A distinct basic fibroblast growth factor (FGF-2)/FGF receptor interaction distinguishes urokinase-type plasminogen activator induction from mitogenicity in endothelial cells

Molecular Biology of the Cell Vol. 7, 369-381, March 1996

A Distinct Basic Fibroblast Growth Factor (FGF-2)/FGF Receptor Interaction Distinguishes Urokinase-type Plasminogen Activator Induction from Mitogenicity in Endothelial Cells Marco Rusnati,*± Patrizia Dell'Era,*t Chiara Urbinati,* Elena Tanghetti,* Maria Luisa Massardi,* Yoshikuni Nagamine,* Eugenio Monti,§ and Marco Presta*I1 *Unit of General Pathology and Immunology and §Unit of Biochemistry, Department of Biomedical Sciences and Biotechnology, School of Medicine, University of Brescia, 25123 Brescia, Italy; and Wriedrich Miescher Institute, Basel, Switzerland Submitted October 12, 1995; Accepted December 27, 1995 Monitoring Editor: Carl-Henrik Heldin

Basic fibroblast growth factor (FGF-2) induces cell proliferation and urokinase-type plasminogen activator (uPA) production in fetal bovine aortic endothelial GM 7373 cells. In the present paper we investigated the role of the interaction of FGF-2 with tyrosinekinase (TK) FGF receptors (FGFRs) in mediating uPA up-regulation in these cells. The results show that FGF-2 antagonists suramin, protamine, heparin, the synthetic peptide FGF-2(112-155), and a soluble form of FGFR-1 do not inhibit FGF-2-mediated uPA up-regulation at concentrations that affect growth factor binding to cell surface receptors and mitogenic activity. In contrast, tyrosine phosphorylation inhibitors and overexpression of a dominant negative TK- mutant of FGFR-1 abolish the uPA-inducing activity of FGF-2, indicating that FGFR and its TK activity are essential in mediating uPA induction. Accordingly, FGF-2 induces uPA up-regulation in Chinese hamster ovary cells transfected with wild-type FGFR-1, -2, -3, or -4 but not with TK- FGFR-1 mutant. Small unilamellar phosphatidyl choline:cholesterol vesicles loaded with FGF-2 increased uPA production in GM 7373 cells in the absence of a mitogenic response. Liposome-encapsulated FGF-2 showed a limited but significant capacity, relative to free FGF-2, to interact with FGFR both at 4°C and 37°C and to be internalized within the cell. uPA up-regulation by liposome-encapsulated FGF-2 was quenched by neutralizing anti-FGF-2 antibodies, indicating that the activity of liposome-delivered FGF-2 is mediated by an extracellular action of the growth factor. Taken together, the data indicate that a distinct interaction of FGF-2 with FGFR, quantitatively and/or qualitatively different from the one that leads to mitogenicity, is responsible for the uPA-inducing activity of the growth factor. INTRODUCTION Basic fibroblast growth factor (FGF-2),1 a Mr 18,000

and Moscatelli, 1992). FGF-2 interacts with high affin-

ity tyrosine kinase (TK) receptors (FGFRs) of the cell

member of the heparin-binding FGF family (Basilico and Moscatelli, 1992), is an angiogenic and neurotrophic molecule, mitogenic for a variety of cell types

surface. The first FGFR to be characterized (FGFR-1/ flg; Lee et al., 1989) was a single membrane-spanning

derived from mesoderm and neuroectoderm (Basilico

Abbreviations used: EGF, epidermal growth factor; FCS, fetal calf serum; FGF-2, basic fibroblast growth factor; FGFR, FGF receptor; FGF-2+SUV, sonicated FGF-2 added to vehicle-loaded liposomes; HSPG, heparan sulfate proteoglycan; PA, plasminogen activator; SUV, small unilamellar vesicles; SUV-FGF-2, FGF2-loaded liposomes; TK, tyrosine kinase; uPA, urokinase-type

The first two authors contributed equally to this work. ¶ Corresponding author: General Pathology, Department of Biomedical Sciences and Biotechnology, via Valsabbina 19, 25123 Brescia, Italy. i 1996 by The American Society for Cell Biology

PA.

369

M. Rusnati et al.

molecule with three extracellular immunoglobulin (Ig)-like domains, an acidic box located between the first and the second Ig-like loop, a transmembrane domain, and an intracellular catalytic TK domain split by a 14-amino acid insertion. Since then, three other genes encoding TK-FGFR have been discovered: FGFR-2/bek (Dionne et al., 1990), FGFR-3 (Keegan et al., 1991), and FGFR-4 (Partanen et al., 1991). Several RNA alternatively spliced variants that structurally differ in the number of Ig-like loops and/or in the absence of the intracellular domain (soluble forms) were also described for FGFR-1 and FGFR-2 (Johnson and Williams, 1993). The growth factor binding site of FGFR appears to be located in the second half of the third Ig-like loop. Three variants of this region, encoded by different exons, have been described: IlIa, IlIb, and IIIc. IIIa sequence seems to be unique for the FGFR-1 soluble receptor form (Johnson et al., 1991), while IlIb and IlIc are found in FGFR-1, FGFR-2, and FGFR-3 membrane spanning molecules (Johnson and Williams, 1993; Chellaiah et al., 1994). At the present time no IlIb/ FGFR-4 variant has been described. Binding studies from several laboratories indicate that IIIb forms are specific for FGF-1 and FGF-7, while IlIc variants show a broader spectrum of ligands (Johnson and Williams, 1993). FGF-2 induces an angiogenic phenotype in cultured endothelial cells that includes, among other responses, cell proliferation, chemotaxis, protease production, integrin expression, and gap-junctional intercellular communication (Moscatelli et al., 1986; Presta et al., 1986; Pepper and Meda, 1992; Klein et al., 1993). The capacity to exert a complex array of biological responses on the same cell type is shared by various TK receptor-binding growth factors such as platelet-derived growth factor (Claesson-Welsh, 1994), epidermal growth factor (EGF) (Chen et al., 1994), and hepatocyte growth factor (Gherardi and Stoker, 1991). This is thought to reflect the capacity of different docking transducer proteins to associate with the activated TK receptor, leading to the switch of multiple intracellular signals (Pawson and Schlessinger, 1993). In agreement with this hypothesis, we demonstrated that the mitogenic activity and urokinase-type plasminogen activator (uPA)-inducing capacity of FGF-2 are mediated by different signal transduction pathways in cultured endothelial GM 7373 cells (Presta et al., 1989). However, we observed also that different FGF-2 mutants devoid of the capacity to up-regulate uPA production in endothelial cells still retain receptor-binding activity and full mitogenic capacity (Isacchi et al., 1991; Presta et al., 1992, 1993), raising the question of the role exerted by TK-FGFR in transducing the uPA-inducing signal. This was examined also in light of the complexity of the mechanism of interaction of FGF-2 with the endothelial cell. Indeed, besides FGFR interaction, FGF-2 370

binds with low affinity and high capacity to heparansulfate proteoglycans (HSPGs) of the cell surface (Basilico and Moscatelli, 1992). HSPGs modulate FGF-2 interaction with FGFR (Yayon et al., 1991) and play a role in mediating FGF-2 internalization (Roghani and Moscatelli, 1992; Rusnati et al., 1993). Relevant to this point, FGF-2 has been proposed to exert its activity also by an intracellular action into the nucleus (Bouche et al., 1987; Dell'Era et al., 1991; Nakanishi et al., 1992; Sherman et al., 1993) and the absolute requirement for cell surface receptor/ligand interaction in mediating the biological activity of the closely related FGF-1 has been questioned (Wiedlocha et al., 1994). These observations prompted us to investigate in more detail the role of FGFR in mediating uPA upregulation induced by FGF-2 in endothelial cells. To this end we evaluated the following: 1) the effect of different FGF-2 antagonists on the receptor-binding capacity and uPA-inducing activity of FGF-2 on endothelial GM 7373 cells; 2) the capacity of FGF-2 to induce uPA up-regulation either in GM 7373 cells transfected with a dominant negative mutant of FGFR or in Chinese hamster ovary (CHO) cells transfected with wild-type or mutagenized FGFR cDNAs; 3) the biological response of GM 7373 cells to liposome-encapsulated FGF-2. The results lead to the conclusion that TK-FGFR/FGF-2 interaction is essential in mediating uPA up-regulation but differs both quantitatively and/or qualitatively from that responsible for the mitogenic activity of the growth factor. MATERIALS AND METHODS Materials Human recombinant FGF-2 was expressed and purified from transformed Escherichia coli cells by heparin-Sepharose chromatography as described (Isacchi et al., 1991). Synthetic peptides representing amino acid residues FGF-2(112-155) and FGF-2(130-155), where amino acid numbering 1-155 is utilized for FGF-2, were a gift of A. Baird (The Whittier Institute, La Jolla, CA). The soluble extracellular form of FGFR-1 was obtained from P. Caccia, Pharmacia-Farmitalia Carlo Erba (Milan, Italy). Egg L-lecithin (phosphatidyl choline), cholesterol from porcine liver, protamine, tyrphostin 23 (AG 18), tyrphostin 47 (RG-50864), and tyrphostin 63 were obtained from Sigma (St. Louis, MO). Suramin was from Bayer AG (Leverkusen, Germany). Polyclonal anti-human FGF-2 antibodies were a gift of D.B. Rifkin (New York University Medical Center, New York, NY) and were purified by FGF-2-Affi-Gel affinity chromatography.

Cell Cultures Fetal bovine aortic endothelial GM 7373 cells were obtained from the NIGMS Human Genetic Mutant Cell Repository (Institute for Medical Research, Camden, NJ). They correspond to the BFA-lc 1BPT multilayered transformed clone described by Grinspan et al. (1983). GM 7373 cells were grown in Eagle's minimal essential medium (MEM) containing 10% fetal calf serum (FCS), vitamins, and essential and nonessential amino acids. CHO cells, obtained from D. Di Lorenzo (Spedali Civili, Brescia, Italy), were grown in Ham's F 12 medium supplemented with 10% FCS.

Molecular Biology of the Cell

FGFR and uPA Up-Regulation

Cell Transfections Plasmids 91023b-flg (murine FGFR-1 cDNA), 91023b-bek (murine FGFR-2 cDNA), pCEP4-flg 1.2 (truncated murine FGFR-1 cDNA),

and pCB7 (a plasmid that carries the neor gene) were provided by A. Mansukhani and C. Basilico (New York University Medical Center). MomFR3SV (murine FGFR-3 cDNA) and LTR 2HX (human FGFR-4 cDNA) were provided by D. Ornitz (Washington University, St. Louis, MO) and K. Alitalo (University of Helsinki, Helsinki, Finland), respectively. All FGFR cDNAs codify for the IIlc variant of the corresponding receptor. Transfections were carried out with the calcium phosphate precipitation protocol (Sambrook et al., 1989). CHO cells were transfected with 91023b-flg, 91023b-bek, MomFR3SV, or LTR2HX in association (10:1, wt:wt) with pCB7 (Kern and Basilico, 1985) and selected with 250 ,tg/ml of G418 (Sigma). Both CHO and GM 7373 cells were transfected with pCEP4-flg 1.2 and selected with 500 or 200 ,tg/ml of hygromycin B (Boehringer Mannheim GmbH, Mannheim, Germany), respectively. For each transfection, resistant clones were tested for 1251-FGF-2 binding capacity (Moscatelli, 1988) and by 125I-FGF-2 cross-linking assay (Presta et al., 1989). Clones bearing the highest receptor number, ranging from 10,000 to 100,000 receptors per cell, were chosen for further experiments.

Cell Proliferation and Plasminogen Activator Assays Cell proliferation assay was performed on GM 7373 cells as described (Presta et al., 1989). Briefly, 70,000 cells/cm2 were seeded in 24-well dishes. After overnight incubation, cells were incubated for 24 h in fresh medium containing 0.4% FCS and 10 ng/ml of FGF-2 in the presence of increasing concentrations of different FGF-2 antagonists. At the end of the incubation, cells were trypsinized and counted in a Burker chamber. Control cultures incubated with no addition or with 10 ng/ml of FGF-2 undergo 0.1-0.2 and 0.7-0.8 cell population doublings, respectively. Cells grown in 10% FCS undergo 1.0 cell population doublings (Presta et al., 1991). To evaluate the uPA-inducing activity of FGF-2, GM 7373 and CHO cell cultures were seeded at 70,000 cells/cm2 and treated with the growth factor in the presence of different antagonists as described for the mitogenicity assay. At the end of the incubation, cell-associated uPA activity was measured using the plasmin chromogenic substrate H-Dnorleucyl-hexahydrotyrosil-lysine-p-nitroanilide-acetate (American Diagnostica, Greenwich, CT) as described (Presta et al., 1989). Steady-state levels of uPA mRNA were evaluated by Northern blot analysis of total RNA (20 ,tg/sample) according to standard procedures (Sambrook et al., 1989) using a human uPA probe (provided by P. Mignatti, University of Pavia, Italy). Uniform loading of total RNA was judged by hybridization of the filters with GAPDH probe.

Preparation and Characterization of FGF-2-loaded Liposomes Small unilamellar phosphatidylcholine:cholesterol vesicles (SUV) containing FGF-2 (SUV-FGF-2) were prepared by ultrasonic irradiation according to standard procedures (New, 1990a). Phosphatidylcholine (240 nmol) and cholesterol (20 nmol) were dried down from a chloroform/methanol solution by a stream of nitrogen. Human recombinant FGF-2 (1.7 or 17 pmol) was added in 500 ,lI of phosphate-buffered saline (PBS). The mixture was vortexed and probe-sonicated in ice with five bursts of 30 s at 40 W, each preceded by a 1-min pause, or until no further optical clearance was observed. The mixture was then loaded onto a heparin-Sepharose column (5 mm x 20 mm) equilibrated with PBS, to remove unencapsulated FGF-2. FGF-2 (50-60%) was routinely recovered associated with phospholipid vesicles in the flow through of the column as determined by tracer 125I-FGF-2. Vol. 7, March 1996

Similar results were obtained when SUV were loaded with different amounts of FGF-2 ranging from 0.5 to 50 pmol. After heparin-Sepharose chromatography, no radioactivity could be immunoprecipitated from liposome-associated 125I-FGF-2 by polyclonal anti-FGF-2 antibodies and protein A-Sepharose beads, indicating that no FGF-2 is present on the liposome surface. This is in keeping with the observation that the basic protein TGF type ,B-1 does not associate with the surface of neutral liposomes (Strassmann et al., 1991). To assess whether encapsulated FGF-2 could be released during a long-term incubation, liposomes were loaded with 1251I-FGF-2, separated from free 1251I-FGF-2 by heparin-Sepharose chromatography, and incubated at 37°C or at 4°C in fresh culture medium or in culture medium pre-conditioned for 24 h by confluent monolayers of endothelial GM 7373 cells. At different time points, ranging from 0 to 48 h, aliquots of the liposome suspension were collected, loaded again onto a heparin-Sepharose column, and the amount of 1251_ FGF-2 bound to the resin beads, representing released FGF-2, was measured. Under all the experimental conditions used, the amount of released 125I-FGF-2 never exceeded 2-4% of 125I-FGF-2 associated with phospholipid vesicles. SUV-FGF-2 was tested for biological activity as described for free FGF-2. Sonicated FGF-2 added to vehicle-loaded preformed vesicles (FGF-2+SUV) to a final concentration equal to that present in the SUV-FGF-2 preparations was used as a positive control in all the assays. For cell binding and internalization experiments, 125I-FGF-2 was encapsulated within phospholipid vesicles as described for unlabeled FGF-2.

125I-FGF-2 Binding and Internalization Assays Human recombinant FGF-2 was labeled with Na125I (37 GBq/ml; Amersham International, Amersham, UK) using lodogen (Pierce Chemical, Rockford, IL) as described (Isacchi et al., 1991). Twentyfour hours after plating in 24-well dishes at the density of 70,000 cells/cm2, cells were washed three times with ice-cold PBS and incubated for 2 h at 4°C in binding medium (serum-free medium containing 0.15% gelatin, 20 mM N-2-hydroxyethylpiperazine-N'-2ethanesulfonic acid, pH 7.5) with 10 ng/ml of liposome-associated or free 1251I-FGF-2. The amount of 125I-FGF-2 bound to high and to low affinity binding sites was evaluated as described (Moscatelli, 1988). Nonspecific binding was measured in the presence of a 100-fold molar excess of unlabeled FGF-2 and subtracted to all values. For internalization experiments, GM 7373 cells were incubated for 2 h at 4°C in binding medium in the presence of liposomeassociated or free 1251-FGF-2 (30 ng/ml) and shifted to 37°C for different periods of time without changing the medium. At the end of the incubation, cell surface-bound 1211-FGF-2 was removed by washing cell cultures with 2.0 M NaCl in 20 mM sodiumacetate, pH 4.0. Internalized 1251-FGF-2 was measured as described (Moscatelli, 1988). In some experiments, GM 7373 cells were incubated as described above in the absence or presence of 50 ,JM chloroquine to prevent lysosomal degradation. After 16 h of incubation at 37°C, cell surface-bound 125I-FGF-2 was removed and the cells were scraped with a rubber policeman and resuspended in 70 ,ul of reducing sample buffer. The cell lysates were sonicated at 50 W for 30 s and boiled, and aliquots containing equivalent amounts of radioactivity (1,000 cpm) were analyzed by SDS/20% PAGE. Gels were dried and exposed to Kodak X-Omat AR films (Eastman Kodak, Rochester, NY) for 3 wk at - 700C.

Down-Regulation of FGF Receptors GM 7373 cells were plated in 24-well dishes at the density of 70,000 cells/cm2. After 24 h, cells were incubated at 370C for different times with culture medium containing 0.4% FCS in the absence or presence of SUV-FGF-2 or FGF-2+SUV (15 ng/ml 371

M. Rusnati et al.

parable to ID50 values for the inhibition of FGF-2 binding to its high affinity receptors (3 ,tg/ml, 10 Ag/ml, and 30 ,ug/ml, respectively). In contrast, these molecules had approximately a 100-fold lower effect on the uPA-inducing activity of FGF-2 (Figure 1). Similar results were obtained when more specific FGF-2 and FGFR antagonists were used. The synthetic peptide FGF-2(112-155) interacts with FGF receptors and acts as a FGF-2 antagonist in 3T3 mouse fibroblasts (Baird et al., 1988). Accordingly, FGF-2(112-155) inhibits 125I-FGF-2/FGFR binding and FGF-2 mitogenic activity also in GM 7373 cells (Figure 2). However, the peptide had no effect on the uPA-inducing activity of the growth factor (Figure 2B). No antagonist activity was exerted by the control peptide FGF-2(130-155) (Baird et al., 1988) in receptor binding, cell proliferation, and uPA-production assays (Figure 2). Finally, the recombinant extracellular domain of FGFR-1 (Bergonzoni et al., 1992) inhibited FGF-2 mitogenic activity and 125I1 FGF-2 binding to FGFR in GM 7373 cells with an ID50 of 3 p,g/ml (Figure 3). The effect was specific for FGF-2 because the soluble receptor did not affect the mitogenic response induced by 30 ng/ml of EGF or by 10% FCS (our unpublished observations). Nevertheless, the soluble receptor did not affect the uPA-inducing activity of FGF-2 (Figure 3). In conclusion, different antagonists affect the binding of FGF-2 to its cell surface receptors and inhibit its mitogenic activity at doses that are ineffective on the uPA-inducing capacity of the growth factor.

FGF-2). At the end of the incubation, the cells were washed twice with ice-cold 2.0 M NaCl in 20 mM sodium acetate, pH 4.0, to remove bound FGF-2, and three times with ice-cold PBS, and then incubated at 4°C in binding medium containing 10 ng/ml of 125I-FGF-2. After 2 h, the amount of 1251I-FGF-2 bound to high affinity binding sites was measured as described (Moscatelli,

1988).

RESULTS FGF-2 Antagonists and uPA Up-Regulation To elucidate the role of the interaction of FGF-2 with endothelial cell surface in mediating uPA up-regulation, the antagonists suramin, heparin, and protamine were evaluated for their capacity to affect FGF-2 activity in GM 7373 cells. These molecules are known to antagonize FGF-2 activity through different mechanisms of inhibition of growth factor/cell surface interaction. Suramin and soluble heparin reversibly bind to the growth factor and inhibit the activity of FGF-2 either by masking the protein (suramin) or by preventing its interaction with cell surface binding sites by a law of mass action (heparin). In contrast, protamine does not bind FGF-2 but competes with the growth factor for cell surface interaction. The three molecules inhibited 125I-FGF-2 binding to low and high affinity sites in GM 7373 cells although with different potencies (Figure 1). The inhibition of FGF-2 binding to high affinity FGFRs was paralleled by a decrease of its mitogenic activity. ID50 values for the inhibition of the mitogenic activity of FGF-2 (2 ,ug/ml for suramin, 3 ,ug/ml for protamine, and 30 ,ug/ml for heparin) were com-

100 Aloft.l

o

1.17. 0.

'1' 80

N0

t -

o._

60

cc

ci

cl e4I

e' 40

Pk

L 20

C 0,01

0,1

1

10

100

suramin ([tg/ml)

0,1

I

10

100

protamine (jg/mi)

1

10

100

1000

heparin (jLg/mi)

Figure 1. Effect of different antagonists on FGF-2 activity. GM 7373 cells were seeded at 70,000 cells/cm2. Twenty-four hours after seeding, half of the cell cultures were incubated in culture medium containing 0.4% FCS and 10 ng/ml FGF-2 in the presence of increasing concentrations of suramin (A), protamine (B), or heparin (C). After 24 h of incubation, the cells were trypsinized and counted (-) or cell-associated uPA activity was measured (0). Parallel cell cultures were incubated with 125I-FGF-2 (10 ng/ml) in the absence or in the presence of the different FGF-2 antagonists. After 2 h of incubation at 4°C, the radioactivity associated with HSPGs (A) and FGFRs (A) was measured. Data are expressed as percent of FGF-2 activity measured in the absence of antagonist. Similar results were obtained in three to four independent experiments. 372

Molecular Biology of the Cell

FGFR and uPA Up-Regulation

C 100

Figure 2. Antagonist activity of FGF-2 fragments. GM 7373 cells were seeded at 70,000 cells/cm2. Then, antagonist activity of synthetic fragments FGF-2(112-155) (closed symbols) and FGF-2(130-155) (open symbols) was evaluated either on the binding of 10 ng/ml I251-FGF-2 to FGFRs (circles in A) or on the mitogenic activity (circles in B) and uPA inducing activity (squares in B) exerted by 10 ng/ml FGF-2, as described in the legend to Figure 1. Data are expressed as percent of the activity exerted by FGF-2 in the absence of FGF-2 fragments. Similar results were obtained in two to three independent experiments.

X

80

= > v

60

- 20 _I 2

m

100 110-IN

0 ""W

80

Iti.

60

co eq1

40-

P64

20

0,01

0,1

I

10

100

soluble FGFR-1 (jg/ml) Figure 3. Effect of soluble FGF receptor on FGF-2 activity. GM 7373 cells were seeded at 70,000 cells/cm2. Twenty-four hours after seeding, half of the cell cultures were incubated in medium containing 0.4% FCS with 10 ng/ml FGF-2 in the absence or presence of increasing concentrations of recombinant soluble FGFR-1 /lfg. After 24 h of incubation, cells were trypsinized and counted (0) or cellassociated uPA activity was measured (0). Remaining cell cultures were incubated with 125I-FGF-2 (10 ng/ml) in the absence or in the presence of the soluble FGF receptor. After 2 h of incubation at 4°C, radioactivity associated with FGFRs was measured (A). Data are expressed as percent of the activity exerted by FGF-2 in the absence of soluble receptor. Similar results were obtained in two independent experiments. Vol. 7, March 1996

*1

. .,r ........, .... ...

0

120-

Iu

t

I

Q 40

FGF-2/FGFR Interaction Is Essential for uPA Up-Regulation Previous experiments had shown that both cell proliferation (Presta et al., 1991) and uPA production (Gualandris and Presta, 1995) represent late responses of GM 7373 cells to stimulation by bFGF, ruling out the possibility that the dissociation between the two biological activities exerted by FGF-2 antagonists may depend on differences in time requirement for FGF-2

aXw

oow

0,1

I

10

100

Peptide (pg/ml) stimulation. Therefore, the results described above different conclusions: either FGFRs in mediating uPA up-regulation by FGF-2 in GM 7373 cells or the mechanism of FGF-2/ FGFR interaction responsible for uPA up-regulation is remarkably different from that responsible for mitoge-

may lead to two are not involved

nicity.

To discriminate between these alternative hypothereceptor negative CHO cells were stably transfected with FGFR-1, the only form of FGFR expressed by GM 7373 cells (Coltrini, unpublished observation), and tested for the capacity to produce uPA after stimulation by FGF-2. As shown in Figure 4A, FGF-2 increases cell-associated uPA activity in the CHOfig7G transfectants with an ED50 equal to 3-10 ng/ml, as is the case for GM 7373 cells (Presta et al., 1989), while EGF, which binds to a different TK-receptor, was ineffective (our unpublished results). Similar results were obtained for various CHO clones expressing amounts of FGFR-1 ranging from 10,000 to 70,000 receptors per cell as determined by Scatchard plot analysis. In these clones no direct correlation existed between receptor number and the extent of uPA upregulation induced by FGF-2 (our unpublished observation). Northern blot analysis confirmed that the increase in PA activity was due to uPA up-regulation (Figure 4B). No increase in cell-associated uPA activity and uPA mRNA expression was observed in mocktransfected CHO cells (Figure 4B). As observed for GM 7373 cells, high concentrations of heparin (100 ,tg/ml) or of soluble FGFR-1 (300 ,ug/ml) did not prevent uPA up-regulation induced by FGF-2 in CHOflg7G cells, and the FGF-2 mutants M1-FGF-2 and M6B-FGF-2, which are unable to induce uPA production in endothelial cells (Isacchi et al., 1991; Presta et al., 1992), were also ineffective in this clone (our unpublished observation). FGF-2 induces uPA production in different cell types besides endothelial cells, including human enses,

373

M. Rusnati et al.

E

A

0,7

:

In

m

0,6

90,5 0,4

*

.>0,3

XC

Figure 4. uPA up-regulation in FGFR-1-transfected CCHO cells. Parental CHO cells and CHOflg7G clone were seeded at 70,000 cells/cm2. Twenty-four hours

after seeding, cell cultures were incubated in medium

lOflg7G __

containing 0.4% (B) FCSof with (A) or 30 ng/ml FGF-2.increasing After 24 hconcentrations of incubation, cell-associated uPA activity was measured (A) or *I RNA was extracted by the guanidine isothiocyanate _______|method, run on formaldehyde gel, and transferred to

0,2

P0,1

o,n

10oI

100

10I

102

+

FGFF-2

FGF-2 (ng/ml)

dometrial adenocarcinoma HEC-1-B cells (Presta et al., 1988). These cells express FGFR-4 as the predominant high affinity FGFR (Coltrini et al., 1995). This prompted us to investigate whether other FGFRs besides FGFR-1 transduce the uPA-inducing signal. To this purpose, CHO cells were transfected with FGFR-2, FGFR-3, or FGFR-4 and selected clones were stimulated with FGF-1, FGF-2, and FGF-4, the three members of FGF family that have been shown to induce uPA production in endothelial cells (Basilico and Moscatelli, 1992). The CHOflg7G clone was used as a positive control. As shown in Figure 5, all the 06

Ri

0,5

1

nitrocellulose. The blot was hybridized with uPA probe (B, top), washed, and rehybridized with

GAPDH probe (B, bottom).

FGFRs are able to transduce the uPA-inducing signal when expressed in CHO cells, and the extent of stimulation exerted by the various FGFs depends on the receptor type. Taken together these observations implicate FGFR in mediating the uPA-inducing activity of FGF-2. Also, the ability of CHO transfectants to respond to FGF-2 antagonists and FGF-2 mutants in a manner similar to endothelial cells indicates that the same distinctive mechanism of FGF-2/FGFR interaction responsible for uPA up-regulation is shared by the two cell types.

0,5 0,

0,4~~~~~~~~~~~~~~~~04 0,30,2 0.1

0.2 1 ~0,0 0,0 0,1 I

10

100

0,1

10

10

FGF (ng/ml) 374

100

100

Figure 5. FGFRs-mediated uPA up-regulation by different FGFs. Stable CHO clones were obtained by transfection with IlIc splice variant of FGFR-1 (R-1, CHOflg7G clone), FGFR-2 (R-2, CHObek7G clone), FGFR-3 (R-3, MomFR3SV 3C clone), and FGFR-4 (R-4, LTR2HX 4D clone). For each clone, cells were seeded at 70,000 cells/cm2. Twenty-four hours after seeding, cell cultures were incubated in medium containing 0.4% FCS with increasing concentrations of FGF-1 plus 1 ,tg/ml heparin (0), FGF-2 (-), FGF-4 (A). After 24 h of incubation, cell-associated uPA activity was measured.

Molecular Biology of the Cell

FGFR and uPA Up-Regulation

uPA Up-Regulation Depends on the Tyrosine Kinase Activity of FGFR To assess whether the TK activity of FGFR is involved in transducing the uPA-inducing signal, tyrosine phosphorylation inhibitors (tyrphostins) were evaluated for their capacity to prevent uPA up-regulation in GM 7373 cells. These molecules had been shown previously to inhibit cell proliferation induced by FGF (Boyer and Thiery, 1993). Increasing concentrations of tyrphostin 23, tyrphostin 47, and tyrphostin 63 were added to GM 7373 cells in the presence of a constant amount of FGF-2. As shown in Figure 6, tyrphostins 23 and 47 inhibited the uPA-inducing activity of FGF-2 with an ID50 of 50-100 ,tg/ml, while tyrphostin 63, which was used as a negative control (Gazit et al., 1989), had no significant effect. On this basis, to confirm the role of the TK activity of FGFR in transducing the uPA-inducing signal in endothelial cells, we transfected GM 7373 cells with a mutated FGFR-1 cDNA carrying a stop codon in the juxtamembrane domain, thus expressing a truncated TK- FGFR devoid of its TK domain and C-terminus (Li et al., 1994). This truncated receptor had been shown to exert a dominant negative effect in NIH 3T3 cells (Li et al., 1994). As shown in Figure 7B, FGF-2 was unable to induce uPA up-regulation in stable transfectants that expressed an excess of truncated FGFR with respect to the endogenous intact receptor (Figure 7A). It must be pointed out that these cells retained the capacity to produce uPA in response to 12-O-tetradecanoyl-phorbol-13 acetate (Figure 7B), indicating the specificity of the dominant negative effect. Accordingly, when CHO cells were transfected with the same mutated FGFR-1 cDNA, FGF-2 did not up-regulate uPA production in selected clones although they expressed twice as much high affinity binding sites than the clones transfected with wild-type FGFR-1 cDNA (our unpublished observation).

Figure 7. Effect of a dominant negative truncated FGFR on FGF-2-mediated uPA up-regulation. GM 7373 cells were transfected with a cDNA encoding a truncated FGFR-1 devoid of its TK domain. Two stable clones (C8 and E6) and nontransfected cells (wt) were compared for 125I-FGF-2 binding by a cross-linking experiment (A). The black arrowhead marks the 125I-FGF-2/truncated FGFR complex while the open arrowhead marks the 125I-FGF-2/ endogenous FGFR complex, evident only after a longer exposure of the gel. Molecular weights are in thousands. (B) GM 7373 cells (gray bars), C8 cells (black bars), and E6 cells (open bars) were seeded at 70,000 cells/cm2. Twenty-four hours after seeding, cell cultures were incubated in medium containing 0.4% FCS with 30 ng/ml of FGF-2 or 100 ng/ml of 12-O-tetradecanoyl-phorbol-13 acetate (TPA). After 24 h of incubation, cell-associated uPA activity was measured. Vol. 7, March 1996

U,7

0.6-

r: 0,5 .eI

0,4

0,30

*a 0,2-

0,c

.~.

1

10

.-1

100

tyrphostin (JM) Figure 6. Effect of tyrosine kinase inhibitor tyrphostins on FGF-2 activity. GM 7373 cells were seeded at 70,000 cells/cm2. Twentyfour hours after seeding, cell cultures were incubated in medium containing 0.4% FCS with 30 ng/ml FGF-2 in the absence or presence of increasing concentrations of tyrphostin 23 (-), tyrphostin 47 (@), or tyrphostin 63 (A). After 24 h, cell-associated uPA activity was measured.

These data point to an absolute requirement for FGFR-TK activity in transducing the intracellular uPA-inducing signal after FGF-2 stimulation.

FGF-2-loaded Liposomes and uPA Up-Regulation Our observations indicate that FGF-2/FGFR interaction is essential for uPA up-regulation. However, this interaction differs from that responsible for mitogenicity, as suggested by the biological properties of different FGF-2 mutants (Isacchi et al., 1991; Presta et al., 1992, 1993) and confirmed by the experiments performed here with various FGF-2 antagonists. Cytokines and growth factors encapsulated within phospholipid vesicles or adsorbed onto their surface

A

r: 200 -

100l-

i

;-4 .

.

_

cl

Wrt C8 E6 FGF-2

IlPA 375

M. Rusnati et al.

retain the capacity to exert their biological activity on cultured cells (Fidler et al., 1985; Debs et al., 1989). Since the delivery of liposomal components to target cells may occur by different mechanisms, including intermembrane transfer, contact release, adsorption, fusion, and phagocytosis/endocytosis (for a review see New et al., 1990), we evaluated the possibility that FGF-2 encapsulated in liposomes may interact with FGFRs differently from free FGF-2, thus shedding some light on the mechanisms of receptor activation responsible for uPA induction. To this purpose, GM 7373 cell cultures were added with increasing amounts of small unilamellar phosphatidyl choline:cholesterol vesicles loaded with FGF-2 (SUV-FGF-2), corresponding to concentrations of growth factor ranging from 3 pg/ml to 100 ng/ml. Equivalent amounts of a suspension of free FGF-2 sonicated and preincubated with empty preformed phospholipid vesicles (FGF-2+SUV) were used as positive controls. As shown in Figure 8, FGF-2+SUV induced both cell proliferation and uPA up-regulation in GM 7373 cells with half-maximal stimulation at 0.1-0.3 ng/ml and 3-10 ng/ml, respectively, similar to that of FGF-2 in the absence of empty phospholipid vesicles (Presta et al., 1989). The activity of FGF2+SUV was totally abrogated by removal of FGF-2 from the suspension by heparin-Sepharose affinity chromatography (our unpublished observation). In contrast, SUV-encapsulated FGF-2 did not elicit a significant mitogenic response but fully retained the capacity to stimulate uPA production (Figure 8). A lack of increase in cell-associated uPA activity was instead observed in cells treated with liposomes loaded with heat-inactivated FGF-2 (our unpublished observations). To define the mechanism of interaction of SUVFGF-2 with endothelial cell surface, GM 7373 cells were incubated at 4°C with 10 ng/ml of 125I-FGF-2 encapsulated within phospholipid vesicles. After 2 h, the amount of radioactivity specifically associated with high and low affinity sites was less than 10% of

0

0,6

that measured when the cells were incubated with the same amount of 125I-FGF-2 added to empty preformed liposomes (Figure 9A). Accordingly, a long-lasting incubation of GM 7373 cells with 15 ng/ml of SUVFGF-2 at 37°C caused a transient, limited down-regulation of FGFRs when compared with the sustained disappearance of high affinity binding sites that follows incubation with FGF-2+SUV (Figure 9B). The rate of internalization of liposome-delivered 1251_ FGF-2 was also evaluated. The results show that only a limited amount of radioactivity was internalized by GM 7373 cells during a 24-h incubation at 37°C with 30 ng/ml of 125I-FGF-2-liposomes, corresponding to approximately 10% of the amount of radioactivity internalized after incubation with 125I-FGF-2 added to empty liposomes (Figure 10A). When the fate of intracellular FGF-2 was analyzed by SDS-PAGE followed by autoradiography of the gel, the degradation pattern of internalized 25I-FGF-2 was very similar for both the liposome-encapsulated and the unencaspulated growth factor (Figure 10B). In both cases, the intracellular degradation of 125I-FGF-2 was inhibited by chloroquine (Figure 10B), indicating that liposome-delivered FGF-2 also entered the lysosomal compartment. Taken together, the data suggest that a limited fraction of liposome-encapsulated FGF-2 interacts with cell surface FGFRs and is internalized within endothelial cells following receptor down-regulation. Accordingly, neutralizing anti-FGF-2 affinity-purified antibodies fully abolished the capacity of SUV-FGF-2 to up-regulate uPA activity in GM 7373 cells (Figure 11). Since no adsorbed growth factor is present on the surface of the phospholipid vesicle (see MATERIAL AND METHODS), this result suggests that liposomeencapsulated FGF-2 becomes exposed to the pericellular environment by a mechanism of contact-release (New, 1990b; New et al., 1990) before interacting with FGFR and being internalized. This interaction is responsible for the uPA-inducing activity of SUV-FGF-2 but is not sufficient to induce a mitogenic response.

B

0 0

0,4

*d 0,4

0 0 *_ 0

I._

0.3

C4

05