Crosstalk between tumor and endothelial cells promotes tumor angiogenesis by MAPK activation of Notch signaling.

Abstract

While significant progress has been made in understanding the induction of tumor vasculature by secreted angiogenic factors, little is known regarding contact-dependent signals that promote tumor angiogenesis. Here, we report that the Notch ligand Jagged1 induced by growth factors via mitogen-activating protein kinase (MAPK) in head and neck squamous cell carcinoma (HNSCC) cells triggered Notch activation in neighboring endothelial cells (ECs) and promoted capillary-like sprout formation. Jagged1-expressing HNSCC cells significantly enhanced neovascularization and tumor growth in vivo. Moreover, the level of Jagged1 was significantly correlated with tumor blood vessel content and associated with HNSCC development. Our results elucidate a novel mechanism by which the direct interplay between tumor cells and ECs promotes angiogenesis through MAPK and Notch signaling pathways.

Full text

Page 1

A R T I C L E
Crosstalk between tumor and endothelial cells promotes tumor
angiogenesis by MAPK activation of Notch signaling
Qinghua Zeng,1Shenglin Li,6Douglas B. Chepeha,2Thomas J. Giordano,3Jong Li,1Honglai Zhang,1
Peter J. Polverini,4Jacques Nor,5Jan Kitajewski,7and Cun-Yu Wang1,*
1Laboratory of Molecular Signaling and Apoptosis, Department of Biologic and Materials Sciences
2Department of Otorhinolaryngology
3Cancer Center Tissue Core
4Department of Oral Medicine, Pathology, and Oncology
5Department of Cariology, Restorative Sciences, and Endodontics, School of Dentistry and Medicine, University of Michigan, Ann
Arbor, Michigan 48109
6Department of Oral and Maxillofacial Surgery, Peking University School of Stomatology, Beijing 100081, People’s Republic of China
7Department of Pathology and Obstetrics and Gynecology, Columbia University, College of Physicians and Surgeons, New York,
New York 10032
*Correspondence: cunywang@umich.edu
Summary
While significant progress has been made in understanding the induction of tumor vasculature by secreted angiogenic
factors, little is known regarding contact-dependent signals that promote tumor angiogenesis. Here, we report that the
Notch ligand Jagged1 induced by growth factors via mitogen-activating protein kinase (MAPK) in head and neck
squamous cell carcinoma (HNSCC) cells triggered Notch activation in neighboring endothelial cells (ECs) and promoted
capillary-like sprout formation. Jagged1-expressing HNSCC cells significantly enhanced neovascularization and tumor
growth in vivo. Moreover, the level of Jagged1 was significantly correlated with tumor blood vessel content and associated
with HNSCC development. Our results elucidate a novel mechanism by which the direct interplay between tumor cells
and ECs promotes angiogenesis through MAPK and Notch signaling pathways.
cer, demonstrating the potential for cancer therapy based upon
blocking angiogenesis (Ferrara et al., 2004). However, targeting
VEGF for human cancer therapy has not been successful in a
variety of other tumor types, suggesting that other factors or
components may also play a critical role in tumor angiogenesis
(Jung et al., 2002; Kerbel and Kamen, 2004). The identification
of those factors and components may have important implica-
tions in human cancer therapy.
Angiogenic growth factors secreted by tumor cells or tumor
stromal cells can directly bind to their receptors on ECs and
stimulate angiogenesis by promoting endothelial sprouting,
branching, differentiation, and survival (Folkman, 2002; Spar-
mann and Bar-Sagi, 2004). For example, VEGF secreted by tu-
mor cells specifically binds to its receptors, VEGF receptors 1
and 2, in ECs (Ferrara et al., 2004). Several important signaling
pathways, including mitogen-activating protein kinase (MAPK)/
extracellular signal-regulated kinase (ERK) and phosphoinositol
3-kinase (PI3K)/Akt, have been found to be induced by angio-
Introduction
Tumor angiogenesis is a complex process in which new blood
vessels are formed in response to interactions between tumor
cells and endothelial cells (ECs), growth factors, and extracel-
lular matrix components. Tumor vessels promote growth and
progression of human solid tumors, including head and neck
squamous cell carcinoma (HNSCC). New tumor blood vessels
penetrate into cancerous growths, supplying nutrients and ox-
ygen and removing waste products (Jung et al., 2002; Folk-
man, 2002; Kerbel and Kamen, 2004; Stupack and Cheresh,
2004). A large number of studies have demonstrated that tu-
mor cells secrete angiogenic growth factors to stimulate EC
proliferation and to induce angiogenesis. Among them, vascu-
lar endothelial growth factor (VEGF) is one of the most potent
angiogenic factors, and it is overexpressed in many human
cancers (Jung et al., 2002). Targeting VEGF for human cancer
therapy has shown promise in the treatment of colorectal can-
S I G N I F I C A N C E
Angiogenesis plays a critical role in human tumor growth and development. It is well known that tumor cells can stimulate angiogen-
esis by secreting proangiogenic factors such as vascular endothelial growth factor (VEGF) and interleukin-8 (IL-8). However, tar-
geting these factors in human cancer therapy has not been very effective, suggesting that other factors or components may also
play a critical role in tumor angiogenesis. In this study, we show that tumor-associated growth factors stimulate the direct interaction
between tumor cells and ECs via MAPK and Notch signaling pathways, thereby promoting tumor angiogenesis and tumor growth.
Our findings underscore the importance of the direct interplay between tumor cells and ECs in tumor angiogenesis, providing a
target for antiangiogenic therapy.
CANCER CELL : JULY 2005 · VOL. 8 · COPYRIGHT © 2005 ELSEVIER INC.DOI 10.1016/j.ccr.2005.06.00413

Page 2

A R T I C L E
genic growth factors in ECs (Shaulian and Karin, 2002). Activa-
tion of these pathways can promote the migration, prolifera-
tion, differentiation, and survival of ECs (Jung et al., 2002). Re-
cently, genetic studies have found that Notch signaling plays a
critical role in vascular formation during early embryonic devel-
opment (Xue et al., 1999; Krebs et al., 2000; Lawson et al.,
2002; Krebs et al., 2004; Gale et al., 2004; Duarte et al., 2004).
Notch was initially identified as a neurogenic gene in Drosoph-
ila (Artavanis-Tsakonas et al., 1999). Extensive studies have
demonstrated that the Notch signaling pathway is highly con-
served and plays a critical role in the specification of cell fate
during development. In human, there are four notch receptor
genes (Notch1 to Notch4) and five ligands, two jagged genes
(jagged1 and jagged2), and three delta genes (delta-like1,
delta-like3, and delta-like4). Notch signaling is triggered by the
direct interaction of cells expressing Notch receptors with cells
that express Notch ligands at their surface. Upon Notch recep-
tor-ligand binding, the Notch intracellular domain (Notch-ICD)
is cleaved from the membrane by γ-secretase and other puta-
tive Notch processing enzymes. Subsequently, Notch-ICD is
translocated to the nucleus, where it binds to RBPJκ [the
mammalian homolog of Su (H), also known as CBF1 or CSL]
to activate transcription of Notch target genes (Lindsell et al.,
1995; Kato et al., 1997; Mumm et al., 2000). Importantly, Notch
receptors and its ligands have been found to be expressed in
ECs (Villa et al., 2001; Liu et al., 2003; Shawber and Kitajewski,
2004). However, the role of Notch signaling in tumor angiogen-
esis has not been explored.
Squamous cell carcinoma (SCC) is one of the most common
cancers in the lung, skin, oral cavity, and head and neck re-
gions. SCC is a very malignant tumor. For example, the 5 year
survival rate for patients with HNSCC is one of the lowest for
any major cancer and has not been significantly improved over
the past decades (Wong et al., 1996; Patel et al., 2001; Foras-
tiere et al., 2001; Mao et al., 2004). The molecular mechanisms
that control the development and progression of HNSCC are
not fully understood. It is known that the MAPK signaling path-
way has been chronically activated in HNSCC (Patel et al.,
2001; Mao et al., 2004). Several growth factors and their recep-
tors, including epidermal growth factor (EGF), transforming
growth factor α (TGFα), and their receptor, EGF receptor
(EGFR), and hepatocyte growth factor (HGF) and its receptor,
c-Met, have also been found to be associated with HNSCC
development and progression (Patel et al., 2001; Zeng et al.,
2002a). All these growth factors can potently promote SCC cell
proliferation by activating the MAPK signaling pathway. Also,
the activation of MAPK by these growth factors induced SCC
cells to secrete proangiogenic factors to promote tumor angio-
genesis (Bancroft et al., 2001). Recently, we have found that
HGF potently stimulated the MAPK signaling pathway to pro-
mote SCC cell survival (Zeng et al., 2002a; Zeng et al., 2002b).
In this study, we identified that MAPK activated by growth fac-
tors including HGF initiated a novel signaling crosstalk be-
tween SCC cells and ECs that promoted tumor angiogenesis
and tumor growth.
Figure 1. HGF induces Jagged1 expression in SCC cells
A and B: HGF-treated SCC9 cells enhanced endothelial sprout formation.
SCC9 cells were untreated or treated with HGF for 4 hr and replaced with
the conditioned EGM media (serum-free). After 24 hr, the conditioned me-
dia were harvested for EC culture. ECs (0.8 × 104) were plated on the
growth factor-reduced Matrigel and incubated with conditioned media.
After 24 hr, cells were fixed with methanol and stained with Diff-Quick solu-
tion II. For coculture experiments, SCC9 cells were untreated or treated
with HGF for 4 hr. Afterwards, ECs (0.8 × 104) and HGF-treated SCC9 cells
(SCC9/H; 0.8 × 104) or control SCC9 cells (0.8 × 104) were plated on the
Matrigel. After 24 hr, cells were fixed and stained with Diff-Quick solution II.
The capillary-like sprouts were counted from five random microscopic
fields (×200) and averaged. The assays were performed in duplicate, and
the results represent mean values ± SD (error bars) from three independent
experiments. Student’s t test was performed to determine statistical signifi-
cance. *p < 0.01, EC + SCC9/H versus EC + CM, EC + SCC9, or EC. Scale
bar, 50 ?m.
C: HGF induced Jagged1 by microarray. Microarray was performed as
described previously (Zeng et al., 2002b).
D: HGF induced Jagged1 in SCC cells by Western blot analysis. SCC1 and
SCC9 cells were treated with HGF for the indicated time periods. Fifty
microgram aliquots of protein extracts were probed with anti-Jagged1
(1:1000) by Western blot analysis. For loading control, the membranes were
stripped and reprobed with anti-α-tubulin (1:7500).
HGF stimulated SCC cells to release angiogenic factors to
stimulate endothelial differentiation, thereby promoting tumor
angiogenesis. Because tumor angiogenesis involves multiple
factors, including ECs, extracellular matrix, and tumor cells, a
growth factor-reduced Matrigel model system for endothelial
differentiation was utilized (Kumar et al., 2004; Liu et al., 2003).
In this model system, upon proangiogenic stimulation, ECs
proliferate, migrate, and organize into capillary-like sprouts that
mimic stages in the development of microvessels in vivo. As
shown in Figures 1A and 1B, the conditioned media from HGF-
treated SCC9 cells moderately enhanced capillary-like network
formation compared with conditioned media from untreated
SCC9 cells for 24 hr. In contrast, if we treated SCC9 cells with
HGF (SCC9/H) and then used these SCC9/H cells in EC cocul-
tures, a marked induction of networks in the Matrigel was ob-
Results
Induction of Jagged1 in SCC cells by growth factors
through MAPK
To further explore the role of growth factors such as HGF in
HNSCC development, initially, we were interested in whether
14
CANCER CELL : JULY 2005

Page 3

A R T I C L E
served compared with untreated SCC9 cells in cocultures for
24 hr. Of note, although some SCC cells may form branching
tubules upon HGF treatment for 1–2 weeks, we did not observe
that HGF induced SCC9 cells to generate capillary-like sprouts
for 24 hr in our model. Moreover, we also confirmed that the
sprouts were derived from ECs, but not SCC cells, in cocul-
tures (Figure S1 in the Supplemental Data available with this
article online). While other factors might be involved in the en-
hancement of network formation, our findings led us to exam-
ine whether the direct interaction between tumor cells and ECs
stimulated a signaling cascade to promote angiogenesis.
Previously, we generated a gene expression profile induced
by HGF in SCC cells using microarray (Zeng et al., 2002b). By
searching our database, interestingly, although secreted angio-
genic factors such as interleukin-8 (IL-8) were upregulated, we
found that Jagged1 was strongly induced by HGF (Figure 1C).
Jagged1 is a ligand for Notch proteins whose signaling plays
a critical role in the specification of cell fate through local cell
interactions. Since ECs express Notch receptors (Villa et al.,
2001; Shawber and Kitajewski, 2004), the induction of Jagged1
in SCC cells might activate the Notch signaling pathway in
ECs. To confirm our microarray data, we examined whether
Jagged1 expression was induced by HGF in SCC cells by
Western blot analysis. As shown in Figure 1D, upon HGF treat-
ment, the expression of Jagged1 in SCC cells was rapidly in-
duced by HGF in a time-dependent fashion as determined by
Western blot analysis.
Recently, we and others found that HGF strongly activated
the MAPK and Akt signaling pathways to inhibit apoptosis in
SCC cells, both of which have been suggested to play a critical
role in the stimulation of gene expression (Zeng et al., 2002a;
Zeng et al., 2002b; Birchmeier et al., 2003; Trusolino et al.,
2001). Therefore, we determined whether HGF-induced Jag-
ged1 expression was dependent on MAPK and/or Akt activa-
tion. As shown in Figure 2, the treatment with the MEK inhibitor
U0126, but not the PI3 kinase inhibitor LY294002, significantly
suppressed HGF-induced Jagged1 expression as determined
by Western blot (Figures 2A and 2B) or Northern blot (Figures
2C and 2D) analysis in SCC9 and SCC14A cells. The specificity
of these inhibitors was confirmed in SCC cells by Western blot
analysis (Zeng et al., 2002a). Since AP-1 is a major effector of
the MAPK signaling pathway (Shaulian and Karin, 2002), we
also examined whether the inhibition of AP-1 activity sup-
pressed Jagged1 expression using SCC cells expressing the
dominant-negative mutant of c-Jun, TAM-67. Previously, we
have demonstrated that TAM-67 was able to inhibit AP activa-
tion induced by HGF in SCC cells (Zeng et al., 2002b). To rule
out clonal variation, we utilized retrovirus-mediated transduc-
tion to generate SCC9 and SCC14A cells that stably expressed
TAM-67 (Figures 2E and 2F, top panel). While Jagged1 was
induced in control SCC cells following HGF treatment, the in-
duction of Jagged1 was suppressed in SCC cells expressing
TAM-67 (Figures 1E and 1F, bottom panel). Consistently, we
found that two close AP-1 binding sites at positions −2488 and
−2616 were present in the region of the jagged1 promoter. Our
chromatin immunoprecipitation (ChIP) assays found that the
AP-1 components c-Jun and c-Fos were recruited to this re-
gion following HGF stimulation (Figure S2).
We also directly examined whether the activation of the
MAPK signaling pathway by EGF or TGFα could induce Jag-
ged1 expression in SCC cells. Importantly, EGFR has been
Figure 2. HGF induces Jagged1 expression through MAPK
A and B: MAPK-dependent Jagged1 expression by Western blot analysis.
Both SCC9 and SCC14A cells were pretreated with U0126 or LY294002 for
30 min and then treated with HGF for the indicated time periods. Fifty
microgram aliquots of protein extracts were probed with anti-Jagged1.
C and D: MAPK-dependent jagged1 expression by Northern blot analysis.
Five microgram aliquots of total RNAs were probed with32P-labeled hu-
man jagged1 cDNA probe. As an internal control, the membrane was
stripped and reprobed with32P-labeled glyceraldehydes-3-phosphate de-
hydrogenase (GAPDH) cDNA probe.
E: The inhibition of AP-1 suppressed Jagged1 expression induced by HGF
in SCC9 cells. To establish SCC9 cells stably expressing TAM-67, cells were
infected with retroviruses expressing TAM-67 or control vector and selected
with G418 (600 ?g/ml) for 10 days. The resistant clones were pooled and
probed with anti-c-Jun. Both SCC9/V and SCC/TAM-67 cells were treated
with HGF (40 ng/ml) for the indicated time periods. Fifty microgram aliquots
of protein extracts were probed with anti-Jagged1.
F: The inhibition of AP-1 suppressed Jagged1 expression induced by HGF
in SCC14A cells. The experimental procedures were performed as de-
scribed in E.
found to be overexpressed in human HNSCC (Mao et al.,
2004). As shown in Figure 3A, the treatment of EGF or TGFα
potently induced the activation of ERK in SCC14A cells as well
as other SCC cells, suggesting that the EGFR functionally
transduced the MAPK signaling cascade in SCC cells. Simi-
larly, the expression of Jagged1 was significantly induced in
these SCC cells by EGF or TGFα (Figure 3B). Moreover, the
inhibition of MAPK, but not Akt, also suppressed Jagged1 ex-
pression in SCC14A cells and other cells (Figure 3C; data not
shown). Taken together, the expression of Jagged1 can be in-
duced by common tumor-associated growth factors that acti-
vate the MAPK signaling pathway.
Promotion of endothelial capillary-like sprout formation
by Jagged1-expressing SCC cells through activating
Notch signaling
Next, we examined whether the induction of Jagged1 in SCC
cells had a functional role in endothelial differentiation. Since
CANCER CELL : JULY 2005 15

Page 4

A R T I C L E
Since the expression of Jagged1 might activate the Notch
signaling pathway in adjacent SCC cells, it was possible that
the Notch-inducible genes in SCC cells stimulated endothelial
differentiation via a paracrine mechanism. To rule out this pos-
sibility, we overexpressed the constitutively activated Notch-
ICD in SCC9 cells (Figure 4E). It is well known that, upon ligand
binding, the Notch-ICD is cleaved and then moved to the nu-
cleus, where it interacts with RBPJκ to activate gene expres-
sion. Previously, there were many studies that have demon-
strated that Notch-ICD expression phenocopied Notch signaling
(Lindsell et al., 1995; Kato et al., 1997; Mumm et al., 2000). In
contrast to SCC9/Jag cells, Notch-ICD-expressing SCC9 cells
did not significantly enhance endothelial sprout formation in
the Matrigel compared with coculture of ECs with SCC9/V cells
(Figures 4C and 4D). The results suggest that the activation of
the Notch signaling pathway in ECs, but not in SCC cells, was
critical for endothelial sprout and network formation. The lucif-
erase reporter assay confirmed that the overexpression of
Notch-ICD stimulated transcription.
Next, we directly determined whether the endothelial sprout
formation was directly dependent on the activation of the
Notch signaling pathway in ECs. Thus, we utilized a highly spe-
cific γ-secretase inhibitor (γ-SI) to inhibit the Notch cleavage
and activation (Dale et al., 2003) in the coculture of ECs and
SCC9/Jag cells. As shown in Figure 4F, γ-SI strongly sup-
pressed SCC9/Jag cell-induced endothelial network formation.
In contrast, we found that γ-SI did not inhibit the proangiogenic
factor IL-8-induced endothelial network formation (Figure S3),
suggesting that γ-SI may specifically inhibit Notch signaling.
The Notch-dependent luciferase reporter assay demonstrated
that γ-SI blocked SCC9/Jag cell-mediated Notch transcription
in ECs (Figure 4B).
To further validate our results, we also stably expressed the
dominant-negative form of Su(H) [(DN-Su(H)] (Artavanis-Tsako-
nas et al., 1999) in ECs or SCC9/Jag cells by retrovirus-medi-
ated transduction, respectively (Figure 4G). Although Su(H)/
RBPJκ-independent activities of Notch are known, Su(H)/
RBPJκ is a major effector of the Notch signaling pathway. As
shown in Figure 4H, overexpression of DN-Su(H) in SCC9/Jag
cells did not have significant effects on endothelial sprout for-
mation in our coculture assay. In sharp contrast, overexpres-
sion of DN-Su(H) in ECs totally abolished SCC9/Jag cell-
induced network formation. Additionally, we found that the
overexpression of Notch-ICD in ECs also promoted the sprout
formation in the Matrigel, which could not be inhibited by
γ-SI or soluble Jagged1-conditioned media (Figure S4). Taken
together, these results suggest that Jagged1-expressing SCC
cells were capable of initiating a crosstalk with ECs through
activating the Notch signaling pathway, thereby inducing endo-
thelial capillary-like sprout networks.
Figure 3. EGF or TGFα induces Jagged1 expression in SCC cells through
MAPK
A: EGF or TGFα activated MAPK. SCC14A cells were pretreated with U0126
or LY294002 and then treated with EGF or TGFα for the indicated time
periods. Fifty microgram aliquots of protein extracts were probed with anti-
phospho-ERK (1:1000). For loading control, the membrane was stripped
and reprobed with anti-ERK (1:1000) or anti-α-tubulin.
B: EGF or TGFα induced Jagged1 expression in SCC cells. SCC cells were
treated with EGF or TGFα for the indicated time periods. The Jagged1 ex-
pression was examined as described in Figure 1.
C: The inhibition of MAPK suppressed Jagged1 expression induced by EGF or
TGFα. The experiments were performed as described in Figures 2A and 2B.
HGF, EGF, or TGFα also induced SCC cells to secrete angio-
genic growth factors (Mao et al., 2004), it was critical to deter-
mine whether the Jagged1-expressing SCC cells alone was
capable of inducing endothelial sprout formation by activating
Notch signaling in ECs. We utilized retrovirus-mediated trans-
duction to stably express Jagged1 in SCC9 cells in which the
level of endogenous Jagged1 was undetectable. Western blot
analysis demonstrated that Jagged1-expressing SCC9 cells
(SCC9/Jag) and control cells (SCC9/V) were generated (Figure
4A). The coculture of ECs with SCC9/Jag cells, but not with
SCC9/V cells, potently activated the Notch-dependent lucifer-
ase reporter in ECs (Figure 4B), confirming that Jagged1-
expressing SCC9 cells could functionally trigger the Notch sig-
naling pathway in their neighboring cells. To determine whether
Jagged1-expressing SCC cells stimulated the capillary-like
network formation, ECs were cocultured with SCC9/Jag cells
or SCC9/V cells on the Matrigel. As shown in Figures 4C and
4D, SCC9/Jag cells induced 4-fold more endothelial sprouts
than SCC-9/V cells. Moreover, these endothelial sprouts in-
duced by SCC9/Jag cells formed a well-connected network.
Endothelial capillary-like network formation induced
by endogenous Jagged1-expressing SCC cells
As demonstrated in Figure 3B, SCC14A cells expressed a high
basal level of endogenous Jagged1. Thus, we wondered
whether the elevated endogenous Jagged1 would promote en-
dothelial differentiation in vitro. As shown in Figures 5A and
5B, like SCC9/Jag cells, coculture of ECs with SCC14A cells
strongly promoted endothelial network formation in the Matri-
gel. To determine whether SCC14A cell-stimulated endothelial
sprout formation was dependent on the Notch signaling path-
16
CANCER CELL : JULY 2005

Page 5

A R T I C L E
way, we added γ-SI to the coculture. As shown in Figures 5A
and 5B, γ-SI totally abolished endothelial tube formation in-
duced by SCC14A. The luciferase assay confirmed that γ-SI
inhibited Notch-dependent transcription induced by SCC14A
cells (Figure 5C). Also, we utilized soluble Jagged1, which was
able to be secreted and to interfere with the interaction be-
tween the normal Notch receptor and ligand (Wu et al., 2001).
SCC14A cells were stably transduced with retroviruses ex-
pressing soluble Jagged1 (S-Jag). As shown in Figure 5D, the
S-Jag proteins were detected not only in cell lysates but also
in cell culture media, demonstrating that S-Jag was secreted
in SCC14A cells. Moreover, S-Jag also inhibited Notch-depen-
dent transcription induced by SCC14A cells (Figure 5C). Com-
pared with SCC14A/V cells, the endothelial network formation
was significantly reduced in coculture of ECs with SCC14A/
S-Jag cells (Figures 5A and 5B). To further confirm our results,
we also determined whether depletion of Jagged1 expression
in SCC14A cells by siRNA inhibited sprout formation in our co-
culture assay. As shown in Figure 5E, Western blot analysis
demonstrated that Jagged1 siRNA, but not luciferase siRNA,
completely suppressed Jagged1 expression in SCC14A cells.
While luciferase siRNA did not interfere with the endothelial
sprout formation in our coculture assay, the depletion of Jag-
ged1 expression in SCC14A potently suppressed the sprout
formation induced by SCC14A cells (Figure 5F).
Finally, we also determined whether Jagged1 expressed in
SCC9 cells was responsible for the capillary-like sprout forma-
tion in our coculture assay upon HGF stimulation. As shown in
Figure 5G, Western blot analysis confirmed that Jagged1
siRNA, but not control luciferase siRNA, completely inhibited
Jagged1 expression in SCC9 cells induced by HGF. Transfec-
tion of luciferase siRNA did not affect the sprout formation in
our coculture assay upon HGF stimulation. The deletion of Jag-
ged1 by siRNA strongly suppressed capillary-like sprout for-
mation stimulated by HGF-treated SCC9 cells (Figure 5H).
Moreover, as shown in Figure 5H, both S-Jag-conditioned me-
dia from SCC14A/S-Jag cells and γ-SI also abolished the
sprout formation induced by HGF-treated SCC9 cells (Figure
5H). Taken together, these results suggested that the endoge-
nous Jagged1 in SCC cells could initiate the contact-depen-
dent signaling in ECs through the Notch receptor, thereby stim-
ulating endothelial differentiation.
Figure 4. Jagged1-expressing SCC cells promote endothelial capillary-like
sprout formation through activating the Notch signaling pathway in ECs
A: SCC9 cells stably expressing Jagged1. SCC9 cells were transduced with
retroviruses expressing Jagged1 or control vector. After selection, the sta-
ble clones were pooled and probed with anti-Jagged1.
B: Jagged1-expressing SCC9 cells activated Notch-dependent transcrip-
tion in ECs. ECs were transfected with Notch-dependent luciferase reporter
for 24 hr and then cocultured with SCC/Jag cells or control cells (SCC/V)
in the absence or presence of γ-SI (50 ?M). The luciferase activities were
measured with a dual luciferase assay system. In some experiments, ECs
were also cotransfected with DN-Su(H). The assay was performed in dupli-
cate, and the results represent mean ± SD (error bars) from three indepen-
dent experiments. Student’s t test was performed to determine statistical
significance. **p < 0.01, EC + SCC9/Jag versus EC or EC + SCC9/V; *p <
0.01, EC + SCC9/Jag versus EC + SCC/Jag + γ-SI or EC/DN-Su + SCC9/Jag.
C and D: Jagged1-expressing SCC cells enhanced capillary-like network
formation. ECs were cultured alone or cocultured with SCC9/V, SCC9/Jag,
or SCC9/ICD (SCC9 expressing Notch-ICD) for 24 hr. The sprouts were
stained and counted as described in Figure 1. Student’s t test was per-
formed to determine statistical significance. **p > 0.05 (no significant differ-
ences), EC + SCC9/ICD versus EC + SCC9/V or EC; *p < 0.01, EC + SCC9/
Jag versus EC + SCC9/ICD. Scale bar, 50 ?m.
E: SCC9 cells expressing Notch ICD. SCC9 cells were stably infected with
retroviruses expressing HA-Notch-ICD or control vector.
F: The inhibition of endothelial tube formation by γ-SI. ECs were cocultured
with SCC9/Jag cells in the presence or absence of γ-SI (50 ?M) for 24 hr.
The assay was performed in duplicate, and the results represent mean ±
SD (error bars) from three independent experiments. Student’s t test was
performed to determine statistical significance. *p < 0.01, +SI versus −SI.
G: ECs or SCC9 cells stably expressing DN-Su(H). ECs or SCC9 cells were
infected with retroviruses expressing HA-DN-Su(H) or control vector. ECs ex-
pressing DN-Su(H) [EC/DN-Su(H)] or SCC9/Jag cells expressing DN-SU(H)
[SCC9/Jag/DN-SU(H)] were confirmed with anti-HA epitope.
Tumor angiogenesis and tumor growth enhanced by
Jagged1-expressing SCC cells in a nude mouse model
To determine whether Jagged1-expressing SCC cells modu-
lated tumor angiogenesis in vivo, we utilized a well-established
SCID mouse model of tumor angiogenesis that mimics human
tumor angiogenesis in vivo using human ECs. For this model
system, both human ECs and SCC cells were mixed with the
Matrigel and seeded in porous poly(L-lactic acid) (PLLA) scaf-
folds. The scaffolds were subcutaneously implanted in the dor-
H: The inhibition of Notch signaling in ECs suppressed sprout formation. EC/
DN-SU(H) cells or control cells (EC/V) were cocultured with SCC/Jag/DN-
SU(H) cells or control SCC cells (SCC/Jag/V) on the Matrigel as described
in Figure 1. The results represent mean ± SD (error bars) from three indepen-
dent experiments. *p < 0.01, EC/DN-Su(H) + SCC9/Jag versus EC/V + SCC9/
Jag, EC/V + SCC9/Jag/DN-Su(H), or EC/V + SCC9/Jag/V.
CANCER CELL : JULY 200517

Page 6

A R T I C L E
sal region of SCID mice, and tumor angiogenesis was exam-
ined by histology and immunostaining (Nor et al., 2001). First,
we examined tumor angiogenesis at an early stage (3 weeks).
As shown in Figure 6A, the scaffold coimplanted with ECs and
SCC9/Jag cells was more vascularized than that with ECs and
SCC9/V cells. Along with tumor formation, immunostaining of
human factor VIII, a marker for ECs, found that significantly
more intratumoral blood vessels were induced in the scaffolds
seeded with the combination of ECs with SCC9/Jag cells com-
pared with SCC9/V cells (Figures 6B and 6C). After 6 weeks,
scaffolds were retrieved, and the coimplants of ECs with
SCC9/Jag cells exhibited more aggressive tumor growth than
those with SCC9/V cells, as determined by tumor volume and
size (Figures 6D and 6E). Finally, human factor VIII immuno-
staining also found that there were three times more blood ves-
sels in tumors from the endothelial coimplants with SCC9/Jag
cells than in those with SCC9/V cells (Figure 6F).
During in vitro cell culture, we observed that both SCC9/V
and SCC9/Jag cells grew at the similar rate. More specifically,
the 5-bromodeoxyuridine (BrdU) incorporation assay indicated
that the rate of DNA synthesis of both cells in vitro was iden-
tical (Figures S5A–S5C). It suggested that the different tumor
growth rates from coimplants were unlikely due to the possi-
bility that Jagged1 directly stimulated SCC cell proliferation.
To determine whether the enhancement of tumor angiogenesis
might affect the proliferation of SCC cells in vivo, tumor-bear-
ing mice were injected with BrdU intraperitoneally, and BrdU
uptakes in tumor cells were examined. Unlike in vitro, the
immunostaining found that significantly greater numbers of
BrdU-positive tumor cells (over 2-fold) were observed in tu-
mors from coimplants of ECs with SCC9/Jag cells compared
with those from coimplants of ECs with SCC9/V cells (Figure
6G and Figure S5D), suggesting that the increasing tumor angi-
ogenesis promoted tumor cell proliferation in vivo. Taken to-
gether, these results from in vivo studies suggest that Jag-
ged1-expressing tumor cells can stimulate tumor angiogenesis
and growth in vivo by interacting with neighboring ECs through
the activation of the Notch signaling pathway.
Figure 5. Endogenous Jagged1-expressing SCC cells enhance endothelial
capillary-like network formation
A and B: γ-SI or soluble Jagged1 inhibited endothelial sprout formation in-
duced by SCC14A cells. ECs were cocultured with SCC14A cells in the
absence or presence of γ-SI or with SCC14A expressing soluble Jagged1.
*p < 0.01, EC + SCC14A versus EC + SCC14A + γ-SI or EC + SCC14A/S-Jag.
Scale bar, 50 ?m.
C: γ-SI or soluble Jagged1 inhibited Notch activation in ECs induced by
SCC14A. ECs were transfected with Notch luciferase reporter for 24 hr and
then cocultured with SCC14A cells in the absence or presence of γ-SI or
SCC14A cells expressing soluble Jagged1 (SCC14A/S-Jag) for 24 hr. The
luciferase activities were measured with a dual luciferase system. **p <
0.01, EC + SCC14A versus EC; *p < 0.01, EC + SCC14A versus EC + SCC14A/
S-Jag, EC + SCC14A + γ-SI, or EC/DN-Su + SCC14A.
D: SCC14A cells stably expressing soluble Jagged1. SCC14A cells were in-
fected with retroviruses expressing HA-soluble Jagged1 (S-Jag) or control
vector. Both supernatants and cell lysates were probed with anti-HA.
E: The depletion of endogenous Jagged1 expression in SCC14A cells by
Jagged1 siRNA. SCC14A cells were transfected with Jagged1 siRNA or lu-
ciferase (Luc) siRNA (100 ?M) for 36 hr, and the expression of Jagged1
proteins was examined by Western blot analysis. For internal controls, the
blot was stripped and rechecked with anti-α-tubulin.
F: The inhibition of Jagged1 expression suppressed the proangiogenic ef-
fects of SCC14A in coculture. SCC14A cells were transfected with Jagged1
siRNA or Luc siRNA for 36 hr. After treatment, Jagged1- or Luc siRNA-trans-
fected SCC14A cells were cocultured with ECs for 24 hr. *p < 0.01, EC +
SCC14A/Luc siRNA versus EC + SCC14A/Jagged1 siRNA.
G: HGF-induced Jagged1 expression was inhibited by Jagged1 siRNA in
SCC9 cells. SCC9 cells were transfected with Jagged1 or Luc siRNA (100
?M) for 36 hr and then treated with HGF for 0, 1, 4, and 8 hr. The level of
Jagged1 expression was examined by Western blot analysis.
H: The inhibition of Notch signaling suppressed the proangiogenic effects
of SCC9 coculture upon HGF stimulation. SCC9 cells were transfected with
Jagged1 or Luc siRNA for 36 hr and then treated with HGF for 4 hr. After
treatment, Jagged1- or Luc siRNA-transfected SCC9 cells were cocultured
with ECs for 24 hr. HGF-treated SCC9 cells were also cocultured with ECs in
the presence of γ-SI or soluble Jagged1-conditioned media from SCC14A/
Jagged1 expression in human HNSCC tissues
Although ECs may also express Notch ligands, our results sug-
gest that tumor cells that express Jagged1 may help to direct
neovasculature during tumor growth. To explore that the Notch
signaling pathway was associated with human tumor angio-
genesis, we also examined whether there was corelationship
between Jagged1 expression and tumor blood vessel forma-
tion in HNSCC tumor tissues by the combination of tissue
microarray (TMA) and immunostaining. Previously, the Notch
signaling components were found to be overexpressed in hu-
man HNSCC by gene profiling (Leethanakul et al., 2000). Thus,
we further compared the level of Jagged1 expression in human
HNSCC with normal epithelial tissues and dysplasias using the
combination of high-density TMA and immunostaining. Repre-
S-Jag cells. *p < 0.01, EC + SCC9 + Luc siRNA + HGF versus EC + SCC9 +
HGF + Jagged1 siRNA, EC + SCC9 + HGF + γ-SI, or EC + SCC9 + HGF +
S-Jag. The results represent mean ± SD (error bars) from three independent
experiments. Student’s t test was performed to determine statistical signifi-
cance.
18
CANCER CELL : JULY 2005

Page 7

A R T I C L E
Figure 7. Jagged1 expression is highly elevated and correlated with human
microvessel formation in human HNSCC tissues
A and B: Jagged1 expression is highly elevated in HNSCC. Human TMA
sections were stained with anti-Jagged1 (1:50). The level of Jagged1 was
scored based on staining intensity of normal or dysplastic epithelial cells
and tumor cells: 0 = negative; 1 = weak; 2 = moderate; 3 = strong. A total
of 350 samples (259 SCCs, 17 normal epithelial tissues, 18 dysplasias, and
56 adjacent muscle and connective tissues) were scored. Of note, some
tissue samples were lost during staining. The average scores were calcu-
lated from each group, and statistical significance was determined by the
Student’s t test. Error bars represent standard deviation. *p < 0.01, dyspla-
sias versus normal epithelial tissues; **p < 0.01, SCC versus dysplasias or
normal epithelial tissues. Scale bar, 25 ?m.
C: Jagged1 expression was correlated with human microvessel formation
in HNSCC. TMA sections were stained with anti-factor VIII (1:200) and anti-
Jagged1 (1:50), respectively. Scale bar, 25 ?m.
Figure 6. Jagged1-expressing SCC cells promote tumor angiogenesis and
tumor growth in a nude mouse model
A: Coimplant of ECs with SCC9/Jag was highly vascularized in vivo. ECs
and SCC9/Jag or SCC9/V cells were mixed with the Matrigel, seeded in
different scaffolds, and then implanted in SCID mice. After 3 weeks, the
implants were harvested and photographed. Left: the coimplant of ECs
and SCC9/Jag cells; right: the coimplant of ECs and SCC9/V cells.
B and C: Jagged1-expressing SCC cells enhanced tumor angiogenesis.
Coimplants of ECs and SCC9/Jag or SCC9/V cells were sectioned and
stained with hematoxylin and eosin (HE) or anti-factor VIII (1:200). The
microvessels were counted from five random fields (×200) and averaged.
The experiments were performed three independent times (five coim-
plants per group), and the results represent mean ± SD (error bars) from 15
coimplants per group. Student’s t test was performed to determine statisti-
cal significance. *p < 0.01. L, low magnification, scale bar, 50 ?m; H, high
magnification, scale bar, 25 ?m.
D–F: Jagged1-expressing SCC cells promoted tumor angiogenesis and tu-
mor growth. The experiments were performed as described in A, and scaf-
folds were harvested 6 weeks after implantation. After tumor volumes and
weights were measured, specimens were fixed, paraffin-embedded, and
sectioned. Serial sections were stained with HE or polyclonal antibodies
against human factor VIII. The microvessels were counted from five ran-
dom fields (×200) and averaged. Two independent experiments were per-
formed (five coimplants per group), and the results represent mean ± SD
(error bars) from ten coimplants per group. Statistical significance was de-
termined by the Student’s t test. *p < 0.01.
G: The promotion of BrdU uptake by SCC cells in tumors derived from the
coimplants of endothelial cells and SCC9/Jag cells. The experiments were
performed as described in A. Mice were injected with BrdU labeling re-
agents (10 ml/kg) intraperitoneally 2 hr before the scaffolds were retrieved.
The sections were incubated with biotinylated anti-BrdU using the Zymed
BrdU staining kit. The BrdU-positive cells were counted from five random
fields (×200) and averaged. Two independent experiments were per-
sentative examples of staining for Jagged1 in normal epithelial
tissues, dysplasias, and SCC are shown in Figure 7A. Mean
Jagged1 staining intensity was significantly increased in hu-
man HNSCC compared with dysplasias and normal epithelial
tissues (p < 0.01). Also, we observed that Jagged1 staining
was modestly stronger in dysplasias than that in normal epithe-
lial tissues (Figures 7A and 7B). However, we did not find that
there was significant difference in Jagged1 staining between
clinical or pathological stages of HNSCC. To determine whether
formed (ten coimplants per group), and the results represent mean ± SD
(error bars) from 20 coimplants per group. Statistical significance was de-
termined by the Student’s t test. *p < 0.01.
CANCER CELL : JULY 200519

Page 8

A R T I C L E
Table 1. The correlation between the expression of Jagged1 and factor VIII in HNSCCa
Factor VIII
Jagged1++++++++++Total
0
+
++
+++
0
7
0
0
1
37
38
8
0
3
50
43
0
0
6
31
1
47
94
82
Tissue microarray (TMA) sections were stained with anti-factor VIII and anti-Jagged1, respectively. The microvessel content was scored based on staining intensity
and area of human factor VIII: + = weak; ++ = moderate; +++ = strong; ++++ = very strong. The level of Jagged1 was scored as described in Figures 7A and 7B. A
total of 224 SCC samples that could be scored for both Jagged1 and factor VIII expression were utilized for our studies.aPearson correlation coefficient analysis. p <
0.01; r = 0.65.
there was corelationship between Jagged1 expression and tu-
mor blood vessel formation in HNSCC tumor tissues, TMA was
also stained with anti-factor VIII antibodies. Our representative
photograph demonstrates that tumor blood vessel formation
was significantly increased in tumors with higher expression of
Jagged1 compared with tumors with low expression of Jag-
ged1 (Figure 7C). Interestingly, in some samples, we observed
that some invasive SCC cells were surrounded by factor VIII-
positive cells. Moreover, the statistical analysis also found that
Jagged1 expression was correlated with blood vessel intensity
in HNSCC tumor tissues (Pearson correlation coefficient; p <
0.01) (Table 1). Our results suggest that HNSCC cells may uti-
lize the Notch signaling pathway to promote tumor angiogen-
esis in vivo and that the level of Jagged1 expression may be
associated with the development of HNSCC.
abolished proangiogenic effects of the SCC cell coculture upon
HGF stimulation. Moreover, the depletion of the high basal level
of Jagged1 in unstimulated SCC cells also abolished the pro-
angiogenic effects of the SCC cell coculture. Our results sug-
gest that contact-dependent Notch signals play a critical role
in endothelial differentiation. However, our findings did not rule
out the possible contribution of the proangiogenic effects of
IL-8 or other secreted proangiogenic factors. In fact, we ob-
served that conditioned media from HGF-treated SCC cells
modestly induced endothelial differentiation in vitro. After the
conditioned medium was concentrated, we found that it could
more strongly promote endothelial differentiation (data not
shown), suggesting that secreted proangiogenic factors func-
tion in a dose-dependent manner. Importantly, we found that
Jagged1 expression was also induced by other growth factors,
including TGFα and EGF, via the activation of MAPK, suggest-
ing that the crosstalk between tumor cells and ECs could be
triggered by these growth factors.
Genetic studies have demonstrated that there is a crosstalk
between the Notch and MAPK signaling pathways in cell fate
determination using a model system of C. elegans and Dro-
sophila (Tsuda et al., 2002; Kumar and Moses, 2001; Shaye
and Greenwald, 2002; Yoo et al., 2004). For example, Berset
et al. (2001) and Yoo et al. (2004) have demonstrated that the
Notch signaling pathway inhibited the MAPK signaling pathway
through the induction of multiple negative regulators of MAPK
during C. elegans vulval development (Shaye and Greenwald,
2002). On the contrary, it was also found that the EGFR-MAPK
activation could downregulate Notch via endocytosis during C.
elegans vulval development. These studies suggest that the
coordination of signals from different pathways must be pre-
cisely integrated for cell fate specification during development.
It has been found that Notch can function either as a tumor
suppressor or as an oncogene, depending on the cellular con-
text (Maillard and Pear, 2003). According to the results from C.
elegans, the inhibition of MAPK by Notch provides a reason-
able explanation for Notch-mediated tumor suppression in
some cases.
However, it is possible that the principal findings elucidated
by studying C. elegans vulval development may not be applica-
ble to human cancer development. Studies by Weijzen et al.
(2002) have found that Notch was also activated in Ras-trans-
formed cells in which MAPK was constitutively activated. The
inhibition of Notch signaling suppressed Ras-mediated trans-
formation. Mailhos et al. (2001) have reported that Dll4 expres-
sion was induced during tumor angiogenesis in a mouse model
Discussion
Our studies demonstrate that contact-dependent Notch signal-
ing triggered by the MAPK signaling pathway plays a critical
role in tumor angiogenesis. Although some early works re-
ported that Notch signaling might inhibit angiogenesis, re-
cently, elegant genetic and molecular studies suggested that
Notch signaling plays a critical role in angiogenesis during de-
velopment (Krebs et al., 2000; Lawson et al., 2002; Duarte et
al., 2004; Krebs et al., 2004; Gale et al., 2004). In Notch or
Jagged1 mutant mouse embryos, vasculogenic formation of
the head, yolk sac, and intersomitic vessel was unaffected. In-
stead, there was a failure to reorganize these rudimentary ves-
sels into large vessels and branching capillaries (Xue et al.,
1999; Krebs et al., 2000). These results suggest that Notch sig-
naling may not be required for vasculogenesis but is essential
for physiological angiogenesis. Currently, how Notch signaling
in angiogenesis is activated during development is not clear.
Our results suggest that the upregulation of Jagged1 expres-
sion in SCC cells may provide a unique mechanism to control
neovascularization. The Jagged1-expressing SCC cells may
not only promote, but also guide angiogenesis, thus providing
nutrients to support the development and growth of HNSCC.
Importantly, we identified a novel mechanism of tumor angi-
ogenesis mediated by the crosstalk between tumor cells and
ECs via the Notch and MAPK signaling pathways. Several
secreted proangiogenic factors, including IL-8 and VEGF, have
been found to be induced by the MAPK or Akt signaling path-
ways (Jung et al., 2002; Sparmann and Bar-Sagi, 2004). Inter-
estingly, the knockdown of Jagged1 expression significantly
20
CANCER CELL : JULY 2005

Page 9

A R T I C L E
Data. For Northern blot analysis, total RNAs were extracted with the Trizol
reagents (Invitrogen) according to the manufacturer’s protocol. Five micro-
grams of total RNAs were resolved on 1.5% agarose formaldehyde gels and
transferred to nitrocellulose membranes overnight. The membranes were
hybridized with32P-labeled full-length Jagged1 cDNA probes released from
pHyTC-Jagged1 plasmids. The probes were prepared with a random-
primed labeling kit (Amersham; Arlington Heights, IL) in the presence of
[α-32P]dCTP (ICN Pharmaceuticals, Costa Mesa, CA) as described pre-
viously (Zeng et al., 2002b).
of xenografted breast adenocarcinoma by unknown mecha-
nisms. Additionally, Notch activation has been found to be as-
sociated with the development of several human cancers such
as salivary gland carcinoma, leukemia, and invasive pancreatic
cancer (Maillard and Pear, 2003). Our studies presented here
defined a unique mode of interplay between Notch and MAPK
signaling pathways, providing a molecular explanation for
Notch-mediated oncogenic transformation. While we found
that constitutive activation of Notch signaling alone in tumor
cells was not sufficient to induce angiogenesis, the upregula-
tion of Jagged1 was capable of initiating crosstalk between
tumor cells and ECs, thereby stimulating tumor angiogenesis
and guiding tumor invasive growth. In support of our findings,
our immunostaining studies have found that Jagged1 expres-
sion was associated with human tumor blood vessel formation.
In our immunostaining studies, we did frequently observe that
Jagged1-expressing SCC cells were surrounded by microves-
sels. Compared with normal epithelial tissues, Jagged1 was
modestly increased in epithelial dysplasia and more highly ex-
pressed in SCC, suggesting that Jagged1 may be associated
with the development of SCC. In human cancer development,
tumor-associated inflammation has been found to play a criti-
cal role in tumor growth and angiogenesis (Sparmann and Bar-
Sagi, 2004). In addition to growth factors, it is also possible
that Jagged1 or other Notch ligands can be activated by non-
specific inflammatory processes (Guo et al., 2004). In future
studies, it will be interesting to explore whether tumor-associ-
ated inflammation promotes tumor angiogenesis via the Notch
signaling pathway or whether tumor-associated inflammatory
cells express the Notch ligands. Nevertheless, our studies sug-
gest that, in addition to the secretion of proangiogenic growth
factors, tumor cells may express Notch ligands to guide angio-
genesis, which supports tumor growth and progression.
The endothelial network formation assay in Matrigel
and siRNA transfection
The reduced Matrigels (125 ?l; BD Systems) were plated in 8-well chamber
slides. The chambers were then incubated at 37°C for 30 min to allow the
Matrigel to polymerize (Kumar et al., 2004). For coculture assay, 0.8 × 104
SCC cells and 0.8 × 104ECs were mixed and added to the top of the
Matrigel in each well. The chambers were then incubated at 37°C for 24 hr.
After incubation, the slides were fixed with methanol and stained with Diff-
Quick solution II (Sigma). The slides were examined, and the sprouts were
counted from five random fields under a microscope (×200).
Cells were transfected with Jagged1 siRNA (100 ?M) or control luciferase
siRNA (100 ?M) overnight mixed with Oligofectamine (cat. 1225-2-011) di-
luted in Opti-Mem (Invitrogen) according to the manufacturer’s instruction.
Thirty-six hours following transfection, cells were treated with HGF for 0, 1,
4, and 8 hr, and the knockdown of Jagged1 expression was confirmed by
Western blot analysis. For coculture experiments, siRNA-transfected SCC
cells were untreated or treated with HGF for 4 hr and washed with PBS.
Afterwards, ECs (0.8 × 104) and siRNA-transfected SCC cells (0.8 × 104)
were plated on the Matrigel. The target sequence for Jagged1 (NM_000214)
siRNA was 5#-GAACAUCACAUUUACCUUUUU-3#. The target sequence for
luciferase is 5#-GCCATTCTATCCTCTAGAG GATG-3#. Both Jagged1 and lu-
ciferase siRNAs (cat. 002099-01-20) were synthesized by Dharmacon.
The SCID mouse model of human tumor angiogenesis assay
PLLA (Sigma) scaffolds were prepared as described previously (Nor et al.,
2001). 0.5 × 106HDMECs and 0.5 × 106SCC9/Jag cells or SCC9/V cells
were mixed with the Matrigel and loaded into scaffolds. Four-week-old fe-
male SCID mice were purchased from Taconic and anesthetized with keta-
mine and xylazine. One scaffold containing the mixture of ECs and SCC9/
Jag cells and another with ECs and SCC9/V cells were implanted subcuta-
neously on the left and right flank region of each mouse, respectively. Three
or six weeks after transplantation, mice were sacrificed, and the scaffolds
were retrieved, immediately measured with calipers, and weighed in an
electronic balance. Afterwards, the scaffolds were fixed with 10% buffered
formalin and embedded in paraffin. The specimens were sectioned at 4 ?M
for histological examination (Nor et al., 2001). For BrdU labeling, mice were
injected with BrdU labeling reagents (10 ml/kg; cat. 103; Zymad) 2 hr before
the scaffolds were retrieved. The care and transplantation of mice was in
accordance with the guidelines of the University of Michigan Committee on
Use and Care of Animals (UCUCA).
Experimental procedures
Cell culture, retroviral infection, and reagents
SCC cell lines were derived from patients with HNSCC in Dr. Thomas Car-
ey’s laboratory at the University of Michigan and were cultured in DMEM
supplemented with 10% FBS from Invitrogen (Grand Island, NY). SCC1 was
derived from patients with HNSCC in the floor of mouth. SCC5, SCC9,
and SCC14A were derived from a patient with HNSCC in the oropharynx
(Takebayashi et al., 2000). The human dermal microvascular ECs (cat. CC-
0288) were purchased from Cambrex (Walkersville, MD) and were cultured
in EGM-2 medium supplemented with growth factors (cat. CC-4147; Cam-
brex). The chemical inhibitors U0126 (cat. 9903) and LY294002 (cat. 9901)
were purchased from Cell Signaling (Beverly, MA). The specific γ-secretase
inhibitor IX (cat. 565770) was purchased from Calbiochem (La Jolla, CA).
The method for retroviral infection is provided in the Supplemental Data.
Immunohistochemistry
Human HNSCC high-density TMA was prepared by the University Michigan
Head and Neck SPORE Tissue Core with Institutional Review Board ap-
proval. Samples (n = 400) from a total of 102 cases with different clinical
and pathological stages were arrayed in slides. At least three tissue cores
(0.6 mm diameter) from tumor tissues were sampled from each case, and
in some cases three tissue cores from adjacent normal tissues were sam-
pled. The TMAs covered the whole spectrum of HNSCC, including normal
epithelial tissues, adjacent normal connective tissues, dysplasias, and pri-
mary SCC and metastatic SCC. The detailed methods for immunostaining
and scoring are described in the Supplemental Data.
Transfection and luciferase reporter assay
Transient transfections were performed by lipofectamine (cat. 18324; Invit-
rogen) according to the manufacturer’s protocol. ECs were plated in a six-
well plate overnight and then cotransfected with the Notch luciferase repor-
ter pGA981-6 (Wu et al., 2001) and pRL-TK Renilla luciferase reporter as an
internal control. Twenty-four hours after transfection, the transfected ECs
were cocultured with SCC/Jag or SCC/V cells for additional 24 hr. Lucifer-
ase activities were measured using a dual luciferase system (Promega) as
described previously (Zeng et al., 2002a).
Western blot and Northern blot analyses
SCC cells were plated in 10 cm tissue culture dishes the day before treat-
ment. Cells were treated with HGF (40 ng/ml; R&D Systems) for the indi-
cated times. Cells were harvested, and whole-cell extracts were prepared.
Western blot analysis was performed as described in the Supplemental
Supplemental data
The Supplemental Data include Supplemental Experimental Procedures
and six figures and can be found with this article online at http://www.
cancercell.org/cgi/content/full/8/1/13/DC1/.
CANCER CELL : JULY 200521

Page 10

A R T I C L E
Kumar, J.P., and Moses, K. (2001). EGF receptor and notch signaling act
upstream of Eyeless/Pax6 to control eye specification. Cell 104, 687–697.
Acknowledgments
We thank Dr. Thomas Carey for SCC cell lines, Chris Strayhorn for histology,
and Dr. Puwan Kumar and Mr. Ryan Benedict for suggestions. This work
was supported by NIH grants R01DE015964, R01DE13848, R01CA100849,
and NSFC30228028 and by the Chang Jiang Scholar Program.
Kumar, P., Miller, A.I., and Polverini, P.J. (2004). p38 MAPK mediates γ-irradi-
ation-induced endothelial cell apoptosis, and vascular endothelial growth
factor protects endothelial cells through the phosphoinositide 3-kinase-Akt-
Bcl-2 pathway. J. Biol. Chem. 279, 43352–43360.
Lawson, N.D., Vogel, A.M., and Weinstein, B.M. (2002). Sonic hedgehog
and vascular endothelial growth factor act upstream of the Notch pathway
during arterial endothelial differentiation. Dev. Cell 3, 127–136.
Received: January 31, 2005
Revised: May 16, 2005
Accepted: June 14, 2005
Published: July 18, 2005
Leethanakul, C., Patel, V., Gillespie, J., Pallente, M., Ensley, J.F., Koontong-
kaew, S., Liotta, L.A., Emmert-Buck, M., and Gutkind, J.S. (2000). Distinct
pattern of expression of differentiation and growth-related genes in
squamous cell carcinomas of the head and neck revealed by the use of
laser capture microdissection and cDNA arrays. Oncogene 19, 3220–3224.
References
Artavanis-Tsakonas, S., Rand, M.D., and Lake, R.J. (1999). Notch signaling:
cell fate control and signal integration in development. Science 284, 770–
776.
Lindsell, C.E., Shawber, C.J., Boulter, J., and Weinmaster, G. (1995). Jag-
ged: a mammalian ligand that activates Notch1. Cell 80, 909–917.
Liu, Z.J., Shirakawa, T., Li, Y., Soma, A., Oka, M., Dotto, G.P., Fairman, R.M.,
Velazquez, O.C., and Herlyn, M. (2003). Regulation of Notch1 and Dll4 by
vascular endothelial growth factor in arterial endothelial cells: implications
for modulating arteriogenesis and angiogenesis. Mol. Cell. Biol. 23, 14–25.
Bancroft, C.C., Chen, Z., Dong, G., Sunwoo, J.B., Yeh, N., Park, C., and
Van Waes, C. (2001). Coexpression of proangiogenic factors IL-8 and VEGF
by human head and neck squamous cell carcinoma involves coactivation by
MEK-MAPK and IKK-NF-κB signal pathways. Clin. Cancer Res. 7, 435–442.
Mailhos, C., Modlich, U., Lewis, J., Harris, A., Bicknell, R., and Ish-Horo-
wicz, D. (2001). Delta4, an endothelial specific Notch ligand expressed at
sites of physiological and tumor angiogenesis. Differentiation 69, 135–144.
Berset, T., Hoier, E.F., Battu, G., Canevascini, S., and Hajnal, A. (2001).
Notch inhibition of RAS signaling through MAP kinase phosphatase LIP-1
during C. elegans vulval development. Science 291, 1055–1058.
Maillard, I., and Pear, W.S. (2003). Notch and cancer: Best to avoid the ups
and downs. Cancer Cell 3, 203–205.
Birchmeier, C., Birchmeier, W., Gherardi, E., and Vande Woude, G.F. (2003).
Met, metastasis, motility and more. Nat. Rev. Mol. Cell Biol. 4, 915–925.
Mao, L., Hong, W.K., and Papadimitrakopoulou, V.A. (2004). Focus on head
and neck cancer. Cancer Cell 5, 311–316.
Dale, J.K., Maroto, M., Dequeant, M.L., Malapert, P., McGrew, M., and
Pourquie, O. (2003). Periodic notch inhibition by lunatic fringe underlies the
chick segmentation clock. Nature 421, 275–278.
Mumm, J.S., Schroeter, E.H., Saxena, M.T., Griesemer, A., Tian, X., Pan,
D.J., Ray, W.J., and Kopan, R.A. (2000). Ligand-induced extracellular cleav-
age regulates γ-secretase-like proteolytic activation of Notch1. Mol. Cell 5,
197–206.
Duarte, A., Hirashima, M., Benedito, R., Trindade, A., Diniz, P., Bekman, E.,
Costa, L., Henrique, D., and Rossant, J. (2004). Dosage-sensitive require-
ment for mouse Dll4 in artery development. Genes Dev. 18, 2474–2478.
Nor, J.E., Christensen, J., Liu, J., Peters, M., Mooney, D.J., Strieter, R.M.,
and Polverili, P.J. (2001). Up-regulation of Bcl-2 in microvascular endothelial
cells enhances intratumoral angiogenesis and accelerates tumor growth.
Cancer Res. 61, 2183–2188.
Ferrara, N., Hillan, K.J., Gerber, H.P., and Novotny, W. (2004). Discovery and
development of bevacizumab, an anti-VEGF antibody for treating cancer.
Nat. Rev. Drug Discov. 3, 391–400.
Folkman, J. (2002). Role of angiogenesis in tumor growth and metastasis.
Sem. Onc. 29, 15–18.
Patel, V., Leethanakul, P.V., and Gutkind, J.S. (2001). New approaches to
the understanding of the molecular basis of oral cancer. Crit. Rev. Oral Biol.
Med. 12, 55–63.
Forastiere, A., Koch, W., Trotti, A., and Sidransky, D. (2001). Head and neck
cancer. N. Engl. J. Med. 345, 1890–1900.
Shaulian, E., and Karin, M. (2002). AP-1 as a regulator of cell life and death.
Nat. Cell Biol. 4, E131–E136.
Gale, N.W., Dominguez, M.G., Noguera, I., Pan, L., Hughes, V., Valenzuela,
D.M., Murphy, A.J., Adams, N.C., Lin, H.C., Holash, J., et al. (2004).
Haploinsufficiency of delta-like 4 ligand results in embryonic lethality due to
major defects in arterial and vascular development. Proc. Natl. Acad. Sci.
USA 101, 15949–15954.
Shawber, C.J., and Kitajewski, J. (2004). Notch function in the vasculature:
insights from zebrafish, mouse and man. Bioessays 26, 225–234.
Shaye, D.D., and Greenwald, I. (2002). Endocytosis-mediated downregula-
tion of LIN-12/Notch upon Ras activation in Caenorhabditis elegans. Nature
420, 686–690.
Guo, D., Zhang, H., Liu, L., Wang, L., Cheng, Y., and Qiao, Z. (2004). Testos-
terone influenced the expression of Notch1, Notch2 and Jagged1 induced
by lipopolysaccharide in macrophages. Exp. Toxicol. Pathol. 56, 173–179.
Sparmann, A., and Bar-Sagi, D. (2004). Ras-induced interleukin-8 expres-
sion plays a critical role in tumor growth and angiogenesis. Cancer Cell 6,
447–458.
Jung, Y.D., Ahmad, S.A., Liu, W., Reinmuth, N., Parikh, A., Stoeltzing, O.,
Fan, F., and Ellis, L.M. (2002). The role of the microenvironment and intercel-
lular cross-talk in tumor angiogenesis. Semin. Cancer Biol. 12, 105–112.
Stupack, D.G., and Cheresh, D.A. (2004). Integrins and angiogenesis. Curr.
Top. Dev. Biol. 64, 207–238.
Kato, H., Taniguchi, Y., Kurooka, H., Minoguchi, S., Sakai, T., Nomura-Oka-
zaki, S., Tamura, K., and Honjo, T. (1997). Involvement of RBP-J in biological
functions of mouse Notch1 and its derivatives. Development 124, 4133–
4141.
Takebayashi, S., Ogawa, T., Jung, K.-Y., Muallem, A., Mineta, H., Fisher,
S.G., Grenman, R., and Carey, T.E. (2000). Identification of new minimally
lost regions on 18q in head and neck squamous cell carcinoma. Cancer
Res. 60, 3397–3403.
Kerbel, R.S., and Kamen, B.A. (2004). The anti-angiogenic basis of met-
ronomic chemotherapy. Nat. Rev. Cancer 4, 423–436.
Trusolino, L., Bertotti, A., and Comoglio, P.M. (2001). A signaling adapter
function for α6β4 integrin in the control of HGF-dependent invasive growth.
Cell 107, 643–654.
Krebs, L.T., Xue, Y., Norton, C.R., Shutter, J.R., Maguire, M., Sundberg, J.P.,
Gallahan, D., Closson, V., Kitajewski, J., Callahan, R., et al. (2000). Notch
signaling is essential for vascular morphogenesis in mice. Genes Dev. 14,
1343–1352.
Tsuda, L., Nagaraj, R., Zipursky, S.L., and Banerjee, U. (2002). An EGFR/
Ebi/Sno pathway promotes Delta expression by inactivating Su(H)/SMRTER
repression during inductive Notch signaling. Cell 110, 625–637.
Krebs, L.T., Shutter, J.R., Tanigaki, K., Honjo, T., Stark, K.L., and Gridley, T.
(2004). Haploinsufficient lethality and formation of arteriovenous malforma-
tions in Notch pathway mutants. Genes Dev. 18, 2469–2473.Villa, N., Walker, L., Lindsell, C.E., Gasson, J., Iruela-Aprispe, M.L., and
22
CANCER CELL : JULY 2005

Page 11

A R T I C L E
Weinmaster, G. (2001). Vascular expression of Notch pathway receptors and
ligands in restricted to arterial vessels. Mech. Dev. 108, 161–164.
Maguire, M., Rand, E.B., Weinmaster, G., and Gridley, T. (1999). Embryonic
lethality and vascular defects in mice lacking the Notch ligand Jagged1.
Hum. Mol. Genet. 8, 723–730.
Weijzen, S., Rizzo, P., Braid, M., Vaishnav, R., Jonkheer, S.M., Zlobin, A.,
Osborne, B.A., Gottipati, S., Aster, J.C., Hahn, W.C., et al. (2002). Activation
of Notch-1 signaling maintains the neoplastic phenotype in human Ras-
transformed cells. Nat. Med. 8, 979–986.
Yoo, A.S., Bais, C., and Greenwald, I. (2004). Crosstalk between the EGFR
and LIN-12/Notch pathways in C. elegans vulval development. Science 303,
663–666.
Wong, D.T., Todd, R., Tsuji, T.R., and Donoff, B. (1996). Molecular biology
of human oral cancer. Crit. Rev. Oral Biol. Med. 7, 319–328.
Zeng, Q., Chen, S., You, Z., Yang, F., Carey, T.E., Saims, D., and Wang, W.
(2002a). Hepatocyte growth factor inhibits anoikis in head and neck
squamous cell carcinoma cells by activation of ERK and Akt signaling inde-
pendent of NFκB J. Biol. Chem. 277, 25203–25208.
Wu, G., Lyapina, S., Das, I., Li, J., Gurney, M., Pauley, A., Chui, I., Deshaies,
R.J., and Kitajewski, J. (2001). SEL-10 is an inhibitor of Notch signaling that
targets Notch for ubiquitin-mediated protein degradation. Mol. Cell. Biol.
21, 7403–7415.
Zeng, Q., McCauley, L.K., and Wang, C.-Y. (2002b). Hepatocyte growth fac-
tor inhibits anoikis by induction of activator protein 1-dependent cyclooxy-
genase-2. J. Biol. Chem. 277, 50137–50142.Xue, Y., Gao, X., Lindsell, C.E., Norton, C.R., Chang, B., Hicks, C., Gendron-
CANCER CELL : JULY 2005 23