ª
Oncogene (2002) 21, 8128 – 8139
2002 Nature Publishing Group All rights reserved 0950 – 9232/02 $25.00
www.nature.com/onc
Rapid tumor development and potent vascularization are independent events
in carcinoma producing FGF-1 or FGF-2
Clotilde Billottet1, Bassam Janji1, Jean-Paul Thiery1 and Jacqueline Jouanneau*,1
1
Laboratory of Cell Morphogenesis and Tumor Progression, UMR 144 CNRS, Institut Curie, Section de recherche, 26 rue d’Ulm,
75248 Paris, cedex 05, France
FGF-1 and FGF-2 are pleiotropic growth factors for
many cell types, operating through the activation of
specific transmembrane FGF receptors (FGFRs). The
role of these factors in tumor progression was investigated, with specific discrimination between their
autocrine and non autocrine cellular activity. The rat
bladder carcinoma NBT-II cells were engineered to
produce FGF-1 or 18 kDa FGF-2 in the presence or
absence of their specific receptor. Non-autocrine cells
that produced FGF-1 or FGF-2 but lacked FGFRs were
epithelial and reminiscent of the parental NBT-II cells.
Whilst autocrine cells, which both constitutively
produced and secreted the growth factor and expressed
FGFRs, had a highly invasive mesenchymal phenotype.
Correspondingly, the autocrine cells were highly tumorigenic in vivo compared to the parental and non-autocrine
cells, which correlated with the increased production of
uPAR and active uPA and increased in vitro invasive
potential. Although all cells produced VEGF, only
tumors derived from cells that produced FGF-1 or
FGF-2 were highly vascularized, suggesting that these
two growth factors could be involved in the angiogenic
process by activating host endothelial cells. As a result of
activation of the FGFR in autocrine cells, changes in cell
morphology and an increase in the invasive and
tumorigenic properties were observed, however no in
vitro or in vivo differential functions between FGF-1 and
FGF-2 could be identified in this system. In conclusion,
our data demonstrates that rapid tumor development is
not dependent upon increased tumor vascularization,
suggesting that ‘basal’ angiogenesis, probably mediated
by VEGF, is sufficient to support tumor growth.
Oncogene (2002) 21, 8128 – 8139. doi:10.1038/sj.onc.
1205935
Keywords: FGF signaling; tumor growth; angiogenesis;
MMPs; uPA/uPAR
*Correspondence: J Jouanneau;
E-mail: [email protected]
Received 10 June 2002; revised 2 August 2002; accepted 5 August
2002
Introduction
Cancer progression and tumor invasion depend critically on many cellular and molecular events, such as the
growth and migration of malignant and endothelial
cells. Various mechanisms cause a loss of cell adhesion
and motility by disrupting basement membrane integrity and degrading the extracellular matrix (ECM). The
migration of cancer cells and the formation of
metastases depend mostly on neovascularization, and
clinical studies have revealed that high vessel density is
correlated with poor prognosis (Weidner, 1999). Tumor
angiogenesis is controlled by positive and negative
modulators produced by cancer, stromal and infiltrating
cells. Many of these modulators are polypeptide growth
factors such as fibroblast growth factor-2 (FGF-2 or
basic FGF) and vascular endothelial growth factor
(VEGF). Endogenous or exogenous in vitro synthesis of
FGF-2 stimulates the proliferation and migration of
endothelial cells and the production of proteases such as
plasminogen activator (PA) and collagenases (Tsuboi et
al., 1990). FGF-2 and VEGF may have synergistic
effects on the induction of angiogenesis both in vitro
and in vivo. FGF-2 may activate the production of
VEGF by endothelial cells, indirectly increasing the
switch to angiogenesis in the tumor (Claffey et al., 2001;
Seghezzi et al., 1998). Clinical studies have shown that
there is a strong correlation between FGF-2, VEGF
levels and vascular density and between metastatic
potential and poor survival in human tumors (Giri et
al., 1999; Schmidt et al., 1999).
Interactions between the different tumor cells and
the surrounding tissue play a crucial role in tumor
growth, invasion, metastasis and angiogenesis. The
production of cancer cell- or stroma-derived factors
and proteases has a major effect on the early oncogenic
process in tumor progression, and often regulates this
cascade of events (Chang and Werb, 2001).
Matrix-metalloproteases (MMPs) and the urokinasetype plasminogen activator/receptor system (uPA/
uPAR) have been shown to be important extrinsic
regulators during cancer progression (Dano et al.,
1999). MMPs constitute a family of over 25 proteins
that degrade ECM components and often release
molecules (growth factors, bioactive fragments of
ECM) that regulate the biological activities of cells
(Bergers and Coussens, 2000; Nelson et al., 2000). Many
studies have shown a positive correlation between
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8129
Table 1
Cells
NBT-II
NSF-14:
NBT-II-FGF-1
bFG22:
NBT-II-FGF-2
DFlg25:
NBT-II-DN-FGFR1*
DFlg25-FGF-1:
cl17, cl29 and cl37
Flg8:
NBT-II-FGFR1
Flg8-FGF-2:
cl1 and cl19
Characteristics of NBT-II and derived NBT-II cells
Transmembrane
receptor FGFR
Constitutive
production of
FGFR2b (endo)
FGFR2b (endo)
Yes
FGF-1
Yes
status autocrine
FGF-2
Yes
status non-autocrine
No
FGFR2b (endo)
FGFR2b (endo) and
DN-FGFR1* (transf)
FGFR2b (endo) and
DN-FGFR1* (transf)
FGFR2b (endo) and
FGFR1 (transf)
FGFR2b (endo) and
FGFR1 (transf)
Response to Response to
FGF-1
FGF-2
Phenotype
Reference
No
No
epithelial
Toyoshima et al., 1971
mesenchymal Jouanneau et al., 1991, 1997
No
epithelial
Jouanneau et al., 1997
No
epithelial
Jouanneau et al., 1999
FGF-1
No
status non-autocrine
Yes
No
epithelial
Yes
epithelial
FGF-2
status autocrine
Yes
mesenchymal
Yes
Jouanneau et al., 1997
endo: endogenous; transf: transfected; DN-FGFR1*: dominant-negative form of FGFR1
tumor progression and the production of multiple
MMPs and their specific endogenous inhibitors called
tissue inhibitor of MMPs (TIMPs) (McCawley and
Matrisian, 2000). In many cases, the overproduction of
MMP-2 and MMP-9 is correlated with an invasive
phenotype in human cancers and ‘angiogenic switch’ in
tumors (Bergers et al., 2000; Fang et al., 2000; Maatta et
al., 2000). The proteolytic system of the plasminogen
activator consists of the urokinase serine protease, uPA
and its transmembrane receptor, uPAR. Active uPA,
linked to its receptor, converts plasminogen to plasmin,
which degrades ECM components and the basement
membrane, and activates certain MMPs and growth
factors. Thus, plasmin production may influence tumor
invasion both directly and indirectly (Andreasen et al.,
1997; Mazzieri et al., 1997). PAIs (plasminogen
activator inhibitors) are specific inhibitors of uPA/
uPAR. Clinical data for human hepatocellular carcinoma has suggested that uPA, uPAR and PAI are all
required for proteolysis during tumor cell invasion and
metastasis (Zheng et al., 2000). Conversely, in a murine
model, absence of PAI-1 prevented tumor progression
and angiogenesis (Bajou et al., 1998). The regulation of
these two proteolytic systems is complex and may
require growth factors. Indeed, FGF-2 has been
reported to stimulate the production of MMPs and of
uPA/uPAR in cancer and endothelial cells (Miyake et
al., 1997; Tsuboi et al., 1990).
FGFs play significant roles in development, woundhealing and tumor progression (Basilico and Moscatelli, 1992; Powers et al., 2000). They induce cell
motility, proliferation and differentiation in a variety of
cell types and favor cell survival. FGF-1 and FGF-2
lack leader sequences for extracellular export, in
contrast with other FGFs that are secreted. The
interaction of FGFs with specific transmembrane
tyrosine kinase receptors FGFRs leads to the transduction of intercellular signals, that can trigger the
proliferation and migration of endothelial cells by
inducing the production of proteases in vitro (Tsuboi et
al., 1990). These two FGFs increase angiogenic activity
in various experimental models (Compagni et al., 2000;
Pili et al., 1997). Experimental approaches and clinical
data have also shown that both FGF-1 and FGF-2
may behave as transforming/oncogenic factors and
may be involved in tumor progression (Forough et al.,
1993; Nesbit et al., 1999; Soslow et al., 1999).
In this study, we investigated the mechanisms by
which FGF-1 and FGF-2 favor tumor progression of
experimental carcinoma. Cell lines autocrine or nonautocrine for these growth factors were generated by
stable transfections and evaluated for their invasiveness, tumorigenic and angiogenic capacities. We found
that autocrine FGF-1 or FGF-2 production promoted
(i) cell scattering and cell motility in an invasion assay
model, (ii) cell invasiveness closely correlated with an
increase in the production and activity of proteases and
(iii) tumorigenicity in this model of carcinoma. This
study also demonstrated that carcinoma cells autocrine
or non-autocrine for FGF-1 or FGF-2 gave rise to
highly vascularized tumors indicating that extensive
tumor angiogenesis was not necessarily correlated with
greater tumor development.
Results
Transfected NBT-II cells produce and secrete FGF-1 or
FGF-2 in a biologically active form
The various cells that were used and selected are
described in Table 1. Potentially autocrine cells possess
a transmembrane receptor FGFR and produce the
growth factor, whereas the non-autocrine cells produce
the growth factor but have no functional FGFR. The
expected FGF was produced in several selected
DFlg25 – FGF-1 and Flg8 – FGF-2 clones (Figure
1Aa,b and 1B), but among them only one transfectant
for each type was presented for subsequent in vitro and
in vivo results: cl37 (DFlg25 – FGF-1) and cl1 (Flg8 –
FGF-2) (for cell nomenclature see Table 1).
Although neither FGF-1 nor FGF-2 possesses a
signal peptide, FGF-2 was detected in the conditioned
medium of cells that produced FGF-2 (Figure 2Aa).
The FGF-2 secreted by cl1 cells was biologically active
against Flg8 cells but, as expected, had no effect on
NBT-II cells (Figure 2Ab,c). No putative secreted
Oncogene
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8130
FGF-1 was detected by Western blotting with conditioned media or by testing the biological activity of the
cell medium (data not shown).
We also tested for potential autocrine activity of the
growth factor in cell aggregate experiments as NBT-II
and their derivatives have been shown previously to
dissociate following the stimulation of cells with
exogenous FGF-1 or FGF-2 (Savagner et al., 1994;
Valles et al., 1990). Furthermore, epithelial Flg8 cells
cultured for 48 h in medium conditioned by bGF22
cells were dissociated whereas polyclonal FGF-2
blocking antibodies abolished this scattering activity
of FGF-2 on Flg8 cells (Jouanneau et al., 1997).
Fluorescent aggregates of NBT-II or Flg8 cells were
deposited on monolayers of NBT-II cells and of cells
producing FGF-1 or FGF-2. As expected, aggregates
on NBT-II cells were not dissociated (Figure 2Ba,d)
whereas signaling between FGF-2-producing cells and
responsive Flg8 cells resulted in the dissociation of the
aggregates upon signaling by the secreted FGF-2
(Figure 2Bf). Similar results were obtained with
NBT-II or Flg8 aggregates in contact with FGF-1producing cells, indicating that FGF-1, even though it
has no signal peptide, can be also exported and, by
paracrine/juxtacrine signaling, can target the transmembrane FGFRs of NBT-II and Flg8 cells, to
mediate a cell dissociation signal (Figure 2Bb,e).
Furthermore, control NBT-II cell aggregates were
dissociated upon addition of exogenous FGF-1 to the
complete medium (20 ng/ml) (data not shown). In
addition, we have shown that the NSF14 cell
phenotype can be partially reverted by using soluble
FGFR1 (Figure 2Ca,b). Thus, both NSF14 and cl1
cells are autocrine for the FGF they produce.
NBT-II cells autocrine for FGF-1 or FGF-2 acquire a
mesenchymal phenotype
NBT-II cells producing endogenous FGF-1 exhibit a
mesenchymal phenotype in vitro, whereas NBT-II cells
producing endogenous FGF-2 remain epithelial, like
parental NBT-II cells (Jouanneau et al., 1991, 1997).
We also found that cl37 cells were not scattered in vitro
(cl17 and cl29 presented the same phenotype) whereas
cl1 cells had a mesenchymal phenotype (cl19 presented
the same phenotype) (Figure 1Ca,b). Autocrine
production of FGF-2 by cl1 carcinoma cells resulted
in a mesenchymal phenotype, similar to that of cells
autocrine for FGF-1 (data not shown). This was
further demonstrated by immunostaining of the
desmosomes, one of the main adhesive junctions
between epithelial cells, and of vimentin, an interFigure 1 Characterization of FGF-producing NBT-II cells. (A)
Production of the growth factor (a,c: FGF-1; b,d: FGF-2) was detected by Western blotting of cell extracts (a,b) and tumor cell extracts (c,d). Recombinant 18 kDa human FGF-2 (100 ng) was
used as the standard. (For cell nomenclature see Table 1). (B) Immunostaining for FGF-1 (a) and FGF-2 (d) in NBT-II cells.
FGF-1 produced by NSF-14 cells (b) and cl37 (c) cells was located in the cytoplasm (red) and the nuclei were stained with
DAPI (blue). FGF-2 was detected in the cytoplasm of bGF22
Oncogene
(e) and cl1 (f) cells. Photographs were taken at 6400 magnification, using a fluorescence microscope coupled to a CCD camera.
(C) a and b show the live cell morphology (a: cl37; b: cl1) under
an inverted microscope at 6200 magnification. Double labeling
for desmosomes (green) and vimentin (red) was performed on
cl37 (c) and cl1 (d) cells. Nuclei were stained with DAPI (blue).
Photographs were taken at 6400 magnification, using a fluorescence microscope coupled to a CCD camera
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8131
mediate filament present in mesenchymal cells. NBT-II
cells non-autocrine for FGF-1 or FGF-2 have
desmosomal junctions, like parental NBT-II cells,
whereas cells autocrine for FGF-2 are dissociated with
mostly internalized desmosomes and like cells autocrine
for FGF-1 (data not shown), they contain vimentin
(Figure 1Cc,d).
Carcinoma cells autocrine for FGF-1 or FGF-2 have
enhanced invasive properties
The invasive potential of the cells was determined by a
Boyden chamber Matrigel invasion assay. Figure 3A
clearly shows that NBT-II carcinoma cells had a
modest invasive potential (3% of the cells), whereas
the proportion of invasive cells was significantly higher
for cells autocrine for FGF-1 or FGF-2 (NSF14 and
cl1 cells, P50.05).
NBT-II cells and FGF-producing NBT-II cells produce
MMPs and TIMP-2
Figure 2 Secretion and biological activity of FGF-1 and FGF-2.
(A) Secretion of FGF-2 by cells (bGF22 and cl1 cells) was detected by Western blotting with conditioned medium (a). NBTII cells and Flg8 cells were cultured for 48 h in the presence of
medium conditioned by FGF-2-producing cells (cl1 cells). Scattering activity of the cl1 cell-conditioned medium on NBT-II cells (b)
and on Flg8 cells (c). Photographs were taken with an inverted
microscope at 6100 magnification. (B) Test for aggregate dissociation: red fluorescent NBT-II and Flg8 aggregates deposited
on monolayers of control cells (a,d: NBT-II cells), monolayers
of FGF-1-producing cells (b,e: NSF14 cells) or on monolayers
of FGF-2-producing cells (c,f: cl1 cells). Nuclei were stained with
DAPI (blue). The white dotted line delimits contours of the aggregates (see Materials and methods). Photographs were taken at
6250 magnification, using a fluorescence microscope coupled to
a CCD camera. (C) Secretion and activity of FGF-1 by producing
cells (NSF14 cells) were demonstrated using soluble recombinant
human FGFR1 extracellular domain (see Materials and methods).
NSF14 cells were cultivated in presence of heparin (a) or heparin
and soluble FGFR1 (b) with a partial reversion of the phenotype
observed. Photographs were taken with an inverted microscope at
6100 magnification
We investigated MMP production and activity by
gelatin zymography. The gelatinolytic activity of progelatinases and gelatinases (MMP-2 and MMP-9) was
detected in the conditioned medium of each cell type
(Figure 3Ba). However, the intensity of the bands
corresponding to MMP-9 and MMP-2 activity was
higher in cells producing FGF-1 or FGF-2 than in
control NBT-II cells. ImageQuant1 analysis showed
that the active form of MMP2 was higher for FGF-1producing NBT-II cells than for other cell types
(Figure 3Bc).
We then investigated if MMP-2 activation could
occur through interaction with MT1 – MMP and
TIMP-2. PCR analyses showed that MT1 – MMP and
TIMP-2 were produced in all cell types (Figure 3Ca,b).
Western blotting for MT1 – MMP, using cellular
extracts of cells stimulated with concanavalin A (Con
A) for 24 h, confirmed the production of MT1 – MMP
and the presence of a 120 kDa band, which may
correspond to a dimer of MT1 – MMP (Figure 3Cc).
Western blot analysis of the conditioned media of the
cells stimulated for 24 h with Con A showed the
production and secretion of TIMP-2 and the presence
of a 85 kDa band for all cell types (Figure 3Cd). The
85 kDa band may correspond to the TIMP-2/MMP-2
complex. Thus, one of the MMP-2 activating processes
may be through MT1 – MMP, as both MT1 – MMP
and TIMP-2 were shown to be present.
Carcinoma cells autocrine for FGF-1 or FGF-2
overproduce uPAR and promote uPA activation
PCR analyses showed that uPA, uPAR and PAI-1
were produced in all cell types, with higher levels of
uPAR production in cells autocrine for FGF-1 or
FGF-2 (Figure 3Da,b). Plasminogen activator production and activity were assessed by zymography (Figure
3Bb). Pro-uPA and uPA were produced in all
conditioned media but the percentage of the active
Oncogene
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8132
form was higher in cells autocrine for FGF-1 or FGF-2
(Figure 3Bc).
Carcinoma cells autocrine for FGF-1 or FGF-2 have
increased tumorigenic capacities
NBT-II cells producing FGF-1 (NSF14 cells) have been
reported to induce the formation of large tumors (2 –
3 cm3 diameter) within 2 weeks of injection, whereas
NBT-II cells producing FGF-2 (bGF22) display similar
tumorigenic behavior to control NBT-II cells (Jouanneau et al., 1997). In this study, cells autocrine for
FGF-2 induced large tumors within 3 weeks, whereas
cells non-autocrine for FGF-1, induced tumors similar
to those induced by control NBT-II cells, 6 – 7 weeks
after injection (Figure 4). The production of these
growth factors in vivo was confirmed by Western
Figure 4 Tumorigenic properties of NBT-II cells and FGF-producing NBT-II cells. Each point corresponds to the mean tumor
size obtained with 5 – 15 animals. There were at least 15 animals
for (1) NBT-II, (2) bGF22 and (4) NSF14 cells; and five animals
for (3) cl37 and (5) cl1 cells. (For details see Materials and
methods)
Figure 3 Characterization of cell invasiveness and proteolytic
systems of NBT-II cells and FGF-producing NBT-II cells. (A)
In vitro invasive potential of NBT-II cells and FGF-producing
cells, assessed by the Boyden chamber assay using Transwell1.
The relative percentage of invasive cells was calculated as described in Materials and methods. For NSF14 and cl1 cells
*P50.05, compared to NBT-II cells. (B) Zymography for gelatinases (a) and PA (b) was performed as described in Materials and
methods. This figure shows the negative image, i.e. black bands
against a clear background. i/aMMP-9: inactive/active MMP-9,
Oncogene
iMMP-2: inactive MMP-2, aMMP-2: active MMP-2, iuPA: inactive uPA and auPA: active uPA. The relative percentages of active
MMP-2 and uPA versus total enzymes were evaluated with
ImageQuant software. Data representative of three independent
experiments (c). (C) PCR for MT1 – MMP (a) and TIMP-2 (b)
using cDNAs from NBT-II cells and FGF-producing NBT-II
cells. Western blot for MT1 – MMP performed with cell extracts
of NBT-II cells, FGF-producing NBT-II cells and FR3T3 cells
as a control. Cells were (+) or were not (7) stimulated with concanavalin A (Con A, 50 mg/ml) in serum-free medium for 24 h (c).
Western blot for TIMP-2 with conditioned medium of cells stimulated with concanavalin A (Con A, 50 mg/ml) in serum-free medium for 24 h (d). Recombinant human TIMP-2 (100 ng) was used
as the standard. (D) PCR for uPA, uPAR, PAI-1 and 18 S (a)
using cDNAs of NBT-II cells and FGF-producing NBT-II cells.
The relative percentages of uPAR versus 18 S internal control
were calculated with ImageQuant software. Data representative
of three independent experiments (b)
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8133
blotting with tumor extracts (Figure 1Ac,d). From
these results, we conclude that autocrine signaling
involving FGF-1/FGFR2b and FGF-2/FGFR1 may be
responsible for the increase in the tumorigenic potential
of NBT-II cells in nude mice.
Tumors obtained with NBT-II cells producing FGF-1 or
FGF-2 display extensive angiogenesis
Tumors derived from NBT-II cells producing FGF-1
or FGF-2 have already been shown to be more
vascularized than tumors obtained with control NBTII cells (Jouanneau et al., 1997). Von Willebrand factor
immunostaining of tumor sections showed that numerous vessels were present throughout the tumors
induced by NBT-II cells and their FGF-producing
derivatives (Figure 5A). However, mean vascular
density was significantly higher in all FGF-producing
tumors than in control NBT-II tumors (Figure 5B).
FGF-1 and FGF-2 may increase tumor angiogenesis
directly, by targeting endothelial cells, and/or indirectly, by promoting the production of other angiogenic
factors such as VEGF. PCR analysis showed that
VEGF was produced in vitro by all cell types (Figure
6A). We also used PCR to check for the production of
VEGF in vivo, using primers that recognize both mouse
and rat VEGF cDNAs. VEGF was produced by all
tumor types (Figure 6C). The PstI and HhaI restriction
patterns for tumor VEGF PCR products were
compared with those for the VEGF PCR products of
NBT-II and NIH3T3 cells (Figure 6B). This comparison showed that the VEGF produced by the various
tumors was essentially of rat origin (Figure 6C).
However, it cannot be excluded that some VEGF of
mouse origin is produced, since mouse VEGF in the
total PCR products cannot be revealed by this
approach if the total is less than 10% (data not
shown). Immunohistochemistry on frozen tumor
sections confirmed the presence of VEGF which is
mostly located in tumor islets but also in the stromal
environment and in the endothelial cells, as shown with
laminin labeling for basal lamina (Figure 5C). These
results indicate that in the NBT-II in vivo system,
VEGF is mostly produced by rat cancer cells, but may
also be produced by mouse stromal cells.
Active gelatinases are produced in the various tumors
The production and activity of MMPs are often
regulated by cross-talk between stromal and cancer
cells. In situ zymography, performed on tumor sections,
indicated that active gelatinases are present principally
in the carcinomatous areas (Figure 7a,c and d),
consistent with the results of in vitro experiments. This
gelatinolytic activity was completely abolished in the
presence of the inhibitor 1,10 phenanthroline (Figure
7b) and was also associated with stromal cells and tumor
vessels (Figure 7e). Thus, active MMPs are produced in
vivo by NBT-II cells and their FGF-producing derivatives, and gelatinolytic activity may be correlated with
tumor progression and tumor angiogenesis.
Figure 5 Vascularization of tumors obtained with NBT-II cells
and FGF-producing NBT-II cells. (A) Staining of frozen tumor
sections (a: NBT-II tumor, b: cl1 tumor, c: cl37 tumor) for the
von Willebrand factor (red). Nuclei were stained with DAPI
(blue). Photographs were taken at 6250 magnification, using a
fluorescence microscope coupled to a CCD camera. (B) Evaluation of tumor vascular density from von Willebrand factor staining (see Materials and methods). For bGF22 tumors *P50.001,
for cl1 and cl37 tumors **P50.01 and for NSF14 tumors
***P=0.1, compared to NBT-II tumors. (C) Staining of frozen
tumor sections (a: NBT-II tumor, b: cl1 tumor, c: cl37 tumor)
for VEGF (red) and laminin (green). Nuclei were stained with
DAPI (blue). Photographs were taken at 6250 magnification,
using a fluorescence microscope coupled to a CCD camera
Oncogene
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8134
Figure 7 Detection of gelatinolytic activity within the tumor. In
situ zymography was performed on frozen tumor sections (a,b:
NBT-II tumor, c: cl37 tumor, d: NSF14 tumor, e: cl1 tumor)
and the production of active gelatinases was detected. The degraded gelatin fluoresced green. Nuclei were stained with DAPI
(blue). A similar experiment was also carried out in the presence
of an MMP inhibitor (250 mM 1,10 phenantroline) (b). Photographs were taken at 6250 magnification, using a fluorescence
microscope coupled to a CCD camera. (e) Enlarged section of a
cl1 tumor showing endothelial cells of a tumor vessel (see arrows)
Figure 6 VEGF production in cells and derived tumors. (A)
PCR for VEGF was performed with cDNAs of NBT-II and
FGF-producing NBT-II cells. 18 S cDNA PCR was performed
as an internal control. (B) Restriction analyses distinguished between rat and mouse VEGF PCR products. HhaI cleaves the
NBT-II rat VEGF PCR product into two fragments (271 bp,
114 bp) and the NIH3T3 mouse VEGF PCR product into two
fragments (213 bp, 170 bp) whereas PstI cleaves only the NBTII rat VEGF PCR product into two fragments (337 bp, 48 bp).
The 48 bp band is not shown. (C) VEGF PCR products from
the various types of tumor were digested with PstI or HhaI
Discussion
FGF-1 and FGF-2 may be involved in cancer
progression and tumor metastasis as a result of their
mitogenic and migratory effects on cancer cells, but
also due to their angiogenic activity. The production of
these factors may be increased in tumors and may be
correlated with the degree of malignancy (Giri et al.,
1999; Soslow et al., 1999; Takahashi et al., 1992).
Although it lacks a signal peptide, FGF-2 is detected
in the conditioned medium of FGF-2-producing cells
and it has been suggested that the molecular system by
which it is exported is energy-dependent (Florkiewicz et
al., 1995). It is unclear whether FGF-1 is exported, as
Oncogene
FGF-1 was not found in the conditioned medium of
FGF-1-producing cells. However, in the presence of
adenovirus expressing soluble FGFRs, the tumor
angiogenesis through FGF-1 is blocked, clearly
demonstrating the secretion and activity of FGF-1
(Compagni et al., 2000). Our results provide evidence
for the secretion of these growth factors as the FGF-1
and FGF-2 produced by NBT-II cells are released and
biologically active. Nonetheless, the process of secretion seems to differ between FGF-1 and FGF-2, and in
our experimental model, FGF-1 may be directly
trapped by the ECM, transmembrane FGFRs or
heparan sulfate.
Experiments involving the transfection of various cell
types with the FGF-1 and FGF-2 cDNAs have shown
that both FGF-1 and FGF-2 have the potential to
behave as transforming and oncogenic factors (Jaye et
al., 1988; Nesbit et al., 1999). These experiments
suggest that the in vitro and in vivo behaviors of
FGF-transfected cells, depend not only on the cell type
and amount of FGF produced, but also on the
possibility to be exported (Forough et al., 1993; Gately
et al., 1995).
Parental NBT-II cells, and their derivatives that are
non-autocrine for FGF-1 or FGF-2, had an epithelial
phenotype in vitro. In the presence of FGF-1, NBT-II
cells are scattered and motile (Valles et al., 1990).
NBT-II cells autocrine for FGF-1 and FGF-2 acquire
a mesenchymal morphology, present a majority of
internalized desmosomes and produce vimentin. Migration and invasion tests performed in Matrigel-coated
Boyden chambers indicated that all the cells were
invasive but that NBT-II cells autocrine for FGF-1 or
FGF-2 were significantly more invasive than the other
cells. Loss of the epithelial phenotype was correlated
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8135
with a greater potential for migration and invasion and
may have resulted from the activation of FGFRs in
autocrine cells (autocrine signaling involving FGF-1/
FGFR2b or FGF-2/FGFR1).
Tumor invasion is associated with proteolytic
degradation of the basement membrane and the
ECM components. Various proteolytic systems are
involved in invasive tumors and increases in the
production of MMPs, uPA or uPAR are correlated
with a more aggressive phenotype in various tumors
(Borgfeldt et al., 2001; Maatta et al., 2000). In
addition, it has been reported that FGF-2 can
stimulate the production of these proteins in cancer
and endothelial cells (Miyake et al., 1997; Tsuboi et al.,
1990).
NBT-II and FGF-producing NBT-II cells expressed
and produced these proteases and their natural
inhibitors (TIMPs and PAIs). MMP-2 and MMP-9
were produced by all cell types, however cells autocrine
for FGF-1 had the highest active MMP-2 ratio. The
activation of MMP-2 has been studied by various
groups (Kinoshita et al., 1998; Lehti et al., 1998) and
recent data have provided evidence for the formation
of a ternary complex, MT1 – MMP/TIMP-2/MMP-2,
as one of the mechanisms of pro-MMP-2 processing
(Hernandez-Barrantes et al., 2000). Studies with our
model did not lead to the direct detection of this
ternary complex but did show the presence of MT1 –
MMP in cell extracts and the release of free TIMP-2
into the conditioned medium. Thus, the production of
MT1 – MMP and TIMP-2 by NBT-II cells may be a
determinant to promote pro-MMP-2 activation during
tumor progression. At low concentrations, TIMP-2
promotes the binding of pro-MMP-2 to MT1 – MMP
and at higher concentrations it may prevent pro-MMP2 activation. Tumor progression critically depends on
the balance between positive and negative regulators
(Lehti et al., 1998).
Pathological studies and basal lamina labeling for
laminin have shown that tumors obtained with cells
autocrine for FGF-1 or FGF-2 are often undifferentiated, whereas tumors obtained with cells nonautocrine for these FGFs are more differentiated, and
resemble parental NBT-II carcinomas. However, most
tumors are heterogeneous and may present both
features. In situ zymography, performed on NBT-II
and FGF-producing NBT-II derivative tumor sections,
showed that the fluorescent, digested gelatin was
located mostly in carcinomatous areas. We observed
no clear difference in gelatinolytic activity between
NBT-II tumors autocrine and non-autocrine for FGF1 and FGF-2. These results correlate with our in vitro
results and indicate that gelatinases are produced by all
the cell types and are processed to a biologically active
form both in vitro and in vivo.
Inactive and active forms of uPA were detected by
zymography in all cell types. NBT-II carcinoma cells
autocrine for FGFs overproduced uPAR and induced
higher levels of pro-uPA activation. Thus uPAR/uPA
are direct or indirect targets for FGF signaling in our
model, leading to an increase in invasive potential.
These results are consistent with those of other studies
reporting that the overproduction of uPAR in some
human cancers results in greater tumor growth and
metastasis (Borgfeldt et al., 2001; Zheng et al., 2000).
But also it seems likely that other proteases are
produced by NBT-II cells or stromal cells and may
serve as mediators between the cancer cells and the
microenvironment (Bergers and Coussens, 2000; Chang
and Werb, 2001).
Invasion is an important step in tumor progression,
which requires the proliferation of cancer cells. Nude
mice injected subcutaneously with NBT-II cells autocrine for FGF-1 and FGF-2 developed large highly
vascularized tumors within 2 weeks, whereas NBT-II
cells non-autocrine for FGFs, like the parental cells,
gave rise to tumors that developed several weeks later.
These results provide evidence that autocrine FGF-1
and FGF-2 act as tumorigenic factors allowing rapid
NBT-II carcinoma development. In previous studies we
have reported that NBT-II cells expressing exclusively
the nuclear localized 24 kDa – FGF-2 are highly
metastatic, but that the tumors obtained were
vascularized similarly to those obtained with parental
NBT-II cells (Okada-Ban et al., 1999).
Tumor progression and the incidence of metastasis
are correlated with vascular density (Weidner, 1999). In
the NBT-II carcinoma model, the in vivo production of
FGF-1 and FGF-2 increases tumor vascularization,
whether the production is autocrine or non-autocrine.
Thus, tumor proliferation is not necessarily correlated
with angiogenic index, and NBT-II tumors are able to
activate an angiogenic switch independent of FGF
signaling. The angiogenesis observed in NBT-II tumors
may be due mostly to the VEGF production of the rat
carcinoma cells whereas, in FGF-producing tumors,
cumulative paracrine signaling of FGFs and VEGF on
endothelial cells probably occurs, increasing tumor
angiogenesis. Recent studies have shown that FGF-2
increases the activation of endothelial cells by
upregulating VEGF production (Claffey et al., 2001;
Seghezzi et al., 1998). Although we did not detect
significant production of mouse VEGF cDNAs in the
tumors, the stimulation of endothelial cells by mouse
VEGF may occur in vivo.
Gelatinases play a crucial role not only in tumor
invasiveness, but also during the tumor angiogenic
switch (Bergers et al., 2000; Fang et al., 2000).
Consistent with this, gelatinolytic activities were
detected by in situ zymography along the length of
the tumor capillaries.
NBT-II cells autocrine for FGF-1 or FGF-2 present
similarly high invasive, tumorigenic and angiogenic
potentials. As FGF-1 and FGF-2 do not activate the
same FGFR, this model does not appear to differentiate between the FGF-1/FGFR2b and FGF-2/
FGFR1 signaling pathways. However, other studies
have reported that FGF-1 and FGF-2 are able to
induce different functions through different pathways
(Boilly et al., 2000).
Taken together, our studies demonstrate that in the
NBT-II experimental carcinoma, extensive tumor
Oncogene
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8136
vascularization is not correlated with an increase in
tumor proliferation.
Clinical data have shown that malignancy index is
often associated with tumor angiogenesis, and recently,
angiogenesis inhibition strategies have emerged as
approaches for the treatment of cancer. However,
tumor development from transformed cells does not
depend only on angiogenic factors. Our results indicate
that an anti-angiogenic approach to reduce or abolish
tumor progression may not be sufficient, as ‘basal’
vascularization may efficiently support tumor development. Improvements in our understanding of the
mechanisms regulating tumor progression may offer
new perspectives for designing therapeutic drug targets
for the treatment of carcinoma. It seems that the
progressive growth of an established tumor beyond
initial steps, which require a ‘basal’ vascularization, is
independent of the vascular density (Giavazzi et al.,
2001). This supports the concept that an efficient
treatment to inhibit the development of solid tumors
should be applied early in tumor progression and that
combined administration of anti-angiogenic agents
with other inhibitors may provide optimal results for
the clinical treatment of certain tumors.
Materials and methods
Reagents
Monoclonal anti-FGF-2 antibodies were purchased from
Transduction Laboratories. Monoclonal and polyclonal antiFGF-1 antibodies were a gift from D Chopin (Hôpital Henri
Mondor, France) and M Jaye (GlaxoSmithKline Pharmaceuticals, USA), respectively. Monoclonal antibodies against
desmosomal proteins, vimentin, and TIMP-2 were purchased
from Boehringer, Amersham Pharmacia Biotech, and
Oncogene Research, respectively. Polyclonal antibodies
against VEGF, and Von Willebrand factor were purchased
from Santa Cruz and DAKO, respectively. Monoclonal
antibodies against MT1 – MMP were provided by MC Rio
(LGME, France) Polyclonal anti-laminin antibodies, protease
inhibitor cocktail, Matrigel, PKH26 cell labeling kit,
concanavalin A (Con A), tetrazolium salt MTT (3-[4,5dimethyl-thiazol-2-y]-2,5-diphenyltetrazolium bromide), gelatin, plasminogen, casein were purchased from Sigma.
Phosphatase inhibitor cocktail was purchased from Calbiochem. Texas Red-conjugated sheep anti-mouse IgG or conjugated donkey anti-rabbit IgG, FITC-conjugated donkey
anti-rabbit IgG, horseradish peroxidase-conjugated sheep
anti-mouse IgG or -conjugated donkey anti-rabbit IgG
(Amersham Pharmacia Biotech), Alexas 488 or Alexas 594conjugated goat anti-mouse IgG (Molecular Probes) and
Texas Red-conjugated donkey anti-goat IgG (Jackson
Immunoresearch Laboratories) were used as second antibodies. Heparin and soluble recombinant human FGFR1
extracellular domain were purchased from Leo and Austral
Biological Laboratories, respectively.
their surface. NBT-II derived cells producing FGF-1 (NSF14
cells) or FGF-2 (bGF22 cells) as well as Flg8 cells expressing
the high affinity FGF-1 and FGF-2 receptor FGFR1 have
been described elsewhere (Jouanneau et al., 1991, 1997). Nonfunctional FGFR expressing cells, DFlg25 cells, generated by
stable transfection with a dominant-negative form of FGFR1
(DN-FGFR1), have been reported previously (Jouanneau et
al., 1999) (for cell nomenclature, see Table 1). FR3T3 cells
(Fisher rat 3T3 fibroblast) provided by F Thierry (Institut
Pasteur, France) were used as a positive control for the
production of the membrane-type metalloprotease MT1 –
MMP. All cell lines were grown in Dulbecco’s Modified
Eagle’s Medium (DMEM) supplemented with 10% fetal
bovine serum, 2 mM glutamine, 100 U/ml penicillin and
100 mg/ml streptomycin (complete medium).
Plasmids and transfection
The pFGF-1 and pFGF-2 expression vectors contain the
entire cDNA coding sequence of the human FGF-1 and the
bovine 18 kDa-FGF-2, respectively (Jouanneau et al., 1991,
1997). pREP4 contains a hygromycin resistance gene
(Invitrogen). To generate NBT-II cells non-autocrine for
FGF-1, DFlg25 cells were cotransfected with pREP4/pFGF-1
plasmids (1 : 5 or 1 : 10 molar ratio). NBT-II cells autocrine
for FGF-2 were obtained by cotransfection of Flg8 cells with
pREP4/pFGF-2 plasmids (1 : 5 or 1 : 10 molar ratio). The in
vitro FGF production in hygromycin B resistant clones was
analysed by Western blotting on cellular extracts of DFlg25FGF-1 clones (cl17, cl29 and cl37) and Flg8-FGF-2 clones
(cl1 and cl19) (see Table 1).
Indirect immunofluorescence staining
Two6104 cells were seeded on a glass coverslip and grown
to subconfluence. For FGF-1 and FGF-2 staining, cells were
fixed in 4% paraformaldehyde in PBS (phosphate bufferedsaline: 136 mM NaCl, 2.7 mM KCl, 8 mM Na2HPO4, 1.8 mM
KH2PO4, pH 7.4), permeabilized with 0.1% Triton-X100
and blocked in 5% fetal bovine serum. Cells were then
incubated with mouse monoclonal anti-FGF-2 antibody
(1 : 25) or anti-FGF-1 (1 : 100) before staining with Texas
Red-conjugated sheep anti-mouse IgG. For desmosomal
components and vimentin staining, cells were fixed in
methanol and acetone at 7208C, and sequentially incubated
with mouse monoclonal anti-desmoglein and anti-desmoplakin antibodies (1 : 25) and then with Alexas 594-conjugated
goat anti-mouse IgG. Labeled cells were then incubated with
mouse monoclonal anti-vimentin antibody (1 : 5) followed by
a second incubation with Alexas 488-conjugated goat antimouse IgG. For VEGF and laminin staining, frozen tumor
sections (7 mm) were fixed in acetone at 7208C and blocked
with 6% fetal bovine serum in TBS (Tris-buffered saline:
100 mM Tris-HCl, 150 mM NaCl, pH 7.4). Sections were
incubated with goat polyclonal anti-VEGF antibody (1 : 25)
and rabbit polyclonal anti-laminin antibody (1 : 200).
Sections were then incubated with Texas-Red-conjugated
donkey anti-goat IgG and FITC-conjugated donkey antirabbit IgG.
Western blot analysis
Cell lines and culture
The epithelial NBT-II cells were derived from a chemicallyinduced rat bladder carcinoma (Toyoshima et al., 1971).
These cells did not produce endogenous FGF-1 or FGF-2,
but possessed a high affinity FGF-1 receptor FGFR-2b on
Oncogene
FGF-1 and FGF-2 Serum-free conditioned media, generated
after 24 h cell culture, were collected, and subconfluent cells
were washed with PBS and scraped off into 2 M NaCl,
0.1 M phosphate buffer pH 7.4 supplemented with protease
inhibitor cocktail. Cell lysates and tumor fragments
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8137
(approximately 100 mg) were crushed in 2 M NaCl, 0.1 M
phosphate buffer pH 7.4 with an Ultraturax and centrifuged
at 12 0006g for 15 min. Samples adjusted to a final
concentration of 0.5 M NaCl were incubated overnight at
48C with heparin-Sepharose beads equilibrated in 0.5 M
NaCl, 0.1 M phosphate buffer pH 7.4. The mixture was
centrifuged and the pellet was resuspended in sample buffer
(62.5 mM Tris-HCl pH 6.8, 2% SDS, 10% glycerol and
0.1% bromophenol blue, 1% b-mercaptoethanol), boiled for
5 min and finally subjected to electrophoresis in a polyacrylamide gel. Proteins were electroblotted onto a PVDF
membrane. Membranes were incubated in 3% nonfat milk
in TBS for FGF-1 or PBS for FGF-2 and then incubated
with either rabbit polyclonal anti-FGF-1 antibody (1 : 500)
or mouse monoclonal anti-FGF-2 antibody (1 : 250). The
membrane was incubated with horseradish peroxidaseconjugated donkey anti-rabbit IgG or -conjugated sheep
anti-mouse IgG.
MT1 – MMP and TIMP-2 Subconfluent cells were incubated for 24 h with serum-free medium in the presence or
absence of 50 mg/ml Con A. Cells were scraped into RIPA
buffer (1% NP40, 0.5% sodium deoxycholate and 0.1%
SDS pH 7.4) supplemented with protease inhibitor cocktail
and lysed. Conditioned media were collected and concentrated by a factor of 40 on Vivaspin 20 columns (cut-off
10 000, Vivascience). The lysate was centrifuged at 40006g
for 15 min and the supernatant was collected. Fifty mg of
protein of lysate or 40 ml of concentrated conditioned
medium were treated as described above. The samples were
subjected to electrophoresis in a polyacrylamide gel and
were transferred. The membrane was blocked with 2% BSA
in PBS and then incubated with mouse monoclonal anti-rat
MT1 – MMP antibody (1 : 1000) or anti-rat TIMP-2 antibody (1 : 20). The membrane was then incubated with
horseradish peroxidase-conjugated sheep anti-mouse IgG.
For all blots, proteins were detected by ECL (Amersham
Pharmacia Biotech).
Evidence for the FGF-1 autocrine signaling
Fluorescent cell aggregate behavior Living NBT-II and Flg8
cells were labeled with PKH26, a red fluorescent cell linker
compound, according to the manufacturer’s procedure. This
fluorescent labeling does not affect cell viability, is stable and
Table 2
could be detected both in vitro and in vivo for up to 14 days
after labeling. Cell aggregates were obtained by growing in a
fluorescently labeled suspension for 48 h at 378C with
shaking (72 r.p.m.). Aggregates were collected and carefully
washed several times with complete medium. The fluorescence of the aggregates was assessed by epifluorescence
microscopy. They were deposited on cell monolayers (NBT-II
cells, NSF14 cells or cl1 cells) grown on glass coverslips and
were incubated further for 3 days at 378C.
Inhibition by soluble FGFR1 In a tissue culture testplate (24wells), 26105 cells per well were grown for 1 week in 500 ml
of complete medium supplemented with 50 mg/ml heparin
and in the presence or absence of 0.5 mg/ml soluble FGFR1.
Cell invasion assay
Invasiveness was determined in a two-compartment Boyden
chamber (8 mm pores) (Transwell, Costar). Complete
medium was added to the upper and lower chambers and
56104 cells were seeded on the top of the membrane coated
with Matrigel (30 mg/200 ml), and incubated for 4 days. The
non-invasive cells present on the upper side of the
membrane and the invasive cells present on the under side
of the membrane and in the lower chamber were counted by
means of a colorimetric test using the tetrazolium salt MTT
(5 mg/ml). The results, expressed as a percentage of invasive
cells were determined from three independent experiments.
Statistical significance was determined by a paired Student ttest.
RT – PCR amplifications
Total RNA was isolated from cell lines and tumors by
RNA PLUSTM extraction kit (Bioprobes Systems). One mg
of total RNA was reverse-transcribed to cDNA using the
RNA PCR kit (AMV-RT, Takara Biomedicals). The
resulting cDNAs were amplified using specific primers for
the genes of interest (see Table 2) together with primers for
the cDNA of 18S ribosomal RNA subunit. Since VEGF
primers recognized both mouse and rat VEGF, VEGF
PCR products generated from tumor cDNAs, were digested
with HhaI, which cleaves both the mouse and rat VEGF
PCR products, or with PstI which only cleaves the rat
VEGF PCR products.
Characteristics of PCR primers
Tm (8C)
No of cycles
Product length
ruPA
55
32
173 bp
ruPAR
65
32
253 bp
rPAI-1
55
35
276 bp
rMT1-MMP
60
35
175 bp
rTIMP-2
55
35
171 bp
rVEGF
55
35
385 bp
r18S
55
18
139 bp
Oligonucleotide primers
sense 5’-GCTATGTGCAAATTGGCCTA-3’
antisense 5’-CTGGTTCTCAACGACAGTGA-3’
sense 5’-GCCCTGGGCCAGGACCTCTG-3’
antisense 5’-GAGAGGTGCAGGATGCACAC-3’
sense 5’-CGGCACTGGTAAATCTTTCC-3’
antisense 5’-GGGTCATCCTTCATAGCAAT-3’
sense 5’-ATGACATCTTCTTGGTGGC-3’
antisense 5’-TGACCCTGACTTGCTTCCAT-3’
sense 5’-GTGACTTTATTGTGCCCTGG-3’
antisense 5’-TGATGCTCTTCTCTGTGACC-3’
sense 5’-TGGACCCTGGCTTTACTGCTG-3’
antisense 5’-GCTTTGTTCTATCTTTCTTTG-3’
sense 5’-GGGGAATCAGGGTTCGATT-3’
antisense 5’-GCCTCGAAAGAGTCCTGTA-3’
Each cycle: denaturation at 948C for 1 min, primer annealing at Tm8C and extension at 728C for 1 min. r: rat
Oncogene
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8138
Zymography
Gelatinases (MMP-2 and MMP-9) and PA activities were
analysed by specific zymography on serum-free conditioned
media, which were concentrated by a factor of 100 on
Centricon YM-30 (Polylabo) for gelatin zymography.
Samples containing 10 mg of protein were assessed for gelatin
zymography, and samples with 0.75 mg of protein for
plasminogen zymography. The samples were mixed with
sample buffer under non-reducing conditions and subjected to
electrophoresis in a polyacrylamide gel containing 0.1%
gelatin for gelatin zymography or in a polyacrylamide gel
containing 7 mg/ml plasminogen and 1 mg/ml casein for
plasminogen zymography. Gels were washed in 37 mM TrisHCl pH 7.6, 2.5% Triton-X100 for 1 h and then incubated in
activation buffer (50 mM Tris-HCl pH 7.6, 200 mM NaCl,
5 mM CaCl2 and 0.02% Brij 35) for 48 h at 378C for gelatin
zymography or (100 mM glycine, pH 8.3) for 24 h at 378C for
plasminogen zymography. The gels were stained with
Coomassie Brilliant Blue R-250, and proteolytic activity
was visualized as clear bands against a blue background.
Tumorigenicity in nude mice
Groups of 6-week-old female nude mice (nu/nu Swiss strain,
Iffa Credo, France) were subcutaneously injected in the flank
with control NBT-II cells or with cells that produced FGF-1
or FGF-2. In each case, a total of 3.56106 cells was injected.
Tumor volumes were monitored twice per week by caliper
measurement. The animals were killed when tumor volume
reached approximately 2 cm3. Tumors were removed and
their angiogenic status was carefully assessed.
Tumor angiogenesis
Fresh tumor fragments were embedded in OCT compound
(Tissu-Tek1) and frozen at 7808C until use. Seven mm
sections were fixed in acetone at 7208C and permeabilized in
0.25% NP 40 in PBS. Sections were satured with 5% nonfat
milk, incubated overnight at 48C with rabbit polyclonal antiVon Willebrand factor antibody (1 : 1000), and then
incubated with Texas-Red-conjugated donkey anti-rabbit
IgG. The mean vascular density was estimated by counting
the labeled vessels on four different areas of each section (by
two independent individuals). Statistical significance was
determined by a paired Student t-test.
In situ zymography
Location of MMP gelatinolytic activity within the tumors
was investigated by in situ zymography with the EnzCheck
Gelatinase Assay Kit (Molecular Probes), as previously
described (Oh et al., 1999). This non-fluorescent substrate is
subject to intramolecular quenching. Gelatinase activity
releases the FITC-gelatin peptides, which become fluorescent.
The areas in which gelatinolytic activity occurs can then be
identified by fluorescence. Unfixed frozen tumor sections
(7 mm) were stained with DAPI in reaction buffer (0.05 M
Tris-HCl, 0.15 M NaCl, 5 mM CaCl2 and 0.2 mM NaN3,
pH 7.6). They were washed once with reaction buffer,
overlaid with 50 ml of fluorescein conjugate DQ gelatin
(40 mg/ml in reaction buffer) and incubated overnight at 378C
in a humid chamber. The same procedure was also performed
in the presence of 250 mM 1,10 phenanthroline inhibitor, as a
negative control.
Acknowledgements
We would like to thank Dr MF Poupon for nude mouse
injections, Dr F Thierry for the FR3T3 cells, Dr MC Rio
for the anti-rat MT1 – MMP antibody, Dr M Jaye and Dr
D Chopin for their anti-FGF-1 antibodies, Dr R Mudge
and Dr M Morgan for useful discussions. This work was
supported by the Centre National de la Recherche
Scientifique and the Institut Curie, by grants from the
Association pour la Recherche sur le Cancer (ARC-9477
and ARC-5902 to J Jouanneau), the Ligue Nationale
Française contre le Cancer (National and Paris committees), the European Commission (B Janji) and the
Groupement des Entreprises Françaises contre le Cancer
(Gefluc).
References
Andreasen PA, Kjoller L, Christensen L and Duffy MJ.
(1997). Int. J. Cancer, 72, 1 – 22.
Bajou K, Noel A, Gerard RD, Masson V, Brunner N, HolstHansen C, Skobe M, Fusenig NE, Carmeliet P, Collen D
and Foidart JM. (1998). Nat. Med., 4, 923 – 928.
Basilico C and Moscatelli D. (1992). Adv. Cancer Res., 59,
115 – 165.
Bergers G, Brekken R, McMahon G, Vu TH, Itoh T, Tamaki
K, Tanzawa K, Thorpe P, Itohara S, Werb Z and Hanahan
D. (2000). Nat. Cell Biol., 2, 737 – 744.
Bergers G and Coussens LM. (2000). Curr. Opin. Genet.
Dev., 10, 120 – 127.
Boilly B, Vercoutter-Edouart AS, Hondermarck H,
Nurcombe V and Le Bourhis X. (2000). Cytokine Growth
Factor Rev., 11, 295 – 302.
Borgfeldt C, Hansson SR, Gustavsson B, Masback A and
Casslen B. (2001). Int. J. Cancer, 92, 497 – 502.
Chang C and Werb Z. (2001). Trends Cell Biol., 11, S37 – S43.
Claffey KP, Abrams K, Shih SC, Brown LF, Mullen A and
Keough M. (2001). Lab. Invest., 81, 61 – 75.
Compagni A, Wilgenbus P, Impagnatiello MA, Cotton M
and Christofori G. (2000). Cancer Res., 60, 7163 – 7169.
Oncogene
Dano K, Romer J, Nielsen BS, Bjorn S, Pyke C, Rygaard J
and Lund LR. (1999). Apmis, 107, 120 – 127.
Fang J, Shing Y, Wiederschain D, Yan L, Butterfield C,
Jackson G, Harper J, Tamvakopoulos G and Moses MA.
(2000). Proc. Natl. Acad. Sci. USA, 97, 3884 – 3889.
Florkiewicz RZ, Majack RA, Buechler RD and Florkiewicz
E. (1995). J. Cell Physiol., 162, 388 – 399.
Forough R, Xi Z, MacPhee M, Friedman S, Engleka KA,
Sayers T, Wiltrout RH and Maciag T. (1993). J. Biol.
Chem., 268, 2960 – 2968.
Gately S, Tsanaclis AM, Takano S, Klagsbrun M and Brem
S. (1995). Neurosurgery, 36, 780 – 788.
Giavazzi R, Giuliani R, Coltrini D, Bani MR, Ferri C,
Sennino B, Tosatti MP, Stoppacciaro A and Presta M.
(2001). Cancer Res., 61, 309 – 317.
Giri D, Ropiquet F and Ittmann M. (1999). Clin. Cancer
Res., 5, 1063 – 1071.
Hernandez-Barrantes S, Toth M, Bernardo MM, Yurkova
M, Gervasi DC, Raz Y, Sang QA and Fridman R. (2000).
J. Biol. Chem., 275, 12080 – 12089.
Jaye M, Lyall RM, Mudd R, Schlessinger J and Sarver N.
(1988). EMBO J., 7, 963 – 969.
FGF-1 and FGF-2 in tumor progression and vascularization
C Billottet et al
8139
Jouanneau J, Gavrilovic J, Caruelle D, Jaye M, Moens G,
Caruelle JP and Thiery JP. (1991). Proc. Natl. Acad. Sci.
USA, 88, 2893 – 2897.
Jouanneau J, Plouet J, Moens G and Thiery JP. (1997).
Oncogene, 14, 671 – 676.
Jouanneau J, Moens G and Thiery JP. (1999). Oncogene, 18,
327 – 333.
Kinoshita T, Sato H, Okada A, Ohuchi E, Imai K, Okada Y
and Seiki M. (1998). J. Biol. Chem., 273, 16098 – 16103.
Lehti K, Lohi J, Valtanen H and Keski-Oja J. (1998).
Biochem. J., 334, 345 – 353.
Maatta M, Soini Y, Liakka A and Autio-Harmainen H.
(2000). Clin. Cancer Res., 6, 2726 – 2734.
Mazzieri R, Masiero L, Zanetta L, Monea S, Onisto M,
Garbisa S and Mignatti P. (1997). EMBO J., 16, 2319 –
2332.
McCawley LJ and Matrisian LM. (2000). Mol. Med. Today,
6, 149 – 156.
Miyake H, Yoshimura K, Hara I, Eto H, Arakawa S and
Kamidono S. (1997). J. Urol., 157, 2351 – 2355.
Nelson AR, Fingleton B, Rothenberg ML and Matrisian
LM. (2000). J. Clin. Oncol., 18, 1135 – 1149.
Nesbit M, Nesbit HKE, Bennett J, Andl T, Hsu MY, Dejesus
E, McBrian M, Gupta AR, Eck SL and Herlyn M. (1999).
Oncogene, 18, 6469 – 6476.
Oh LY, Larsen PH, Krekoski CA, Edwards DR, Donovan F,
Werb Z and Yong VW. (1999). J. Neurosci., 19, 8464 –
8475.
Okada-Ban M, Moens G, Thiery JP and Jouanneau J. (1999).
Oncogene, 18, 6719 – 6724.
Pili R, Chang J, Muhlhauser J, Crystal RG, Capogrossi MC
and Passaniti A. (1997). Int. J. Cancer, 73, 258 – 263.
Powers CJ, McLeskey SW and Wellstein A. (2000). Endocr.
Relat. Cancer, 7, 165 – 197.
Savagner P, Valles AM, Jouanneau J, Yamada KM and
Thiery JP. (1994). Mol. Biol. Cell, 5, 851 – 862.
Schmidt NO, Westphal M, Hagel C, Ergun S, Stavrou D,
Rosen EM and Lamszus K. (1999). Int. J. Cancer, 84, 10 –
18.
Seghezzi G, Patel S, Ren CJ, Gualandris A, Pintucci G,
Robbins ES, Shapiro RL, Galloway AC, Rifkin DB and
Mignatti P. (1998). J. Cell Biol., 141, 1659 – 1673.
Soslow RA, Nabeya Y, Ying L, Blundell M and Altorki NK.
(1999). Histopathology, 35, 31 – 37.
Takahashi JA, Fukumoto M, Igarashi K, Oda Y, Kikuchi H
and Hatanaka M. (1992). J. Neurosurg., 76, 792 – 798.
Toyoshima K, Ito N, Hiasa Y, Kamamoto Y and Makiura S.
(1971). J. Natl. Cancer Inst., 47, 979 – 985.
Tsuboi R, Sato Y and Rifkin DB. (1990). J. Cell Biol., 110,
511 – 517.
Valles AM, Boyer B, Badet J, Tucker GC, Barritault D and
Thiery JP. (1990). Proc. Natl. Acad. Sci. USA, 87, 1124 –
1128.
Weidner N. (1999). J. Pathol., 189, 297 – 299.
Zheng Q, Tang ZY, Xue Q, Shi DR, Song HY and Tang HB.
(2000). J. Cancer Res. Clin. Oncol., 126, 641 – 646.
Oncogene