JOURNAL OF CELLULAR PHYSIOLOGY 149:50-59 (1991)

Interaction of Vasculotropin/Vascular Endothelial Cell Growth Factor With Human Umbilical Vein Endothelial Cells: Binding, Internalization, Degradation, and Biological Effects A. BIKFALVI,* C. SAUZEAU, H. M O U K A D I R I , J. M A C L O U F , N. BUSSO, M. BRYCKAERT, 1. PLOUET, AND G. TOBELEM lNSERM U 150, Hopital LariboisiPre, 75010 Paris (A.B., C.S., I.M., M.B., C.T.), lNStRM U 86, Centre des Cordeliers, 75006 Paris (H.M., I.P.), and laboratoires Claxo, 9 I940 Les Ulis (N.B.), France Vasculotropinivascular endothelial cell growth factor (VASIVEGF) i s a newly purified growth factor with a unique specificity for vascular endothelial cells. We have investigated the interactions of VASNEGF with human umbilical vein endothelial cells (HUVE cells). 'Lsl-VAS/VEGF was bound to HUVE cells in a saturable manner with a half-maximum binding at 2.8 ngiml. Scatchard analysis did show two classes of high-affinity binding sites. The first class displayed a dissociation constant of 9 p M with 500 sitesicell. The dissociation constant and the number of binding sites of the second binding class were variable for different HUVE cell cultures (K, = 179 & 101 pM, 5,850 2,950 sitesicell). Half-maximal inhibition of '"I-VASIVEGF occurred with a threefold excess of unlabeled ligand. Basic fibroblast growth factor (bFGF) and heparin did not compete with '251VASNEGF binding. In contrast, suramin and protamin sulfate completely displaced '251-VAS/VEGF binding from HUVE cells. VASiVEGF was shown to be internalized in HUVE cells. Maximum internalization (55% of total cell-associated radioactivity) was observed after 30 min. '"I-VASIVEGF was completely degraded 2-3 hr after binding. At 3 hr, the trichloroacetic acid (TCA)-soluble radioactivity accumulated in the medium was 60% of the total radioactivity released by HUVE cells. No degradation fragment of 'L'I-VAS/VEGF was observed. Chloroquine completely inhibited degradation. VASNEGF was able to induce angiogenesis in vitro in HUVE cells. However, it did not significantly modulate urokinase-type plasminogen activator (u-PA), tissue-type plasminogen activator (t-PA), plasminogen activator inhibitor (PAI-l), and tissue factor (TF). Prostacyclin production was only stimulated at very high VASNEGF concentrations. Taken together, these results indicate that VASiVEGF might be a potent inducer of neovascularization resulting from a direct interaction with endothelial cells. The angiogenic activity seems to be independent of the plasminogen activator or inhibitor system.

*

Vasculotropinivascular endothelial cell growth factor (VASIVEGF) is a recently identified growth factor presenting so far a unique specificity for vascular endothelial cells. This factor has been isolated from the tumoral cell line AtT 20 derived from mouse anterior pituitary (Plouet e t al., 1989) and bovine normal folliculo-stellate cells (Ferrara and Henzel, 1989; Gospodarowicz et al., 1989). Concomitantly, three other groups have isolated a similar factor from various cell lines (Connolly et al., 1989a,b; Conn et al., 1989; Levy et al., 1990). VASIVEGF is a homodimer of 45 kDa containing 164 amino acids. The cloning of the gene provided the complete sequence which showed that VASiVEGF is a new member of the platelet-derived growth factorisis (PDGFisis) family (Leung et al., 1989; 0 1991 WILEY-LISS, INC

Received March 5, 1991; accepted May 8, 1991. "To whom reprint requests/correspondence should be addressed: Department of Cell Biology, New York University Medical Center, 550 First Avenue, New York, NY 10076. H. Moukadiri and J. Plouet's present address is: ATIP, Centre de Recherches en Biochimie et Genetique Cellulaire, 114 route de Narbonne, 3100 Toulouse, France. Abbreviations used: bFGF, basic fibroblast growth factor; DNTB, di-thio bis nitrobenzoate; HUVE, human umbilical vein; PAI-1, plasminogen activator inhibitor-1; PCA, procoagulant activity; PDGF, platelet-derived growth factor; PGFla, prostaglandin Fla; TF, tissue factor; TGF-p, transforming growth factor-p; TPA, 12-0 tetradecanoylphorboll3-acetate; t-PA, tissue-type plasminogen activator; uPA, urokinase-type plasminogen activator; VAS, vasculotropin; VEGF, vascular endothelial cell growth factor; VPF, vascular permeability factor.

VASCULOTROPIN EFFECT ON HUVE CELLS

Keck et al., 1989; Tischer et al., 1989; Conn et al., 1990). The overall homology with A and B chains of PDGF was respectively 15 and 18%. The cDNA sequence predicted the existence of two more peptides of 189 and 121 amino acids generated by alternative splicing. The 121-amino-acid form displays a deletion of 44 amino acids between positions 116 and 159, and the 189 protein possesses a insertion of 22 amino acids in position 116. These three putative chains contain a signal sequence which would allow the secretion via the classical secretory pathway. VASiVEGF stimulates the growth of bovine capillary endothelial cells, aortic endothelial cells, and human umbilical vein endothelial cells. It fails to elicit any mitogenic response in a wide variety of non-vascular cells including corneal endothelial cells, vascular smooth muscle cells, baby hamster kidney cells (BHK 21), and cells from the granulosa and adrenal cortex. VASiVEGF has also been described to induce vascular permeability and was therefore also named vascular permeability factor (VPF) (Connolly et al., 1989a,b). Although VASiVEGF enhances the proliferation of endothelial cells in vitro to a lower extent than basic fibroblast growth factor (bFGF), i t appeared to induce neovascularization in the corneal pocket assay as well as bFGF (Favard et al., in preparation). It is not known whether this effect of VASiVEGF on angiogenesis is direct or indirect. We have therefore looked a t various steps controlling VASiVEGF-dependent angiogenesis. In this study we describe the interactions between VASiVEGF and human umbilical vein endothelial cells (HUVE cells). HUVE cells were chosen for this study since they respond to various angiogenic factors including 1 2 - 0 tetradecanoylphorbol 13-acetate (TPA) and are able to form capillary-like structures in vitro (Montesano and Orci, 1987). Binding studies of VASI VEGF on endothelial cells have shown the presence of two classes of high-affinity binding sites with a dissociation constant of 2 and 48 pM and a low number of binding sites (Plouet and Moukadiri, 1990a,b). A receptor of 185 kDa has been characterized by cross-linking experiments (Plouet and Moukadiri, 1990b; Vaisman et al., 1990). These studies have been conducted on bovine endothelial cells but only a little information is presently available concerning the interaction of VASIVEGF with human endothelial cells. Also, internalization and degradation of VAS bound to endothelial cells has not yet been reported. The major protease-antiprotease system used by the vascular endothelium for invasion during angiogenesis is the plasminogen activator and inhibitor system (Montesano and Orci, 1985, 1987; Montesano et al., 1986; Pepper e t al., 1990). It is modulated by several growth and angiogenic factors including bFGF and transforming growth factor-p (TGF-P) (Montesano et al., 1986; Saksela et al., 1987; Pepper et al., 1990). We have therefore investigated whether VASIVEGF would be able to stimulate plasminogen activators or inhibitor. In addition, we have looked into the putative role of VASIVEGF on prostacyclin production and tissue factor expression, since, beside their role in angiogenesis,

51

endothelial cells respond to various cytokines by increasing both of these activities (Bevilacqua et al., 1984, 1986; Brevario et al., 1990; Zavoico et al., 1990). MATERIALS AND METHODS Cell culture media and reagents All media and reagents for cell culture were obtained from Boeringer (Mannheim, FRG). ‘“I-Na (specific activity 250 millicuriesiml) were from the “Comissariat de 1’Energie Atomique” (Gif sur Yvette, France). PD 10 column and heparin-sepharose were from Pharmacia (Uppsala, Sweden). Twelve-0-tetradecanoylphorbol13acetate, bovin serum albumin (BSA), Chloramin T, gelatin, and chloroquine were from Sigma (St. Louis). All reagents for electrophoresis were from Bio Rad (Richmond, VA). All plasticware for cell cultures were purchased from Flow (McLean, VA). Basic fibroblast growth factor (bFGF) was a gift of J. Abraham (Calbio, Mountain View, CA). Collagen type 1 from rat tail, interleukin-la (11-l), and epidermal growth factor (EGF) were from Collaborative Research (Bedford, MA). Suramin was from Mobay Inc. and heparin from Choay institut (Paris, France). Tissue plasminogen activator (t-PA) kit, urokinase-type plasminogen activator (u-PA)kit, and plasminogen activator inhibitor-1 (PAI-1) kit were from American Diagnostica (Greenwich, CT). All other materials were research grade. Cell cultures Human umbilical vein endothelial (HUVE) cells were isolated and cultured according to the method of Jaffe et al. (1973). The cells were cultured in medium 199 supplemented with 10% human serum and passaged in a split ratio of 1 : 2. All experiments were carried out between passages 1 and 3. Purification of VASIVEGF Vasculotropinivascular endothelial cell growth factor was purified from the murine ATt2O cell line as described (Plouet et al., 1989). Murine ATt20-derived VAS is a homodimer of 45 kDa and corresponds to the 164-amino-acid isoform of VASIVEGF. The purity was monitored by SDS-PAGE electrophoresis providing a single band of 45 kDa under unreduced conditions.

Iodination of VASlVEGF Vasculotropinlvascular endothelial cell growth factor was iodinated according to the chloramine T method (Greenwood et al., 1963). One microgram of VASiVEGF was incubated in 0.1 M phosphate buffer in saline (PBS) with 0.8 mCi of 1251-Nain the presence of 5 pl chloramin T (2 mgiml). The reaction was stopped after 30 seconds by adding 10 pl sodium bisulfite (1mgiml). NaI was added after 5 min and the unbound lZ5Iwas eliminated by chromatography on a Sephadex PD 10 column. The specific activity was 200,000 cpming VASI VEGF. Bioactivity of iodinated VASIVEGF was measured by comparison with the bioactivities of native VASIVEGF.

52

BIKFALVI ET AL.

Cell surface binding of 1251-VAS/VEGF HUVE cells were grown to confluence in 12-well plates, rinsed thoroughly with binding buffer (Medium 199 containing 20 mM Hepes, and gelatine 0.2%, pH 7.4) and preincubated with the same binding buffer at 4°C for 15 min. Binding was carried out on a n orbital shaker. Increasing concentrations of 1251-VAS/VEGF were added to the wells in a final volume of 250 pl with or without 400 ngiml unlabeled VASiVEGF. The binding was stopped after the periods of time indicated in the legends of the figures by washing the cells five times with cold binding buffer. The cells were solubilized with a solution of 0.5 ml 2% Triton X-100/10% glycerolil mgiml of bovine serum albumin and the radioactivity was counted with a Beckman 7000 gamma counter. Non-specific binding was determined in the presence of 400 ngiml unlabeled VASiVEGF. Specific binding was determined by substracting nonspecific binding from total binding. The data were analysed according to Scatchard's procedure (1948). Competition experiments were done by incubating the cells with a fixed concentration (0.5 ngiwell) of lZ5IVAS/VEGF and increasing concentrations of cold ligand or competitors.

were run in duplicates and repeated a t least three times. Angiogenesis in vitro In vitro angiogenesis experiments were carried according to Montesano and Orci (1985). Briefly, collagen type I from rat tail (2 mgiml) was mixed with MEM 10 X and sodium bicarbonate (11.7 mgiml) (ratio 7:1:2) and distributed into wells of 24-well plates. After gelification a t 37°C for 10 min, HUVE cells (60,0001 well) were layered a t the top of the gel. At confluence, VASiVEGF was added to the wells to induce tube formation. Photomicrographs were taken during the assay and a t the end of the assay.

Tissue f a c t o r (TF) activity Procoagulant activity (PCA) was measured in a single-step clotting assay performed by robot (Biomek, Beckman). Briefly, second-passaged HUVE cells plated in 96-well plates (40,000 cells/well) were incubated for 5 hr with different concentrations of VASIVEGF or appropriated controls. Cells were washed twice with calcium free PBS. After freezing and thawing in 10 mM Tris-HC1 buffer (pH 7.4) containing 1 mM EDTA, coagulation was initiated by addition of 80 ~1of human pooled citrated platelet-free plasma and 80 pl 30 mM Internalization studies of '251-VASNEGF CaC1,. Clot formation was monitored a t 540 nm. The Internalization of bound ligand was determined ac- kinetics of the reaction was recorded and the time cording to the method of Haigler et al. (1985). The cells required to reach a fixed optical density (clotting time) were washed five times after 2 h r of binding at 4°C with was then determined. The results were expressed as cold binding buffer and incubated a t 37°C with 1251- milliunits of TF defined by conversion from standard VASiVEGF in the presence or absence of 400 ngiml curves (log-log plot) developed with a preparation of unlabeled ligand. After specified time intervals, the rabbit brain thromboplastin. This PCA was totally cells were washed for 3 min with ice-cold 150 mM TF-dependent as assessed by using factor VIILdeficient NaClI20 mM acetic acid buffer (pH 3).The radioactivity plasma instead of normal plasma or by preincubating of the incubation medium, the acetic wash, and the cell lysates with anti-TF antibodies. Triton-X 100 extract were then determined in a gamma Assay f o r prostacyclin production counter. Second-passaged HUVE cells were incubated in 12Degradation studies of lZ5I-VAS/VEGF well plates for 4 to 48 h r with either medium 199 Degradation of internalized 1251-VASiVEGFwas de- containing 0.4% BSA or 10% FCS with or without termined by incubating the cells after 2 hours of either VASiVEGF or appropriated controls. At the end binding a t 4"C, with fresh binding medium at 37°C in of the incubation period, the accumulation of the major the presence or absence of chloroquine (50 FM). The prostacyclin metabolite, 6-Keto-PGF1, in the medium radioactivity was measured after specified time inter- was determined by ELISA as previously described vals in the medium and in the Triton-X 100 extract. (Pradelles et al., 1985). Trichloracetic acid (TCA) solubility was measured by Briefly, 96-well microtiter plates were coated with TCA precipitation (12%, v/v) at 4°C. After several monoclonal anti-rabbit antibody. Standard or adequate hours, the samples were centrifuged at 2,200g for 20 dilutions of the test samples, tracer (acetylcholinestmin and the radioactivity was determined in the TCA- erase-labeled 6-Keto-PGF1,), and specific anti-6-Ketoprecipitable and -soluble fraction. In parallel experi- PGF,, antibody were incubated a t 4°C for 40 hr. ments, the incubation medium was removed and the Separation of bound from free was done by washing five cells were solubilized with 200 pl of sodium dodecyl times with washing buffer (0.05 M sodiumphosphate, sulfate (SDS) buffer containing 2 mM EDTA and stored pH 7.4,0.05% Tween 20) by using a n automatic washer. a t -20°C until electrophoresis. All samples were Ellman reagent (7.5 x M acetylcholine iodine, heated for 15 min at 100°C in SDS sample buffer 5 x M 5,5' dithiobis 2-nitrobenzoate (DTNB), 0.01 containing 5% beta-mercaptoethanol. SDS-polyacryla- M phosphate buffer, pH 7.4) was added and hydrolysis mide gel electrophoresis (SDS-PAGE) was performed of acetylcholine was determined with DTNB by meaaccording to Laemmli (1970). After electrophoresis, the suring the absorbance a t 412 nm. Calculations were gels were dryed, stained with Coomassie brilliant blue, done by using a n on-line computer. and subjected to autoradiography at -70°C for a week Plasminogen activator a s s a y by using Kodak X-AR films and Dupont Lightning Plus intensifying screens. All experimental measurements All experiments were performed by using HUVE for cell surface binding, degradation, or internalization cells passage 2 or 3 plated into 12-well plates (400,000

VASCULOTROPIN EFFECT ON HUVE CELLS 10000

(t-PA), and plasminogen activator inhibitor type-1 (PAI-1) ELISA assays were conducted according to the protocols provided by the manufacturer (American Diagnostica, Greenwich, CT).

1

RESULTS HUVE cells bind, internalize, and degrade VASIVEGF The kinetics of '251-VAS/VEGF binding was determined by incubating confluent HUVE cells at 4°C for different time intervals with a fixed concentration of lZ5I-VAS/VEGFwith or without 400 ngiml unlabeled ligand (non-s ecific binding) (Fig. 1).It appeared that a plateau of1 51-VASiVEGFbinding was reached after 2 hr. Half-maximum binding was reached a t 15 min. Non-specific binding a t the plateau phase was between 7 and 35%. In order to determine the saturability of lZ5I-VASi VEGF binding, HUVE cells were incubated a t 4°C with increasing concentrations of 1251-VASIVEGF in the presence or absence of a n excess of unlabeled ligand (non-specific binding) (Fig. 2). Saturation of binding ocurred a t 2.5 ngiwell (10 ngiml) with a half-maximal binding of 0.7 ngiwell (2.8 ngiml). Non-specific binding a t saturating concentrations of 1251-VASIVEGF was usually not more than 25%. When the binding data were analyzed according to Scatchard (1948) two classes of high-affinity binding sites (HA, and HA,) were found (Fig. 2 inset, which is representative for the experiments). The dissociation constant (K,) and the number of binding sites for HA, sites were respectively 9 pM and 500 sitesicell and for HA, sites 179 pM and 5,850 sitesicell assuming a apparent molecular mass of 45 kDa for VASIVEGF. However, variability of the HA, binding values was observed (SD for K, 2 101, SD for number of sitesicell * 2,950) depending on different cell cultures. The competition of binding was measured by incubating HUVE cells a t 4°C with a fixed concentration of 1251-VASIVEGF(0.5 ngiml) and increasing concentrations of unlabeled VASIVEGF. Half-maximum inhibition occurred a t 1.5 ngiml (threefold the tracer value) (Fig. 3). VASiVEGF is a heparin-binding protein. We tested therefore whether or not the binding of lZ5IVASIVEGF to HUVE cells was modulated by heparin. Figure 3 shows that heparin was not able to inhibit significantly the binding of 1251-VASiVEGF.Suramin (100 pgirnl) and protamin (20 Fgiml), known to inhibit the interaction of a variety of growth factors with their receptors, were shown to block completely the receptor binding of '"I-VASIVEGF (Fig. 3). Basic FGF (Fig. 3) and EGF (data not shown) were not capable of competing with 1251-VASiVEGFbinding to HUVE cells. Thus, the conditions of cell surface binding of VASIVEGF in human endothelium are similar to that observed in bovine endothelial cells (Plouet and Moukadiri, 1990a,b; Vaisman et al., 1990). Internalization of bound VASiVEGF was determined according to the method of Hai ler et al. (1980). Cells were incubated a t 4°C with "'1-VASIVEGF with or without a n excess of unlabeled compound (non-specific binding) and shifted to 37°C. Resistance of the cellassociated radioactivity to acidic treatment was determined after the time intervals indicated in Figure 4.

B

O

P 0

60

30

9 0 120 150 180 210 240 270 300 330 360

l i m e (min)

Fig. I . Kinetics of '251-VAS/VEGF binding to HUVE cells. HUVE cells were incubated with 0.5 ngiml (100,000 cpm) '2511-VAS/VEGF in presence or absence of a n excess of unlabeled ligand (400 ngiml). The experiments were performed as indicated in Materials and Methods. Cell-associated radioactivity (m -m), non-specific binding (E----U),

0.00

r

0

53

BOYnd (nplvsll) i

1

2

3

4

5

V A S I VEGF (nglwell) Fig. 2. Concentration dependence of '"I-VASIVEGF binding to HUVE cells. HUVE cells were incubated with increasing concentrations of 1251-VAS/VEGFwith or without an excess of unlabeled ligand (400 ngiml). The experiments were performed as indicated in Materials and Methods. Scatchard analysis is shown in the inset.

cellsiwell). The cells were incubated for 4 and 24 h r in medium 199 containing 2.5% FCS and different concentrations of either VASiVEGF or appropriate controls. At the end of the incubation period, the medium was aspirated and the cells were extracted with a 10 mM Tris-HC1 buffer (pH 8.1) containing 0.5% Triton X-100. The incubation medium and cell extract were stored at - 20°C until use. Urokinase-type plasminogen activator (u-PA), tissue-type plasminogen activator

BIKFALVI ET AL.

54

C

Fi Q

u

01 0

I

i

10

20

unlabeled VASIV EGF(ng/ml)

100

=

20 2 0 0 1 0 0

u@ml

Wml uS/d

Fig, 3. Competition of 1251-VAS/VEGFbinding to HUVE cells. HUVE cells were incubated at 4°C with a fixed concentration of 'Z51-VAS/VEGF(0.5 ngiml) and either increasing concentrations of unlabeled ligand (0-0) or either heparin (100 pgiml), suramin (100pg/ml), protamin (20 pg/ml), or bFGF (200 ngiml). The experiments were achieved as indicated in Materials and Methods.

These experiments demonstrated a n increase of the radioactivity in the Triton-X 100-extractable material. The maximum in the Triton-X 100-extractable radioactivity was obtained at 30 min followed by a subsequent decline. In parralel, the radioactivity in the medium increased. This was most certainly due to the dissociation of a fraction of bound li and from its receptor and to the release of degraded 851-VASIVEGF from the cells. In contrast, the radioactivity contained in the acidic wash decreased progressively to attain 22% and 7% of total radioactivity at 30 min and 2 h r respectively. When the data were plotted as percentages of total cell-associated radioactivity a t 30 min (total of radioactivity bound after acidic wash and radioactivity in the acidic wash), it appeared that a maximum of 55% of binding was reached a t 30 min which remained stable a t subsequent time intervals (Fig. 4B). Internalized '251-VAS/VEGF was found to be degraded in HUVE cells (Fig. 5A-C). In the absence of chloroquine, cell-associated radioactivity decreased progressively with a concomitant increase of radioactivity in the medium. The minimum for the cellassociated radioactivity (30% of total radioactivity) and the maximum for the radioactivity found in the medium (64% of total radioactivity) were attained a t 2 and 3 hr. In the presence of chloroquine (50 pM), significantly less radioactivity (25% of total radioactivity) was released in the medium and only a small decrease of the cell-associated radioactivity was observed (70% total cell-associated radioactivity). A plateau from 1hr onward was observed for both released and cell-associated radioactivity.

When the radioactivity found in the medium was analysed for its TCA precipitability, it was demonstrated that 60% and 66% of the released radioactivity became TCA soluble after 3 and 6 h r respectively (Fig. 5C). Chloroquine was capable of inhibiting the presence of TCA-soluble radioactivity to 12% and 22% after 3 and 6 h r respectively (Fig. 3C). Analysis of the degradation pattern by SDS-PAGE demonstrated only the presence of native '251-VAS/ VEGF which decreased progressively with time (Fig. 6). No degradation fragments were seen. No decrease of 1251-VASIVEGFwas observed a t 4°C and a t 37°C in the presence of chloroquine (data not shown).

VASIVEGF induces angiogenesis in vitro in HUVE cells The effect of VASIVEGF on angiogenesis in vitro was tested by using the angiogenic assay on collagen type I gels described by Montesano and Orci (1985) (Fig. 7). This assay has been used to establish the angiogenic effect of different soluble factors (Montesano et al., 1986; Montesano and Orci, 1987). VASIVEGF was able to induce angiogenesis in vitro. After 1day sprouting of HUVE cells was seen. Tube formation began a t day 3 and was clearly apparent a t day 7. This effect was similar to that observed for TPA and bFGF, and no significant tube formation was seen at day 7 in the absence of VASIVEGF or with 2 pglml cycloheximide (data not shown). From this result it can be concluded that VASIVEGF directly induced angiogenesis in vitro.

55

VASCULOTROPIN EFFECT ON HUVE CELLS 'OO-1

A

5 0 ~ 0 1A

0

10 2 0 3 0

40

50 60

70 80 90 100110120130

0

60

120

Time (min)

180

240

300

360

240

300

360

240

300

380

Time (rnin)

..-P

60

XI/

0

1 I-

40

101

o ! .... 0

I

20

. - . . , . . . ' " . " . . . " 1 " . ' 1 . -

40

60

80

100

120

Time (min) Fig. 4. Internalization of '"I-VASIVEGF in HUVE cells. HUVE cells were incubated for 2 hr a t 4°C with lZ5I-VAS/VEGF with or without an excess of unlabeled ligand (400 ng/ml) and shifted to 37°C. The experiments were achieved as indicated in Materials and Methin the cell extract ods. A Radioactivity in the medium (0-01, (A----A), and in the acidic wash (B----W). B: Results are expressed as percentages of total cell-associated radioactivity, in the acidic wash (W - - - - w), and in the cell extract ( 0 -0 ) .

Effect of VASNEGF on plasminogen activators, plasminogen activator inhibitor type 1, tissue factor, and prostacyclin production Table 1 summarizes the effect of VASIVEGF on urokinase-type plasminogen activator (u-PA), tissuetype plasminogen activator (t-PA), plasminogen activator inhibitor type 1 (PAI-11, tissue factor (TF), and prostacyclin production. Plasminogen activators and inhibitor were measured by ELISA in the medium and in cell extract of HUVE cells stimulated for 4 to 24 h r by VASIVEGF using monoclonal antibodies against human u-PA, t-PA, and PAI-1. The results demonstrated that VASiVEGF a t concentrations up to 200 nglml did not exhibit a significant effect on both plasminogen activators and on their inhibitor PAI-1. Table 1 shows the results of 24-hr-stimulated HUVE cells. TF activity was measured as procoagulant activity

0

60

120

180

Time (min)

0

60

120

180

Time (min)

Fig. 5. Degradation of '251-VAS/VEGFin HUVE cells. HUVE cells were incubated with '251-VAS/VEGF for 2 hr at 4°C and shifted to 37°C. The radioactivities were determined after several time intervals in the medium I.-.( and the cell extract ( n - - - - O ) in the presence (B) or absence (A) of chloroquine (50 pM). The medium was subsequently precipitated with 12% TCA and the radioactivity was determined in the TCA-soluble and -precipitable fraction. The data were plotted as percentages of TCA-soluble radioactivity in the presence (A----A) or absence (A-A) of chloroquine (C).

56

BIKFALVI ET AL.

kDa

3121.5*

0

30

60

120 rnin

Fig. 6. Degradation pattern of internalized '2sI-VASIVEGF. HUVE cells were incubated for 2 hr at 4°C with '"I-VASIVEGF and shifted to 37°C. After several time intervals, the cells were extracted with SDS buffer and the samples were run under reduced conditions on a 15% PAA gel, colored, dried, and exposed for autoradiography. The standards were stained with Coomassie blue.

by a single-step clotting assay in cell lysates from HUVE cells incubated with VASIVEGF or controls for 5 hr. VASiVEGF was unable to induce TF in HUVE cells. No effect was seen a t higher VASIVEGF concentrations up to 200 ngIml. In addition, VASIVEGF was not capable of modulating TF activity induced by IL-1 (1UIml) or TPA (10 ng/ml) (data not shown). Prostacyclin production was measured by determination of the amount of its major metabolite, 6-KetoPGF,,, accumulated in the medium. VASiVEGF did not show a n enhancement or decrease of prostacyclin production measured after 4 hr. In comparison, thrombin (1 UIml) and IL-1 (1 Uiml) did increase 6-KetoPGF,, in the medium by 7-fold and 6-fold over unstimulated control. However, after 48 h r prostacyclin production was significantly increased by VASIVEGF but only a t a concentration of 200 ngIml (2-fold over unstimulated control).

DISCUSSION In this study we have shown that human umbilical vein endothelial cells (HUVE cells) bind, internalize, and vasculotro~inivascular growth factor (VASIVEGF). Binding was time and

C

D7

D1

Fig. 7. Angiogenesis in vitro induced by VASIVEGF. Confluent HUVE cells plated on collagen type I gels were stimulated at confluence with 100 ngiml VAS. Photographs were at days 1, 3, and 7.

57

VASCULOTROPIN EFFECT ON HUVE CELLS TABLE 1. Comparative effects of V A S N E G F on u-PA, t-PA, PAI-1, TF, and PGFlo production'

VASNEGF (1-200 ng/ml)

t-PA

u-PA

PAI-1

TF

105115% (CE) 102-112% (CM)

102-11l'Jh (CE)

98-102% (CM)

85-133%

PGF1 rv

210% (at 200 ndml)

'The experiments were performed a s indicated in Materials and Methods. u-PA, t-PA, and PAI-1 were measured in the medium (CM) or cellular extract (CEj from cells stimulated with VASIVEGF or TPA (10 nglml) for 24 hr. TPA stimulated u-PA 1.8-fold (CE),t-PA 4.6- (CE) and 9.7-fold (CM),and PAI-1 1.4-fold (CM) in comparison to unstimulated control. T F was determined in the cell extract after 5 hr stimulation with VAWVEGF, TPA (10 ng/ml), or IL-1 (1 U/ml). TPA and IL-1 stimulated T F 7.5-fold and 6.2-fold respectivelyin comparison to unstimulatedcontgrol. Prostacyclin was determined in the medium after stimulation with VAS/VEGF, thrombin (1U/mlj, IL-1 (1 U/mlj, or arachidonic acid (25 uM) for 4 or 48 hr. Stimulation of PGFlu production by thrombin, IL-1, or arachidonic acid a s 7.2-fold, &fold, and 7-fold. respectively, in compa ison to unstimulated co trol Control values means of three different exper'mentsj were respectively u-P$, 0.47 ng/lO 6cells (CE);tPA, 6 ng/lO'cells (CFj, 9.5 ng/lO cells (CM);PAI-1,412.5 ng/lO B cells (CM);TF, 1.7 mU/4 X 10 cells; PGFlu, 0.5 ng/105 cells (4 hr), 11 ng/lO" cells (48 hr). Results are indicated a s percentages of stimulation in comparison to control (control a s 100%).

6

concentration dependent and was saturable. The values of the binding constants are in agreement with the values observed for bovine endothelial cells, published during the preparation of this article (Plouet and Moukadiri, 1990a,b; Vaisman et al., 1990). Two classes of binding sites were observed. The K, and the number of the second class of high-affinity binding sites in HUVE cells were found to be highly variable, which seems to depend on different cell cultures. However, Connolly et al. (198913) have reported the presence of a single class of binding sites in HUVE cells and bovine aortic endothelial cells (BAEC) with a dissociation constant of 50 pM. No information on the number of binding sites was provided. This discrepancy might be due to the low number of the first class of high-affinity binding sites which would not have been detected. The binding was found to be specific since bFGF or EGF did not compete. PDGF has been described to be structurally related to VASiVEGF. However, we have not used PDGF for competition with 1251-VASIVEGF binding since HUVE cells do not exhibit receptors for that growth factor (Heldin et al., 1981). VASIVEGF tightly binds heparin and is isolated by heparin-Sepharose chromatography. We have therefore investigated whether heparin was able to modulate the binding of '"I-VASIVEGF. No effect of heparin on the binding of 1251-VASlVEGFto HUVE cells was observed. This data suggested that in contrast to bFGF, which binds to high-affinity receptors and low-affinity proteoheparan sulfate binding sites (Moscatelli, 1987, 1988; Kieffer et al., 1990), VAS did not seem to bind t o this type of sites. Suramin and protamin are basic proteins which were demonstrated to block the binding of a variety of growth factors including PDGF and bFGF to their receptors (Williams et al., 1984; Coffrey et al., 1987; Neufeld and Gospodarowicz, 1987). Protamin has also been shown to inhibit angiogenesis in vitro (Tavlor and Folkman. 1982). Suramin and Drotamin we& found to block 'completely the bindhg of '"I-VASIVEGF to HUVE cells. 1251-VAS/VEGFbound to HUVE cells was found to be rapidly internalized. Internalization of VASiVEGF reaches a maximum of 55% after 30 min. Internalization of bFGF has been reported to be slower than for VASNEGF (Moscatelli, 1988; Bikfalvi et al., 1989). These differences might be due to the absence of

binding of VASIVEGF to slow internalizing low-affinity binding sites as has been demonstrated for bFGF (Moscatelli, 1988). Internalized 1251-VASiVEGFwas rapidly degraded in HUVE cells. Chloroquine was capable of inhibiting degradation, suggesting that degradation occurs through the lysosomal pathway as has been described for bFGF (Bikfalvi et al., 1989; Moenner et al., 1989).In the absence of chloroquine no clear degradation fragment was observed for 1251-VAS/VEGF. These data suggest that VASiVEGF is rapidly degraded by lysosomal enzymes into small fragments which are not detected by SDS-PAGE. Degradation has been reported to be different for bFGF. Degradation was slow for bFGF and did not reach a plateau even after 6 h r (Bikfalvi et al., 1989). lZ5I-bFGF has also been shown to be degraded into fragments of different sizes ranging from 6 to 15 kDa (Bikfalvi et al., 1989; Moenner et al., 1989). It has been demonstrated that VASIVEGF induces angiogenesis in vivo (Plouet et al., 1989; Levy et al., 1990; Connolly et al., 1989a). In order to investigate whether this angiogenic effect was direct or indirect, the angiogenic activity was tested in a n in vitro angiogenic assay according to Montesano and Orci (1985). This assay has been used for different angiogenic factors including TPA, bFGF, and TGF-beta (Montesano et al., 1986; Montesano and Orci, 1985, 1987; Madri et al., 1988; Pepper et al., 1990). The results clearly indicated that VASIVEGF possesses a direct angiogenic activity in HUVE cells. Sprouts appeared rapidly and cord formation began after 2 days. Cords with branches were visible at day 7. These data agree well with the high efficiency of VASiVEGF in inducing the migration of endothelial cells in vitro (Favard et al., in preparation). The plasminogen activator system is the major protease-antiprotease system required for invasion and has been reported to be modulated by angiogenic factors including bFGF and TGF-P (Montesano et al., 1986; Saksela et al., 1987; Pepper et al., 1990). In order to test whether this system is stimulated by VASIVEGF, HUVE cells were incubated with VASi VEGF and the amounts of u-PA, t-PA, and PAI-1 accumulated in the cell extract or medium were measured by ELISA. The absence of stimulation by VASI

58

BIKFALVI ET AL

VEGF argued against the involvement of plasminogen activator or inhibitor in angiogenesis induced by VASi VEGF. Nevertheless, bovine capillary endothelial cells, in contrast to HUVE cells, contain much more u-PA than t-PA (Saksela et al., 1987). Our results do not rule out completely a n effect of VASIVEGF on at least u-PA in capillary endothelium and further experiments should be performed with capillary endothelial cells. It has been postulated that the modulation of prostacyclin production might be involved in angiogenesis (Ziche et al., 1982; Voelkel, 1989). Conflicting results have been reported regarding the effect of FGF on prostacyclin production (Hasegawa et al., 1988; Kuwashima et al., 1988; Weksler, 1990). In order to test whether VASiVEGF was able to modulate prostacyclin production, HUVE cells were stimulated with VASi VEGF and the ability to induce 6-Keto-PGF1, in these cells was tested. VASIVEGF, a t maximal mitogenic concentration (1-2 ngiml), was not able to elicit a n increase in the prostacyclin production. VASIVEGF at concentrations of 200 ngiml was able to stimulate significantly its production. The reason for the requirement of such high concentrations which are beyond maximal receptor occupancy is unknown. Interleukin-1 (IL-1) and TNFa are two potent cytokines which play a n important role in inflammation and induce procoagulant activity on the endothelial cell surface, thereby stimulating the activation of the extrinsic coagulation pathway (Bevilacqua et al., 1984, 1986). Both cytokines have been demonstrated to play a role in angiogenesis (Frater-Schroder et al., 1987; Cozzolino e t al., 1990). Since we have recently described lymphokine properties of VASiVEGF (Moukadiri et al., 1990), we wondered whether it could modulate, like IL-1, procoagulant activity. VASiVEGF did not induce tissue factor activity in contrast to IL-1 and TPA. These data rule out the possibility that a t least ATt20-drived VASiVEGF representing the 164-amino-acid isoform possesses a cytokine-like effect on tissue factor expression. In summary, our results indicate that human endothelial cells bind, internalize, and degrade VASiVEGF and that these cells respond to this growth factor. However, VASiVEGF was not found to be clearly degraded into fragments of different sizes and was shown not to significantly stimulate plasminogen activators or inhibitor, prostacyclin production, or tissue factor activity. Angiogenesis induced by this growth factor could possibly be independent of the plasminogen activation system. Basic FGF has also been described to be angiogenic in vitro and to stimulate the plasminogen activator system (Montesano et al., 1986; Saksela et al., 1987; Pepper et al., 1990). However, bFGF does not contain a signal peptide and it is not secreted in the medium. It is therefore unlikely that its in vivo angiogenic role might be exerted through the classical secretory pathway. However, under certain pathological circumstances like irradiation (Witte et al., 1989), endotoxin application (Gadjusek and Carbon, 19891, or mechanically induced membrane disruptions (McNeil et al., 1989) significant amounts of bFGF are released from the cells. VASIVEGF, in contrast, contains a signal

sequence and is secretable under basal conditions (Ferrara and Henzel, 1989; Keck et al., 1989; Tischer et al., 1989). It would be a better candidate for a soluble angiogenic factor playing a role in vivo during physiological angiogenesis. In support of such a role, Phillips et al. (1990) have recently demonstrated by in situ hybridization in the ovary a temporal relationship between VAS expression and growth of capillary vessels. In light of the present result, VASiVEGF could be one of the physiological angiogenic factors acting directly on its endothelial cell target, inducing capillary cord formation.

ACKNOWLEDGMENTS This work was supported by grants from “1’Associationde la Recherche contre le Cancer (ARC)” and “la Ligue contre le Cancer.” A.B. was a recipient of a fellowship from “la Fondation pour la Recherche Medicale (FRM)”;C.S. was a recipient from Lipha; and H.M. a recipient from “la Fondation Merieux.” The authors thank Dr. D. Hicks for critically reading the manuscript. LITERATURE CITED Bevilacqua, M.P., Pober, J.S., Majeau, G.R., Cotran, R.S., and Gimbrone, M.A. (1984) Interleukin-1 (IL-1) induces biosynthesis and cell surface expression of procoagulant activity in human vascular endothelial cells. J. Exp. Med., 16Ot618-623. Bevilacqua, M.P., Pober, J.S., Majeau, G.R., Fiers, W., Cotran, R.S., and Gimbrone, M.A. (1986) Recombinant tumor necrosis factor induces procoagulant activity in cultured human vascular endothelium: Characterization and comparison with the actions of interleukin. Proc. Natl. Acad. Sci. USA, 83t4533-4537. Bikfalvi, A,, Dupuy, E., Inyang, A.L., Fayein, N., Leseche, G., and Courtois, Y. (1989) Binding, internalization and degradation of basic fibroblast growth factor in human microvascular endothelial cells. Exp. Cell Res., 181t75-84. Brevario, F., Proserpio, P., Bertocchi, F., Lampugnani, M.G., Montovani, A,, and Dejana, E. (1990) Interleukin-1 stimulates prostacyclin production in cultured human endothelial cells by increasing arachidonic acid mobilization and conversion. Arteriosclerosis, 10(1):129-134. Coffrey, R.J., Leof, E., Shipley, G.D., and Moses, H. (1987) Suramin inhibition of growth factor receptor binding and mitogenicity in AKR-2B cells. J . Cell. Physiol., 132:143-148. Conn, G., Bayne, M.L., Soderman, P.W., Kxok, P.W., Sullivan, K.A., Palisi, T.M., Hope, D.A., and Thomas, K.A. (1990) Amino acid and cDNA sequence of a vascular endothelial cell mitogen which is homologous to platelet derived-growth factor. Proc. Natl. Acad. Sci. USA, 87t2628-2632. Conn, G., Soderman, D.G., Schaeffer, M.T., Wile, M., Hatcher, V.B., and Thomas, K.A. (1989) Purification of a glycoprotein vascular endothelial cell mitogen from rat-glioma derived cell line. Proc. Natl. Acad. Sci. USA, 87:1323-1327. Connolly, D.T., Heuvelman, D., Nelson, R., Olander, J.V., Eppler, B.L., Delfino, J.J., Siegel, N.R., Leimgruber, R.M., and Feder, J. (1989a) Tumor vascular permeability factor stimulates endothelial cell growth and angiogenesis. J. Clin. Invest., 84:1470-1478. Connolly, D.T., Olander, J.V., Heuvelman, D., Nelson, R., Monsell, R., Siegel, N., Haymore, B.L., Leimgruber, R., and Feder, N. (1989b) Human vascular permeability factor. J. Biol. Chem., 264:20172024. Cozzolino, F., Torcia, M., Aldinucci, D., Ziche, M., Almerigogna, F., Bani, D., and Stern, D. (1990) Interleukin 1 is a n autocrine regulator of human endothelial cell growth. Proc. Natl. Acad. Sci. USA, 87t6487-6491. Ferrara, N., and Henzel, W.J. (1989) Pituitary follicular cells secrete a novel heparin-binding growth factor specific for vascular endothelial cells. Biochem. Biophys. Res. Commun., 16It851-858. Frater-Schroder, M., Risau, W., Hallman, R., Gautschi, P., and Bohlen, P. (1987) Tumor necrosis factor type a,a potent inhibitor of endothelial cell growth in vitro, is angiogenic in vivo. Proc. Natl. Acad. Sci. USA, 845277-5281.

VASCULOTROPIN EFFECT ON HUVE CELLS Gadjusek, C., and Carbon, S. (1989) Injury-induced release of basic fibroblast growth factor from bovine aortic endothelium. J. Cell. Physiol., 139t570-579. Gospodarowicz, D., Abraham, J.,and Schilling, J . (1989)Isolation and characterization of a endothelial cell mitogen produced by pituitaryderived folliculo-stellated cells. Proc. Natl. Acad. Sci. USA, 86r7311-7315. Greenwood, F.C., Hunter, W.H., and Glover, J.S. (1963) '251-labeled growth hormone of high specific radioactivity. Biochem. J., 8 9 r l l P 123. Haigler, H.X., Maxfield, F.R., Willingham, M.C., and Pastan, I. (1980) Dansylcadaverine inhibits internalization of '251-Epidermal Growth Factor in Balb 3T3 cells. Biol. Chem., 255t1239-1241. Hasegawa, N., Yamamoto, M., and Yamamoto, K. (1988) Stimulation of cell growth and inhibition of prostacyclin production by herparin in human umbilical vein endothelial cells. J. Cell. Physiol., 137:603-607. Heldin, K.-H., Westermark, B., and Wasteson, A. (1981) Specific receptors for platelet derived growth factor on cells derived from connective tissue and glia. Proc. Natl. Acad. Sci. USA, 78t36643668. Jaffe, E.A., Nachman, R.L., Becker, C.G., and Minick, C.R. (1973) Culture of human endothelial cells derived from umbilical veins. J. Clin. Invest., 52t2745-2756. Keck, P.J., Hauser, S.D., Krivi, G., Sanzo, K., Warren, T., Feder, J., and Connolly, D.T. (1989) Vascular permeability factor, an endothelial cell mitogen related to PDGF. Science, 246:1309-1312. Kieffer, M.C., Stephans, J.C., Crawford, K., Okino, K., and Barr, P. (1990) Ligand affinity cloning and structure of a cell surface heparan sulfate proteoglycan that binds basic fibroblast growth factor. Proc. Natl. Acad. Sci. USA, 87t6984-6989. Kuwashima, L., Greaber, J., and Glaser, B. (1988) Stimulation of endothelial cell prostacyclin release by retina-derived growth factors. Invest. Ophthalmol. Vis. Sci., 29:1213-1220. Laemmli, V.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, 227t680-685. Leung, D.W., Cachianes, G., Kuang, W.J., Goedel, D.W., and Ferrara, N. (1989) Vascular endothelial cell growth factor is a secreted angigenic mitogen. Science, 246:13061312. Levy, A.L., Tamargo, R., Brem, H., and Nathans, D. (1990) An endothelial growth factor from the mouse neuroblastoma cell line. Growth Factors, 2t9-19. Madri. J., Pratt. B.M.. and Tucker, A. (1988) Phenotypic modulation of endothelial cells by transforming growth fact;; beta depends upon the composition and organization of the extracellular matrix. J . Cell Biol., 106t1375-1384. McNeil, P.L., Muthukrishnan, L., Warder, E., and D'Armore, P. (1989) Growth factors are released by mechanically wounded endothelial cells. J. Cell Biol., 109t811-822. Moenner, M., Gannoun-Zaki, L., Badet, J., and Barritault, D. (1989) Internalization and limited processing of basic fibroblast growth factor on Chinese hamster lung fibroblast. Growth Factors, 1t115123. Montesano, R., and Orci, L. (1985) Tumor promoting phorbol ester induces angiogenesis in vitro. Cell, 42,469437. Montesano, R., Vassalli, J.D., Baird, A,, Guillemin, R., and Orci, L. (1986) Basic fibroblast growth factor induces angiogenesis in vitro. Proc. Natl. Acad. Sci. USA, 84:7297-7301. Montesano, R., and Orci, L. (1987) Phorbol esters induce angiogenesis in vitro from large vessel endothelial cells. J. Cell. Physiol., 130:284-291. Moscatelli, D. (1987) High and low affinity binding sites for basic fibroblast growth factor on cultured cells: absence of a role of low affinity binding in the stimulation of plasminogen activator production by bovine capillary endothelial cells. J . Cell. Physiol., 131t123-130.

59

Moscatelli, D. (1988) Metabolism of receptor bound and matrix bound basic fibroblast growth factor by bovine capillary endothelial cells. J . Cell Biol., 107t753-759. Moukadiri, H., Favard, C., Praloran, V., and Plouet, J. (1990) Vasculotropin, a new angiogenic growth factor. Pathol. Biol. (in press). Neufeld, G., and Gospodarowicz, D. (1987) Protamin sulfate inhibits mitogenic activities of extracellular matrix and fibroblast growth factor, but potentiates that of epidermal growth factor. J. Cell. Physiol., 132t287-294. Pepper, M.S., Belin, D., Montesano, R., Orci, L., and Vassalli, J.-D. (1990)Transforming growth factor beta 1 modulates basic fibroblast growth factor-induced proteolytic and angiogenic properties of endothelial cells in vitro. J. Cell Biol., 111t743-755. Phillips, H.S., Haines, J., Leung, D., and Ferrara, N. (1990) Vascular endothelial growth factor is expressed in rat corpus luteum. Endocrinology, 12 7t965-967, Plouet, J.,and Moukadiri, H. (1990a)Specific binding of vasculotropin to bovine brain capillary endothelial cells: description of a radioreceptor assay. Biochimie, 72:51-55. Plouet, J., and Moukadiri, H. (1990b) Characterization of the receptor of vasculotropin on bovine adrenal cortex derived capillary endothelial cells. J. Biol. Chem., 265:22071-22074. Plouet, J., Schilling, J., and Gospodarowicz, D. (1989) Isolation and characterization of a newly identified endothelial cell mitogen produced by AtT-20 cells. EMBO J., 8t3801-3806. Pradelles, P., Grassi, J.,and Maclouf, J. (1985) Enzyme immunoassay of eicosanoids using acetylcholine esterase as label: an alternative to radioimmunoassay. Anal. Chem., 57t1170-1173. Saksela, O., Moscatelli, D., and Rifkin, D. (1987) The opposing effects of basic fibroblast growth factor and transforming growth factor beta on the regulation of plasminogen activator activity in capillary endothelial cells. J. Cell Biol., 105:957-964. Scatchard, G. (1948)The attraction of proteins for small molecules and ions. Ann. NY Acad. Sci., 51t660-672. Taylor, S., and Folkman, J. (1982) Protamin is an inhibitor of angiogenesis. Nature, 297:307-312. Tischer, E., Gospodarowicz, D., Mitchell, R., Silva, M., Schilling, J., Lau, K., Grisp, T., Fiddes, J.C., and Abraham, J.A. (1989) Vascular endothelial cell growth factor: a new member of the platelet-derived growth factor family. Biochem. Biophys. Res. Commun., 165t11981206. Vaisman, N., Gospodarowicz, D., and Neufeld, G. (1990) Characterization of the receptor of Vascular Endothelial Growth Factor. J. Biol. Chem., 265:19461-19466. Voelker, N.F. (1989) Role of platelet-derived growth factor and other growth factors in inflamation. In: Mediators of the inflamatory process. P.M. Henson and B. Murphy, eds. Elsevier Science Publisher, Amsterdam, Chap. 13. pp. 45-56. Wecksler, B.B. (1990) Heparin and acidic fibroblast growth factor interact to decrease prostacyclin synthesis in human endothelial cells by affecting both, prostaglandin H synthetase and prostacyclin synthetase. J. Cell. Physiol., 142t51P522. Williams, R.T., Trimble, P.M., Lavin, M.F., and Sunday, M.E. (1984) Platelet derived growth factor receptors from a high affinity state in membrane preparations. J. Biol. Chem., 259:5287-5294. Witte, L., Fuks, Z., Haimovitz-Friedman, A,, Vlodavsky, I., Goodman, D.S., and Eldor, A. (1989) Effects of irradiation on the release of growth factors from cultured bovine, porcine and human endothelial cells. Cancer Res., 49:1253-1260. Zavoico, G.B., Hrbolich, J.K., Gimbrone, M.A., and Schafer, A. (1990) Enhancement of thrombin and ionomycin stimulated prostacyclin and platelet acting factor production in cultured endothelial cells by tumor promoting phorbol ester. J. Cell. Physiol., 143r596-609. Ziche, M., Jones, J.,and Gullino, P.M. (1982) Role of prostaglandin E l and copper in angiogenesis. J. Natl. Cancer Inst., 69t475-482.