MOLECULAR AND CELLULAR BIOLOGY,
Copyright © 1999, American Society for Microbiology. All Rights Reserved.
Mar. 1999, p. 1784–1799 Vol. 19, No. 3
The Gab1 PH Domain Is Required for Localization of Gab1 at
Sites of Cell-Cell Contact and Epithelial Morphogenesis
Downstream from the Met Receptor Tyrosine Kinase
CHRISTIANE R. MAROUN,1MARINA HOLGADO-MADRUGA,2ISABELLE ROYAL,1
MONICA A. NAUJOKAS,1TANYA M. FOURNIER,3ALBERT J. WONG,2,4AND MORAG PARK1,3,5*
Departments of Medicine,1Oncology,5and Biochemistry,3Molecular Oncology Group, Royal Victoria Hospital,
McGill University, Montreal, Quebec, Canada H3A 1A1, and Departments of Microbiology and Immunology2
and Pharmacology,4Kimmel Cancer Institute, Philadelphia, Pennsylvania 19107
Received 10 August 1998/Returned for modification 23 September 1998/Accepted 30 November 1998
Stimulation of the hepatocyte growth factor (HGF) receptor tyrosine kinase, Met, induces mitogenesis,
motility, invasion, and branching tubulogenesis of epithelial and endothelial cell lines in culture. We have
previously shown that Gab1 is the major phosphorylated protein following stimulation of the Met receptor in
epithelial cells that undergo a morphogenic program in response to HGF. Gab1 is a member of the family of
IRS-1-like multisubstrate docking proteins and, like IRS-1, contains an amino-terminal pleckstrin homology
domain, in addition to multiple tyrosine residues that are potential binding sites for proteins that contain SH2
or PTB domains. Following stimulation of epithelial cells with HGF, Gab1 associates with phosphatidylinositol
3-kinase and the tyrosine phosphatase SHP2. Met receptor mutants that are impaired in their association with
Gab1 fail to induce branching tubulogenesis. Overexpression of Gab1 rescues the Met-dependent tubulogenic
response in these cell lines. The ability of Gab1 to promote tubulogenesis is dependent on its pleckstrin
homology domain. Whereas the wild-type Gab1 protein is localized to areas of cell-cell contact, a Gab1 protein
lacking the pleckstrin homology domain is localized predominantly in the cytoplasm. Localization of Gab1 to
areas of cell-cell contact is inhibited by LY294002, demonstrating that phosphatidylinositol 3-kinase activity is
required. These data show that Gab1 is an important mediator of branching tubulogenesis downstream from
the Met receptor and identify phosphatidylinositol 3-kinase and the Gab1 pleckstrin homology domain as
crucial for subcellular localization of Gab1 and biological responses.
Hepatocyte growth factor/scatter factor (HGF) is a mesen-
chymally derived factor that stimulates a wide variety of cellu-
lar responses through activation of the Met receptor tyrosine
kinase. HGF is a potent mitogen for primary hepatocytes and
renal tubule cells (29, 43, 77), stimulates epithelial cell disso-
ciation and invasion, and acts as an initiating signal for an
intrinsic cellular morphogenic program of kidney, breast, and
lung epithelium grown in matrix cultures (41, 59). In vivo, HGF
is a potent angiogenic factor (21) and is involved in organ
regeneration (39) as well as tumorigenesis (4, 53, 58, 64).
Moreover, recent studies have demonstrated a role for Met
and HGF in the development of the liver and placenta, the
development and innervation of skeletal muscle, and in direct-
ing of the growth of axonal cones (9, 37, 57, 65, 74).
Using receptor chimeras, we and others have demonstrated
that the Met cytoplasmic domain is sufficient to mediate the
pleiotropic biological responses attributed to HGF in epithelial
cells (32, 70, 80) and that these events require Met protein
tyrosine kinase activity (70, 79). Phosphorylated tyrosine resi-
dues in the noncatalytic cytoplasmic domains of receptor ty-
rosine kinases act as specific binding sites for Src homology 2
(SH2) and phosphotyrosine binding domain-containing pro-
teins, and these in turn transduce intracellular signals (re-
viewed in reference 47). While signaling pathways downstream
from receptor tyrosine kinases involved in a mitogenic re-
sponse have been characterized in detail, until recently little
was known about the signaling pathways involved in cell dis-
sociation, motility, and morphogenesis. Toward this end, the
characterization of signaling pathways downstream from the
Met receptor has been essential.
Upon stimulation with HGF, the Met receptor cytoplasmic
domain becomes highly phosphorylated on tyrosine residues
(52, 79). Structure-function analyses have shown that two ty-
rosine residues within the carboxyl terminus (Y1349 and
Y1356), which are highly conserved between other members of
the Met receptor tyrosine kinase gene family, Sea and Ron
(54), are crucial for cell scatter and branching morphogenesis
in Madin-Darby canine kidney (MDCK) epithelial cells (32,
70, 79, 80). Tyrosine 1356 forms a multisubstrate binding site,
coupling the Met receptor with the Grb2 and Shc adapter
proteins, the p85 subunit of phosphatidylinositol 3-kinase
(PI3K), phospholipase C?1 (PLC?1), and the phosphatase
SHP2 (11, 13, 15, 16, 48, 79).
From a search for Met-specific substrates that could be im-
plicated in branching morphogenesis, we have recently identified
the Grb2-associated binder 1 (Gab1) as the major phosphor-
ylated protein in epithelial cells that undergo a morphogenic
program in response to HGF (44). Gab1 was initially identified
in a library screen as a Grb2 binding protein and is phosphor-
ylated downstream from the epidermal growth factor receptor,
the insulin receptor, and the TrkA receptor (24, 25). More
recently, it has been shown that interleukin-3 (IL-3), IL-6, and
alpha and gamma interferons also induce the tyrosine phos-
phorylation of Gab1 (62). Gab1 is a member of the IRS-1
family of multisubstrate binding proteins, which includes
IRS-1, IRS-2, p62dok, and DOS (6, 23, 49, 73, 75). While these
proteins lack enzymatic activities, they are thought to function
* Corresponding author. Mailing address: Molecular Oncology
Group, Royal Victoria Hospital, 687 Pine Ave. West, Rm H5.10,
Montreal, Quebec, Canada H3A 1A1. Phone: (514) 842-1231 ext.
5845. Fax: (514) 843-1478. E-mail: firstname.lastname@example.org.
as multisubstrate docking proteins by virtue of their ability to
associate with multiple signaling molecules.
Murine Gab1 contains eighteen tyrosine residues, some of
which, if phosphorylated, provide potential binding sites for
SH2 or PTB domain-containing proteins (24, 25, 62). In addi-
tion, Gab1 contains several proline rich regions which could
interact with SH3 domain-containing proteins. The greatest
homology observed with IRS-1 family members lies within the
N terminus of Gab1, which contains a pleckstrin homology
(PH) domain, suggesting a conserved role for the PH domain
within these proteins (5, 24). While IRS-1 contains a phospho-
tyrosine binding domain involved in its recruitment to the
insulin receptor, Gab1 lacks such a domain. In vivo, Gab1 is
thought to be recruited to the Met receptor predominantly
indirectly, via the Grb2 adapter protein, through the interac-
tion with the carboxy-terminal SH3 domain of Grb2 (3, 14, 44)
and association of the Grb2 SH2 domain with Y1356 of the
multisubstrate binding site in the Met receptor. In addition, a
direct interaction between Gab1 and Y1349 in the Met recep-
tor was observed in the yeast two-hybrid system (69) and oc-
curs to a lesser extent in vivo (44). This requires a proline-rich
domain in Gab1, defined as the Met binding domain (69).
The observations that overexpression of Gab1 in neuronal
cells promotes cell survival downstream from the TrkA recep-
tor and that overexpression of Gab1 in epithelial cells pro-
motes a morphogenesis program support a role for Gab1 as an
important signaling molecule in multiple cell types (25, 69).
Interestingly, MDCK cells expressing Met receptor mutants
that fail to associate with Grb2 are unable to form branching
tubules upon Met activation, suggesting that Grb2-dependent
signaling pathways are involved in the morphogenic activities
of HGF (16). We show here that overexpression of Gab1
rescues the tubulogenesis defect in cells expressing these Met
mutants, consistent with the function of Gab1 lying down-
stream from Grb2. Importantly, this provides an assay system
to investigate the structural domains of Gab1 required for
Met-dependent branching tubulogenesis. From structure-func-
tion analyses, we have established that the PH domain of Gab1
is essential for the ability of Gab1 to support branching mor-
phogenesis downstream from the Met receptor tyrosine kinase.
We demonstrate that the PH domain is required to target
Gab1 to the proximity of the cellular membrane at sites of
cell-cell contact and that this localization is also dependent on
the activity of PI3K. This study provides evidence for a key role
for PI3K and the Gab1 PH domain in determining Gab1 cel-
lular localization and biological responses downstream from
the Met receptor.
MATERIALS AND METHODS
Cell culture and DNA transfections. MDCK cells were maintained in Dulbec-
co’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum
(FBS). The generation of MDCK cell lines expressing wild-type colony-stimu-
lating factor 1 (CSF)-Met receptor and mutants thereof by retroviral infection
has been described previously (16, 80). For the generation of stable cell lines
expressing wild-type and mutant HA-tagged Gab1, the Gab1 cDNA was cloned
into the pCDNA1.1 vector and cotransfected with a PLX SH vector, which
confers resistance to hygromycin, by the calcium phosphate method as described
elsewhere (72). Cell lines were selected in hygromycin (300 ?g/ml). For tran-
sient-transfection assays, 293T cells were seeded at 106/100-mm petri dish and
transfected 24 h later with 2 ?g of plasmid DNA encoding wild-type Gab1
without or with CSF-Met cDNA by the calcium phosphate precipitation method
(72). At 16 h later, the cells were washed twice in DMEM medium lacking FBS
and then cultured for another 48 h in medium containing 10% FBS; they were
Antibodies and reagents. Antibodies raised in rabbit against a C-terminal
peptide of human Met were used (51). Anti-p85 was kindly provided by T.
Pawson, Mount Sinai Hospital, University of Toronto. Antiphosphotyrosine
(4G10) was obtained from Upstate Biotechnology Inc., Lake Placid, N.Y. An-
ti-HA (HA.11) was purchased from BABCO, Richmond, Calif. Anti-py PY20
was purchased from Transduction Laboratories. Anti-SHP2 was kindly provided
by G.-S. Feng, Indiana University School of Medicine, and anti-E-cadherin (3G8)
was provided by M. Pasdar, University of Alberta. The PI3K inhibitor LY294002
and the PLC inhibitor U73122 were purchased from Biomol, Plymouth Meeting,
Pa. The p85 and p110 constructs were obtained from A. Klippel (31). CY3-
conjugated goat anti-mouse immunoglobulin h (IgG) was purchased from Jack-
son ImmunoResearch Laboratories, Inc. The generation of Gab1?PI3K mutant
protein was described elsewhere (25). The Gab1?PH domain mutant encom-
passes amino acids 116 to 695 of the murine Gab1 cDNA (24). It was constructed
by performing PCR on full-length Gab1 cDNA by using a 5? primer, GTGGG
ATCCTCGGATTCAATCCCACAGAAGAA, and a 3? primer, CGAATTCAC
TTCACATTCTTGG. The PCR product was cloned into the BamHI and EcoRI
sites of pcDNA1.1 downstream from an in-frame HA tag.
HGF stimulation of MDCK cell lines expressing wild-type and mutant Gab1.
Cells were seeded at 106per 100-mm dish. At 24 h later, they were washed once
with DMEM and then starved for 24 h in 10 ml of DMEM containing 0.02%
FBS. HGF was added at 100 U/ml in 2 ml for the indicated times. The cells were
immediately lysed in 1 ml of lysis buffer (50 mM HEPES [pH 7.4], 150 mM NaCl,
10% glycerol, 0.1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 1 ?g
each of leupeptin and aprotinin per ml, 1 mM Na3VO4).
Immunoprecipitations and Western blotting. MDCK cell lysates (2 mg of total
protein) or 293T cell lysates (50 ?g) were incubated with antibodies as indicated
in the figures for 1 h at 4°C with gentle rotation. A 20-?l volume of a 50% slurry
of either protein A or protein G-Sepharose was added for an additional 1 h to
collect immune complexes. Following three washes in lysis buffer, proteins were
resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-
PAGE) and transferred to a nitrocellulose membrane. The membranes were
blocked for 1 h with 3% bovine serum albumin in TBST (10 mM Tris-HCl [pH
7.4], 2.5 mM EDTA, 150 mM NaCl, 0.1% Tween 20) and then for 1 h with
primary antibody (1:1,000). Following five washes in TBST, the proteins were
revealed with secondary anti-mouse (Jackson ImmunoResearch Laboratories,
Inc.) or protein A (Gibco) conjugated to horseradish peroxidase. The proteins
were visualized with an enhanced chemiluminescence detection system (Amer-
PI3K assay. The PI3K assay was performed as previously described (15).
Briefly, 3 mg of 293T cell lysates was subjected to immunoprecipitation with
either PY20, anti-Met, or anti-HA for 1 h at 4°C. Immune complexes were
collected with protein A-Sepharose for the first two antibodies and protein
G-Sepharose for the last antibody. Immunoprecipitates were washed three times
with lysis buffer, once with phosphate-buffered saline (PBS), once with 100 mM
Tris-HCl (pH 7.5)–0.5 M LiCl, once with distilled H2O, once with 20 mM
Tris-HCl (pH 7.5)–100 mM NaCl–1 mM EDTA, and once with kinase buffer (20
mM Tris-HCl [pH 7.5], 100 mM NaCl, 0.5 mM EGTA). The beads were incu-
bated for 10 min at room temperature with 50 ?l of kinase buffer containing
phosphatidylinositol (0.2 mg/ml). Then 20 ?Ci of [?-32P]ATP and 20 mM MgCl2
were added for 10 min at room temperature. The reactions were stopped upon
the addition of 150 ?l of chloroform–methanol–11.6 M HCl (50:100:1). The
lipids were extracted with 100 ?l of chloroform, and the organic phase was
washed as described elsewhere (55), resuspended in 15 ?l of chloroform, spotted
on a silica gel 60 thin-layer chromatography plate (Merck), and resolved in
chloroform–methanol–28% ammonium hydroxide–water (86:76:10:14) for 1 h.
Phosphorylated phospholipids were visualized following autoradiography and
quantitated with a Fuji Bas 1000 phosphorimage analyzer.
Collagen assays. The ability of MDCK cells to form branching tubules was
assayed as previously described (79). Briefly, 5 ? 103cells were resuspended in
500 ?l of collagen solution (Vitrogen 100 [Celtrix]) prepared as specified by the
manufacturer and layered over 350 ?l of the collagen solution in a 24-well plate.
The cells were maintained in Liebowitz medium containing 5% FBS and allowed
to form cysts for 5 to 7 days. For stimulations, HGF or recombinant human CSF
(rhCSF-1) (5 U/ml) (kindly provided by Genetics Institute, Boston, Mass.) was
added to the Liebowitz medium containing 5% FBS. Tubules were apparent by
light microscopy 5 to 10 days after the addition of stimuli. The medium was
changed every 4 days, and photographs were taken 14 to 20 days later on Kodak
TMY400 films at a magnification of ?10. For the quantitation of the morpho-
genic response, 60 colonies in each of six independent cultures (wells) were
scored for their ability to form branching tubules (structures whose length is five
times their diameter), and the results were plotted as the average number of cysts
able to undergo tubulogenesis per culture per 100.
Immunofluorescence. MDCK cells overexpressing wild-type Gab1 or the Gab1
mutants were plated for the indicated times on glass coverslips (Bellco Glass
Inc.) in a 24-well dish (Nunc) in DMEM containing 10% FBS. For stimulations,
104MDCK cells overexpressing wild-type Gab1 or Gab1?PH mutant protein
were plated overnight in 10% serum-containing medium and 50 U of HGF per
ml was added for 15 min at 37°C. The cells were fixed in 2% paraformaldehyde
in PBS for 30 min at room temperature, washed twice in PBS, and incubated for
10 min in PBS containing 50 mM ammonium chloride. Following one additional
wash in PBS, the cells were treated for 10 min at room temperature with PBS
containing 0.1% Triton X-100 and 5% FBS (buffer A). Anti-HA (1:300 in buffer
A) was then added to the cells, and after three washes in the same buffer,
CY3-conjugated goat anti-mouse IgG (1:2,000) was added for 10 min, and the
cells were given three washes in buffer A. For CSK treatments, cells were
incubated for 10 min at room temperature in a buffer containing 10 mM piper-
VOL. 19, 1999Gab1 PH DOMAIN IN Gab1 LOCALIZATION AND FUNCTION1785
azine-N,N?-bis(2-ethanesulfonic acid) (PIPES, pH 7.0), 300 mM sucrose, 50 mM
NaCl, 3 mM MgCl2, 0.5% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 1
?g each of leupeptin and aprotinin per ml, and 1 mM Na3VO4. Following two
washes in PBS, the cells were fixed and labeled with anti-HA in buffer A by the
method described above. To induce the disruption of cell-cell adhesion, cells
were cultured for 3 days in medium containing 10% FBS, washed once in PBS,
incubated for 2 h in medium containing 5 mM EGTA as described by Takaishi
et al. (63), and then fixed and labeled as described above. For the treatments with
LY294002 and U73122, cells were incubated for 2 h at 37°C in 50 ?M LY294002
or 2 ?M U73122, washed twice, fixed in 2% paraformaldehyde in PBS, and
labeled as described above. The glass coverslips were mounted onto slides in
Immunofluore medium (ICN) and visualized with a Nikon Labophot-2 epifluo-
rescence microscope. Photographs were taken with Kodak TMZ3200 film.
pEGFP Gab1 and microinjections. Wild-type Gab1, the Gab1?PH mutant,
and the Gab1PH domain (amino acids 1 to 116) cDNAs were subcloned as
BamHI-EcoRI fragments into the BglII-EcoRI sites in the multiple-cloning site
of pEGFP-C2 (Clontech), downstream of GFP. Plasmids expressing these fusion
proteins were microinjected at 50 ?g of DNA per ml into nuclei of MDCK cells
by using an Eppendorf microinjector. At 2 h after injection, the cells were fixed
in 2% paraformaldehyde in PBS and visualized as described above. For the
injection of constructs expressing the p110 and p85 subunits of PI3K (31),
MDCK cells were serum starved in medium containing 0.02% FBS for 24 h prior
to the microinjections. These constructs were microinjected into the nuclei of
cells at 100 ?g/ml. The pEGFP-Gab1 construct was comicroinjected where
indicated in the figures, and cells were visualized 4 h after the injection.
Gab1 associates with multiple substrates downstream from
the Met receptor tyrosine kinase. We have previously demon-
strated that Gab1 is the predominant protein phosphorylated
following stimulation of the Met receptor in epithelial cells
(44). Gab1 associates with the p85 subunit of PI3K and the
phosphatase SHP2 downstream from the insulin, EGF, and
TrkA receptors (24, 25). To assess whether these signaling
proteins associate with Gab1 following Met activation, we have
generated MDCK epithelial cell lines stably expressing HA
epitope-tagged wild-type Gab1 and assayed the ability of HA-
Gab1 to associate with p85 and SHP2 in a Met-dependent
fashion. Stimulation of HA-Gab1-expressing cell lines with
HGF resulted in an increase in the phosphorylation of HA-
Gab1 (Fig. 1A). Importantly, immunoprecipitation of HA-
Gab1 followed by Western blotting with antibodies specific for
SHP2 or the p85 subunit of PI3K revealed that HGF induced
an increase in the association of these proteins with Gab1 (Fig.
1A). Therefore, in a similar manner to the insulin, EGF, and
TrkA receptors, tyrosine phosphorylation of Gab1 downstream
from the Met receptor was accompanied by an increase in the
ability of Gab1 to associate with the p85 subunit of PI3K and
with SHP2. These results indicate that Gab1 functions as a
multisubstrate docking protein coupling the Met receptor with
multiple signaling pathways in epithelial cells.
Binding of the p85 subunit of PI3K with IRS-1 correlated
with activation of PI3K downstream from the insulin receptor
(2, 42). Similarly, Gab1 was associated with activation of PI3K
downstream from the TrkA receptor in PC12 cells (25). PI3K
is activated downstream from the Met receptor (15, 55). How-
ever, although the Met receptor and oncoprotein can bind p85
directly, low levels of PI3K activity are coimmunoprecipitated
with anti-Met compared to the levels in coimmunoprecipita-
tions with antiphosphotyrosine (anti-PY) (15a). To establish if
Met-stimulated PI3K activity is associated with Gab1, 293T
cells were transiently transfected with HA-Gab1 alone, Met
alone, or HA-Gab1 and Met together (Fig. 1B). Five indepen-
dent transfections were performed, and the results of a repre-
sentative experiment are shown in Fig. 1B through D. Over-
expression of Met results in the induction of Met activation
and phosphorylation in the absence of ligand (44), and West-
ern blot analysis of anti-HA-Gab1 immunoprecipitates with
anti-PY revealed robust tyrosine phosphorylation of Gab1 only
in cells coexpressing Met (Fig. 1C). Significantly, in cells coex-
pressing HA-Gab1 and Met, anti-HA-Gab1 immunoprecipi-
tates contained 50% of the PI3K activity observed in anti-PY
immunoprecipitates whereas low levels of PI3K activity (5%)
were observed in anti-Met immunoprecipitates (Fig. 1C).
Thus, a significant portion of PI3K activated downstream from
the Met receptor is associated with Gab1 (Fig. 1D).
Overexpression of Gab1 rescues the branching tubulogen-
esis defect of MDCK cells expressing mutant Met receptors.
Structure-function studies with chimeric Met receptors con-
taining the extracellular domain of the CSF-1 receptor fused to
the transmembrane and cytoplasmic domains of Met revealed
that Met receptor mutants with impaired Gab1 association fail
to induce branching tubules following stimulation of the Met
receptor (16, 79). To examine the biological function of Gab1,
we have overexpressed HA-tagged Gab1 in cell lines express-
ing wild-type CSF-Met receptors that scatter and form branch-
ing tubules in response to CSF (80) and in cells expressing
CSF-Met receptor mutants that fail to form branching tubules
in response to CSF, including CSF-Met receptor multisub-
strate binding-site mutants (Y1356F and Y1349/1356F) that
have lost their ability to bind to multiple cellular substrates (16,
79) and a CSF-Met receptor mutant (N1358H), that is specif-
ically unable to bind Grb2 (16).
Twenty independent clones of each HA-Gab1-transfected
cell line were selected, and five showing similar levels of Gab1
expression were characterized in detail and assayed for branch-
ing tubulogenesis. The level of expression of Gab1 is shown for
two representative clones of each cell line (Fig. 2A), and the
tubulogenic response is shown for one clone (Fig. 2C). The
tubulogenic response was quantitated in Fig. 2B, where the
response to HGF is an indication of the total number of cysts
capable of undergoing tubulogenesis. In cells expressing wild-
type CSF-Met and Gab1, no branching tubules were observed
in the absence of Met stimulation (Fig. 2C, clone 4). These
cells formed cysts, as did parental cells, when grown in a col-
lagen matrix, and they formed branching tubules (structures
whose length is five times their width) only when the endoge-
nous Met or the chimeric CSF-Met protein was stimulated
with either HGF or CSF-1, respectively (Fig. 2B and C, clone
4). Importantly, in cells expressing the CSF-Met multisubstrate
binding-site mutant (Y1356F), which fail to form tubules in
response to CSF-1 (Fig. 2B and C, Y1356F ? vector), overex-
pression of Gab1 rescued the tubulogenesis defect in the pres-
ence of CSF-1 in the five clones analyzed (Fig. 2B and C, clone
7). Moreover, overexpression of Gab1 rescued the branching
tubulogenesis defect in 10 clones from two independent
MDCK cell lines, expressing a CSF-Met receptor mutant that
fails to bind only Grb2 (Fig. 2B and C, N1358H). In each case,
overexpression of Gab1 rescued tubule formation in 75% of
the cysts that were able to undergo tubulogenesis (50 to 60% of
preformed cysts responded to CSF-1 by forming tubules, com-
pared with 70 to 80% in response to HGF). However, cells
expressing vector alone were unable to form tubules in re-
sponse to CSF-1 (Fig. 2B and C, N1358H ? vector). These
results demonstrate that overexpression of Gab1 was sufficient
to complement the branching tubulogenesis defect of cells
expressing Met receptor mutants that fail to bind Grb2 and, as
a consequence, show reduced ability to recruit Gab1 (44). In
contrast, overexpression of Gab1 failed to rescue the branching
tubulogenesis defect of cells expressing a CSF-Met Y1349/
1356F mutant that fails to bind Gab1 (Fig. 2B and C, compare
Y1349/1356F Vector with clone 3) demonstrating that Y1349
contributed to the ability of Gab1 to rescue tubulogenesis in
the Y1356F or N1358H CSF-Met mutants.
Gab1 is a substrate for the EGF receptor but fails to pro-
mote EGF-dependent epithelial branching tubulogenesis.
1786MAROUN ET AL. MOL. CELL. BIOL.
Overexpression of Gab1 promoted branching tubulogenesis
induced through the HGF receptor, Met (Fig. 2). MDCK cells
have abundant EGF receptors, and EGF induces tyrosine
phosphorylation of Gab1 and its association with PI3K and
SHP2 (24) but does not induce morphogenesis in MDCK cells.
Therefore, we investigated whether EGF stimulation could
induce branching tubulogenesis in MDCK cells overexpressing
Gab1. As shown in Fig. 3A, while HGF stimulated the forma-
tion of branching tubules, EGF did not. To investigate whether
this could be related to qualitative or quantitative differences
in the phosphorylation of Gab1, in response to HGF or EGF,
MDCK cells overexpressing Gab1 were stimulated with either
growth factor for the time indicated (Fig. 3B). Gab1 phosphor-
ylation in response to HGF was observed as early as 1 min
following stimulation, reached a maximum at 15 to 30 min, was
sustained for 1 h, and then returned to baseline within 2 h. In
contrast, EGF-mediated Gab1 phosphorylation increased and
then declined rapidly, reaching baseline within 15 min of stim-
ulation (Fig. 3B). The distinct pattern of phosphorylation of
Gab1 downstream from HGF and EGF was not due to differ-
FIG. 1. Association of Gab1 with cellular substrates is dependent on Met-mediated tyrosine phosphorylation of Gab1. (A) Stable MDCK cell lines expressing
HA-Gab1 or vector control were serum starved in 0.02% FBS for 24 h and subsequently stimulated with HGF (100 U/ml) for 15 min. Cell lysates were subjected to
immunoprecipitation (ip) with anti-HA followed by blotting with anti-PY. Products of parallel precipitations were blotted with anti-p85, anti-SHP2, or anti-HA. (B)
293T cells were transiently transfected with plasmids encoding epitope-tagged wild-type Gab1, with or without a Met chimera composed of the extracellular domain
from CSF receptor fused to the Met transmembrane and intracellular domains. Cells were serum starved 24 h prior to harvest. Lysates were immunoprecipitated with
anti-Met, anti-PY (PY20), or anti-HA. PI3K activity was determined as described in Materials and Methods. Phosphatidylinositol phosphates were separated by
thin-layer chromatography on silica gel plates and revealed by autoradiography. The position of phosphatidylinositol 3-phosphate is shown (PIP), as is the origin (Ori).
(C) Lysates from 293T cells transfected as described in panel B were subjected to immunoprecipitation with anti-HA or anti-Met and blotted with anti-HA, anti-Met,
or anti-PY as indicated. (D) The incorporated radioactivity in the phosphatidylinositol 3-phosphate was quantitated with a Fujix BAS 1000 image analyzer, and the
results are plotted on the bar graph as relative PI3K activity.
VOL. 19, 1999Gab1 PH DOMAIN IN Gab1 LOCALIZATION AND FUNCTION1787
FIG. 2. Gab1 rescues branching tubulogenesis in MDCK cells expressing mutant CSF-Met receptors that fail to bind Grb2. MDCK cell lines expressing either
wild-type (WT) CSF-Met or the CSF-Met mutants Y1349/1356F, Y1356F, or N1358H were stably transfected with vector or wild-type HA-tagged Gab1. (A) Lysates
from two representative lines of each experimental group were subjected to immunoprecipitation with anti-HA, and proteins resolved by SDS-PAGE were transferred
to a nitrocellulose membrane and immunoblotted with anti-HA. (B) Quantitation of the tubulogenic response following stimulation with HGF and CSF in cell lines
1788MAROUN ET AL.MOL. CELL. BIOL.
ent levels of expression of HA-Gab1 in the different experi-
mental groups (Fig. 3B). Therefore, a sustained phosphoryla-
tion of Gab1 is observed downstream from HGF whereas only
transient phosphorylation of Gab1 is observed downstream
The PH domain of Gab1 is required for Met-dependent
branching tubulogenesis. To identify domains in Gab1 that are
critical for rescue of the tubulogenic response, mutant Gab1
proteins were expressed in the cell lines expressing the CSF-
Met N1358H mutant, which fails to bind Grb2 and fails to
expressing either CSF-Met or mutants thereof alone (open bars) or together with Gab1 (solid bars) was undertaken as described in Materials and Methods. The
responses are plotted as the percentage of cysts that have undergone branching tubulogenesis. The values were derived from three independent experiments done in
duplicate. None of the cysts formed tubules in the absence of stimulation. (C) Cell lines were grown in collagen for 5 days, during which they formed cysts. rhCSF-1
(5 U/ml) was added, and 14 days later branching tubules were visualized at a magnification of ?10 and photographs taken with Kodak TMY400 film. A representative
cell line for each group is shown.
FIG. 3. Overexpression of Gab1 does not mediate branching tubulogenesis downstream from EGF. (A) Control MDCK cells or MDCK cells overexpressing Gab1
were plated in a collagen matrix for 5 days. HGF (5 U/ml) or EGF (20 ng/ml) was added to the cultures, and photographs were taken 14 days later at a magnification
of ?10. (B) MDCK cells overexpressing Gab1 were serum starved for 24 h prior to stimulation with either 100 U of HGF per ml or 100 400 ng of EGF per ml for the
indicated time. Gab1 was immunoprecipitated (ip) with anti-HA. Proteins were resolved by SDS-PAGE, transferred to nitrocellulose, and probed with anti-PY or
anti-HA. WT, wild type.
VOL. 19, 1999Gab1 PH DOMAIN IN Gab1 LOCALIZATION AND FUNCTION 1789
induce tubule formation. Since a significant portion of the
Met-induced PI3K activity was associated with Gab1 (Fig. 1B)
and since PI3K is essential for cell dissociation and tubulogen-
esis (8, 55), we first examined a Gab1 mutant unable to bind
PI3K. Sequences downstream from tyrosines 447, 472, and 589
in the carboxy-terminal portion of Gab1 contain putative bind-
ing sites for SH2 domains of the p85 subunit of PI3K, and
substitution of these residues with phenylalanine residues re-
sults in the loss of p85 binding to Gab1 (25) (see Fig. 5B). To
investigate whether the ability of Gab1 to bind p85 is essential
for its ability to rescue tubulogenesis, MDCK cell lines express-
ing the CSF-Met N1358H Grb2-mutant were stably transfected
with the Gab1?PI3K mutant (Fig. 4). Five of seven indepen-
dent clones tested formed branching tubules in response to
CSF-1. A representative clone is shown in Fig. 4C (?PI3K-1).
Quantitation of the response in this clone revealed that while
60% of the preformed cysts were able to undergo morphogenic
changes, an extensive branching tubulogenic response, where
the individual tubule length was at least five times its width,
was observed with 20% of the cysts. Thus, while wild-type
Gab1 rescued tubulogenesis in all 10 cell lines tested, expres-
sion of the Gab1?PI3K mutant rescued to an intermediate
level, suggesting that although the ability of Gab1 to associate
with PI3K was not essential for the ability of Gab1 to rescue
tubulogenesis, it was required for efficient rescue.
PH domains are found in many proteins with a broad spec-
FIG. 4. The PH domain of Gab1 is essential for Met-mediated branching tubulogenesis. (A) MDCK cells expressing the N1358H CSF-Met mutant protein
[N1358H(17)] were transfected with plasmids encoding for Gab1 ?PI3K (?PI3K, clones 6 and 1). Two independent N1358H CSF-Met mutant expressing lines
[N1358H(17) and N1358H(1)] were transfected with Gab1 ?PH-encoding plasmids [?PH(1) clones 8 and 6, ?PH(17) clones 1 and 2]. Lysates were subjected to
immunoprecipitation (ip) and blotting with anti-HA. (B) The tubulogenic response was quantitated in cells expressing Gab1?PI3K and ?PH, and results from
representative clones are plotted as the percentage of cysts that have formed branching tubules in response to HGF or CSF-1. Solid bars represent results from cysts
that have undergone a complete tubulogenic response; i.e., the tubule length is at least five times the size of the width; the hatched bar represents a partial response.
(C) Cells expressing Gab1 mutants were plated in a collagen matrix and allowed to form cysts for 5 days. rh-CSF or HGF (5 U/ml) was added, and 14 days later
branching tubules were visualized by light microscopy at a magnification of ?10. Representative lines are shown.
1790MAROUN ET AL.MOL. CELL. BIOL.
trum of activities (reviewed in references 20, 22, and 34). De-
spite differences in the amino acid sequences of various PH
domains, they adopt a common fold composed of seven ?
sheets and one ? helix (34, 50). Evidence for in vitro binding to
membrane phospholipids has been presented for a number of
PH domains (17–19, 22, 35, 50, 56). This interaction has pro-
vided a possible mechanism through which these proteins
could be recruited to the membrane and/or activated. To study
the role of the PH domain in Gab1, a truncated Gab1 protein
that lacks this domain (Gab1?PH) was assayed for its ability to
rescue branching tubulogenesis in cell lines expressing the
CSF-Met N1358H Grb2-mutant receptor. Two independent
CSF-Met N1358H Grb2 mutant receptor-expressing cell lines,
N1358H(1) and N1358H(17) were transfected with the
Gab1?PH mutant protein. As shown in Fig. 4A, stable cell
lines expressed the Gab1?PH protein to high levels. Although
some change in the morphology of the cysts was occasionally
observed, in all 10 independent clones derived from each of the
two CSF-Met N1358H Grb2-mutant cell lines, the Gab1?PH
mutant was unable to promote branching tubulogenesis fol-
Gab1?PH-8 and Gab1?PH-1, are shown in Fig. 4B and C).
Furthermore, expression of Gab1?PH did not interfere with
the intrinsic ability of these cells to form branching tubules,
since stimulation of the endogenous, wild-type Met receptor
with HGF induced branching tubules (Fig. 4B and C). This
suggested that overexpression of the Gab1?PH protein did not
inhibit the tubulogenic response. Importantly, these results
indicate that Gab1 requires its PH domain to rescue branching
tubulogenesis downstream from the Met receptor in MDCK
Since the Gab1 mutant that lacks the PH domain failed to
rescue tubulogenesis in cell lines expressing the N1358H CSF-
Met mutant, we determined whether this reflected the inability
of the Gab1?PH mutant protein to be phosphorylated by
and/or be recruited to the CSF-Met receptor. The ability of
Gab1 to coimmunoprecipitate with CSF-Met or the CSF-
N1358H Met mutant was investigated following transient-
transfection assays in 293T cells. Wild-type CSF-Met coimmu-
noprecipitated with the Gab1?PH and Gab1?PI3K mutant
proteins as efficiently as with wild-type Gab1 (Fig. 5A). The
N1358H CSF-Met receptor mutant, as described previously,
associated less efficiently with Gab1 than did wild-type CSF-
Met (44). Importantly, both the Gab1?PI3K and Gab1?PH
mutants were comparable to wild-type Gab1 in their efficiency
to coimmunoprecipitate with N1358H CSF-Met (Fig. 5A).
Stripping of the blots and reblotting with anti-HA showed that
similar levels of Gab1 were expressed in the different experi-
mental groups (Fig. 5B).
To test the ability of the Gab1?PH mutant protein to be
phosphorylated, the CSF-Met N1358H cell lines expressing the
Gab1?PH mutant protein (Gab1?PH-8 and Gab1?PH-1 [Fig.
5B]) were stimulated with CSF-1 and the extent of tyrosine
phosphorylation of HA-Gab1?PH was determined by Western
blotting with anti-PY. Activation of CSF-Met resulted in an
increase in the level of phosphorylation of Gab1?PH protein,
and this increase was comparable to that observed for the
Gab1?PI3K mutant (Fig. 5B). Further, the phosphorylation
kinetics of Gab1?PH following Met activation (Fig. 5C) par-
alleled that observed for wild-type Gab1 (Fig. 3B). Phosphor-
ylation of Gab1?PH was observed within 1 min of stimulation
with HGF, and was maintained for 60 min, reaching baseline
2 h poststimulation (Fig. 5C and 3B). Moreover, as expected
from its phosphorylation, Gab1?PH coimmunoprecipitated
with the p85 subunit of PI3K and SHP2 (Fig. 5B). Taken
together, these results indicate that the association and the
phosphorylation of Gab1 with the Met receptor was not de-
pendent on the Gab1 PH domain. Further, the Gab1 PH do-
main is not required for its ability to associate with signaling
proteins (p85 and SHP2) following activation of the Met re-
Localization of Gab1 to sites of cell-cell contact requires its
PH domain. PH domains have been implicated in the recruit-
ment of multiple proteins to the plasma membrane (10, 40, 46,
68). Gab1 was previously shown to localize to sites of cell-cell
contact in epithelial cells (69). To investigate the cellular lo-
calization of Gab1 and Gab1 mutants, MDCK cells expressing
wild-type Gab1, or mutants thereof, were grown on glass cov-
erslips and subjected to indirect immunofluorescence with an-
ti-HA. Fluorescence microscopy revealed that in tight colonies
of MDCK cells, the majority of Gab1 was localized at the cell
periphery, more specifically at sites of cell-cell contacts, al-
though some Gab1 protein was detected in the cytoplasm (Fig.
6A). This is in accord with the results of Weidner et al. (69).
Importantly, the Gab1?PH mutant did not localize to areas of
cell-cell contact but instead localized in the cytoplasm (Fig.
6A). In contrast, the Gab1?PI3K mutant protein localized at
sites of cell-cell contact in a similar manner to wild-type Gab1
protein (Fig. 6A).
The PH domains of SOS, the guanine nucleotide exchange
factor for Ras, and PLC? are sufficient for membrane targeting
(7, 10), whereas deletion of the PH domain from PLC?1 re-
sulted in the localization of this protein to the cytoplasm (46).
To determine if the Gab1PH domain is sufficient for subcellu-
lar localization, we have generated vectors expressing the
pEGFP-Gab1PH domain, pEGFP-Gab1?PH, and pEGFP-
Gab1?PI3K fusion proteins. Microinjection of these vectors
into the nuclei of MDCK cells in colonies indicated that the
PH domain of Gab1 was sufficient for localization to cell-cell
contact sites (Fig. 6B). Taken together, these results indicate
that the PH domain of Gab1 directs the subcellular localization
of this protein to the cell periphery.
The localization of Gab1 to areas of cell-cell contact was
similar to that of cell-cell adhesion proteins such as E-cad-
herin. However, unlike E-cadherin, Gab1 protein at the
MDCK cell periphery was localized in a compartment that is
Triton X-100 soluble. Treatment of cell lines expressing wild-
type Gab1 prior to fixation with CSK buffer, which solubilizes
proteins not stably associated with the cytoskeleton, revealed
that E-cadherin remained insoluble and associated with the
cytoskeleton whereas Gab1 was not stably associated with the
cytoskeleton or a detergent insoluble compartment (Fig. 6C,
?CSK). Furthermore, when cells were grown in media con-
taining low Ca2?concentrations, a condition that causes cell
dissociation (45), E-cadherin relocalized to the cytoplasm
whereas the membrane-proximal localization of Gab1 was not
significantly altered, suggesting that at low extracellular Ca2?
concentrations, Gab1 was not coupled to E-cadherin (Fig. 6C,
Localization of Gab1 to sites of cell-cell contacts requires
PI3K activity. Increasing evidence suggests that PH domains
may bind membrane phospholipids in vitro (17–19, 22, 35, 50,
56). This is supported by the crystal structure of several PH
domains (12, 28, 78) and provides a molecular basis through
which PH domain-containing proteins could be targeted to
membranes. The recruitment of the PH domain of the Ras
exchange factor SOS to the cell periphery is regulated by se-
rum (7). Since Gab1 was expressed at the periphery when cells
were grown in 10% serum, we determined whether its local-
ization was serum dependent. MDCK cell colonies plated on
glass coverslips for 48 h were starved in 0.02% FBS for 24 h
and subsequently subjected to indirect immunofluorescence
VOL. 19, 1999Gab1 PH DOMAIN IN Gab1 LOCALIZATION AND FUNCTION1791
followed by microscopy. While Gab1 was located in a mem-
brane-proximal compartment in the presence of 10% serum,
serum depletion resulted in redistribution of Gab1 to the cy-
toplasm (Fig. 7A). Moreover, in cells cultured for 18 h, where
cells remained as single cells or small colonies, Gab1 was
localized in the cytoplasm even in the presence of high serum
concentrations (Fig. 7A).
Several PH domain-containing proteins can bind the prod-
ucts of PI3K or PLC? (17–22, 34, 50, 56). Therefore, it is
possible that serum-stimulated PI3K activity is required for
Gab1 cellular localization. To test this possibility, we have used
pharmacological inhibitors that inhibit PI3K or phospholipase
activity and the generation of membrane phospholipids. Treat-
ment of MDCK cells expressing wild-type Gab1 with the PI3K
inhibitor LY294002 (67) for 2 h resulted in redistribution of
Gab1 to the cytoplasm, whereas in cells treated with vehicle
alone (dimethyl sulfoxide) Gab1 remained localized at cell-cell
contact sites (Fig. 7B). In contrast, treatment of these cells with
FIG. 6. The PH domain of Gab1 is required for the localization of Gab1 to sites of cell-cell contact. (A) MDCK cells stably transfected with wild-type Gab1,
Gab1?PH, or Gab1?PI3K were grown on glass coverslips in DMEM containing 10% FBS. The cells were fixed in 2% paraformaldehyde and then labeled with anti-HA
followed by CY3-conjugated anti-mouse antiserum. (B) Plasmids encoding pEGFP-Gab1, pEGFP-Gab1?PH, or pEGFP-Gab1PH were microinjected into nuclei of
MDCK cells grown in DMEM plus 10% FBS. At 2 h following the microinjections, cells were visualized. (C) MDCK cells overexpressing Gab1 were either fixed in
2% paraformaldehyde (?), treated for 10 min in CSK buffer prior to fixation (?CSK), or treated for 2 h in 5 mM EGTA-containing media (Low [Ca2?]). The cells
were subsequently labeled with anti-HA or anti-E-cadherin as indicated on the figure. Photographs were taken at a magnification of ?100.
FIG. 5. Gab1 mutant proteins associate with Met and are phosphorylated following Met activation. (A) 293T cells were transiently transfected with wild-type (WT)
CSF-Met or the N1358H CSF-Met mutant, together with wild-type Gab1, Gab1?PI3K, or Gab1?PH. Lysates were subjected to immunoprecipitation (ip) with anti-HA,
and proteins were resolved by SDS-PAGE (8% polyacrylamide), transferred to a nitrocellulose membrane, and blotted with anti-Met. The blot was stripped and
reprobed with anti-HA. (B) MDCK cells expressing N1358H CSF-Met and either Gab1?PI3K or Gab1?PH were stimulated with 2 ?g of CSF-1 per ml for 15 min at
37°C. Lysates were subjected to immunoprecipitation with anti-HA and blotting with anti-PY. The blots were stripped and reprobed with anti-p85 and then with
anti-SHP2. A parallel blot was probed with anti-HA. (C) MDCK cells expressing Gab1?PH protein were stimulated for the indicated time, and lysates were subjected
to immunoprecipitation with anti-HA. Proteins were resolved by SDS-PAGE and, following transfer to nitrocellulose, were blotted with either anti-PY or anti-HA as
1792 MAROUN ET AL.MOL. CELL. BIOL.
VOL. 19, 1999 Gab1 PH DOMAIN IN Gab1 LOCALIZATION AND FUNCTION1793
an inhibitor of phospholipases, U73122, had no detectable
effect on Gab1 distribution (Fig. 7B). This demonstrates that
either a lipid(s) generated by PI3K or another effector down-
stream from PI3K was responsible for Gab1 recruitment and
stabilization at cell-cell junctions. To establish whether PI3K
or its downstream effector(s) is involved in the recruitment of
Gab1 to the membrane, vectors encoding the p110 or the p85
subunits of PI3K (31) were comicroinjected with pEGFP-Gab1
into nuclei of MDCK cells in established colonies. Overexpres-
sion of p85 and p110 has been shown to correlate with en-
hanced PI3K activity in the absence of stimulation (31). Under
serum-starved conditions, pEGFP-Gab1 was expressed in the
cytoplasm (Fig. 7C). Importantly, comicroinjection of pEGFP-
Gab1 with p110 and p85 resulted in localization of a fraction of
pEGFP-Gab1 to areas of cell-cell contact, supporting a role for
PI3K in the recruitment of Gab1 to these sites (Fig. 7C,
pEGFP-Gab1 ? p85/p110).
Met-mediated recruitment of Gab1 to the membrane is not
dependent on the Gab1 PH domain. Activation of the Met
receptor resulted in the association of Gab1 (directly or indi-
rectly) with Met (Fig. 5); therefore, the prediction followed
that Met activation recruits Gab1 to the vicinity of the cell
membrane. To address if Gab1 is recruited to the cell periph-
ery following Met activation, we have examined Gab1 local-
ization following HGF stimulation of 18-h cultures of MDCK
FIG. 7. The localization of Gab1 is serum and PI3K dependent. (A) MDCK
cells (104) stably expressing HA-Gab1 were grown in DMEM plus 10% FBS for
72 or 18 h as indicated. For serum starvation experiments, cells were first grown
for 48 h in 10% FBS and then transferred for 24 h to medium containing 0.02%
FBS. The cells were fixed in 2% paraformaldehyde, and the localization of Gab1
was determined following indirect immunofluorescence labeling with anti-HA
followed by CY3-conjugated anti-mouse antiserum. (B) HA-Gab1-expressing
MDCK cells were treated for 2 h at 37°C either with 50 ?M LY294002 or with
2 ?M U73122. The cells were subsequently fixed in 2% paraformaldehyde and
labeled with anti-HA followed by CY3 anti-mouse antiserum. (C) MDCK cells
were serum starved for 24 h prior to microinjection with plasmids encoding the
p85 and p110 subunits of PI3K, together with pEGFP-Gab1 or control pEGFP
plasmid. Photographs were taken 4 h following the injections. DMSO, dimethyl
1794 MAROUN ET AL.MOL. CELL. BIOL.
cells, a condition where Gab1 is localized in the cytoplasm (Fig.
7). Gab1 was translocated from the cytoplasm to a membrane-
proximal localization within 15 min of stimulation with HGF.
Moreover, the Gab1?PH mutant protein translocated to the
membrane vicinity in a similar manner to wild-type Gab1 (Fig.
8). This was consistent with the ability of both wild-type Gab1
and the Gab1?PH mutant to associate with Met, as deter-
mined biochemically (Fig. 5B). Taken together, these results
suggest that there are two mechanisms through which Gab1
could be recruited to the membrane; the first is dependent on
the Gab1 PH domain and PI3K (Fig. 7), and the second is
based on the recruitment of Gab1 to Met and is independent
of the Gab1 PH domain (Fig. 8).
HGF stimulates a wide variety of cellular processes through
activation of the Met receptor tyrosine kinase in epithelial
cells. In some cells HGF is a mitogenic factor, whereas in
others it stimulates cell dissociation and invasion and acts as an
initiating signal for an intrinsic cellular morphogenic program.
How the Met receptor orchestrates these distinct events was
unclear. Here we show that Gab1 acts as a multisubstrate
docking protein that is required for branching tubulogenesis in
epithelial cells downstream from the Met receptor. From struc-
ture-function studies, we found that the Gab1 PH domain was
essential for the subcellular localization of Gab1 to areas of
cell-cell contact and for branching tubulogenesis.
Gab1 is the major phosphorylated protein in MDCK cells
following activation of Met and acts to couple the Met receptor
with the p85 subunit of PI3K and associated PI3K activity as
well as with SHP2 and PLC?1 (Fig. 1 and data not shown).
Gab1 associates with similar substrates following stimulation
of cells with EGF, insulin, nerve growth factor (NGF), and
IL-3, indicating that Gab1 acts to target multiple receptors to
common downstream signaling pathways (24, 25, 62). Gab1 is
important for NGF-dependent PC12 cell survival (25), and in
a similar manner, it functions as a critical signal transducer for
branching tubulogenesis downstream from the Met receptor.
Met receptor mutants that are impaired in their ability to
associate with Gab1 fail to induce branching tubulogenesis in
epithelial cells (16, 70, 79). Gab1 is recruited predominantly to
the Met receptor in vivo to a Grb2 binding site at Y1356
(YVNV), at least in part through the Grb2 adapter protein (3,
14, 44), and to a lesser extent to Y1349 through a direct
interaction (44, 69). Overexpression of Gab1 rescued the
branching tubulogenesis defect of cells expressing mutant Met
receptors with reduced Gab1 binding capacity (Y1356F and
N1358H [Fig. 2B and C]) but not cells expressing mutant Met
receptors that fail to bind Gab1 (Y1349/1356F [Fig. 2B and
C]). This indicates that the recruitment and phosphorylation of
Gab1 by Met is essential and that when Gab1 is overexpressed,
Y1349 is sufficient for Gab1 recruitment and signaling (Fig. 2B
and C and 5A and B).
These data support those of Weidner et al. that Gab1 is an
important mediator of morphogenesis in epithelial cells (69).
However, while these authors have demonstrated that overex-
FIG. 8. Gab1 is recruited to the membrane folowing Met activation. MDCK cells (104) expressing wild-type HA-Gab1 or HA-Gab1?PH mutant proteins were
grown overnight in medium containing 10% FBS. They were then stimulated with 50 U of HGF per ml at 37°C for 15 min. Following fixation, the cells were labeled
with anti-HA and photographs were taken at a magnification of ?100.
VOL. 19, 1999Gab1 PH DOMAIN IN Gab1 LOCALIZATION AND FUNCTION1795
pression of Gab1 promoted branching tubulogenesis in the
absence of stimulation of the Met receptor, in our experience,
overexpression of Gab1 in 40 independent cell lines examined
was not sufficient to induce branching tubulogenesis in the
absence of Met activation (Fig. 2). This discrepancy may simply
reflect a difference in MDCK cell lines, expression levels of
Gab1, or the criteria used to score positive tubulogenesis. In
our assays, the ability to induce tubulogenesis was scored pos-
itive only when more than 50% of cystlike structures formed
tubules (Fig. 2 and 4). Moreover, although Gab1 is phosphor-
ylated downstream from the EGF receptor in MDCK cells
(Fig. 3), overexpression of Gab1 did not promote branching
tubulogenesis in response to EGF (Fig. 3). Thus, Gab1 phos-
phorylation per se is not sufficient to induce branching tubu-
logenesis in these cells.
The selectivity of the Gab1-dependent tubulogenic response
downstream from Met and not the EGF receptor may suggest
that a Met-specific substrate, in addition to Gab1, is required
for branching tubulogenesis. Alternatively, we show that al-
though the amplitude of phosphorylation of Gab1 is similar in
response to HGF and EGF, HGF induces a prolonged tyrosine
phosphorylation of Gab1 (60 min [Fig. 3B]) whereas Gab1
phosphorylation downstream from EGF is transient (15 min
[Fig. 3B]). Similarly, a sustained activation of the MAPK path-
way correlated with differentiation of MDCK cells in response
to HGF (30) and of PC12 cells in response to NGF (38),
whereas a transient activation was observed in either cell line
in response to EGF and correlated with proliferation. Thus,
the differentiation and tubulogenic responses in MDCK cells
may require the sustained phosphorylation of Gab1 induced
downstream of the Met receptor, indicating that the duration
of Gab1 phosphorylation downstream from multiple extracel-
lular signals may be an important factor in the Gab1 dependent
The ability of Gab1 to rescue branching tubulogenesis in cell
lines expressing Met receptor mutants provided a means by
which we could undertake a structure-function analysis of
Gab1. An intact Gab1 PH domain, but not Gab1-associated
PI3K activity, was essential for rescue of branching tubulogen-
esis downstream from Met receptor mutants (Fig. 4). The
observation that PI3K activity was required for cell dissociation
and branching tubulogenesis downstream from the Met recep-
tor (8, 55), had suggested that Gab1-associated PI3K activity
may be required for rescue of tubulogenesis. However, since
the Gab1?PI3K mutant only partially rescued tubulogenesis
(Fig. 4B and C), this may indicate a requirement for Gab1-
associated PI3K activity for efficient rescue. Alternatively, the
Met-associated or high basal cellular levels of PI3K observed
in 5% serum used for these assays compensate for the lack of
association of Gab1 with PI3K.
Gab1 is most homologous to the IRS-1 family of proteins in
the PH domain (5, 25). The PH domain of IRS-1, in addition
to its PTB domain, is required for efficient phosphorylation of
IRS-1 by the insulin receptor (5, 76). Moreover, the Gab1 PH
domain, but not those of spectrin, PLC?, and ?ARK, can
functionally substitute for the IRS-1 PH domain, suggesting a
conserved modulatory function for the PH domains of IRS1
family proteins (5). The inability of the Gab1?PH mutant to
rescue the tubulogenesis defect downstream from Met recep-
tor mutants may reflect the possibility that the Gab1?PH mu-
tant was not phosphorylated by the Met receptor. However, in
MDCK cell lines, Met activation resulted in tyrosine phosphor-
ylation of the Gab1?PH mutant, and this mutant associated
with cellular substrates to a similar extent to wild-type Gab1
(Fig. 5). Thus, in agreement with previous data implicating
Grb2 and the proline-rich Gab1 Met binding domain in re-
cruitment of Gab1 to the Met receptor (Fig. 8) (44, 69), the
Gab1 PH domain is not essential for phosphorylation of Gab1
by Met, suggesting a distinct role for the Gab1 PH domain in
Met-regulated signal transduction.
An intact Gab1 PH domain was required for subcellular
localization of Gab1 to areas of cell-cell contact in colonies of
MDCK cells (Fig. 6). Increasing evidence supports a role for
PH domains in the regulated targeting of proteins to cell mem-
branes through their interactions with inositol phospholipids
and/or additional interactions with proteins (17–19, 22, 28, 33,
35, 50, 56). The amino acids implicated in phospholipid bind-
ing are highly conserved among PH domains (12, 28, 36),
including that of Gab1, supporting the possibility that the Gab1
PH domain also interacts with membrane phospholipids. Con-
sistent with this, Gab1 is present at sites of cell-cell contact only
in the presence of high serum concentrations (Fig. 7A). More-
over, the inhibition of PI3K by LY294002 resulted in the re-
localization of Gab1 to the cytoplasm, even in the presence of
high serum concentrations (Fig. 7B). The PH domains of the
Rac exchange factor Tiam1, the ARF exchange factor family,
and PLC? bind preferentially to products of PI3K (phospha-
tidylinositol 3,4-bisphosphate or phosphatidylinositol 3,4,5-
triphosphate [1, 61]), implicating PI3K-dependent phospholip-
ids in the targeting of these proteins to the plasma membrane
(10, 60, 66). Similarly, in in vitro binding studies, the Gab1 PH
domain shows greatest affinity for phosphatidylinositol 3,4,5-
triphosphate (PI3P) (25a), suggesting that the PI3K-dependent
localization of Gab1 to sites of cell-cell contact involves the
interaction of the Gab1 PH domain with PIP3 in the mem-
brane. In support of this, overexpression of the p110 and p85
subunits of PI3K, which show elevated PI3K activity in the
absence of stimulation (31), induces the translocation of Gab1
from the cytosol to the membrane in cells maintained at low
serum concentrations (Fig. 7C).
The localization of Gab1 at sites of cell-cell contact requires
extensive cell-cell interactions established after 3 days of cul-
ture, rather than initial cell-cell contacts observed after 18 h of
culture (Fig. 7A). Unlike Tiam1, which is localized to E-cad-
herin-containing adherens junctions at sites of cell-cell contact
in MDCK cells (26, 60), Gab1 is not associated with a deter-
gent-insoluble compartment. Moreover, once localized to the
proximity of the plasma membrane, the disruption of cell-cell
interactions in media containing a low concentration of Ca2?
does not cause the redistribution of Gab1 to the cytoplasm as
is the case for E-cadherin (Fig. 6C), suggesting that once re-
cruited to the membrane, Gab1 can be retained at the mem-
brane at high serum concentrations. However, it is possible
that like Tiam1, the localization of Gab1 to areas of cell-cell
contact requires both a lipid binding and protein-protein in-
teraction, and further structure-function studies are required
to distinguish these requirements (60).
Our data indicate that Gab1 is recruited to the cell mem-
brane by two distinct mechanisms. One, where Gab1 is local-
ized to areas of cell-cell contact at high serum concentrations,
is dependent on the formation of extensive cell-cell interac-
tions in aged cultures, the Gab1 PH domain, and requires PI3K
activity (Fig. 9A). The second, via recruitment of Gab1 to the
Met receptor, is independent of the Gab1 PH domain (Fig. 8
and 9B) and PI3K activity (results not shown). However, re-
cruitment of Gab1 to the Met receptor and its phosphorylation
by the receptor are insufficient for tubulogenesis in the absence
of the Gab1 PH domain. Therefore, while Met activation can
result in Gab1 localization to the membrane, the Gab1 PH
domain may act to stabilize its interaction with membrane
bound phospholipids and enable Gab1 to potentiate and/or
compartmentalize the signal downstream from Met (Fig. 9C),
1796 MAROUN ET AL.MOL. CELL. BIOL.
although we cannot rule out additional functions for the Gab1
PH domain. The elucidation of the mechanism(s) by which the
Gab1 PH domain targets Gab1 to sites of cell-cell contact and
the role of the PH domain in epithelial morphogenesis medi-
ated by Gab1 will provide important insights into how these
processes are normally regulated and how the organized epi-
thelial architecture is altered in human cancers.
This research was supported by an operating grant from the Na-
tional Cancer Institute of Canada with funds from the Canadian Can-
cer Society (to M.P.), an American Cancer Society Grant, and Na-
tional Institutes of Health grants NS 34514 and CA69495 (to A.J.W.).
Financial support was provided by the Medical Research Council as
Fellowships (to C.R.M. and I.R.), a fellowship from the Ministerio de
Educacion y Ciencia of Spain (to M.H.-M.), and the Royal Victoria
Hospital Research Institute as a studentship (to T.M.F.). M.P. is a
Scientist of the Medical Research Council of Canada.
We are grateful to Genetics Institute for recombinant CSF-1, T.
Pawson for anti-p85, G. S. Feng for anti-SHP2, M. Pasdar for anti-E-
cadherin, and members of the Park laboratory and A. Nepveu for
1. Auger, K. R., L. A. Serunian, S. P. Soltoff, P. Libby, and L. C. Cantley. 1989.
PDGF-dependent tyrosine phosphorylation stimulates production of novel
polyphosphoinositides in intact cells. Cell 57:167–175.
2. Backer, J. M., M. J. Myers, Jr., S. E. Shoelson, D. J. Chin, X. J. Sun, M.
Miralpeix, P. Hu, B. Margolis, E. Y. Skolnik, and J. Schlessinger. 1992.
Phosphatidylinositol 3?-kinase is activated by association with IRS-1 during
insulin stimulation. EMBO J. 11:3469–3479.
3. Bardelli, A., P. Longati, D. Gramaglia, M. C. Stella, and P. M. Comoglio.
1997. Gab1 coupling to the HGF/Met receptor multifunctional docking site
requires binding of Grb2 and correlates with the transforming potential.
4. Bellusci, S., G. Moens, G. Gaudino, P. Comoglio, T. Nakamura, J. P. Thiery,
and J. Jouanneau. 1994. Creation of an hepatocyte growth factor/scatter
factor autocrine loop in carcinoma cells induces invasive properties associ-
ated with increased tumorigenicity. Oncogene 9:1091–1099.
5. Burks, D. J., S. Pons, H. Towery, J. Smith-Hall, M. G. Myers, Jr., L. Yenush,
and M. F. White. 1997. Heterologous pleckstrin homology domains do not
couple IRS-1 to the insulin receptor. J. Biol. Chem. 272:27716–27721.
6. Carpino, N., D. Wisniewski, A. Strife, D. Marshak, R. Kobayashi, B. Still-
man, and B. Clarkson. 1997. p62dok: A constitutively tyrosine-phosphory-
lated, GAP-associated protein in chronic myelogenous leukemia progenitor
cells. Cell 88:197–204.
7. Chen, R. H., S. Corbalan-Garcia, and D. Bar-Sagi. 1997. The role of the PH
domain in the signal-dependent membrane targeting of Sos. EMBO J. 16:
8. Derman, M. P., M. J. Cunha, E. J. Barros, S. K. Nigam, and L. G. Cantley.
1995. HGF-mediated chemotaxis and tubulogenesis require activation of the
phosphatidylinositol 3-kinase. Am. J. Physiol. 268:1211–1217.
9. Ebens, A., K. Brose, E. D. Leonardo, M. G. Jr. Hanson, F. Bladt, C. Birch-
meier, B. A. Barres, and M. Tessier-Lavigne. 1996. Hepatocyte growth fac-
tor/scatter factor is an axonal chemoattractant and a neurotrophic factor for
spinal motor neurons. Neuron 17:1157–1172.
10. Falasca, M., S. K. Logan, V. P. Lehto, G. Baccante, M. A. Lemmon, and J.
Schlessinger. 1988. Activation of phospholipase C? by PI 3-kinase-induced
PH domain-mediated membrane targeting. EMBO J. 17:414–422.
11. Faletto, D. L., D. R. Kaplan, D. O. Halverson, E. M. Rosen, and G. F. Vande
Woude. 1993. Signal transduction in c-met mediated motogenesis. Hepato-
cyte growth factor: scatter factor. EXS 65:107–130.
12. Ferguson, K. M., M. A. Lemmon, J. Schlessinger, and P. B. Sigler. 1995.
Structure of the high affinity complex of inositol triphosphate with a phos-
pholipase C pleckstrin homology domain. Cell 83:1037–1046.
13. Fixman, E. D., T. M. Fournier, D. M. Kamikura, M. A. Naujokas, and M.
Park. 1996. Pathways downstream of Shc and Grb2 are required for cell
transformation by theTpr-Met
14. Fixman, E. D., M. Holgado-Madruga, L. Nguyen, D. M. Kamikura, T. M.
Fournier, A. J. Wong, and M. Park. 1997. Efficient cellular transformation by
the Met oncoprotein requires a functional Grb2 binding site and correlates
with phosphorylation of the Grb2-associated proteins Cbl and Gab1. J. Biol.
15. Fixman, E. D., M. A. Naujokas, G. A. Rodrigues, M. F. Moran, and M. Park.
1995. Efficient cell transformation by the Tpr-Met oncoprotein is dependent
upon tyrosine 489 in the carboxy-terminus. Oncogene 10:237–249.
15a.Fixman, E. D., I. Royal, and M. Park. Unpublished observations.
16. Fournier, T., D. Kamikura, K. Teng, and M. Park. 1996. Branching tubulo-
genesis, but not scatter of Madin-Darby canine kidney cells requires a func-
tional Grb2 binding site in the Met receptor tyrosine kinase. J. Biol. Chem.
17. Franke, T. F., D. R. Kaplan, L. C. Cantley, and A. Toker. 1997. Direct
regulation of the Akt proto-oncogene product by phosphatidylinositol-3,4-
bisphosphate. Science 275:665–668.
18. Frech, M., M. Andjelkovic, E. Ingley, K. K. Reddy, J. R. Falck, and B. A.
Hemmings. 1997. High affinity binding of inositol phosphates and phospho-
inositides to the pleckstrin homology domain of RAC/protein kinase B and
their influence on kinase activity. J. Biol. Chem. 272:84474–84481.
19. Garcia, P., R. Gupta, S. Shah, A. J. Morris, S. A. Rudge, S. Scarlata, V.
Petrova, S. McLaughlin, and M. J. Rebecchi. 1995. The pleckstrin homology
domain of phospholipase C-?1 binds with high affinity to phosphatidylinosi-
tol 4,5-biphosphate in bilayer membranes. Biochemistry 34:16228–16234.
20. Gibson, T. J., M. Hyvo ¨nen, A. Musacchio, M. Saraste, and E. Birney. 1994.
PH domain: the first anniversary. Trends Biochem. Sci. 19:349–353.
21. Grant, D. S., H. K. Kleinman, I. D. Goldberg, M. M. Bhargava, B. J.
Nickoloff, J. L. Kinsella, P. Polverini, and E. M. Rosen. 1993. Scatter factor
induces blood vessel formation in vivo. Proc. Natl. Acad. Sci. USA 90:1937–
22. Harlan, J. E., P. J. Hajduk, H. S. Yoon, and S. W. Fesik. 1994. Pleckstrin
homology domains bind to phosphatidylinositol 4,5-bisphosphate. Nature
FIG. 9. Gab1 is recruited to the cell surface by two distinct mechanisms. (A) One mechanism depends on the Gab1 PH domain and serum and requires PI3K
activity. (B) The second mechanism is based on Met activation and is independent of the Gab1 PH domain (see Discussion for details). (C) Once recruited to and
phosphorylated by Met, Met-associated or Gab1-associated PI3K activity is predicted to stabilize Gab1 association with the membrane.
VOL. 19, 1999Gab1 PH DOMAIN IN Gab1 LOCALIZATION AND FUNCTION1797
23. Herbst, R., P. M. Caroll, J. D. Allard, J. Schilling, T. Raabe, and M. A.
Simon. 1996. Daughter of sevenless is a substrate of the phosphotyrosine
phosphatase Corkscrew and functions during sevenless signaling. Cell 85:
24. Holgado-Madruga, M., D. R. Emlet, D. K. Moscatello, A. K. Godwin, and
A. J. Wong. 1996. A Grb2-associated docking protein in EGF- and insulin-
receptor signaling. Nature 379:560–564.
25. Holgado-Madruga, M., D. K. Moscatello, D. R. Emlet, R. Dieterich, and A. J.
Wong. 1997. Grb2-associated binder-1 mediates phosphatidylinositol 3-ki-
nase activation and the promotion of cell survival by nerve growth factor.
Proc. Natl. Acad. Sci. USA 94:12419–12424.
25a.Holgado-Madruga, M., and A. J. Wong. Submitted for publication.
26. Hordijk, P. L., J. P. ten Klooster, R. A. van der Kammen, F. Michiels, L. C.
Oomen, and J. G. Collard. 1997. Inhibition of invasion of epithelial cells by
Tiam1-Rac signaling. Science 278:1464–1466.
27. Hu, Q., A. Klippel, A. J. Muslin, W. J. Fantl, and L. T. Williams. 1995.
Ras-dependent induction of cellular responses by constitutively active phos-
phatidylinositol-3 kinase. Science 268:100–102.
28. Hyvo ¨nen, M., M. J. Macias, M. Nilges, H. Oschkinat, M. Saraste, and M.
Wilmanns. 1995. Structure of the binding site for inositol phosphates in a PH
domain. EMBO J. 14:4676–4685.
29. Kan, M., G. Zhang, R. Zarnegar, G. Michalopoulos, Y. Myoken, W. L.
McKeehan, and J. I. Stevens. 1991. Hepatocyte growth factor/hepatopoietin
A stimulates the growth of rat kidney proximal tubule epithelial cells
(RPTE), rat nonparenchymal liver cells, human melanoma cells, mouse
keratinocytes and stimulates anchorage-independent growth of SV-40 trans-
formed RPTE. Biochem. Biophys. Res. Commun. 174:331–337.
30. Khwaja, A., K. Lehmann, B. M. Marte, and J. Downward. 1998. Phospho-
inositide 3-kinase induces scattering and tubulogenesis in epithelial cells
through a novel pathway. J. Biol. Chem. 273:18793–18801.
31. Klippel, A., J. A. Escobedo, M. Hirano, and L. T. Williams. 1994. The
interaction of small domains between the subunits of phosphatidylinositol
3-kinase determines enzyme activity. Mol. Cell. Biol. 14:2675–2685.
32. Komada, M., and N. Kitamura. 1993. The cell dissociation and motility
triggered by scatter factor/hepatocyte growth factor are mediated through
the cytoplasmic domain of the c-Met receptor. Oncogene 8:2381–2390.
33. Konishi, H., S. Kuroda, and U. Kikkawa. 1994. The pleckstrin homology
domain of RAC protein kinase associates with the regulatory domain of
protein kinase C?. Biochem. Biophys. Res. Commun. 205:1770–1775.
34. Lemmon, M. A., M. Falasca, K. M. Ferguson, and J. Schlessinger. 1997.
Regulatory recruitment of signaling molecules to the cell membrane by
pleckstrin homology domains. Trends Cell Biol. 7:237–242.
35. Lemmon, M. A., K. M. Ferguson, R. O’Brien, P. B. Sigler, and J. Schless-
inger. 1995. Specific and high-affinity binding of inositol phosphates to an
isolated pleckstrin homology domain. Proc. Natl. Acad. Sci. USA 92:10472–
36. Lemmon, M. A., K. M. Ferguson, and J. Schlessinger. 1996. PH domains:
diverse sequences with a common fold recruit signaling molecules to the cell
surface. Cell 85:621–624.
37. Maina, F., M. C. Hilton, C. Ponzetto, A. M. Davies, and R. Klein. 1997. Met
receptor signaling is required for sensory nerve development and HGF
promotes axonal growth and survival of sensory neurones. Genes Dev. 11:
38. Marshall, C. J. 1995. Specificity of receptor tyrosine kinase signaling: tran-
sient versus sustained extracellular signal-regulated kinase activation. Cell
39. Matsumoto, K., and T. S. Nakamura. 1993. Roles of HGF as a pleiotropic
factor in organ regeneration. EXS 65:225–249.
40. Michiels, F., J. C. Stam, P. L. Hordijk, R. A. van der Kammen, L. Ruuls-
vanStalle, C. A. Feltkamp, and J. G. Collard. 1997. Regulated membrane
localization of Tiam1, mediated by the NH2-terminal pleckstrin homology
domain, is required for Rac-dependent membrane ruffling and C-Jun NH2-
terminal kinase activation. J. Cell Biol. 137:387–398.
41. Montesano, R., G. Schaller, and L. Orci. 1991. Induction of epithelial tubu-
lar morphogenesis in vitro by fibroblast-derived soluble factors. Cell 66:697–
42. Myers, M. G. Jr., J. M. Backer, X. J. Sun, S. Shoelson, P. Hu, J. Schless-
inger, M. Yoakim, B. Schaffhausen, and M. F. White. 1992. IRS-1 activates
phosphatidylinositol 3?-kinase by associating with src homology domains of
p85. Proc. Natl. Acad. Sci. USA 89:10350–10354.
43. Nakamura, T. 1991. Structure and function of hepatocyte growth factor.
Prog. Growth Factor Res. 3:67–85.
44. Nguyen, L., M. Holgado-Madruga, C. Maroun, E. D. Fixman, D. Kamikura,
T. Fournier, A. Charest, M. L. Tremblay, A. J. Wong, and M. Park. 1997.
Association of the multisubstrate docking protein Gab1 with the hepatocyte
growth factor receptor requires a functional Grb2 binding site involving
tyrosine 1356. J. Biol. Chem. 272:20811–20819.
45. Pasdar, M., and W. J. Nelson. 1989. Regulation of desmosome assembly in
epithelial cells: kinetics of synthesis, transport, and stabilization of desmo-
glein I, a major protein of the membrane core domain. J. Cell Biol. 109:
46. Paterson, H. F., J. W. Savopoulos, O. Perisic, R. Cheung, M. V. Ellis, R. L.
Williams, and M. Katan. 1995. Phospholipase C ?1 requires a pleckstrin
homology domain for interaction with the plasma membrane. Biochem. J.
47. Pawson, T. 1995. Protein modules and signaling networks. Nature 373:573–
48. Ponzetto, C., A. Bardelli, Z. Zhen, F. Maina, P. dalla Zonca, S. Giordano, A.
Graziani, G. Panayotou, and P. M. Comoglio. 1994. A multifunctional dock-
ing site mediates signaling and transformation by the hepatocyte growth
factor/scatter factor receptor family. Cell 77:261–271.
49. Raabe, T., J. Riesgo-Escovar, X. Liu, B. S. Bausenwein, P. Deak, P. Maroy,
and E. Hafen. 1996. DOS, a novel pleckstrin homology domain-containing
protein required for signal transduction between sevenless and RAS1 in
Drosophila. Cell 85:911–920.
50. Rameh, L. E., A.-K. Arvidsson, K. L. Carraway III, A. D. Couvillon, G.
Rathbun, A. Crompton, B. VanRenterghem, M. P. Czech, K. S. Ravichand-
ran, S. J. Burakoff, D. S. Wang, C. S. Chen, and L. C. Cantley. 1997. A
comparative analysis of the phosphoinositide binding specificity of pleckstrin
homology domains. J. Biol. Chem. 272:22059–22066.
51. Rodrigues, G. A., M. A. Naujokas, and M. Park. 1991. Alternative splicing
generates isoforms of the met receptor tyrosine kinase which undergo dif-
ferential processing. Mol. Cell. Biol. 11:2962–2970.
52. Rodrigues, G. A., and M. Park. 1994. Autophosphorylation modulates the
kinase activity and oncogenic potential of the Met receptor tyrosine kinase.
53. Rong, S., M. Bodescot, D. Blair, J. Dunn, T. Nakamura, K. Mizuno, M. Park,
A. Chan, S. Aaronson, and G. F. Vande Woude. 1992. Tumorigenicity of the
met proto-oncogene and the gene for hepatocyte growth factor. Mol. Cell.
54. Ronsin, C., F. Muscatelli, M. G. Mattei, and R. Breathnach. 1993. A novel
putative receptor protein tyrosine kinase of the met family. Oncogene
55. Royal, I., and M. Park. 1995. Hepatocyte growth factor induced scatter of
Madin-Darby canine kidney cells requires phosphatidylinositol 3-kinase.
J. Biol. Chem. 270:27780–27787.
56. Salim, K., M. J. Bottomly, E. Querfurth, M. J. Zvelebil, I. Gout, R. Scaife,
R. L. Margolis, R. Gigg, C. I. Smith, P. C. Driscoll, M. D. Waterfield, and G.
Panayotou. 1996. Distinct specificity in the recognition of phosphoinositides
by the pleckstrin homology domains of dynamin and Bruton’s tyrosine ki-
nase. EMBO J. 15:6241–6250.
57. Schmidt, C., F. Bladt, S. Goedecke, V. Brinkmann, W. Zschiesche, M.
Sharpe, E. Gherardi, and C. Birchmeier. 1995. Scatter factor/hepatocyte
growth factor is essential for liver development. Nature 373:699–702.
58. Schmidt, L., F. M. Duh, F. Chen, T. Kishida, G. Glenn, P. Choyke, S. W.
Scherer, Z. Zhuang, I. Lubensky, M. Dean, R. Allikmets, A. Chidambaram,
U. R. Bergerheim, J. T. Feltis, C. Casadevall, A. Zamarron, M. Bernues, S.
Richard, C. J. Lips, M. M. Walther, L. C. Tsui, L. Geil, M. L. Orcutt, T.
Stackhouse, J. Lipan, L. Slife, H. Brauch, J. Decker, G. Niehans, M. D.
Hughson, H. Moch, S. Storkel, M. I. Lerman, W. M. Linehan, and B. Zbar.
1997. Germline and somatic mutations in the tyrosine kinase domain of the
MET proto-oncogene in papillary renal carcinomas. Nat. Genet. 16:68–73.
59. Soriano, J. V., M. S. Pepper, T. Nakamura, L. Orci, and R. Montesano. 1995.
Hepatocyte growth factor stimulates extensive development of branching
duct-like structures by cloned mammary gland epithelial cells. J. Cell Sci.
60. Stam, J. C., E. E. Sander, F. Michiels, F. N. van Leeuwen, H. E. Kain, R. A.
van der Kammen, and J. G. Collard. 1997. Targeting Tiam1 to the plasma
membrane requires the cooperative function of the N-terminal pleckstrin
homology domain and an adjacent protein interaction domain. J. Biol.
61. Stephens, L. R., K. T. Hughes, and R. F. Irvine. 1991. Pathway of phophati-
dylinositol (3,4,5)-trisphosphate synthesis in activated neutrophils. Nature
62. Takahashi-Tezuka, M., Y. Yoshida, T. Fukada, T. Ohtany, Y. Yamanaka, K.
Nishida, M. Hibi, and T. Hirano. 1998. Gab1 acts as an adapter molecule
linking the cytokine receptor gp130 to ERK mitogen-activated protein ki-
nase. Mol. Cell. Biol. 18:4109–4117.
63. Takaishi, K., T. Sasaki, H. Kotani, H. Nishioka, and Y. Takai. 1997. Regu-
lation of cell-cell adhesion by Rac and Rho small G proteins in MDCK cells.
J. Biol. Chem. 139:1047–1059.
64. Takayama, H., W. J. LaRochelle, R. Sharp, T. Otsuka, P. Kriebel, M. Anver,
S. A. Aaronson, and G. Merlino. 1997. Diverse tumorigenesis associated with
aberrant development in mice overexpressing hepatocyte growth factor/scat-
ter factor. Proc. Natl. Acad. Sci. USA 94:701–706.
65. Uehara, Y., O. Minowa, C. Mori, K. Shiota, J. Kuno, T. Noda, and N.
Kitamura. 1995. Placental defect and embryonic lethality in mice lacking
hepatocyte growth factor/scatter factor. Nature 373:702–705.
66. Venkateswarlu, K., P. B. Oaty, J. M. Tavare ´, and P. J. Cullen. 1998. Insulin-
dependent translocation of ARNO to the plasma membrane of adipocytes
requires phosphatidylinositol 3-kinase. Curr. Biol. 8:463–466.
67. Vlahos, C. J., W. F. Matter, K. Y. Hui, and R. F. Brown. 1994. A specific
inhibitor of phosphatidylinositol 3-kinase, 2-(4-morpholinyl)-8-phenyl-4H-1-
benzopyran-4-one (LY294002). J. Biol. Chem. 269:5241–5248.
1798MAROUN ET AL.MOL. CELL. BIOL.