Research Article

Gonadotropin-Releasing Hormone Promotes Ovarian Cancer Cell

Invasiveness through c-Jun NH2-Terminal Kinase–Mediated
Activation of Matrix Metalloproteinase (MMP)-2 and MMP-9
1 2 1
Lydia W.T. Cheung, Peter C.K. Leung, and Alice S.T. Wong
1
Department of Zoology, University of Hong Kong, Hong Kong and 2Department of Obstetrics and Gynecology, University of British
Columbia, Vancouver, British Columbia, Canada

Abstract However, a role of GnRH in ovarian epithelial cell migration/

invasion has not been established and the mechanism by which
Gonadotropin-releasing hormone (GnRH) receptor is present
GnRH contributes to ovarian cancer invasiveness is also unknown.
in 80% of ovarian cancer, and numerous studies have provided
Ovarian cancer is the most lethal gynecologic malignant tumor
evidence for a role of GnRH in cell proliferation. In this study,
in the Western world. This is in part due to the fact that most (60%)

Downloaded from http://aacrjournals.org/cancerres/article-pdf/66/22/10902/2555753/10902.pdf by guest on 21 October 2022

the effect of GnRH on the invasion potential of ovarian cancer
cells was investigated. In vitro migration and cell invasion
assays with the ovarian cancer cell lines Caov-3 and OVCAR-3
revealed the biphasic nature of GnRH; low concentrations of
GnRH agonist (GnRHa) increased the cell motility and
invasiveness of these cells, but at increased concentrations,
the stimulatory effect was insignificant. Reverse transcription-
PCR, Western blot, and gelatin zymography showed that the
expression of metastasis-related proteinases, matrix metal-
loproteinase (MMP)-2 and MMP-9, was up-regulated and
activated by GnRHa. Moreover, we observed that GnRHa was
able to transactivate the MMP-2 and MMP-9 promoters. The
invasive/migratory phenotype activated by GnRHa can be
blocked by specific inhibitors or neutralizing antibodies to
MMP-2 and MMP-9. Knockdown of the GnRH receptor using
small interfering RNA significantly inhibited the GnRH-
induced MMP activation, invasion, and migration. In addition,
we showed that the c-Jun NH2-terminal kinase, but not
extracellular signal-regulated kinase 1/2 or p38 mitogen-
activated protein kinase, signaling pathway was critical for
GnRH-mediated up-regulation of MMP, cell invasion, and
motility. These results indicate for the first time an expanded
role for GnRH in other aspects of ovarian tumor progression,
such as metastasis, via activation of MMP and the subsequent
increase in cell migration and invasion. (Cancer Res 2006;
66(22): 10902-10)
Introduction
The hypothalamic decapeptide gonadotropin-releasing hormone
(GnRH) plays a key role in the ovary and ovarian cancer (reviewed
in ref. 1). GnRH receptor expression has been shown in 80% of
human ovarian tumors and numerous cancer cell lines, and the
analogues inhibit proliferation of the GnRH receptor-bearing
tumor cells both in vivo and in vitro, supporting evidence for a
direct antiproliferative effect (1–3). Of particular interest, levels of
GnRH receptor seem to be associated with cancer grading and have
been reported to be elevated in advanced-stage (stages III and IV)
ovarian carcinomas (4). This opens the possibility that GnRH could
directly regulate tumor progression of ovarian cancer cells.
Requests for reprints: Alice S.T. Wong, Department of Zoology, University of
Hong Kong, 4S-14 Kadoorie Biological Sciences Building, Pokfulam Road, Hong Kong.
Phone: 852-2299-0865; Fax: 852-2559-9114; E-mail: awong1@hku.hk.
I2006 American Association for Cancer Research.
doi:10.1158/0008-5472.CAN-06-2217
of the patients are diagnosed with already widespread intraperi-
toneal metastasis and ascites and the lack of effective therapies
for advanced-stage disease (5). Thus, there is a need for new
therapeutic targets and a better understanding of the mechanisms
involved in the spread of ovarian carcinoma. The factors that
regulate the metastatic process of ovarian cancer, however, are
poorly understood.
Matrix metalloproteinases (MMP), a family of secreted or
transmembrane enzymes, can collectively digest almost all
extracellular matrix (ECM) and basement membrane components
(6, 7). Thus, MMPs are largely implicated in promoting angiogen-
esis and tumor metastasis. Above all, MMP-2 and MMP-9, which
can degrade collagen IV, the major ECM component of the
basement membranes, have been suggested to be critical for the
invasive and metastatic potential in ovarian carcinoma. Elevated
expression of MMP-2 and MMP-9 has been detected in ovarian
cancer ascites, tissues, and cancer cells in culture (8–11), and
experimental metastasis is suppressed by a synthetic MMP
inhibitor (12). MMP activity is often controlled by a specific
inhibitor known as the tissue inhibitor of metalloproteinases
(TIMP). TIMP-1 selectively binds pro-MMP-9, whereas TIMP-2
associates with pro-MMP-2 in a crucial step in the cell-mediated
activation of MMPs (13). By inhibiting active MMPs, TIMPs
inhibit cell invasion in vitro and tumorigenesis and metastasis
in vivo (14–16).
In the present study, we show for the first time a role for GnRH
in the invasive phenotype and motility of human ovarian cancer
cells. We also present evidence suggesting that the activation of
metastasis-related proteinases, especially MMP-2 and MMP-9,
through the c-Jun NH2-terminal kinase (JNK) mediated signaling
pathway in this process.
Materials and Methods
Cell culture. The human ovarian epithelial carcinoma cell lines Caov-3
and OVCAR-3 (kindly provided by Dr. N. Auersperg, University of British
Columbia, Vancouver, British Columbia, Canada) were grown in medium
199:105 (Sigma, St. Louis, MO) supplemented with 5% fetal bovine serum
and 100 units/mL penicillin-streptomycin (Invitrogen, San Diego, CA) in a
humidified atmosphere of 5% CO2 at 37jC.
Antibodies, reagents, and plasmids. GnRH agonist (GnRHa; D-Ala6)
was purchased from Sigma. PD98059, SB203850, SP600125, MMP-2 inhibitor
(OA-Hy), MMP-9 inhibitor (SB-3CT), and monoclonal antibodies to
MMP-2 and MMP-9 were purchased from Calbiochem (San Diego, CA).
Anti-TIMP-1 and anti-TIMP-2 were from Chemicon International, Inc.
(Temecula, CA). Anti-extracellular signal-regulated kinase 1/2 (ERK1/2),

Cancer Res 2006; 66: (22). November 15, 2006 10902 www.aacrjournals.org
 GnRH Stimulates Ovarian Cancer Invasion

Figure 1. GnRH promotes ovarian cancer

cell migration and invasion. A, ovarian
cancer cells (1.5  106/mL) were seeded
on a Matrigel-precoated filter (8-Am pore)
of Transwell chambers in the presence or
absence of increasing concentrations
(0.1 nmol/L to 1 Amol/L as indicated) of
GnRHa. B, cell motility under same
treatment was measured through uncoated
filters, except that conditioned medium
from NIH-3T3 fibroblast culture was used in
the lower chamber as chemoattractant.
After 24 hours of incubation, cells in the

Downloaded from http://aacrjournals.org/cancerres/article-pdf/66/22/10902/2555753/10902.pdf by guest on 21 October 2022

upper side of the filter were removed and
the invaded or migrated cells were fixed,
stained, and counted. Left, representative
pictures. Columns, mean number of
invaded or migrated cells of five fields of
triplicate wells from three independent
experiments; bars, SD. C, expression of
GnRH receptor (GnRHR ) in ovarian cancer
cell lines was detected by semiquantitative
PCR using primers specific for GnRH
receptor and h-actin. D, cells were
transiently transfected with vector
expressing short hairpin RNA directed
against GnRH receptor (2 Ag) or
nonspecific siRNA (NS siRNA ) as control.
Transfected cells were collected for
invasion assay through Matrigel (left) and
migration assay (right ) through uncoated
filter in the presence or absence of
0.1 nmol/L GnRHa. *, P < 0.05; **,
P < 0.005, compared with untreated
controls.

anti-p38 mitogen-activated protein kinase (MAPK), anti-JNK, anti-
phosphorylated ERK1/2, anti-phosphorylated p38 MAPK, and anti-
phosphorylated JNK were obtained from Cell Signaling Technology, Inc.
(Beverly, MA). The dominant-negative mutant of human JNK was kindly
provided by Dr. R. Davis (University of Massachusetts Medical School,
Worcester, MA). The MMP-2 promoter region between positions 1,659 and
+57 was cloned into promoterless luciferase reporter vector pGL3-Basic as
described previously (17). The MMP-9 promoter plasmid was a generous
gift of Dr. D. Boyd (M. D. Anderson Cancer Center, Houston, TX).
Western blot analysis. Equal amounts of protein (50 Ag) were subjected
to SDS-PAGE (7.5%) and then transferred to nitrocellulose membrane.
Membranes were blocked using 5% nonfat dry milk in PBS containing 0.05%
Tween 20 and incubated overnight at 4jC with the appropriate primary
antibody. The immunocomplex was detected with horseradish peroxidase–
conjugated IgG as secondary antibody (Bio-Rad, Hercules, CA), and blots
were developed with an enhanced chemiluminescence kit (Amersham
Biosciences, Little Chalfont, United Kingdom). The experiment was
repeated at least twice with independent protein samples.
Small interfering RNA transfection. The pSUPER vector encoding 19-
mer hairpin small interfering RNA (siRNA) duplex specific to GnRH
receptor sequence 5¶-AGATCCGAGTGACGGTTAC-3¶ (nucleotides 107-125
downstream of the start codon) has been described elsewhere (18). As a
nonspecific siRNA control, scrambled siRNA duplex was used. Transfection
was done using LipofectAMINE 2000 reagent (Invitrogen) following the
manufacturer’s instruction.
Gelatin zymography. The activities of MMP-2 and MMP-9 in the
conditioned medium were determined by gelatin zymography. Briefly, the
media were collected and clarified by centrifugation to remove cells and
debris. Samples (80 Ag) were loaded under nonreducing conditions onto
SDS-polyacrylamide gel polymerized with 1 mg/mL gelatin. Following
electrophoresis, the gels were washed with 2.5% Triton X-100 to remove SDS
and then incubated in a developing buffer [50 mmol/L Tris-HCl buffer
(pH 7.4), 10 mmol/L CaCl2] overnight at 37jC. Gels were stained with 0.25%
Coomassie Brilliant Blue R-250 and destained in the same solution without
dye. Gelatinase activity was visualized as clear bands against the blue-
stained gelatin background. Molecular sizes were determined from mobility
using gelatin zymography standards (Chemicon International). Three
individual experiments were conducted with independent protein samples.
Semiquantitative reverse transcription-PCR. Total RNA was isolated
from cultured cells using the Trizol reagent (Invitrogen) according to the
manufacturer’s procedure. First-strand cDNA synthesis was done from
2.5 Ag of total RNA using SuperScript Reverse Transcriptase (Life
Technologies). The cDNA was used for subsequent PCR using primers
specific for human GnRH receptor, MMP-2, MMP-9, TIMP-1, and TIMP-2 as

www.aacrjournals.org 10903 Cancer Res 2006; 66: (22). November 15, 2006
Cancer Research
reported previously (19, 20). h-Actin was used as internal control to
normalize the relative amount of cDNA in each reaction. The logarithmic
phase of PCR amplification for each target gene had been determined in
initial experiments. The reproducibility of the quantitative measurements
was evaluated by three independent cDNA syntheses and PCR ran from
each preparation of RNA.
Reporter gene assay. Cells were transiently transfected with 1 Ag of the
construct using LipofectAMINE (Invitrogen) as directed by the manufac-
turer. After 6 hours, the cells were incubated with or without GnRHa
(0.1 nmol/L) for an additional 24 hours. The pSV-h-galactosidase plasmid
was cotransfected as internal control. Luciferase units were calculated as
luciferase activity/h-galactosidase activity and are presented as the
mean FSD of three individual experiments with triplication. The fold
change was calculated by comparison with the promoterless luciferase
vector (pGL3-Basic).
Invasion and migration assays. Twenty-four-well Transwell inserts
with 8-Am pores coated with Matrigel (50 AL/well; BD Biosciences, Palo
Alto, CA) were used to assess cell invasion. Cells (1  105) in serum-free
medium, with or without GnRHa, were seeded in triplicate in the upper
chamber. Serum-free medium was placed in the lower wells. The chambers
were incubated for 24 hours at 37jC in a 5% CO2 atmosphere. Cells that had
Cancer Res 2006; 66: (22). November 15, 2006 10904
not penetrated the filter were wiped out, and invaded cells on the lower
surface of the filter were fixed with ice-cold methanol and stained with 0.5%
crystal violet. Results are presented as the mean number of invaded cells of
five fields FSD of three independent experiments. To assess cellular
migration potential, the protocol described above was used, except that
Matrigel was omitted and 3T3 fibroblast conditioned medium was placed in
the lower chamber as a chemoattractant.
Statistical analysis. Statistical analysis was carried out using ANOVA
followed by Tukey’s post hoc test (GraphPad Software, San Diego, CA).
P < 0.05 was considered statistically significant.
Results
GnRH stimulates ovarian tumor cell migration and inva-
sion. During metastasis, invasion of basement membrane by tumor
cells is thought to be a critical event (21). To study whether
expression of GnRH receptor is associated with metastasis of
Downloaded from http://aacrjournals.org/cancerres/article-pdf/66/22/10902/2555753/10902.pdf by guest on 21 October 2022
ovarian cancer cells, the effect of GnRH on cellular motility and
invasive activity was evaluated. Because native GnRH has a
relatively short half-life both in vivo and in vitro, GnRHa, in
Figure 2. GnRH induces production and
synthesis of MMP-2 and MMP-9. A, cells
were grown for 24 hours in the presence or
absence of increasing doses of GnRHa
(0.1 nmol/L to 1 Amol/L). Equal amounts
of protein (50 Ag) were then analyzed by
Western blot using specific antibodies
recognizing MMP-2 and MMP-9.
Immunoblotting for h-actin was included as
loading controls. The signal intensity was
determined by densitometry, and the levels
of MMP-2 and MMP-9 protein were
normalized against that of h-actin.
B, conditioned medium from treated cells
was also collected and analyzed for MMP
activities by gelatin zymography. Arrows,
gelatinase activities corresponding to
pro-MMP-2, active MMP-2, pro-MMP-9,
and active MMP-9. Lane 1, a human
MMP-2 and MMP-9 standard was used as
a positive control. C, cells were transiently
transfected with vector expressing short
hairpin RNA targeted against GnRH
receptor or nonspecific siRNA as control.
Left, aliquots of conditioned medium
normalized for same total protein
concentration were subjected to gelatin
zymography; right, cell lysates (50 Ag) were
analyzed by Western blot for the presence
of MMP-2 and MMP-9 proteins.
Immunoblotting for h-actin was included as
loading controls. D, cells were grown in
the presence or absence of 0.1 nmol/L
GnRHa for 24 hours and total protein was
harvested for Western blot analysis using
specific antibodies against TIMP-1,
TIMP-2, and h-actin. Experiments were
repeated thrice. Columns, mean; bars, SD.
*, P < 0.05, statistically significant
difference when treated samples were
compared with untreated controls.
www.aacrjournals.org
 GnRH Stimulates Ovarian Cancer Invasion

Figure 3. GnRH increases MMP-2 and

MMP-9 mRNA synthesis through modulation
of transcription but not mRNA stability.
A, cells were grown for 24 hours in the
presence or absence of increasing doses
of GnRHa (0.1 nmol/L to 1 Amol/L). Cells
were then harvested for semiquantitative
PCR with specific primers as described in
Materials and Methods. h-Actin served as an
internal control. B, cells were pretreated with
0.1 nmol/L GnRHa for 3 hours followed by
the post-treatment with 4 Ag/mL ActD over a
time course of 0, 1, 2, 4, and 6 hours. Cells
in the absence of GnRHa pretreatment
served as control. Total RNA was then
extracted and reversed transcribed followed
by PCR with MMP-2, MMP-9, and h-actin
sequence-specific primers. The signal
intensity was determined by densitometry,
and the level of MMP-2 and MMP-9 mRNA
was normalized against that of h-actin.

Downloaded from http://aacrjournals.org/cancerres/article-pdf/66/22/10902/2555753/10902.pdf by guest on 21 October 2022

The MMP-2 and MMP-9 mRNA levels before
the addition of ActD were arbitrarily set to be
100%. C, cells (5  105) were plated in a
six-well plate and transfected with 1 Ag of
reporter constructs of MMP-2 and MMP-9
promoter. All cells were cotransfected with
0.5 Ag of pSV-h-galactosidase construct.
Eight hours after transfection, cells were
incubated with 0.1 nmol/L GnRHa for a
further 24 hours and harvested. Luciferase
and h-galactosidase activities were assayed,
and the luciferase activity of each sample
was normalized by h-galactosidase activity.
The relative luciferase activities were
calculated relative to the pGL3-Basic vector
(pGL3), which was arbitrarily assigned a
value of ‘‘1.’’ Experiments were repeated
thrice. Columns, mean; bars, SD.
*, P < 0.05; **, P < 0.005, compared with
untreated controls.

which Gly6 has been substituted by a D-amino acid, have been
commonly used to resist degradation and to increase potency (22).
The GnRHa D-Ala6 used in this study exhibits 3.5 to 4.5 times
higher potency and 3 to 8 times higher stability than the native
hormone in vitro (22, 23). Caov-3 and OVCAR-3 cells, which express
functional GnRH receptors (24), were treated with increasing
concentrations of the GnRHa (D-Ala6), and their invasive capacities
were assessed using Transwells with filters coated with Matrigel.
Our results showed that GnRHa induced a dose-dependent
biphasic regulation on invasion: cells treated with GnRHa at low
doses of 0.1 to 10 nmol/L showed significantly higher numbers of
invaded cells (P < 0.05), whereas the response to higher doses of
GnRHa (100 nmol/L to 1 Amol/L) was insignificant (Fig. 1A).
We also assessed in vitro migration in response to GnRH
stimulus using a Transwell migration assay. GnRHa dose
dependently stimulated the migration of ovarian cancer cells
through uncoated porous filter, with no significant effect at a
concentration of 100 nmol/L and 1 Amol/L and a maximal effect at
0.1 nmol/L (P < 0.05; Fig. 1B).
GnRH receptor siRNA inhibits ovarian tumor cell invasion
and migration. Numerous studies have shown that alteration of
GnRH receptor mRNA levels is involved in the regulation of ovarian
responsiveness to GnRH (19, 25, 26). We therefore examined
whether the effect of GnRH on ovarian cancer invasion was
associated with altered GnRH receptor gene expression. As shown
in Fig. 1C, treatment with the GnRHa induced a biphasic regulation
pattern for GnRH receptor mRNA in both Caov-3 and OVCAR-3.
Low doses of GnRHa (0.1-10 nmol/L) increased GnRH receptor
mRNA levels, with a maximal effective dose of 0.1 nmol/L (80.5%
increase); thereafter, the levels decreased in cells treated with
GnRHa at 100 nmol/L to 1 Amol/L. Cotreatment of cells with the
GnRH antagonist antide prevented the biphasic effects of GnRHa,
whereas antide alone had no effect, indicating the specificity of the
response (data not shown).
To further confirm the role of GnRH receptor on the invasive
properties of ovarian cancer cells, we inhibited the expression of
GnRH receptor using vector-based RNA interference assay. The
effectiveness of this siRNA to deplete GnRH receptor expression
was confirmed by semiquantitative reverse transcription-PCR (RT-
PCR) and Western blot analysis (data not shown) and was
consistent with the previous study (18). Importantly, siRNA-
mediated depletion of GnRH receptor, but not nonspecific siRNA,
reduced invasion to near basal levels (80.8% inhibition; Fig. 1D, left).
Similarly, we showed that depletion of GnRH receptor inhibited the
cell migration by 85.2% (Fig. 1D, right), confirming a direct
involvement of GnRH receptor in the acquisition of a migratory
and invasive phenotype in ovarian cancer cells.
GnRH induces MMP-2 and MMP-9 expression and enhances
their activities. We next determined whether GnRH-stimulated
cell invasion resulted from elevated levels of MMPs, which are well-
documented ECM-degrading enzymes and whose activity is asso-
ciated with tumor invasiveness (27). MMP protein and enzymatic
activities were measured by Western blotting and gelatin
zymography using conditioned medium from Caov-3 (Fig. 2) and
OVCAR-3 (data not shown). As shown in Fig. 2A, the effect of
GnRHa on MMP production was dose dependent because low

www.aacrjournals.org 10905 Cancer Res 2006; 66: (22). November 15, 2006
Cancer Research
concentrations (0.1-10 nmol/L) of GnRHa resulted in an up-
regulation of MMP-2/MMP-9 expression (P < 0.05), whereas high
concentration (100 nmol/L to 1 Amol/L) had minimal response
(Fig. 2A). Consistently, gelatin zymography showed stronger lytic
zones at the molecular mass corresponding to latent and active
forms of MMP-2 (72-kDa and 66-kDa forms) and MMP-9 (92-kDa
and 86-kDa forms) in cells treated with 0.1 nmol/L GnRHa with
respect to untreated cells, showing their elevated secretion and
enzyme activities by GnRHa (Fig. 2B). These results matched well
with those observed in cellular invasion (Fig. 1), suggesting that
GnRH-mediated induction of MMP-2 and MMP-9 may play a role
in GnRH-enhanced cell invasion/migration. To ascertain whether
GnRH plays a critical role in the induction of MMP-2 and MMP-9,
we monitored the regulation of these proteinases after transfection
of a GnRH receptor-targeted siRNA. As shown in Fig. 2C, both the
activity and expression of MMP-2 and MMP-9 were clearly reduced
in the presence of the GnRH receptor siRNA. In contrast, control
experiments with nonspecific siRNA showed negligible effect.
MMP expression is often coordinately regulated with production
of their endogenous inhibitors, TIMPs (21). However, there was no
change in the expression of TIMP-1 and TIMP-2 irrespective of the
presence of different concentration of GnRHa as shown by Western
blot analysis (Fig. 2D).
GnRH modulates transcription activities of MMP-2 and
MMP-9. To determine whether the observed alteration in MMP-2
and MMP-9 was the result of changes in gene expression, RT-PCR
Cancer Res 2006; 66: (22). November 15, 2006 10906
was done. In accordance with changes in protein expression,
MMP-2 and MMP-9 mRNA expression were found to be signi-
ficantly enhanced with 0.1 nmol/L GnRHa (5.3-fold for MMP-2 and
2.5-fold for MMP-9 compared with untreated control cells; P < 0.05;
Fig. 3A).
To test whether the effect of GnRH on MMP-2 and MMP-9
mRNA expression was the result of increased mRNA stability, we
did actinomycin D (ActD) chase experiments to determine the half-
life of MMP-2 and MMP-9 mRNA. Cells were preincubated with
0.1 nmol/L GnRHa for 3 hours. Then, ActD (4 Ag/mL) was added to
stop transcription and RNA was isolated after 0, 1, 2, 4, and 6 hours.
The amount of gelatinase mRNA compared with h-actin was
determined by RT-PCR. As shown in Fig. 3B, GnRHa did not affect
the decay rate of the MMP-2 and MMP-9 mRNA.
To assess whether GnRHa causes transcriptional activation at
the MMP-2 and MMP-9 promoters, human MMP-2 or MMP-9
Downloaded from http://aacrjournals.org/cancerres/article-pdf/66/22/10902/2555753/10902.pdf by guest on 21 October 2022
promoter-driven luciferase constructs were transiently transfected
into Caov-3 and OVCAR-3 cells in either the presence or the
absence of 0.1 nmol/L GnRHa for 24 hours. The results showed
that exposure of cells to GnRHa increased the activities of MMP-2
and MMP-9 promoters by 10- and 3-fold, respectively (P < 0.005;
Fig. 3C). Moreover, these increases in promoter activities were
completely abolished by GnRH receptor siRNA treatment but not
the nonspecific siRNA (Fig. 3C). These results collectively show a
role for GnRH in the transcriptional activation of MMP-2 and
MMP-9.
Figure 4. Inhibition of MMP-2 and MMP-9
suppresses cell invasion and migration in
response to GnRH. Cells were left untreated
(control) or pretreated with OA-Hy
(15 Amol/L), SB-3CT (10 Amol/L), or
neutralizing MMP-9 antibody (anti-MMP-9 ;
15 Ag/mL) for 30 minutes. A, pretreated
cells (1.5  105) were seeded on
Matrigel-coated (50 Ag/mL) Transwell
chambers (8-Am pore) in the presence or
absence of 0.1 nmol/L GnRHa. After
24 hours, cells on the upper surface were
removed; the invaded cells were stained by
crystal violet and counted. B, in a parallel
experiment, cells that migrated through
uncoated filters were counted. Columns,
mean number of invaded/migrated cells of
five fields of triplicate wells from three
independent experiments; bars, SD.
**, P < 0.005, compared with untreated
controls.
www.aacrjournals.org

Inhibition of MMP blocks GnRH-enhanced invasion and
migration in ovarian cancer cells. To evaluate the potential
contribution of MMP-2 and MMP-9 in the GnRH-induced invasion
of ovarian cancer cells, we blocked the gelatinase activities using
potent MMP-2 (OA-Hy, 15 Amol/L) or MMP-9 (SB-3CT, 10 Amol/L)
specific inhibitors or MMP-9 neutralizing antibody (15 Ag/mL) in
the presence or absence of 0.1 nmol/L GnRHa. Specific inhibition of
MMP-2 by OA-Hy reversed the stimulatory effect of GnRH on cell
invasion (by 81.8% in Caov-3 and 82.9% in OVCAR-3), whereas
concomitant treatment with MMP-9 inhibitor SB-3CT or anti-
MMP-9 significantly blocked the increase in invaded cells after
GnRHa treatment (by 75.6% and 72.7%, respectively, in Caov-3 and
70.3% and 73.1%, respectively, in OVCAR-3; Fig. 4A). Similar results
were obtained in migration assays as shown in Fig. 4B. Together,
these experiments show a crucial role for both MMP-2 and MMP-9
in the GnRH-dependent invasion and migration.
Prolonged activation of JNK by GnRH. Because studies have
shown that the MAPK pathway is critical for the activation of gene
expression by GnRH in some cell types (28, 29), we asked if they
also play a role in GnRH-induced human ovarian cancer cell
invasion and/or migration. We first examined the levels of phos-
phorylated (active) forms of MAPK family members (ERK1/2,
JNK, and p38 MAPK) in Caov-3 and OVCAR-3 cells treated with
0.1 nmol/L GnRHa for increasing time course of 0, 5, 15, 30, 60, and
120 minutes. As shown, addition of GnRHa to the cells significantly
increased the phosphorylation of ERK1/2, JNK, and p38 MAPK
(Fig. 5). Interestingly, whereas prominent (15-fold) and sustained
activation (60 minutes) of JNK was shown in GnRH-activated cells
(P < 0.005), activation of ERK1/2 and p38 MAPK was rapid
and transient, reaching a maximum of approximately 2- to 3-fold
15 minutes after GnRHa addition and decline thereafter (Fig. 5).
Figure 5. Time course activation of
ERK1/2, p38 MAPK, and JNK by GnRH
stimulation. Cells were exposed to 0.1
nmol/L GnRHa for the indicated durations.
Activities of ERK1/2, p38 MAPK, and
JNK were determined by Western
blotting using antibodies specific for
phosphorylated, activated forms of ERK1/2
(A), p38 MAPK (B), and JNK (C). A to C,
bottom, Western blots using antibodies to
total ERK1/2, p38 MAPK, and JNK. The
relative intensity of phosphorylated forms of
ERK1/2, p38 MAPK, and JNK was
normalized to total values of ERK1/2, p38
MAPK, and JNK, which were determined by
densitometric analysis. Three identical
experiments independently done yielded
similar results. *, P < 0.05; **, P < 0.005,
compared with untreated controls.
www.aacrjournals.org 10907
GnRH Stimulates Ovarian Cancer Invasion
GnRH-induced invasion and motility are JNK dependent. To
assess the functional significance of GnRH-activated ERK1/2, JNK,
and p38 MAPK in invasion and motility, we asked if interfering with
their activation by specific inhibitors diminishes GnRH-mediated
in vitro invasiveness and motility. As shown, treatment of 50 Amol/L
SP600125, a specific JNK inhibitor, significantly reduced the number
of both invaded (Fig. 6A) and migrated (Fig. 6B) cells to near basal
levels. In contrast, inhibition of GnRH-stimulated ERK1/2 and p38
MAPK activation by 50 Amol/L PD98059 or 10 Amol/L SB203580,
respectively, had no effect (Fig. 6A and B). These results indicate a
critical role for JNK, but not ERK1/2 or p38 MAPK, in GnRH-induced
invasive phenotype and migration in ovarian cancer cells.
To additionally investigate the role for JNK in cellular invasion
and migration, we transfected Caov-3 and OVCAR-3 cells with a
dominant-negative JNK (DN-JNK) construct in which the dual
phosphorylation motif Thr-Pro-Tyr was mutated to Ala-Pro-Phe
Downloaded from http://aacrjournals.org/cancerres/article-pdf/66/22/10902/2555753/10902.pdf by guest on 21 October 2022
(30). Specific inactivation of JNK in DN-JNK-transfected cells was
confirmed by Western blot analysis (Fig. 6A, inset). Both the
invasive phenotype and cell migration were clearly reduced in
DN-JNK transfectants as shown in Fig. 6A and B, showing the
significance of the JNK pathway in GnRH-induced invasion and
migration in Caov-3 and OVCAR-3 cells.
GnRH-mediated up-regulation of MMP-2 and MMP-9
involves JNK. Next, we asked whether JNK mediates GnRH
induction of MMP-2 and MMP-9 activities. As shown in Fig. 6C,
both basal and GnRH-stimulated proform and active form of the
MMP-2 and MMP-9 production were inhibited with 50 Amol/L
SP600125. GnRH-induced MMP-2 and MMP-9 promoter activities
were also inhibited by targeted inhibition of JNK using SP600125 or
DN-JNK (Fig. 6D). Consistent with the effects on invasion and cell
migration, pretreatment with either PD98059 or SB203580 showed
Cancer Res 2006; 66: (22). November 15, 2006
Cancer Research

Figure 6. Inhibition of JNK inhibits

GnRH-induced cell invasion, migration,
and MMP expression. Cells were
pretreated with the ERK1/2 inhibitor
PD98059 (PD ; 50 Amol/L), the p38 MAPK
inhibitor SB203580 (SB ; 10 Amol/L), or the
JNK inhibitor SP600125 (SP ; 50 Amol/L)
for 30 minutes. In some cases, cells were
transiently transfected with either the
control vector (pcDNA-3 ) or DN-JNK
construct in the presence or absence of
GnRHa. Inset, whole-cell lysates were
analyzed for the levels of phosphorylated
and total forms of JNK by Western blot
analysis. A, cells were then allowed to
invade through the Matrigel-coated filters
for 24 hours in the presence or absence of
GnRHa (0.1 nmol/L). B, migration through

Downloaded from http://aacrjournals.org/cancerres/article-pdf/66/22/10902/2555753/10902.pdf by guest on 21 October 2022

uncoated filters of the cells was measured.
Columns, mean number of invaded/
migrated cells of five fields of triplicate
wells from three independent experiments;
bars, SD. C, cells were either left untreated
(control) or pretreated with the indicated
kinase inhibitor for 30 minutes followed by
0.1 nmol/L GnRHa addition for another
24 hours. The MMP gelatinolytic activities
secreted into the medium were examined
by gelatin zymography. D, cells were
transiently transfected with either the
MMP-2 or the MMP-9 promoter/reporter
construct. Six hours after transfection, cells
were pretreated with the JNK inhibitor
SP600125 (50 Amol/L) for 30 minutes and
then incubated with 0.1 nmol/L GnRHa.
In some cases, cells were cotransfected
with DN-JNK as indicated. After 24 hours of
incubation, the cells were lysed and
luciferase activities were determined.
Luciferase values were normalized
for transfection efficiency (luciferase/
h-galactosidase ratios). The relative
luciferase activities were calculated relative
to the pGL3-Basic vector, which was
arbitrarily assigned a value of 1. Columns,
mean of three independent experiments;
bars, SD.

no inhibition on GnRH-mediated MMP induction (data not dependent manner, high concentrations were not as efficient.
shown). These results suggest that activation of JNK is required Alteration of GnRH receptor numbers or gene expression is
for enhanced MMP-2 and MMP-9 expression. involved in mechanisms underlying the biphasic effect of GnRH in
the pituitary and ovary (19, 25, 26, 31). In ovarian carcinoma cells,
Discussion we showed previously (25) and in this study that low doses of
GnRH up-regulated its receptor, whereas high doses decreased it.
There is much evidence that implicates an important role for
This also argues for a role for GnRH receptor to activate
GnRH in regulating the proliferation for a wide range of hormone-
intracellular signals necessary for cellular responses. Notably, the
dependent malignancies, including ovarian cancer (1). However,
effects of GnRH on JNK and MMP-2/MMP-9 activation were
little is known about its role in tumor invasion, metastasis, and
consistent with changes in the GnRH receptor mRNA levels. When
other aspects of cancer progression. In this study, we report for the
GnRH receptor mRNA was up-regulated, GnRH enhanced JNK and
first time that GnRH may contribute to the metastatic dissemina-
MMP-2/MMP-9 activities; however, when GnRH receptor mRNA
tion of ovarian cancer cells by promoting the expression and/or
was unaffected or down-regulated, GnRH had no effect on JNK3
processing of metastasis-related MMP, with subsequent down-
and MMP-2/MMP-9. Our finding that siRNA-mediated down-
stream changes in migratory and invasive behavior, providing an
regulation of the GnRH receptor inhibited MMP-dependent
insight into the prospect of developing targeted therapy for ovarian
invasion and cellular migration further upholds this view.
carcinoma.
The present study showed a biphasic effect of GnRH on cellular
migration and invasion. Whereas lower concentrations of the
3
GnRHa stimulated cellular migration and invasion in a dose- Unpublished data.

Cancer Res 2006; 66: (22). November 15, 2006 10908 www.aacrjournals.org
 GnRH Stimulates Ovarian Cancer Invasion

Metastasis is a complex phenomenon that requires several
specific steps, such as decreased adhesion, increased motility, and
proteolysis. In prostate cancer, GnRH has been shown to regulate
cell motility through its interaction with the small GTPases Rac1,
Cdc42, and RhoA, which are involved in the regulation of actin
polymerization (32). Our data show that another mechanism by
which GnRH may promote tumor cell migration and invasion is
through increased expression and proteolytic activities of MMP-2
and MMP-9 that specifically degrade the basement membrane.
MMP-2 and MMP-9 have been suggested to play critical roles in
ovarian cancer metastasis, and up-regulation of MMP-2 and
MMP-9 is associated with increased invasion and poor prognosis
in late-stage or invasive ovarian cancer patients (8–11). In addition
to their enzymatic activities, MMP-2 and MMP-9 can also promote
tumor cell migration by influencing cytoskeletal organization
through their association with different families of adhesion
The kinetic profile revealed differential regulation of ERK1/2, p38
MAPK, and JNK by GnRH with a sustained signaling through the
JNK pathway. The duration of kinase activation is a major
determinant for signal outcome in different cellular contexts (39),
and its influence may have included invasiveness. In support
of this hypothesis, GnRH-stimulated MMP expression and cell
invasion were attenuated by specific inhibition of JNK but not
ERK1/2 and p38 MAPK. Consistently, a recent report shows that
activated JNK levels are much higher in the malignant pleural
and peritoneal effusions of patients with advanced-stage ovarian
carcinomas, suggesting that activation of JNK may contribute to
a more invasive phenotype (40). However, unlike ERK1/2 and p38
MAPK, a role for JNK in tumor invasion or migration has not
been widely reported. JNK signaling was shown recently to
regulate MMP-2 production and invasiveness in human gliomas
(41). The JNK pathway targets multiple transcription factors,

Downloaded from http://aacrjournals.org/cancerres/article-pdf/66/22/10902/2555753/10902.pdf by guest on 21 October 2022

receptors (33). They have also been reported to promote ascites
formation in ovarian cancer cells via vascular endothelial growth
factor production (34). Surprisingly, activation of MMP-2 and
MMP-9 was not related to a decrease in the specific inhibitor of
TIMP-1 and TIMP-2 in the GnRH-induced ovarian cancer cell
invasion and motility. There was, however, an increase in MMP/
TIMP ratios, which also in favor of ECM degradation.
One of the novel findings of the present study is the demons-
tration that GnRH is a direct transcriptional activator of MMP-2
and MMP-9 synthesis. MMP-2 and MMP-9 expression are reported
to be regulated at many levels, including gene activation, mRNA
stability, proenzyme activation, and inactivation by endogenous
inhibitors in a complex fashion by numerous oncogene and tumor
suppressor pathways and conditions of hypoxia (35–38). In this
study, we showed that GnRH increased MMP-2 and MMP-9 mRNA
expression, and such expression was not due to changes in mRNA
stabilities. We further showed that GnRH directly induced MMP-2
and MMP-9 promoter activities. This is the first demonstration that
MMP-2 and MMP-9 are target genes of GnRH receptor activation
and adding GnRH as a new member of MMP-2 and MMP-9
transcriptional modulators.
Our results identify a JNK signaling pathway that mediates
GnRH-stimulated MMP expression, secretion, and cell invasion.
including c-Jun and c-Fos (42), ATF (43), and PEA (44), and
putative binding sites for these DNA-binding proteins are present
in the MMP promoters (45). Whether these putative regulatory
elements participate in the GnRH-dependent activation of the
MMP-2 and MMP-9 genes remains to be determined.
In summary, our findings emphasize the potential role of GnRH
in promoting cellular migration and metastasis-related proteinase
activities in GnRH receptor-expressing ovarian tumors. This
information provides a mechanistic rationale for the observed
GnRH receptor overexpression in advanced-stage ovarian carcino-
mas. The promising role for GnRH in invasion of ovarian cancer
cells in vitro merits further investigation in vivo, which may provide
additional insight into its potential as a therapeutic molecular
target to decrease metastasis.
Acknowledgments
Received 6/16/2006; revised 8/24/2006; accepted 9/12/2006.
Grant support: Hong Kong Research Grants Council grants 7484/04M and 7599/
05M (A.S.T. Wong) and Canadian Institutes of Health Research grant 42441 (P.C.K.
Leung).
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.

References
1. Kang SK, Choi KC, Yang HS, Leung PC. Potential role
of gonadotrophin-releasing hormone (GnRH)-I and
GnRH-II in the ovary and ovarian cancer. Endocr Relat
Cancer 2003;10:169–77.
2. Irmer G, Burger C, Muller R, et al. Expression of the
messenger RNAs for luteinizing hormone-releasing
hormone (LHRH) and its receptor in human ovarian
epithelial carcinoma. Cancer Res 1995;55:817–22.
3. Volker P, Grundker C, Schmidt O, Schulz KD, Emons G.
Expression of receptors for luteinizing hormone-releas-
ing hormone in human ovarian and endometrial
cancers: frequency, autoregulation, and correlation with
direct antiproliferative activity of luteinizing hormone-
releasing hormone analogues. Am J Obstet Gynecol
2002;186:171–9.
4. Chien CH, Chen CH, Lee CY, Chang TC, Chen RJ, Chow
SN. Detection of gonadotropin-releasing hormone
receptor and its mRNA in primary human epithelial
ovarian cancers. Int J Gynecol Cancer 2004;14:451–8.
5. Naora H, Montell DJ. Ovarian cancer metastasis:
integrating insights from disparate model organisms.
Nat Rev Cancer 2005;5:355–66.
6. Mignatti P, Rifkin DB. Biology and biochemistry of
proteinases in tumor invasion. Physiol Rev 1993;73:
161–95.
7. Fishman DA, Liu Y, Ellerbroek SM, Stack MS.
Lysophosphatidic acid promotes matrix metalloprotei-
nase (MMP) activation and MMP-dependent invasion in
ovarian cancer cells. Cancer Res 2001;61:3194–9.
8. Moser TL, Young TN, Rodriguez GC, Pizzo SV, Bast RC,
Jr., Stack MS. Secretion of extracellular matrix-degrading
proteinases is increased in epithelial ovarian carcinoma.
Int J Cancer 1994;56:552–9.
9. Young TN, Rodriguez GC, Rinehart AR, Bast RC, Jr.,
Pizzo SV, Stack MS. Characterization of gelatinases
linked to extracellular matrix invasion in ovarian
adenocarcinoma: purification of matrix metalloprotei-
nase 2. Gynecol Oncol 1996;62:89–99.
10. Fishman DA, Bafetti LM, Banionis S, Kearns AS,
Chilukuri K, Stack MS. Production of extracellular
matrix-degrading proteinases by primary cultures of
human epithelial ovarian carcinoma cells. Cancer 1997;
80:1457–63.
11. Davidson B, Goldberg I, Gotlieb WH, et al. The
prognostic value of metalloproteinases and angiogenic
factors in ovarian carcinoma. Mol Cell Endocrinol 2002;
187:39–45.
12. Davies B, Brown PD, East N, Crimmin MJ, Balkwill FR.
A synthetic matrix metalloproteinase inhibitor decreases
tumor burden and prolongs survival of mice bearing
human ovarian xenografts. Cancer Res 1993;53:2087–91.
13. Brew K, Dinakarpandian D, Nagase H. Tissue
inhibitors of metalloproteinases: evolution, structure,
and function. Biochim Biophys Acta 2000;1477:267–83.
14. Khokha R, Denhardt DT. Matrix metalloproteinases
and tissue inhibitor of metalloproteinases: a review of
their role in tumorigenesis and tissue invasion. Invasion
Metastasis 1989;9:391–405.
15. Albini A, Melchiori A, Santi L, Liotta LA, Brown PD,
Stetler-Stevenson WG. Tumor cell invasion inhibited by
TIMP-2. J Natl Cancer Inst 1991;83:775–9.
16. DeClerck YA, Perez N, Shimada H, Boone TC, Langley
KE, Taylor SM. Inhibition of invasion and metastasis in
cells transfected with an inhibitor of metalloproteinases.
Cancer Res 1992;52:701–8.
17. Bian J, Sun Y. Transcriptional activation by p53 of the
human type IV collagenase (gelatinase A or matrix
metalloproteinase 2) promoter. Mol Cell Biol 1997;17:
6330–8.
18. An BS, Selva DM, Hammond GL, Rivero-Muller A,
Rahman N, Leung PC. Steroid receptor coactivator-3 is
required for progesterone receptor trans-activation of
target genes in response to gonadotropin-releasing

www.aacrjournals.org 10909 Cancer Res 2006; 66: (22). November 15, 2006
Cancer Research
hormone treatment of pituitary cells. J Biol Chem 2006;
281:20817–24.
19. Kang SK, Choi KC, Cheng KW, Nathwani PS, Auersperg
N, Leung PC. Role of gonadotropin-releasing hormone as
an autocrine growth factor in human ovarian surface
epithelium. Endocrinology 2000;141:72–80.
20. Zhou HY, Wong AS. Activation of p70S6K induces
expression of matrix metalloproteinase 9 associated
with hepatocyte growth factor-mediated invasion in
human ovarian cancer cells. Endocrinology 2006;147:
2557–66.
21. Liotta LA, Stetler-Stevenson WG. Tumor invasion
and metastasis: an imbalance of positive and negative
regulation. Cancer Res 1991;51:5054–9s.
22. Karten MJ, Rivier JE. Gonadotropin-releasing hor-
mone analog design. Structure-function studies toward
the development of agonists and antagonists: rationale
and perspective. Endocr Rev 1986;7:44–66.
23. Yang X, Rojanasakul Y, Wang L, Ma JY, Ma JK.
Enzymatic degradation of luteinizing hormone releasing
hormone (LHRH)/[D-Ala6]-LHRH in lung pneumocytes.
Pharm Res 1998;15:1480–4.
24. Kimura A, Ohmichi M, Kurachi H, et al. Role of
mitogen-activated protein kinase/extracellular signal-
regulated kinase cascade in gonadotropin-releasing
hormone-induced growth inhibition of a human ovarian
cancer cell line. Cancer Res 1999;59:5133–42.
25. Kang SK, Cheng KW, Nathwani PS, Choi KC, Leung
PC. Autocrine role of gonadotropin-releasing hormone
and its receptor in ovarian cancer cell growth.
Endocrine 2000;13:297–304.
26. Peng C, Fan NC, Ligier M, Vaananen J, Leung PC.
Expression and regulation of gonadotropin-releasing
hormone (GnRH) and GnRH receptor messenger
ribonucleic acids in human granulosa-luteal cells.
Endocrinology 1994;135:1740–6.
27. Coussens LM, Werb Z. Matrix metalloproteinases
and the development of cancer. Chem Biol 1996;3:
895–904.
Cancer Res 2006; 66: (22). November 15, 2006
28. Bonfil D, Chuderland D, Kraus S, et al. Extracellular
signal-regulated kinase, Jun N-terminal kinase, p38, and
c-Src are involved in gonadotropin-releasing hormone-
stimulated activity of the glycoprotein hormone
follicle-stimulating hormone h-subunit promoter.
Endocrinology 2004;145:2228–44.
29. Luo X, Ding L, Chegini N. Gonadotropin-releasing
hormone and TGF-h activate MAP kinase and differen-
tially regulate fibronectin expression in endometrial
epithelial and stromal cells. Am J Physiol Endocrinol
Metab 2004;287:E991–1001.
30. Derijard B, Hibi M, Wu IH, et al. JNK1: a protein
kinase stimulated by UV light and Ha-Ras that binds
and phosphorylates the c-Jun activation domain. Cell
1994;76:1025–37.
31. Kaiser UB, Jakubowiak A, Steinberger A, Chin WW.
Regulation of rat pituitary gonadotropin-releasing
hormone receptor mRNA levels in vivo and in vitro .
Endocrinology 1993;133:931–4.
32. Enomoto M, Utsumi M, Park MK. Gonadotropin-
releasing hormone induces actin cytoskeleton remodel-
ing and affects cell migration in a cell-type-specific
manner in TSU-Pr1 and DU145 cells. Endocrinology
2006;147:530–42.
33. Sanceau J, Truchet S, Bauvois B. Matrix metal-
loproteinase-9 silencing by RNA interference triggers
the migratory-adhesive switch in Ewing’s sarcoma cells.
J Biol Chem 2003;278:36537–46.
34. Belotti D, Paganoni P, Manenti L, et al. Matrix
metalloproteinases (MMP9 and MMP2) induce the
release of vascular endothelial growth factor (VEGF)
by ovarian carcinoma cells: implications for ascites
formation. Cancer Res 2003;63:5224–9.
35. Sehgal I, Thompson TC. Novel regulation of type IV
collagenase (matrix metalloproteinase-9 and -2) acti-
vities by transforming growth factor-h1 in human
prostate cancer cell lines. Mol Biol Cell 1999;10:407–16.
36. Baruch RR, Melinscak H, Lo J, Liu Y, Yeung O,
Hurta RA. Altered matrix metalloproteinase expres-
10910
sion associated with oncogene-mediated cellular
transformation and metastasis formation. Cell Biol
Int 2001;25:411–20.
37. Nair RR, Boyd DD. Expression cloning of novel
regulators of 92 kDa type IV collagenase expression.
Biochem Soc Trans 2005;33:1135–6.
38. Munoz-Najar UM, Neurath KM, Vumbaca F, Claffey
KP. Hypoxia stimulates breast carcinoma cell invasion
through MT1-MMP and MMP-2 activation. Oncogene
2006;25:2379–92.
39. Marshall CJ. Specificity of receptor tyrosine kinase
signaling: transient versus sustained extracellular signal-
regulated kinase activation. Cell 1995;80:179–85.
40. Davidson B, Konstantinovsky S, Kleinberg L, et al.
The mitogen-activated protein kinases (MAPK) p38 and
JNK are markers of tumor progression in breast
carcinoma. Gynecol Oncol 2006;102:453–61.
41. Hu B, Jarzynka MJ, Guo P, Imanishi Y, Schlaepfer DD,
Cheng SY. Angiopoietin 2 induces glioma cell invasion
by stimulating matrix metalloprotease 2 expression
through the avh1 integrin and focal adhesion kinase
signaling pathway. Cancer Res 2006;66:775–83.
Downloaded from http://aacrjournals.org/cancerres/article-pdf/66/22/10902/2555753/10902.pdf by guest on 21 October 2022
42. Cirillo G, Casalino L, Vallone D, Caracciolo A, De
Cesare D, Verde P. Role of distinct mitogen-activated
protein kinase pathways and cooperation between Ets-2,
ATF-2, and Jun family members in human urokinase-
type plasminogen activator gene induction by interleu-
kin-1 and tetradecanoyl phorbol acetate. Mol Cell Biol
1999;19:6240–52.
43. Gupta S, Campbell D, Derijard B, Davis RJ. Tran-
scription factor ATF2 regulation by the JNK signal
transduction pathway. Science 1995;267:389–93.
44. O’Hagan RC, Tozer RG, Symons M, McCormick F,
Hassell JA. The activity of the Ets transcription factor
PEA3 is regulated by two distinct MAPK cascades.
Oncogene 1996;13:1323–33.
45. Chakraborti S, Mandal M, Das S, Mandal A,
Chakraborti T. Regulation of matrix metalloproteinases:
an overview. Mol Cell Biochem 2003;253:269–85.
www.aacrjournals.org