Oncogene (2001) 20, 8125 ± 8135
ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00
www.nature.com/onc
Downregulation of E-cadherin and Desmoglein 1 by autocrine hepatocyte
growth factor during melanoma development
Gang Li1,2, Helmut Schaider1, Kapaettu Satyamoorthy1, Yasushi Hanakawa3, Koji Hashimoto3
and Meenhard Herlyn*,1
1
The Wistar Institute, 3601 Spruce Street, Philadelphia, Pennsylvania, PA 19104, USA; 2Program of Cell and Molecular Biology,
Biomedical Graduate Studies, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, PA 19104, USA;
3
Department of Dermatology, School of Medicine, Ehime University, Ehime, Japan
During melanoma development, transformed cells evade
keratinocyte-mediated control by downregulating cell
adhesion molecules. This study investigated the regulation of cell adhesion by hepatocyte growth factor (HGF)
in melanoma. Melanocytes and two melanoma lines,
WM164 and WM35, expressed normal level E-cadherin
and Desmoglein 1, whereas most melanomas (18 out of
20) expressed no E-cadherin and signi®cantly reduced
Desmoglein 1. Overexpression of dominant negative Ecadherin and Desmoglein in melanocytes demonstrated
that both molecules contribute to adhesion between
melanocytes and keratinocytes. In contrast to melanocytes, most melanomas expressed HGF. All melanocytic
cells expressed the HGF receptor c-Met, and autocrine
HGF caused constitutive activation of c-Met, MAPK
and PI3K. When autocrine activation was induced with
HGF-expressing adenovirus, E-cadherin and Desmoglein
1 were decreased in melanocytes, WM164 and WM35.
MAPK inhibitor PD98059 and PI3K inhibitor wortmannin partially blocked the downregulation, suggesting that
both pathways are involved in this process. c-Met was
coimmunoprecipitated with E-cadherin, Desmoglein 1
and Plakoglobin, suggesting that they form a complex
(es) that acts to regulate intercellular adhesion.
Together, the results indicate that autocrine HGF
decouples melanomas from keratinocytes by downregulating E-cadherin and Desmoglein 1, therefore frees
melanoma cells from the control by keratinocytes and
allows dissemination of the tumor mass. Oncogene
(2001) 20, 8125 ± 8135.
Keywords: HGF; autocrine; E-cadherin; Desmoglein;
melanoma
Introduction
The incidence of cutaneous melanoma has risen rapidly
in the last several decades (Parker et al., 1997; Ries et
al., 2000). Melanoma is notorious for its propensity to
*Correspondence: M Herlyn; E-mail: [email protected]
Received 3 August 2001; revised 27 September 2001; accepted 9
October 2001
metastasize and its poor response to current therapeutic regimens. The transition from benign lesions to
invasive, metastatic cancer occurs through a complex
process involving changes in expression and function of
oncogenes or tumor suppressor genes (Meier et al.,
1998), as well as changes that provide tumor cells with
the ability to overcome cell ± cell adhesions and
microenvironmental controls and to invade surrounding tissues and translocate to distant tissues and organs
(Johnson, 1999; Li and Herlyn, 2000).
In the human epidermis, melanocytes residing at the
basement membrane are interspersed among basal
keratinocytes. E-cadherin is physiologically expressed
on the cell surface of keratinocytes and melanocytes,
and is the major adhesion molecule between the two
cell types (Hsu et al., 1996; Tang et al., 1994). A
progressive loss of E-cadherin expression occurs during
melanoma development (Danen et al., 1996; Hsu et al.,
1996; Scott and Cassidy, 1998; Silye et al., 1998).
Restoration of E-cadherin expression in melanoma
cells results in restored keratinocyte-mediated growth
control and downregulated expression of invasionrelated adhesion receptors (Hsu et al., 2000).
Besides E-cadherin-mediated adherens junctions,
desmosomes are also believed to play an important role
in maintaining human skin homeostasis (Garrod, 1993).
Desmosomes contain specialized cadherin adhesion
molecules (Desmogleins and Desmocollins) associating
with various cytoplasmic proteins such as Desmoplakins and Plakoglobin (Garrod, 1993; Koch and Franke,
1994; Kowalczyk et al., 1999; North et al., 1999).
Desmosomes provide the cells with binding domains for
intermediate ®laments of the cytokeratin network and
are required for tissue organization (Bornslaeger et al.,
1996; Kouklis et al., 1994; Kowalczyk et al., 1997;
Smith and Fuchs, 1998; Stappenbeck and Green, 1992).
Plakoglobin (g-catenin) is also part of the cadherin ±
catenin complex in adherens junctions (Aberle et al.,
1994; Butz and Kemler, 1994; Lewis et al., 1997) and
may mediate crosstalk between adherens junctions and
desmosomes (Lewis et al., 1997). Reduction or loss of
desmosomes may contribute to the invasive and
metastatic behavior of various tumors, for example
transitional cell carcinoma of the bladder (Conn et al.,
1990) and squamous cell carcinomas (Depondt et al.,
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
8126
1999; Krunic et al., 1996; Shinohara et al., 1998).
Reduction in desmosome formation was correlated with
invasion and with reduction in E-cadherin staining
(Shinohara et al., 1998). Therefore, desmosomes appear
to have tumor-suppressor properties. In fact, overexpressing desmosomal cadherins in squamous cell
carcinoma cells inhibited invasiveness (De Bruin et al.,
1999; Tada et al., 2000).
Hepatocyte growth factor (HGF) is a multifunctional cytokine acting as mitogen, motogen and
morphogen for many epithelial cells (Gherardi et al.,
1989; Nakamura et al., 1989; Stoker et al., 1987)
through its tyrosine kinase receptor c-Met (Bottaro et
al., 1991b), which is present in epithelial cells and
melanocytes (Bottaro et al., 1991a; Sonnenberg et al.,
1993). HGF is physiologically secreted by cells of
mesenchymal origin and acts on neighboring epithelial
cells through a paracrine loop (Iyer et al., 1990; Rosen
et al., 1994). However, coexpression of HGF and cMet has been identi®ed in a variety of transformed
cells and tumors both in vitro and in vivo and shown to
be involved in tumor development and invasion
(Bellusci et al., 1994; Moriyama et al., 1999; Rong et
al., 1992; Tamatani et al., 1999; To and Tsao, 1998;
Vande Woude et al., 1997).
HGF/c-Met signaling has been implicated in the
development of melanoma (Hendrix et al., 1998; Natali
et al., 1993; Rusciano et al., 1998). HGF stimulates
proliferation and motility of human melanocytes
(Halaban et al., 1992; Imokawa et al., 1998; Kos et al.,
1999). In transgenic mice that ubiquitously expressed
HGF, ectopic localization of melanocytes and hyperpigmentation in skin were observed (Takayama et al., 1996)
and melanoma arose spontaneously (Kos et al., 1999;
Otsuka et al., 1998; Takayama et al., 1997). In these
mice, ultraviolet radiation-induced carcinogenesis was
accelerated (Noonan et al., 2000). It is suggested that cMet autocrine activation induced development of
malignant melanoma and acquisition of the metastatic
phenotype (Otsuka et al., 1998). However it is unknown
whether the autocrine loop actually exists in human
melanoma, and the mechanisms by which HGF induces
human melanoma remain unclear.
The purpose of the study was to address the potential
role of HGF in human melanoma. Our results indicate
that an HGF autocrine loop disrupts adhesion between
melanocytes and keratinocytes by downregulating Ecadherin and Desmoglein 1, resulting in decoupling
melanocytic cells from the control by keratinocytes,
which may allow uncontrolled proliferation of cancer
cells and dissemination of the tumor mass.
Results
Desmoglein 1 is downregulated in melanoma cell lines,
and its expression levels correlates with those of
E-cadherin
During melanoma development, there is a progressive
loss of E-cadherin expression (Hsu et al., 1996; Li and
Oncogene
Herlyn, 2000). We hypothesized that cell-cell adhesions
have to be disrupted before malignant cells can migrate
and metastasize to remote locations; and that besides
E-cadherin-mediated adherens junctions, other types of
adhesive junctions have to be attenuated as well. Since
Desmosomes are important in maintaining human skin
homeostasis (Green and Gaudry, 2000), they are likely
to be a major target of cell ± cell detachment during
transformation. Western blots (Figure 1a) showed that
both keratinocytes (FK113) and primary melanocytes
(FOM73) expressed high levels of Desmoglein 1, while
the majority of melanoma cells (18 out of 20, only
eight lines shown) expressed very little Desmoglein 1.
There are exceptions, however. Two lines, WM35 and
WM164, did express Desmoglein 1 at levels comparable to those in keratinocytes or melanocytes. These
two cell lines are exceptional in that, unlike other
melanoma cells, they are E-cadherin positive (Figure
1a), suggesting that there is a correlation between the
levels of E-cadherin and those of Desmoglein 1 in
melanoma cells. Desmoglein 2 and Desmoglein 3 were
not detectable by Western blots in either normal
melanocytes or melanoma cells of various stages (data
not shown). Since Plakoglobin (g-catenin) is shared by
both adherens junctions and desmosomes (Cowin et
Figure 1 Expression of Desmoglein 1, E-cadherin and Plakoglobin in melanocytes and melanoma cells. (a) Immunoblotting
analyses of Desmoglein 1, E-cadherin and Plakoglobin (g-catenin)
expression in melanocytes (FOM73) and melanoma cell lines
(WM9, 1205Lu, WM278, WM793, WM35, WM115, WM164 and
WM239). Epidermal keratinocytes (FK113) were used as positive
control for E-cadherin and Desmoglein 1. 15 mg/lane whole-cell
protein extracts were separated on an 8% SDS-polyacrylamide gel
and then transferred to PVDF membrane. The same blot was
probed with antibodies against of E-cadherin, Desmoglein 1 and
Plakoglobin (g-catenin). b-Actin was used as a loading control. (b)
Indirect immuno¯uorescence of Desmoglein 1 expression in
melanocytes (FOM73) and representative melanoma cells WM35
and 1205Lu
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
8127
al., 1986; Franke et al., 1989; Kowalczyk et al., 1999;
Mathur et al., 1994; Roh and Stanley, 1995;
Troyanovsky and Leube, 1998), and loss of Plakoglobin in murine melanoma cell lines has been reported
(Jouneau et al., 2000), we also tested its expression in
melanoma cells. Although reduced expression and
absence were seen in some cell lines (Figure 1a), there
was no clear trend. Immuno¯uorescence staining
showed that in Desmoglein-positive cells (FOM73
and WM35), Desmoglein 1 was concentrated at the
cell ± cell contact areas whereas in Desmoglein-low cells
such as 1205Lu, no distinct staining was observed
(Figure 1b).
Desmoglein 1 functions as a co-receptor for E-cadherin in
mediating cell-cell adhesion between melanocytes and
keratinocytes
To test the hypothesis that desmosomes function
together with adherens junctions to mediate cell ± cell
adhesion between melanocytes and keratinocytes, we
expressed dominant negative E-cadherin (E-cadDEC)
and Desmoglein 3 (Dsg3DEC) (Hanakawa et al., 2000)
in melanocytes. Dsg3DEC was originally made to
target Desmoglein 3, but later shown to be able to
disrupt Desmoglein 1 as well (Hanakawa et al.,
unpublished observation). After con®rming their expression in melanocytes (Figure 2a,b), we used cell
adhesion assays to test their role in melanocytekeratinocyte interactions (Figure 2c). E-cadDEC signi®cantly (42% reduction) disrupted adhesion between
the two cell types, whereas Dsg3DEC alone didn't
make a signi®cant di€erence. Intriguingly, when both
Dsg3DEC and E-cadDEC were introduced to melanocytes, a strong disruption (79% reduction) was
obtained (Figure 2c). This suggests that disruption of
Desmoglein alone is apparently not enough to break
cell ± cell adhesion, however it can further weaken the
interaction if E-cadherin-mediated adhesion is also lost.
In the search for the possible mechanisms underlying
the downregulation of Desmoglein and E-cadherin, we
failed to detect gene deletion or promoter methylation
in our panel of melanoma lines (data not shown). This
is in contrast to what has been shown in other cancer
types (for review, see Hirohashi, 1998), but is
consistent with observations made by others in
melanoma (Poser et al., 2001). Therefore, we investigated whether abnormal growth factor expression
caused the downregulation of E-cadherin and/or
Desmoglein.
Melanoma cells, but not normal melanocytes, express
HGF, which causes constitutive activation of c-Met,
MAPK and PI3K in melanoma cells
During screening of growth factors, we found that the
majority of melanoma cells express HGF (Figure 3a,b).
In normal skin, HGF is secreted by cells of
mesenchymal origin such as ®broblasts (Stoker et al.,
1987) (FF2441 in Figure 3a,b), but not melanocytes
(FOM73). Therefore it is surprising to see that HGF
Figure 2 Both E-cadherin and Desmoglein 1 mediate adhesion
between melanocytes and keratinocytes. (a) Expression of EcadDEC in melanocytes co-infected with AxCANCre and
AxCALNLEcadDEC adenoviruses (20 pfu/cell). Immunoblotting
detected both endogenous E-cadherin (notice the reduced level of
endogenous E-cadherin in E-cadDEC-expressing cells) and
dominant negative E-cadDEC. (b) Expression of Dsg3DEC in
melanocytes co-infected with AxCANCre and AxCALNLDsg3DEC at 20 pfu/cell each. Immunoblotting detected c-Myc tag on
Dsg3DEC. (c) Melanocytes infected with indicated adenoviruses
were pre-labeled with a ¯uorescent dye DiI, resuspended in assay
medium, added to keratinocyte monolayer and allowed to adhere
for 30 min. Numbers of adherent cells per high power ®eld in
triplicate wells were counted under a ¯uorescence microscope
autocrine loop exists in melanoma cells. The two lines
(WM35 and WM164) with high E-cadherin and
Desmoglein 1 expression showed little HGF expression. We then further tested the expression and status
of the HGF receptor, c-Met. All cells, including
keratinocytes, melanocytes and melanoma cells, expressed c-Met (Figure 4a), and apparently there was no
strong correlation between HGF expression levels and
c-Met abundance. When melanocytes and melanoma
cells were treated with additional HGF, a dramatic
increase in tyrosine-phosphorylated c-Met (Figure 4b),
accompanied by activation of MAPK and AKT
(Figure 4c), was seen, suggesting that c-Met as well
as its downstream signaling molecules are functional in
these cells. The levels of c-Met did not change after
HGF treatment (Figure 4b and data not shown).
Constitutively phosphorylated c-Met could be detected
even in the absence of exogenous HGF (Figure 4b,
Oncogene
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
8128
Figure 3 Expression of HGF by melanocytes and melanoma
cells. (a) RT ± PCR analysis of HGF transcripts in melanocytes
and melanoma cells. Fibroblasts (FF2441) and keratinocytes
(FK113) were included as positive control and negative control,
respectively. b-Actin was used as internal control. (b) Secreted
HGF was detected by ELISA. Conditioned media were collected
at indicated time points. The concentrations shown represent
average values from four repeats and normalized to 105 cells/2 ml
medium for each assay
WM278 and 1205Lu). The constitutive activation is
caused by endogenous HGF, because a HGF blocking
antibody could reduce c-Met phosphorylation (Figure
5a). On the other hand, since WM164 and WM35
expressed little HGF (Figure 3), the endogenous levels
of c-Met activation were low (Figure 4b). Melanocytes
contained no constitutively active c-Met (Figure 4b).
Constitutive activation of MAPK and AKT has been
observed in melanoma both in vitro and in vivo
(Satyamoorthy et al. unpublished data); (Li et al.,
2001). To test the hypothesis that autocrine HGF is, at
least in part, responsible for it, we used HGF blocking
antibody to treat 1205Lu melanoma cells. After HGF
neutralization, a signi®cant decrease in phosphorylation of both MAPK (Figure 5b) and AKT (Figure 5c)
was obtained. Similar results were seen in WM115 cells
(data not shown).
Sustained HGF stimulation causes downregulation of both
E-cadherin and Desmoglein 1 in melanocytic cells through
PI3K and MAPK pathways
Increased proliferation of human melanocytes in vitro
by HGF has previously been reported (Halaban et al.,
1992; Imokawa et al., 1998; Matsumoto et al., 1991).
However, its mitogenic activity became obvious only
when it acted together with other growth factors such
Oncogene
Figure 4 Characterization of c-Met in melanocytes and melanoma cells. (a) Immunoblotting analyses of c-Met expression in
melanocytes (FOM73) and melanoma cell lines (WM9, 1205Lu,
WM278, WM793, WM35, WM115, WM164 and WM239).
Epidermal keratinocytes (FK113) were used as positive control.
15 mg/lane whole-cell protein extracts were separated on an 8%
SDS-polyacrylamide gel and then transferred to PVDF membrane. b-Actin was used as a loading control. Of the two major
bands the upper one is c-Met precursor (180 kDa) and the lower
one is b-chain of c-Met (145 kDa). (b) Phosphorylation of c-Met
in melanocytes and melanoma cells. Melanocyte (FOM73) and
melanoma cells were infected with Ad.CMV.rhHGF (20 pfu/cell)
or LacZ/Ad5 (20 pfu/ml). Forty-eight hours later, cells were lysed
and immunoprecipitated with anti-c-Met antibody. The immunoprecipitated proteins were separated by SDS-polyacrylamide gel
electrophoresis and hybridized against anti-phosphotyrosine antibody. (c) MAPK and AKT activation by exogenous HGF in
melanocytes. FOM73 cells were cultured in melanocyte medium
with reduced FBS (1%). Human recombinant HGF was added to
the culture medium at a ®nal concentration of 100 ng/ml. Twenty
minutes later cells were lysed and the lysates were subject to
Western blots using anti-p44/42 MAPK, anti-phospho-p44/42
MAPK (Thr202/Tyr204), anti-Akt and anti-phospho-Akt
(Ser473) antibodies
as bFGF (Halaban et al., 1992). We did not observe
signi®cant growth stimulation in either melanocytes or
melanoma cells after HGF treatment alone (data not
shown). On the contrary, a dramatic decrease in the
expression levels of Desmoglein 1 and E-cadherin were
detected in melanocytes, WM164 and WM35 melanoma cells exposed to continuous HGF stimulation for
three days (Figure 6). An adenoviral vector
Ad.CMV.rhHGF (Phaneuf et al., 2000) was then used
at 20 pfu/cell (corresponding to 30 ng/ml at 72 h) to
introduce an autocrine loop in these cells and to ensure
sustained stimulation. Under the same condition,
expression of Connexin 43 did not change (Figure 6).
Noticeably, stimulation of melanocytes with one pulse
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
8129
Figure 6 HGF autocrine induces downregulation of both Ecadherin and Desmoglein 1 in melanocytes and E-cadherinpositive, Desmoglein-positive melanoma cells. Melanocyte
(FOM73) and melanoma lines WM164, WM35 were infected
with Ad.CMV.rhHGF (20 pfu/cell) or LacZ/Ad5 (20 pfu/ml).
72 h (FOM73) or 96 h (melanoma cells) later, whole-cell protein
extracts were subject to immunoblotting using anti-E-cadherin,
anti-Desmoglein 1 and anti-Connexin 43 antibodies. b-Actin was
used as a loading control. The numbers represent the quantitated
intensities relative to corresponding control experiments
Figure 5 HGF autocrine causes constitutive activation of c-Met,
MAPK and PI3K. 1205Lu melanoma cells were cultured in W489
medium supplemented with 2% FBS until 90% con¯uence. Then
the cells were starved in `protein free' medium for 24 h before
antibodies were added. Both control antibody (IGF-1 neutralizing
antibody) and HGF neutralizing antibody were initially added at
10 mg/ml, then replenished by adding additional 5 mg/ml every
24 h. PD98059 (10 mM) and wortmannin (30 mM) were added in
control samples to con®rm the speci®city. At selected time points,
cells were lysed and the lysates were subject to immunoprecipitation (for c-Met activation) and Western blots using antiphosphotyrosine, anti-p44/42 MAPK, anti-phospho-p44/42
MAPK (Thr202/Tyr204), anti-Akt and anti-phospho-Akt
(Ser473) antibodies
of recombinant protein at 50 ng/ml over 5 and 12 h did
not alter E-cadherin or Desmoglein expression (data
not shown), suggesting that continuous stimulation is
necessary for the observed changes. Because it is
known that HGF stimulation can initiate multiple
signal transduction pathways, including the MAPK
and PI3K cascades (Ponzetto et al., 1994), we were
interested to know whether the downregulation of
adhesion molecules depends on these pathways. When
HGF neutralizing antibodies (10 mg/ml) were added to
Ad.CMV.rhHGF transduced melanocytes, downregulation of E-cadherin and Desmoglein 1 was abolished
(Figure 7). The same e€ect was obtained in wortmannin (30 mM) treated cells (Figure 7). PD98059 (10 mM
and 50 mM, data not shown) was less e€ective in
blocking the downregulation (Figure 7), suggesting that
although both PI3K and MAPK are involved, PI3K
pathway might be more critical.
Desmoglein 1, E-cadherin, c-Met and Plakoglobin can be
co-immunoprecipitated from melanocytic cells
It has been previously shown that c-Met is co-localized
with b-catenin and E-cadherin at regions of cell ± cell
contact in human colon and breast cancer cell lines
(Hiscox and Jiang, 1999a; Kamei et al., 1999).
Immuno¯uorescence assays showed that c-Met, Ecadherin and Desmoglein 1 have similar subcellular
distribution in melanocytic cells (Figure 8a). Since all
of the above molecules are membrane-bound proteins,
membranous co-localization doesn't necessarily suggest
direct physical interaction. To test the possibility that
physical interactions underlie the crosstalk between
HGF/c-Met signaling and adhesion receptor regulation, we investigated the complex formation by
immunoprecipitation analyses. We were able to detect
Desmoglein 1, Plakoglobin, c-Met and E-cadherin in
the complexes pulled down by either anti-Desmoglein 1
antibody or anti-Plakoglobin antibody (Figure 8b) in
keratinocytes (FK113) and melanocytes (FOM73).
WM278 expresses no E-cadherin, but the other
components apparently also form complexes. The
speci®city of the interaction was con®rmed by probing
the same blots with anti-IGF-R1 antibody. IGF-R1
was not detected in the immunoprecipitants, but it was
Oncogene
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
8130
Figure 7 Downregulation of adhesion molecules by HGF is
MAP kinase and PI 3-kinase-dependent. Melanocytes (FOM73)
were transduced with Ad.CMV.rhHGF at 20 pfu/cell. In the
neutralizing antibody-treated group, antibody (10 mg/ml) were
added 12 h after transduction and replenished with additional
5 mg/ml for every 24 h. In inhibitors-treated groups, wortmannin
(30 mM) or PD98059 (10 mM) were added 12 h after transduction.
Seventy-two hours after adenoviral transduction, whole-cell
protein extracts were subject to immunoblotting using anti-Ecadherin and anti-Desmoglein 1 antibodies. b-Actin was used as a
loading control. The numbers represent the quantitated intensities
relative to corresponding control experiments
detectable in whole cell extract (data not shown). To
test whether HGF treatment has an e€ect on the
physical interaction, WM35 and WM164 melanoma
cells were treated with 100 ng/ml HGF for 30 min
before immunoprecipitation (Figure 8c). Using anti-cMet antibody, we were able to pull down E-cadherin
and Desmoglein 1, which further demonstrates the
existence of c-Met/E-cadherin/Desmoglein 1 complexes. HGF activation apparently has no substantial
e€ect on the complex formation (Figure 8c).
Discussion
Previous studies have documented the functional
importance of cell adhesion molecules in normal skin
homeostasis and in melanoma development (Hsu et al.,
2000; Johnson, 1999; Li and Herlyn, 2000). Here, we
provide evidence that crosstalk between cell adhesion
molecules and growth factor HGF is responsible for
dysregulated skin homeostasis and tumor development.
Speci®cally, the results demonstrate that: (1) Desmoglein 1 functions as a co-receptor for E-cadherin in
mediating cell adhesion between melanocytes and
keratinocytes. Although both E-cadherin and Desmoglein 1 are highly expressed in normal melanocytes,
they are absent or substantially reduced in melanoma
cells. (2) Melanoma cells, but not normal melanocytes,
express HGF, which causes sustained activation of the
HGF receptor c-Met and its downstream e€ectors
MAPK and AKT/PKB. (3) Prolonged HGF stimulation causes downregulation of E-cadherin and Desmoglein 1, which is MAPK- and PI3K-dependent. (4)
Desmoglein 1, E-cadherin, Plakoglobin and c-Met can
be co-immunoprecipitated from melanocytic cells,
which may underline the mechanism by which EOncogene
Figure 8 Coimmunopricipitation of c-Met, E-cadherin, Desmoglein 1 and Plakoglobin. (a) Subcellular localization of Ecadherin, Desmoglein1 and c-Met in WM35 melanoma cells by
indirect immuno¯uorescence. (b) Keratinocytes (FK113), melanocytes (FOM73) and melanoma line WM278 were lysed and
immunoprecipitated using either anti-Desmoglein 1 or antiPlakoglobin antibodies. Immunoprecipitants were separated on
gel, transferred onto PVDF membrane and probed with antiDesmoglein 1, anti-Plakoglobin, anti-c-Met and anti-E-cadherin
antibodies. (c) WM35 and WM164 melanoma cells were treated
with 100 ng/ml HGF or vehicle for 30 min, followed by cell lysis
and immunoprecipitation with anti-c-Met antibodies. Immunoprecipitants were separated on gel, transferred onto PVDF
membrane and probed with anti-E-cadherin, anti-Desmoglein 1,
anti-c-Met antibodies
cadherin and Desmoglein 1 are coordinately downregulated by c-Met activation.
Adhesive interactions between cells are dynamic and
regulated during tissue development and homeostasis
(Garrod et al., 1996; Green and Gaudry, 2000;
Gumbiner, 1996; Li and Herlyn, 2000; Steinberg,
1996; Takeichi, 1995). They not only glue cells
together, but also regulate intercellular and intracellular signaling pathways that control cell motility,
growth and di€erentiation. Cell scattering requires the
attenuation or dissociation of cell ± cell adhesion (Bauer
et al., 1998; Drubin and Nelson, 1996; Gumbiner,
1996). Previous studies have shown a progressive loss
of E-cadherin expression during melanoma development (Hsu et al., 1996; Johnson, 1999). In the present
study, we, for the ®rst time, demonstrated that
Desmoglein expression is greatly reduced in melanoma
cells. Although the levels of desmocollin expression in
melanoma cells are not clear, downregulation or
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
dysfunction of Desmoglein itself should be enough to
attenuate adhesion (Amagai et al., 1992, 1994, 1996;
Armstrong et al., 1999; Koch et al., 1997; Rickman et
al., 1999; Stanley, 1993). Since Desmoglein can
contribute to the adhesion mainly mediated by Ecadherin, reduction or loss of Desmoglein and Ecadherin at the same time may free the melanocytic
cells from microenvironmental control by keratinocytes, which contributes to the invasive and metastatic
behavior of melanoma (Hsu et al., 2000; Li and
Herlyn, 2000; Shinohara et al., 1998).
In contrast to other cancer types (for review, see
Hirohashi (1998)), our study and others (Poser et al.,
2001) showed that loss of E-cadherin in melanoma cell
lines does not usually involve mutations in the Ecadherin gene or promoter methylation. This prompted
us to investigate the possibility that abnormal growth
factor expression causes the downregulation of Ecadherin and/or Desmoglein. During the screen, we
found that a majority of melanoma cells tested, but not
normal primary melanocytes, synthesize and secrete
HGF. Physiologically, HGF is secreted by cells of
mesenchymal origin and acts as mitogen, motogen and
morphogen for many epithelial cells (Gherardi et al.,
1989; Nakamura et al., 1989; Stoker et al., 1987) and is
therefore considered a paracrine factor. Although the
c-Met signaling pathway has been implicated in human
melanomas (Hendrix et al., 1998; Natali et al., 1993),
there has been no report on how c-Met is activated or
the source of its ligand HGF. Our study showed that
HGF can be provided by not only ®broblasts
(paracrine), but also, more importantly, melanoma
cells themselves (autocrine). The latter appears to play
a much more important role in melanoma development
because ®broblasts constantly express HGF, even
under normal conditions. Endogenously expressed
HGF causes sustained activation of c-Met and its
downstream e€ectors MAPK and PI3K. Furthermore,
when E-cadherin- and Desmoglein-positive cells were
exposed to prolonged HGF treatment, the expression
levels of the two adhesion molecules were greatly
reduced. Interestingly, pulse treatment with HGF
didn't cause downregulation of either E-cadherin or
Desmoglein, suggesting that continuous stimulation is
necessary for the observed changes. It has been shown
previously that HGF could disrupt intercellular
junctions in normal and tumor cells (Hiscox and Jiang,
1999b; Stoker et al., 1987; Takeichi et al., 1994).
However, in those cases, HGF decreased cell ± cell
contact without apparent loss of E-cadherin expression, which may represent a rapid e€ect of HGF
stimulation occurring via post-translational modi®cation.
It has been shown that activation of both MAPK
(p42/p44) and PI3K is required for HGF-induced
motility (Marshall, 1996; Ponzetto et al., 1994;
Potempa and Ridley, 1998; Royal and Park, 1995).
Using the MAPK inhibitor PD98059 and the PI3K
inhibitor wortmannin, we showed that both pathways
are involved in the downregulation. However, the
downstream molecule that is directly responsible for
the downregulation is unclear. Among the candidate
genes are zinc ®nger transcription factors Snail and/or
Slug. Slug has been shown to play a critical role in
HGF-induced desmosome dissociation in a bladder
carcinoma cell line (Savagner et al., 1997). In
melanoma cells, activation of Snail expression plays
an important role in downregulation of E-cadherin
(Poser et al., 2001). The role of Snail/Slug in HGFinduced downregulation of E-cadherin and Desmoglein
is now under further investigation.
One important aspect of HGF is its mitogenic
activity (Vande Woude et al., 1997). However, we did
not detect a signi®cant increase in melanocytic cell
proliferation after either pulse or prolonged HGF
stimulation, which is in concert with previous reports
(Halaban et al., 1993; Rusciano et al., 1998), suggesting
that for melanocytic cells, HGF is not a potent
mitogen (Halaban et al., 1992; Tamatani et al., 1999).
The results also suggest that growth and motility are
separable events mediated by di€erent factors, pathways and mechanisms. IGF-1 (Rodeck et al., 1987) and
bFGF (Halaban et al., 1992; Meier et al., 2000) are
most important mitogens for melanocytic cells.
Besides the downregulation of E-cadherin and
Desmoglein, an HGF autocrine loop in melanoma
has other important implications. For example, a
recent study showed that activation of avb3 can be
induced by HGF (Trusolino et al., 1998). Upregulation of avb3 has been implicated in melanomas
(Albelda et al., 1990; Hsu et al., 1998; Van Belle et al.,
1999). This has implications for invasion processes,
particularly in the context that activation of c-Met by
HGF leads to invasiveness of melanocytic cells.
Another important observation made in this study is
the physical interaction among c-Met, E-cadherin,
Desmoglein 1 and Plakoglobin. It is not clear,
however, whether any one component can interact
with all others directly. It has been shown previously
that c-Met is co-localized with b-catenin and Ecadherin at regions of cell ± cell contact in human
colon and breast cancer cell lines (Hiscox and Jiang,
1999a; Kamei et al., 1999). Interestingly, c-erbB-2 also
co-immunoprecipitates with E-cadherin-catenin (b- and
g-catenin) complex (Kanai et al., 1995; Ochiai et al.,
1994), and EGF receptor physically interacts with bcatenin (Hoschuetzky et al., 1994), suggesting that the
physical interaction between tyrosine kinase receptors
and cell adhesion molecules may be a common scheme
underlying the functional crosstalk and regulation
between the two families of molecules (Klymkowsky
and Parr, 1995).
8131
Materials and methods
Cell culture
Human melanoma cells were isolated from clinically and
histologically de®ned lesions (Satyamoorthy et al., 1997).
Cells were maintained in medium W489, a 4 : 1 mixture of
MCDB153 and L15, supplemented with 2% heat-inactivated
Oncogene
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
8132
FBS and insulin (5 mg/ml) in a 378C, 5% CO2 atmosphere at
constant humidity. For growth in `protein-free' media, FBS
and insulin were omitted from the medium. Primary human
dermal ®broblasts were initiated as explant cultures from
trypsin-treated and epidermis-stripped neonatal foreskin, and
maintained in Dulbecco's Modi®ed Eagle's medium (DMEM)
supplemented with 10% FBS. Human melanocytes were
isolated from foreskin and maintained in MCDB153 medium
supplemented with FBS, endothelin-3 (ET-3) and stem cell
factor (SCF). Human keratinocytes were isolated from
foreskin and maintained in SFM (GIBCO-BRL). Transcomplementing 293 cells, a cell line immortalized and transformed
by adenovirus E1a and E1b, were obtained from the Vector
Core at the Institute for Human Gene Therapy (University of
Pennsylvania, Philadelphia, PA, USA) and grown in DMEM
with 10% FBS. All tissue culture reagents were purchased
from Sigma Chemical Co. (St. Louis, MO, USA) and
GIBCO-BRL (Gaithersburg, MD, USA).
Antibodies and reagents
Mouse anti-E-cadherin, -Desmoglein 1, -g-Catenin (Plakoglobin) and -phosphotyrosine (PY20) mAbs were from Transduction Laboratories, Inc. (Lexington, KY, USA). Mouse
anti-b-actin antibody was purchased from Sigma (St. Louis,
MO, USA). Anti-IGF-R1, anti-c-Met (c-12) and anti-c-Myc
(9E10) antibodies were from Santa Cruz Biotechnology Inc.
(Santa Cruz, CA, USA). Anti-FITC-and Cy3-conjugated
secondary antibodies were from Jackson Research Laboratories, Inc. (West Grove, PA, USA). HGF neutralizing
antibody, recombinant human HGF and HGF ELISA kit
were all purchased from R&D Systems (Minneapolis, MN,
USA). Anti-Akt, anti-phospho-Akt (Ser473), anti-p44/42
MAPK and anti-phospho-p44/42 MAPK (Thr202/Tyr204)
antibodies were purchased from Cell Signaling Technology,
Inc (Beverly, MA, USA). Inhibitors PD98059 and wortmannin were purchased from Sigma (St. Louis, MO, USA).
Adenoviral vectors
The recombinant adenoviral vector Ad.CMV.rhHGF was
kindly provided by Dr James Wilson of the University of
Pennsylvania, Philadelphia, PA, USA (Phaneuf et al., 2000).
The adenoviral vectors AxCANCre (Kanegae et al., 1995,
1996; Miyake et al., 1996), AxCALNLDsg3DEC and
AxCALNLDEcadDEC were described (Hanakawa et al.,
2000). Plaque-puri®ed virus propagated in 293 cells was
puri®ed by ultracentrifugation in a cesium chloride gradient.
Viral titer was assayed by plaque formation in permissive 293
cells. Control adenoviral vector containing the LacZ cDNA
(LacZ/Ad5) was purchased from Vector Core at the Institute
for Human Gene Therapy, University of Pennsylvania,
Philadelphia, PA, USA.
Viral infection of melanocytes and melanoma cells
Optimal viral concentrations for infection were determined as
the amounts of virus required to yield the desired levels of
gene expression without apparent alteration in cellular
phenotype or toxicity. Subcon¯uent melanocytes and melanoma cells were infected with adenoviruses for 2 ± 4 h at 378C
in protein-free W489 medium. Viral suspension was then
replaced with culture medium. Because dominant negative
Desmoglein-3 vector AxCALNLDsg3DEC and E-cadherin
vector AxCALNLEcadDEC were made using the Cre-loxP
system (Hanakawa et al., 2000), AxCANCre was co-infected
with AxCALNLDsg3DEC or AxCALNLEcadDEC at a ratio
Oncogene
of 1 : 1, respectively. Cells were allowed to recover for at least
24 h before assay.
Immunoblotting
For immunoblotting, cells were washed with PBS and
harvested in RIPA bu€er. Total protein concentrations were
measured using the bicinchoninic acid (BCA) assay (Pierce
Chemical Co. Rockford, IL, USA). Samples were loaded at
15 mg/lane and separated on 8% SDS-polyacrylamide gels
and transferred to polyvinylidene di¯uoride (PVDF) membranes and probed with speci®c primary antibodies. To detect
the signal, peroxidase-conjugated secondary antibody was
added, followed by exposure using enhanced chemiluminescence (ECL) (Amersham, Arlington Heights, IL, USA). Some
of the immunoblots were quanti®ed using NIH Image
(version 1.61).
Immunofluorescence
Melanocytes and melanoma cells were seeded in 8-well
chamber slides (Lab-Tek, Nunc, Inc., Naperville, IL, USA).
Cells were washed with cold PBS, ®xed in 4% paraformaldehyde at 48C for 5 min, permeabilized with 0.5% Triton X-100
in PBS for 20 min at room temperature, blocked with 5%
BSA, and incubated with primary antibody (anti-Desmoglein
IgG1, dilution 1 : 50), followed by biotin-conjugated goat
anti-mouse IgG1 and Cy3-conjugated streptavidin. Primary
and secondary antibody incubations were performed at room
temperature for 1 h with three washings in between. Cells
were mounted in anti-fade medium Gel Mount (Biomeda
Corp.) and examined by ¯uorescence microscopy.
Immunoprecipitation
To investigate the status of c-Met phosphorylation, subcon¯uent melanoma cells were infected with LacZ or
Ad.CMV.rhHGF (20 pfu/cell) and cultured for 48 h. In
neutralizing antibody-treated group, HGF neutralizing antibody (®nal concentration 10 mg/ml) was added to serum-free
medium 24 h before lysis. Cells were washed with cold PBS
containing 1 mM Na3VO4, and rapidly scraped into 200 ml
PBS containing 1% Triton X-100, 1% NP40, 1 mM Na3VO4,
1 mM phenylmethyl sulfonyl ¯uoride (PMSF), and protease
inhibitors (1 mg/ml leupeptin, aprotinin and pepstatin). After
incubation on ice for 15 min, the extracts were separated by
centrifugation at 13 000 g for 10 min. The supernatants were
used for immunoprecipitation using an anti-c-Met antibody
(5 mg) for 1 h at 48C with shaking. Protein A ± G sepharose
CL-4B beads (Pharmacia Biotech, Uppsala, Sweden) were
then added and incubated for 12 h. Samples were washed
four times with lysis bu€er, boiled in Laemmli bu€er
containing b-mercaptoethanol, and subjected to electrophoresis on an 8% SDS-polyacrylamide gel. Separated proteins
were transferred onto PVDF membrane and detected by antiphosphotyrosine primary antibody (PY20) and peroxidaseconjugated secondary antibody. Signals were detected using
ECL.
To investigate c-Met/Desmoglein/E-cadherin complexes,
cells cultured in their optimal media were lysed in 10 mM
Tris-HCl (pH 7.4), 1% NP-40, 1% Triton X-100, 2 mM
CaCl2, 150 mM NaCl, 1 mM PMSF and protease inhibitors
(1 mg/ml leupeptin, aprotinin and pepstatin). The extracts
were immunoprecipitated using either anti-Desmoglein 1 or
anti-Plakoglobin antibody in the same condition described
above. Immunoprecipitants were separated on gels, transferred onto PVDF membrane and probed with anti-
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
Desmoglein 1, -Plakoglobin, -c-Met and -E-cadherin antibodies.
Detection of HGF expression by RT ± PCR and ELISA
Total RNA was isolated from the cell lines using Trizol
Reagent (Gibco-BRL). Reverse transcription was carried
using the Superscript II system (Gibco-BRL). PCR primers
for ampli®cation of HGF were: forward primer: 5'-TCCCCATCGCCATCCCC-3'; reverse primer: 5'-CACCATGGCCTCGGCTGG-3'. The expected size of PCR product is
749 bp. Internal control primers (forward: 5'-TGACGGGGTCACCCACACTGTGCCCATCTA-3; reverse: 5'-CTAGAAGCATTTGCGGTGGACGATGGAGGG-3') ampli®ed a
650 bp fragment of b-actin. PCR was performed using
1.5 U Taq DNA polymerase (Gibco-BRL), 0.4 mM dNTPs
(Pharmacia), 2 mM MgCl2 in 16PCR bu€er (Gibco-BRL).
PCR was started with a 5-min denaturation at 958C, after
which ampli®cation was performed for 30 cycles of
denaturation at 948C for 1 min, annealing at 658C for
1.5 min, and elongation at 728C for 1 min. Samples were
analysed by electrophoresis in 1.5% agarose gels containing
0.5 mg/ml ethidium bromide.
To detect secreted HGF by cells, conditioned media were
collected, centrifuged to remove cellular debris, diluted to
appropriate concentrations and subjected to enzyme linked
immunosorbent assay (ELISA). ELISA was performed
according to the conditions suggested by the manufacturer
(R&D Systems, Minneapolis, MN, USA).
Cell adhesion assay
Melanocytes or E-cadherin-positive, Desmoglein-positive
melanoma cells infected with indicated adenoviruses were
pre-labeled with a ¯uorescent dye DiI (10 mg/ml; Molecular
Probes, Eugene, OR, USA) for 2 h, washed with HBSS and
harvested by treatment with 0.01% trypsin in HBSS containing 1 mM calcium for 30 min at 378C. Under these
conditions, cadherins are speci®cally protected from proteolytic digestion. Cells were then resuspended in assay medium
(HBSS containing 1% bovine serum albumin and 1 mM
calcium). About 5000 cells were added to keratinocyte
monolayer in 8-well chamber slides and allowed to adhere
for 30 min. After removal of nonadherent cells, slides were
®xed. Numbers of adherent cells per high power ®eld in
triplicate wells were counted under a ¯uorescence microscope.
8133
Abbreviations
AKT, murine thymoma viral (v-akt) oncogene homologue;
bFGF, basic ®broblast growth factor; DMEM, Dulbecco's
modi®ed Eagle's medium; Dsg3DEC, Desmoglein 3 mutant
with extracellular domain deleted; E-cadDEC, E-cadherin
mutant with extracellular domain deleted; ELISA, enzyme
linked immunosorbent assay; FBS, fetal bovine serum;
HGF, hepatocyte growth factor; IGF-R1, insulin-like
growth factor receptor 1; MAPK, mitogen-activated
protein kinase; PI3K, phosphatidyl-inositol-3-kinase.
Acknowledgments
We thank Dr James Wilson (University of Pennsylvania)
for providing recombinant adenoviral vector
Ad.CMV.rhHGF and Dr Izumu Saito (University of
Tokyo, Japan) for AxCANCre. We also thank Rena Finko
for help in cell culture and Ruth Snyder for reading the
manuscript. This work was supported by National Institutes of Health Grants CA-25874, CA-76674, CA-47159,
CA-80999 and CA-10815.
References
Aberle H, Butz S, Stappert J, Weissig H, Kemler R and
Hoschuetzky H. (1994). J. Cell. Sci., 107, 3655 ± 3663.
Albelda SM, Mette SA, Elder DE, Stewart R, Damjanovich
L, Herlyn M and Buck CA. (1990). Cancer Res., 50, 6757 ±
6764.
Amagai M, Karpati S, Klaus-Kovtun V, Udey MC and
Stanley JR. (1994). J. Invest. Dermatol., 102, 402 ± 408.
Amagai M, Karpati S, Prussick R, Klaus-Kovtun V and
Stanley JR. (1992). J. Clin. Invest., 90, 919 ± 926.
Amagai M, Koch PJ, Nishikawa T and Stanley JR. (1996). J.
Invest. Dermatol., 106, 351 ± 355.
Armstrong DK, McKenna KE, Purkis PE, Green KJ, Eady
RA, Leigh IM and Hughes AE. (1999). Hum. Mol. Genet.,
8, 143 ± 148.
Bauer A, Lickert H, Kemler R and Stappert J. (1998). J. Biol.
Chem., 273, 28314 ± 28321.
Bellusci S, Moens G, Gaudino G, Comoglio P, Nakamura T,
Thiery JP and Jouanneau J. (1994). Oncogene, 9, 1091 ±
1099.
Bornslaeger EA, Corcoran CM, Stappenbeck TS and Green
KJ. (1996). J. Cell. Biol., 134, 985 ± 1001.
Bottaro DP, Rubin JS, Faletto DL, Chan AM, Kmiecik TE,
Vande Woude GF and Aaronson SA. (1991b). Science,
251, 802 ± 804.
Bottaro DP, Rubin JS, Faletto DL, Chan AM, Kmiecik TE,
Vande Woude GF and Aaronson SA. (1991a). Science,
251, 802 ± 804.
Butz S and Kemler R. (1994). FEBS Lett., 355, 195 ± 200.
Conn IG, Vilela MJ, Garrod DR, Crocker J and Wallace
DM. (1990). Br. J. Urol., 65, 176 ± 180.
Cowin P, Kapprell HP, Franke WW, Tamkun J and Hynes
RO. (1986). Cell, 46, 1063 ± 1073.
Danen EH, de Vries TJ, Morandini R, Ghanem GG, Ruiter
DJ and van Muijen GN. (1996). Melanoma. Res., 6, 127 ±
131.
De Bruin A, Muller E, Wurm S, Caldelari R, Wyder M,
Wheelock MJ and Suter MM. (1999). Cell. Adhes.
Commun., 7, 13 ± 28.
Depondt J, Shabana AH, Florescu-Zorila S, Gehanno P and
Forest N. (1999). Eur. J. Oral. Sci., 107, 183 ± 193.
Drubin DG and Nelson WJ. (1996). Cell, 84, 335 ± 344.
Franke WW, Goldschmidt MD, Zimbelmann R, Mueller
HM, Schiller DL and Cowin P. (1989). Proc. Natl. Acad.
Sci. USA, 86, 4027 ± 4031.
Garrod D, Chidgey M and North A. (1996). Curr. Opin. Cell.
Biol., 8, 670 ± 678.
Garrod DR. (1993). Curr. Opin. Cell. Biol., 5, 30 ± 40.
Gherardi E, Gray J, Stoker M, Perryman M and Furlong R.
(1989). Proc. Natl. Acad. Sci. USA, 86, 5844 ± 5848.
Green KJ and Gaudry CA. (2000). Nat. Rev. Mol. Cell. Biol.,
1, 208 ± 216.
Gumbiner BM. (1996). Cell, 84, 345 ± 357.
Oncogene
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
8134
Oncogene
Halaban R, Rubin JS, Funasaka Y, Cobb M, Boulton T,
Faletto D, Rosen E, Chan A, Yoko K, White W, Cook C
and Moellmann G. (1992). Oncogene, 7, 2195 ± 2206.
Halaban R, Rubin JS and White W. (1993). Exs., 65, 329 ±
339.
Hanakawa Y, Amagai M, Shirakata Y, Sayama K and
Hashimoto K. (2000). J. Cell. Sci., 113, 1803 ± 1811.
Hendrix MJ, Seftor EA, Seftor RE, Kirschmann DA,
Gardner LM, Boldt HC, Meyer M, Pe'er J and Folberg
R. (1998). Am. J. Pathol., 152, 855 ± 863.
Hirohashi S. (1998). Am. J. Pathol., 153, 333 ± 339.
Hiscox S and Jiang WG. (1999a). Biochem. Biophys. Res.
Commun., 261, 406 ± 411.
Hiscox S and Jiang WG. (1999b). Anticancer Res., 19, 509 ±
517.
Hoschuetzky H, Aberle H and Kemler R. (1994). J. Cell.
Biol., 127, 1375 ± 1380.
Hsu MY, Meier FE, Nesbit M, Hsu JY, Van Belle P, Elder
DE and Herlyn M. (2000). Am. J. Pathol., 156, 1515 ±
1525.
Hsu MY, Shih DT, Meier FE, Van Belle P, Hsu JY, Elder
DE, Buck CA and Herlyn M. (1998). Am. J. Pathol., 153,
1435 ± 1442.
Hsu MY, Wheelock MJ, Johnson KR and Herlyn M. (1996).
J. Investig. Dermatol. Symp. Proc., 1, 188 ± 194.
Imokawa G, Yada Y, Morisaki N and Kimura M. (1998).
Biochem. J., 330, 1235 ± 1239.
Iyer A, Kmiecik TE, Park M, Daar I, Blair D, Dunn KJ,
Sutrave P, Ihle JN, Bodescot M and Vande Woude GF.
(1990). Cell Growth Di€er., 1, 87 ± 95.
Johnson JP. (1999). Cancer Metastasis Rev., 18, 345 ± 357.
Jouneau A, Yu YQ, Pasdar M and Larue L. (2000). Pigment
Cell. Res., 13, 260 ± 272.
Kamei T, Matozaki T, Sakisaka T, Kodama A, Yokoyama
S, Peng YF, Nakano K, Takaishi K and Takai Y. (1999).
Oncogene, 18, 6776 ± 6784.
Kanai Y, Ochiai A, Shibata T, Oyama T, Ushijima S,
Akimoto S and Hirohashi S. (1995). Biochem. Biophys.
Res. Commun., 208, 1067 ± 1072.
Kanegae Y, Lee G, Sato Y, Tanaka M, Nakai M, Sakaki T,
Sugano S and Saito I. (1995). Nucleic Acids Res., 23,
3816 ± 3821.
Kanegae Y, Takamori K, Sato Y, Lee G, Nakai M and Saito
I. (1996). Gene, 181, 207 ± 212.
Klymkowsky MW and Parr B. (1995). Cell, 83, 5 ± 8.
Koch PJ and Franke WW. (1994). Curr. Opin. Cell. Biol., 6,
682 ± 687.
Koch PJ, Mahoney MG, Ishikawa H, Pulkkinen L, Uitto J,
Shultz L, Murphy GF, Whitaker-Menezes D and Stanley
JR. (1997). J. Cell. Biol., 137, 1091 ± 1102.
Kos L, Aronzon A, Takayama H, Maina F, Ponzetto C,
Merlino G and Pavan W. (1999). Pigment Cell. Res., 12,
13 ± 21.
Kouklis PD, Hutton E and Fuchs E. (1994). J. Cell. Biol.,
127, 1049 ± 1060.
Kowalczyk AP, Bornslaeger EA, Borgwardt JE, Palka HL,
Dhaliwal AS, Corcoran CM, Denning MF and Green KJ.
(1997). J. Cell. Biol., 139, 773 ± 784.
Kowalczyk AP, Bornslaeger EA, Norvell SM, Palka HL and
Green KJ. (1999). Int. Rev. Cytol., 185, 237 ± 302.
Krunic AL, Garrod DR, Smith NP, Orchard GS and Cvijetic
OB. (1996). Acta Derm. Venereol., 76, 394 ± 398.
Lewis JE, Wahl JK, Sass KM, Jensen PJ, Johnson KR and
Wheelock MJ. (1997). J. Cell. Biol., 136, 919 ± 934.
Li G and Herlyn M. (2000). Mol. Med. Today, 6, 163 ± 169.
Li G, Satyamoorthy K and Herlyn M. (2001). Cancer Res.,
61, 3819 ± 3825.
Marshall CJ. (1996). Curr. Opin. Cell. Biol., 8, 197 ± 204.
Mathur M, Goodwin L and Cowin P. (1994). J. Biol. Chem.,
269, 14075 ± 14080.
Matsumoto K, Tajima H and Nakamura T. (1991). Biochem.
Biophys. Res. Commun., 176, 45 ± 51.
Meier F, Nesbit M, Hsu MY, Martin B, Van Belle P, Elder
DE, Schaumburg-Lever G, Garbe C, Walz TM, Donatien
P, Crombleholme TM and Herlyn M. (2000). Am. J.
Pathol., 156, 193 ± 200.
Meier F, Satyamoorthy K, Nesbit M, Hsu MY, Schittek B,
Garbe C and Herlyn M. (1998). Front. Biosci., 3, D1005 ±
D1010.
Miyake S, Makimura M, Kanegae Y, Harada S, Sato Y,
Takamori K, Tokuda C and Saito I. (1996). Proc. Natl.
Acad. Sci. USA, 93, 1320 ± 1324.
Moriyama T, Kataoka H, Koono M and Wakisaka S. (1999).
Int. J. Mol. Med., 3, 531 ± 536.
Nakamura T, Nishizawa T, Hagiya M, Seki T, Shimonishi
M, Sugimura A, Tashiro K and Shimizu S. (1989). Nature,
342, 440 ± 443.
Natali PG, Nicotra MR, Di Renzo MF, Prat M, Bigotti A,
Cavaliere R and Comoglio PM. (1993). Br. J. Cancer, 68,
746 ± 750.
Noonan FP, Otsuka T, Bang S, Anver MR and Merlino G.
(2000). Cancer Res., 60, 3738 ± 3743.
North AJ, Bardsley WG, Hyam J, Bornslaeger EA,
Cordingley HC, Trinnaman B, Hatzfeld M, Green KJ,
Magee AI and Garrod DR. (1999). J. Cell. Sci., 112,
4325 ± 4336.
Ochiai A, Akimoto S, Kanai Y, Shibata T, Oyama T and
Hirohashi S. (1994). Biochem. Biophys. Res. Commun.,
205, 73 ± 78.
Otsuka T, Takayama H, Sharp R, Celli G, LaRochelle WJ,
Bottaro DP, Ellmore N, Vieira W, Owens JW, Anver M
and Merlino G. (1998). Cancer Res., 58, 5157 ± 5167.
Parker SL, Tong T, Bolden S and Wingo PA. (1997). CA
Cancer J. Clin., 47, 5 ± 27.
Phaneuf D, Chen SJ and Wilson JM. (2000). Mol. Med., 6,
96 ± 103.
Ponzetto C, Bardelli A, Zhen Z, Maina F, dalla Zonca P,
Giordano S, Graziani A, Panayotou G and Comoglio PM.
(1994). Cell, 77, 261 ± 271.
Poser I, Dominguez D, de Herreros AG, Varnai A, Buettner
R and Bosserho€ AK. (2001). J. Biol. Chem., 25, 25.
Potempa S and Ridley AJ. (1998). Mol. Biol. Cell., 9, 2185 ±
2200.
Rickman L, Simrak D, Stevens HP, Hunt DM, King IA,
Bryant SP, Eady RA, Leigh IM, Arnemann J, Magee AI,
Kelsell DP and Buxton RS. (1999). Hum. Mol. Genet., 8,
971 ± 976.
Ries LA, Wingo PA, Miller DS, Howe HL, Weir HK,
Rosenberg HM, Vernon SW, Cronin K and Edwards BK.
(2000). Cancer, 88, 2398 ± 2424.
Rodeck U, Herlyn M, Menssen HD, Furlanetto RW and
Koprowsk H. (1987). Int. J. Cancer, 40, 687 ± 690.
Roh JY and Stanley JR. (1995). J. Invest. Dermatol., 104,
720 ± 724.
Rong S, Bodescot M, Blair D, Dunn J, Nakamura T, Mizuno
K, Park M, Chan A, Aaronson S and Vande Woude GF.
(1992). Mol. Cell. Biol., 12, 5152 ± 5158.
Rosen EM, Nigam SK and Goldberg ID. (1994). J. Cell.
Biol., 127, 1783 ± 1787.
Royal I and Park M. (1995). J. Biol. Chem., 270, 27780 ±
27787.
Rusciano D, Lin S, Lorenzoni P, Casella N and Burger MM.
(1998). Tumour Biol., 19, 335 ± 345.
Autocrine HGF disrupts cell adhesion in melanoma
G Li et al
Satyamoorthy K, DeJesus E, Linnenbach AJ, Kraj B,
Kornreich DL, Rendle S, Elder DE and Herlyn M.
(1997). Melanoma. Res., 7(Suppl 2): S35 ± S42.
Savagner P, Yamada KM and Thiery JP. (1997). J. Cell.
Biol., 137, 1403 ± 1419.
Scott GA and Cassidy L. (1998). J. Invest. Dermatol., 111,
243 ± 250.
Shinohara M, Hiraki A, Ikebe T, Nakamura S, Kurahara S,
Shirasuna K and Garrod DR. (1998). J. Pathol., 184,
369 ± 381.
Silye R, Karayiannakis AJ, Syrigos KN, Poole S, van
Noorden S, Batchelor W, Regele H, Sega W, Boesmueller
H, Krausz T and Pignatelli M. (1998). J. Pathol., 186,
350 ± 355.
Smith EA and Fuchs E. (1998). J. Cell Biol., 141, 1229 ± 1241.
Sonnenberg E, Meyer D, Weidner KM and Birchmeier C.
(1993). J. Cell Biol., 123, 223 ± 235.
Stanley JR. (1993). Adv. Immunol., 53, 291 ± 325.
Stappenbeck TS and Green KJ. (1992). J. Cell Biol., 116,
1197 ± 1209.
Steinberg MS. (1996). Dev. Biol., 180, 377 ± 388.
Stoker M, Gherardi E, Perryman M and Gray J. (1987).
Nature, 327, 239 ± 242.
Tada H, Hatoko M, Tanaka A, Kuwahara M and
Muramatsu T. (2000). J. Cutan. Pathol., 27, 24 ± 29.
Takayama H, La Rochelle WJ, Anver M, Bockman DE and
Merlino G. (1996). Proc. Natl. Acad. Sci. USA, 93, 5866 ±
5871.
Takayama H, LaRochelle WJ, Sharp R, Otsuka T, Kriebel P,
Anver M, Aaronson SA and Merlino G. (1997). Proc.
Natl. Acad. Sci. USA, 94, 701 ± 706.
Takeichi M. (1995). Curr. Opin. Cell Biol., 7, 619 ± 627.
Takeichi M, Watabe M, Shibamoto S and Ito F. (1994).
Princess Takamatsu Symp., 24, 28 ± 37.
Tamatani T, Hattori K, Iyer A, Tamatani K and Oyasu R.
(1999). Carcinogenesis, 20, 957 ± 962.
Tang A, Eller MS, Hara M, Yaar M, Hirohashi S and
Gilchrest BA. (1994). J. Cell Sci., 107, 983 ± 992.
To CT and Tsao MS. (1998). Oncol. Rep., 5, 1013 ± 1024.
Troyanovsky SM and Leube RE. (1998). Subcell. Biochem.,
31, 263 ± 289.
Trusolino L, Serini G, Cecchini G, Besati C, AmbesiImpiombato FS, Marchisio PC and De Filippi R. (1998).
J. Cell. Biol., 142, 1145 ± 1156.
Van Belle PA, Elenitsas R, Satyamoorthy K, Wolfe JT,
Guerry, DT, Schuchter L, Van Belle TJ, Albelda S, Tahin
P, Herlyn M and Elder DE. (1999). Hum. Pathol., 30,
562 ± 567.
Vande Woude GF, Je€ers M, Cortner J, Alvord G, Tsarfaty
I and Resau J. (1997). Ciba. Found. Symp., 212, 119 ± 130.
8135
Oncogene