs Show Prominent Changes in F-actin
Distribution and Increase in Caveola InternalizationTo
examine the consequences of ITSN-2L deficiency in caveola
endocytosis, we applied an siRNA approach. We used Dharmacon custom SMARTpool reagents comprising four different
siRNA oligonucleotides designed for specific and efficient
down-regulation of ITSN-2L gene expression in ECs. Efficient
down-regulation was achieved at 72 h post-siRNA transfection,
as detected by Western blot analysis of total lysates prepared
from control and siRNA-transfected cells (Fig. 5C). Densitometric analysis of immunoreactive bands (Fig. 5D) indicated
about 75 85% decrease of ITSN-2L protein expression compared with nontransfected ECs and ECs transfected with the
siCONTROLTM functional and nontargeting siRNA sequences.
ITSN-2L siRNA oligonucleotide 1 was the most efficient and
was used in all studies performed. RT-PCR analysis of RNA
extracted from ECs 72 h post-siRNA treatment, as described
previously (14), also indicated a specific silencing of ITSN-2L
gene expression (Fig. 5E, lane c). ITSN-2L siRNA transfection
SEPTEMBER 18, 2009 VOLUME 284 NUMBER 38

down-regulated ITSN-2L but had no effect on mRNA of the

other two ITSN isoforms expressed by lung ECs. Fluorescence
microscopy analysis of ECs monolayers transfected with 100 nM
siGlo cyclophilin B siRNA showed numerous cells with perinuclear punctate fluorescent staining at 48 h post-transfection,
indicative of high efficiency of siRNA transfection (data not
shown).
Next, we applied the biotin assay for caveola internalization
to both control (Fig. 5F, panel f1) and ECs depleted of ITSN-2L
(Fig. 5F, panel f2) as described above. Biotin-NeutrAvidin staining indicated a strong punctate pattern throughout the cell
under both experimental conditions. Quantification by ELISA
of the biotin amounts internalized by ITSN-2L siRNA-transfected cells indicated a 62% increase in internalization of biotinylated cell surface proteins relative to controls (Fig. 5B).
To gain insights into the functional effects of ITSN-2L deficiency, we analyzed the status of cellular actin in control (Fig.
6A, panel a1) and ITSN-2L siRNA-transfected ECs (Fig. 6A,
panels a2 and a3) by phalloidin-Alexa 488 labeling. Down-regulation of ITSN-2L led to marked changes in EC shape and
F-actin distribution. ITSN-2L-depleted cells lost their polygoJOURNAL OF BIOLOGICAL CHEMISTRY

25959

ITSN-2L and Caveola Endocytosis

functional effects of DH-PHITSN-2L
expression in ECs, a common
approach to assess the function of
multimodular proteins (22, 23). The
approach allowed us to investigate
the effects exclusively ascribed to
the DH-PH domain, thus eliminating the possible influence or regulatory effects of the neighboring
sequences. Under conditions of
adequate expression and subcellular
localization, the DH-PH fragment
promoted specific activation of
Cdc42 and augmented actin polymerization. The DH-PHITSN-2Ltransfected cells displayed a wider
peripheral actin band, numerous
filopodia-like structures, and disruption of IEJs. Moreover, the trafficking of caveolae is inhibited indicating that the wider and perhaps
the denser cortical network of actin
functions as a molecular barrier,
preventing the inward movement of
caveolae. The transport of caveolae
away from the PM may require
changes in F-actin, with local actin
remodeling and possibly depolyFIGURE 6. ITSN-2L depletion causes F-actin re-distribution and decreases Cdc42 activation. A, phalloidinmerization to facilitate caveola
488 staining of PFA-fixed control (panel a1) and ITSN-2L siRNA-transfected ECs (panel a2). Control cells exhibit
polygonal shape, intact junctions, cortical actin band (arrows), and some stress fibers (arrowheads), whereas movement. Consistent with this
transfected cells are elongated and have disorganized cortical actin and enhanced stress fiber formation idea, stabilization of actin filaments
(panels a2 and a3). B, GST pulldowns for Cdc42, Rac1, and RhoA activation followed by Western blotting.
C, densitometry quantification shows a decrease of Cdc42 activation upon depletion of ITSN-2L. Bars, 50 m by treatment of ECs with Jasp, a
(panels a1 and a2) 10 m (panel a3).
cyclic peptide that binds to and stabilizes filamentous actin in vitro and
enhances actin polymerization in
nal shape and became elongated with numerous intercellular vivo (27), may explain the observed inhibition of caveola intergaps. The dense peripheral actin bands were disorganized and nalization. Furthermore, because the PM of all cells is linked to
disrupted. Dense organized network of stress fibers present an underlying actin cytoskeleton, augmented actin polymerizathroughout the cytoplasm and aligned with the long axis of the tion at this level may cause increased rigidity of the PM that may
cell became the most prominent actin structure (Fig. 6A, panel affect caveola formation. These observations strongly suggest
a3). Similar shift from cortical actin to stress fibers due to lack that actin polymerization processes must be the subject of an
of Cdc42 activity and some filopodia formation were seen in the especially fine temporal and spatial coordination to facilitate
caveola endocytosis.
DH-PH-(1339 1347)-transfected ECs.
We also show that siRNA-mediated ITSN-2L knockdown
To determine whether actin cytoskeletal changes were associated with activation of small Rho GTPases, we performed causes a significant decrease in Cdc42 activation that coincides
activity assays for Cdc42, Rac1, and Rho (Fig. 6B). Densitomet- with significant remodeling of actin cytoskeletal organization.
ric analysis of blots from three different experiments indicated Interestingly, disorganization of the dense peripheral actin
that depletion of ITSN-2L caused a quantitative decrease in band triggered the increase of caveola internalization. ITSN-2L
Cdc42 activation (Fig. 6C). However, the relative activities of deficiency removed the cortical actin network assumed to conRac1 and RhoA remained the same. Our findings showing that trol caveola trafficking, and thus led to the recorded up-reguladisorganization of cortical actin network correlated with abun- tion of caveola internalization. Moreover, ITSN-2L deficiency
dant and thicker stress fibers and with no shift between F-and may cause weakening of the PM-associated actin cytoskeleton
G-actin support the hypothesis that ITSN-2L controls the and less membrane rigidity favoring caveola formation. These
results suggest that in ECs ITSN-2L is required for the organilocalized architectural organization of actin filaments in ECs.
zation of cortical actin as peripheral dense band and actin
DISCUSSION
membrane cytoskeleton. Interestingly, the catalytically dead
To investigate the functional importance of ITSN-2L in DH-PH domain, which compromised significantly activation of
caveola endocytosis, we first analyzed the morphological and Cdc42 in DH-PH-(1339 1347)-transfected cells, caused a

25960 JOURNAL OF BIOLOGICAL CHEMISTRY

VOLUME 284 NUMBER 38 SEPTEMBER 18, 2009

ITSN-2L and Caveola Endocytosis

dramatic shift of F-actin pools from cortical actin band to stress
fibers. The disorganization of cortical actin network correlated
with abundant and thicker stress fibers, and no shift between
F-actin and G-actin strongly suggests that ITSN-2L, as a GEF
for Cdc42, may control the cellular distribution and architectural organization of actin filaments in different subcellular
scaffolds. Whereas the effect of ITSN-2L siRNA likely reflected
a selective cytoskeletal action, it would have been too naive to
consider that the ITSN-2L deficiency would be limited to
cytoskeletal modification. ITSN-2L, similar to its neuronal
counterpart, may bind a plethora of regulatory molecules,
interactions that may affect the efficacy of caveola internalization. ITSN-2L may bind dynamin that interacts with the actin
machinery via cortactin and syndapin (28 30). Both dynamin
and N-WASP interact via their proline-rich domains with subsets of the Src homology 3 domains of ITSN-2L (2), an observation that raises the possibility of competitive binding and
thereby, the possibility of ITSN-2L as a molecular control for
both caveola endocytic function and actin polymerization.
Alternatively, ITSN-2L may hetero-oligomerize via the central
coiled-coil region with the other two endothelial ITSN isoforms
and work synergistically by controlling the recruitment of
dynamin at the neck region of caveolae. Because ITSN isoforms
and their splice variants have a distinct expression pattern (5,
17), they may also play isoform-specific functions, which are
still to be determined.
We conclude that changes in actin polymerization, as regulated by experimental manipulation of ITSN-2L protein
expression, may compromise caveola endocytosis, indicating
that efficient caveola endocytosis requires a finely tuned coupling of caveola endocytic function and actin polymerization
machinery. ITSN-2L is a universally expressed GEF for Cdc42
and displays some distinctiveness that makes it an ideal candidate to functionally link Cdc42 activation and subsequent actin
polymerization to caveola endocytosis.
AcknowledgmentsWe are grateful to Dr. Susana de la Luna (Medical and Molecular Genetics Center, IRO, Barcelona, Spain) for providing ITSN-2L cDNA and Dr. Tatyana Voyno-Yasenetskaya (University of Illinois, Chicago) for the V12-Cdc42 plasmid.
REFERENCES
1. Kaksonen, M., Toret, C. P., and Drubin, D. G. (2006) Nat. Rev. Mol. Cell
Biol. 7, 404 414
2. McGavin, M. K., Badour, K., Hardy, L. A., Kubiseski, T. J., Zhang, J., and

SEPTEMBER 18, 2009 VOLUME 284 NUMBER 38

Siminovitch, K. A. (2001) J. Exp. Med. 194, 17771787

3. Hussain, N. K., Jenna, S., Glogauer, M., Quinn, C. C., Wasiak, S., Guipponi,
M., Antonarakis, S. E., Kay, B. K., Stossel, T. P., Lamarche-Vane, N., and
McPherson, P. S. (2001) Nat. Cell Biol. 3, 927932
4. Malacombe, M., Ceridono, M., Calco, V., Chasserot-Golaz, S., McPherson, P. S., Bader, M. F., and Gasman, S. (2006) EMBO J. 25, 3494 3503
5. Pucharcos, C., Estivill, X., and de la Luna, S. (2000) FEBS Lett. 478, 4351
6. Whitehead, I. P., Campbell, S., Rossman, K. L., and Der, C. J. (1997) Biochim. Biophys. Acta 1332, F1F23
7. Liu, X., Wang, H., Eberstadt, M., Schnuchel, A., Olejniczak, E. T., Meadows, R. P., Schkeryantz, J. M., Janowick, D. A., Harlan, J. E., Harris, E. A.,
Staunton, D. E., and Fesik, S. W. (1998) Cell 95, 269 277
8. Cerione, R. A., and Zheng, Y. (1996) Curr. Opin. Cell Biol. 8, 216 222
9. Pruitt, W. M., Karnoub, A. E., Rakauskas, A. C., Guipponi, M., Antonarakis, S. E., Kurakin, A., Kay, B. K., Sondek, J., Siderovski, D. P., and Der, C. J.
(2003) Biochim. Biophys. Acta 1640, 61 68
10. Snyder, J. T., Worthylake, D. K., Rossman, K. L., Betts, L., Pruitt, W. M.,
Siderovski, D. P., Der, C. J., and Sondek, J. (2002) Nat. Struct. Biol. 9,
468 475
11. Merrifield, C. J., Qualmann, B., Kessels, M. M., and Almers, W. (2004) Eur.
J. Cell Biol. 83, 1318
12. Tuma, P. L., and Hubbard, A. L. (2003) Physiol. Rev. 83, 871932
13. Palade, G. E. (1953) J. Histochem. Cytochem. 1, 188 211
14. Predescu, S. A., Predescu, D. N., Knezevic, I., Klein, I. K., and Malik, A. B.
(2007) J. Biol. Chem. 282, 17166 17178
15. Predescu, S. A., Predescu, D. N., Timblin, B. K., Stan, R. V., and Malik, A. B.
(2003) Mol. Biol. Cell 14, 49975010
16. Patterson, R. L., van Rossum, D. B., and Gill, D. L. (1999) Cell 98,
487 499
17. Pucharcos, C., Casas, C., Nadal, M., Estivill, X., and de la Luna, S. (2001)
Biochim. Biophys. Acta 1521, 111
18. Bayless, K. J., and Davis, G. E. (2002) J. Cell Sci. 115, 11231136
19. Hufner, K., Higgs, H. N., Pollard, T. D., Jacobi, C., Aepfelbacher, M., and
Linder, S. (2001) J. Biol. Chem. 276, 3576135767
20. Moreau, V., Tatin, F., Varon, C., and Genot, E. (2003) Mol. Cell. Biol. 23,
6809 6822
21. Predescu, S. A., Predescu, D. N., Shimizu, K., Klein, I. K., and Malik, A. B.
(2005) J. Biol. Chem. 280, 37130 37138
22. Tong, X. K., Hussain, N. K., de Heuvel, E., Kurakin, A., Abi-Jaoude, E.,
Quinn, C. C., Olson, M. F., Marais, R., Baranes, D., Kay, B. K., and McPherson, P. S. (2000) EMBO J. 19, 12631271
23. Schafer, D. A. (2002) Curr. Opin. Cell Biol. 14, 76 81
24. Nobes, C. D., and Hall, A. (1995) Cell 81, 53 62
25. Bretscher, M. S., and Lutter, R. (1988) EMBO J. 7, 4087 4092
26. Bubb, M. R., Senderowicz, A. M., Sausville, E. A., Duncan, K. L., and Korn,
E. D. (1994) J. Biol. Chem. 269, 14869 14871
27. Sun, X., and Whittaker, G. R. (2007) Cell. Microbiol. 9, 16721682
28. Merrifield, C. J., Perrais, D., and Zenisek, D. (2005) Cell 121, 593 606
29. Kessels, M. M., and Qualmann, B. (2004) J. Cell Sci. 117, 30773086
30. Zhu, J., Zhou, K., Hao, J. J., Liu, J., Smith, N., and Zhan, X. (2005) J. Cell Sci.
118, 807 817

JOURNAL OF BIOLOGICAL CHEMISTRY

25961