JOURNAL OF CELLULAR PHYSIOLOGY 147:362-373 (1991)

Culture-Induced Increase in Acidic and Basic Fibroblast Growth Factor Activities and Their Association With the Nuclei of Vascular Endothelial and Smooth Muscle Cells EDITH SPEIR,* JOACHIMSASSE, SHASHI SHRIVASTAV, AND WARD CASSCELLS Cardiology Branch, National Heart, lung, and Blood Institute, National institutes of Health, Bethesda, Maryland 20892 (E.S., S.S., W.C.1, and Shiner', Hospital for Crippled Children, Tampa, Florida U.5.) 336 12 The activity of acidic and basic fibroblast growth factor-like mitogens (aFGF, bFGF) extracted from cultured bovine aortic endothelial (BAEC) and rat aortic smooth muscle cells (SMC) was compared with that of freshly isolated cells from the same tissues. Extracts of subendothelial extracellular matrix (ECM) and cell lysates of cultured BAEC contained 4-fold more bFGF-like activity than the extracts of fresh cells. ECM and cell lysates of SMC yielded 10-fold more bFGF-like activity than the fresh cell lysates. We consistently find aFCF-like activity in both cell types. In the case of BAEC, cultured cells and ECM contained 3-fold more aFGF-like activity when compared with freshly isolated cells, whereas in cultured SMC, aFGF-like activity in cell and ECM extracts was 8-fold higher than in fresh cell extracts. The mitogens extracted from cell lysates and from the ECM are closely related to aFGF or bFGF by the criteria that they bind to heparin-sepharose and elute at 1 . I M (aFGF) or 1.5 M (bFGF) NaCI, have molecularweights of about 18,000, and react with anti-aFGF (1.1 M), or anti-bFGF (1.5M) antibodies when analyzed by Western blots and by radioimmunoassay specific for aFGF and bFGF. This mitogenic activity i s inhibited by neutralizing antibodies to aFGF and bFGF. i n addition, the column fractions are potent mitogens for Balbic 3T3 fibroblasts. Acidic and basic FGF-like mitogenic activity could also be extracted from the cell nuclei. The subcellular localization of both FGFs was visualized in both nuclei and cytoplasm with immunoperoxidase. Compared with primary SMC, secondary SMC had an increased capacity to bind '*llaFGF to high affinity receptors, while binding to freshly isolated BAEC and SMC was negligible. We conclude that FGFs are present at low levels in freshly isolated cells and that propagation in cell culture provides a stimulus for production of these mitogens.

Acidic and basic fibroblast growth factors (aFGF, bFGF) are closely related mitogens of Mr 15,00025,000 which have an amino acid sequence homology of 55% (Bohlen et al., 1985). Both polypeptides are mitogenic for the same wide variety of mesenchymal and neuroectodermal cells in vitro, and are morphogenic and angiogenic in vivo. aFGF and bFGF have been isolated from neural tissues (Gospodarowicz,1987) and from rat heart (Speir et al., 1988), and bFGF is also found in various other normal and neoplastic tissues (Gospodarowicz, 1987). These tissues are heavily vascularized, suggesting that cells derived from the vascular system might be responsible for aFGF and bFGF production. However, a high concentration of bFGF is also found in a nonvascular tissue like cartilage (Sasse et al., 1987). Thus, it cannot be presumed that vessels are the sole source of bFGF in vivo. Cultured aortic endothelial cells have been shown to synthesize bFGF (Vlodavsky et al., 19871, and smooth muscle cells in vitro express aFGF (Winkles et al., 1987) as well as bFGF mRNA, contain bFGF protein, 0 1991 WILEY-LISS, INC.

and presumably express the receptor because they proliferate in response to FGFs (Gospodarowicz et al., 1988). These and other studies, using cultured cells, indicate that FGFs may influence vascular cell function and proliferation. However, the growth of cells in culture is a poor imitation of the finely tuned growth of cells in any tissue. Extrapolation of in vivo levels from in vitro results is difficult; for example, FGF mRNAs are readily found in cultured cells (Schweigerer et al., Received August 28, 1990; accepted December 18, 1990. *To whom reprint requestsicorrespondence should be addressed. Abbreviations used: aFGF and bFGF, acidic and basic fibroblast growth factor; PDGF, platelet-derived growth factor; RIA, radioimmunoassay; BAEC, bovine aorta endothelial cells; SMC, rat aorta smooth muscle cells; ECM, extracellular matrix; hraFGF and hrbFGF, human recombinant acidic and basic fibroblast growth factor; Ab, antibodies; DiI-Ac-LDL, acetylated low density lipoprotein labeled with 1,1'dioctadecyl-l-3,3,3,3'-tetramethylindocarbocyanine perchlorate.

FGFs IN VASCULAR EC AND SMC EX VIVO VS. IN VITRO

1987a,b; Murphy et al., 1988; Winkles et al., 1987; Mansson et al., 1990) but are very difficult to detect in normal tissue. With the exception of recent studies showing the presence of mRNAs for aFGF and bFGF in whole adult rat aorta (Sarzani et al., 1989) and bFGF protein in regenerating endothelium in denuded rat carotids (Lindner et al., 1989),there is a paucity of data on local production of FGFs by vascular tissue or cells in vivo. We hypothesized that vascular smooth muscle and endothelial cells, which are normally quiescent in vivo, might produce one or both of the FGFs in response to the stimuli of propagation in culture. MATERIALS AND METHODS Isolation of bovine aortic endothelial cells Aortas were submerged in sterile, cold saline with 500 Uiml penicillin, 500 pgiml streptomycin, and 1.25 pgiml Fungizone (5% PSF),and were transported on ice from the slaughterhouse. The tissue was washed several times in sterile PBS15% PSF before the vessels were cut longitudinally. Cells were harvested with a cell scraper, using short, gentle strokes, taking care not to scrape the same surface twice (Gospodarowicz et al., 1976). Cell clumps were broken up by repeated titurating with a 2 ml pipet. Endothelial cells from 15 aortas were collected in DMEM (Biofluids), and were centrifuged, washed with PBS/l% PSF, pelleted, and resuspended in PBS. Viability was confirmed by trypan blue exclusion and cells were counted with a hemocytometer. An aliquot of the bovine aortic endothelial cells (BAEC) was grown in culture and fed every 48 hours with 1 ngiml of bFGF or 10 ngiml of aFGF (gifts from L. Cousens and P. Barr, Chiron Corp. Emeryville, CA), until confluence. Cells were passaged at 1:lOO weekly and 1 x lo7cells (passage 3-9) were dissociated 3 days after reaching confluence and 3 days after the last FGF feeding, with 0.05% trypsin/0.02% EDTA (Biofluids) in Puck’s saline and extracted as described. We used the vital stain DiI-Ac-LDL (Acetylated Low Density Lipoprotein labeled with 1,l‘dioctadecyl-1-3,3,3,3’-tetramethyl-indocarbocyanine perchlorate, Biomedical Technologies Inc., Voyta et al., 1984) to identify BAEC. At confluence, BAEC formed the typical monolayer with “cobblestone”pattern. Isolation of rat aortic smooth muscle cells Rat aortic smooth muscle cells (SMC)were harvested from 20 male Sprague-Dawley rats (200-250g, Taconic Farms) by enzymatic digestion of the thoracic aortas previously stripped of adventitia by blunt dissection and denuded of endothelium by rubbing (Owens et al., 1986). 3-5 x lo7 fresh cells were obtained. Four T-75 flasks were seeded with l o 4 cells/cm2 in growth medium (M199, Biofluids) with 10% fetal bovine serum (Biofluids) and 100 U1ml penicillini100 pgiml streptomycin. The cells were s lit after forming a monolayer at a 1:4 ratio. 2-4 x 107pfresh cells or nearly confluent cultured cells at passage 3-9 were extracted for growth factor activity. SMC were purified by “preplating” for 4-5 hours. Fibroblasts and endothelial cells attached to the plastic during that time. SMC were in the supernatant and were transferred to another dish. SMC

363

formed the typical “hill and valley” pattern at confluence. Attachment rate for SMC was 3 0 4 0 % of seeded cells. Since DiI-Ac-LDL stain was negative the cells were judged to be free of endothelial cells. Serum starved sparse SMC or EC were incubated with antia-smooth muscle-l-actin antibody HHF-35 (123000) using the ABC method (Vector Labs), immunoperoxidase and diaminobenzidine. (The antibody was a gift from Dr. A. Gown). EC and SMC treated with mouse IgG (Cappel) were negative controls. Extraction of cells and isolated nuclei and subsequent heparin-sepharose chromatography Freshly scraped BAEC or SMC isolated by collagenase, cultured trypsin-dissociated cells (1-4 x lo7) or isolated BAEC or SMC nuclei (2 x 10’) were suspended in 1 M NaC1, 0.01 M Tris pH 7.4, disrupted by 3 freezeithaw cycles followed by 3 passages through a 25 gauge hypodermic needle. Nuclei ( 3 4 x 10 were isolated from 10 (24 x 24 cm) nearly confluent dishes of BAEC (passage 5) and SMC (passage 10) as described by Soler et al. (1989). Cell or nuclear extracts were centrifuged (48,00Og,30 minutes); the supernatant was diluted with 0.01 M Tris pH 7.4 to adjust the NaCl concentration to 0.5 M and, then, batch-adsorbed overnight at 4°C to 0.2 ml heparin-sepharose (Pharmacia) equilibrated with 0.5 M NaC1, 0.01M Tris pH 7.4. The next day the batch was transferred to a disposable 11 ml column (BioRad). After washing with 10 column volumes of 0.6 M NaCl/O.OlM Tris pH 7.4, bioactivity was eluted b stepwise addition of 1.1M, 1.5 M and 3 M NaC110.01M qris pH 7.4. Six 0.5 ml-fractions were collected for each NaCl concentration and aliquots tested for stimulation of DNA synthesis in 3T3 cells. Peak fractions were tested by RIA and Western blot, as described below. Extraction of extracellular matrix (ECM) from BAEC or SMC ECM was exposed by lysing the cell layer with 0.05% Triton X-100 and 20 mM NH40H in phosphate-buffered saline (PBS), followed by 4 washes in PBS (Vlodavsky et al., 1987). FGF activity was extracted from the ECM by incubating with collagenase (50 kg/ml, CLS 2, Worthington) for 2 hours at 37°C in PBS (Vlodavsky et al., 1987). Insoluble material was removed by centrifugation (48,OOOg)for 30 minutes. The extracts were batch-adsorbed to heparin-se harose (0.5 ml) equilibrated with 0.5 M NaC1lO.O1 M h i s pH 7.4, then processed as described for cell lysates. The small contribution of FGFs adsorbed to the ECM from the cell lysate has been found negligible (Vlodavsky et al., 1987). Bioassays for FGF activity aFGF and bFGF growth factor activities for 3T3 cells were measured as described (Hauschka et al., 1986). Briefly, Balbic 3T3 murine fibroblasts, clone A 31 (American Type Culture Collection) were grown to 60-75% confluence in 75 cm2 flasks (Corning) in DMEM with 10%calf serum (Colorado Serum Co.), and 1%PSF, in a humidified 37°C incubator with 10% C02. Cells were washed with 10 ml of Puck‘s saline, dislodged with 5 ml of 0.05%trypsin-EDTA in 5 ml Puck’s,

364

SPEIR ET AL

and either split at a ratio of 1:lOOO or plated in 96-well dishes at 10,000 cells/well in DMEM/10%calf serum for the mitogen assays. Cells in the 96-well plates formed confluent monolayers after 3 days and were maintained without medium change for 8-11 days. These quiescent cells can be stimulated by serum or FGFs t o resume DNA synthesis and are extremely sensitive to both aFGF and bFGF, with bFGF consistently showing 2-4 times higher activity than aFGF. Crude extracts, pure growth factors, or column eluates (0.5-lpl) were added to the 10-day old cells once, followed by 1pCiiwell of L3HIthymidine (6.7 Ciimmol, American Radiochemical Corp.). Cells were harvested 48 hours later using a PhD brand cell harvester (Waltham, MA). A unit of 3T3 cell growth factor activity was defined as the amount of growth factor in 0.2 ml required to stimulate half maximal DNA synthesis in 3T3 cells. Serum was used as standard and human recombinant aFGF (hraFGF) or hrbFGF were tested as positive controls. In our culture conditions, 0.5 ng hrbFGF or 2 ng of hr aFGF equalled 1 serum unit.

10-fold in Centricon filter units (10,000 Mr exclusion, Amicon).The filters were pre-soaked overnight in 0.2% gelatiniPB 0.05% Tween, washed with PBS before the sample was added, then centrifuged according t o manufacturer's instructions. Concentrated samples (14 pl), 5-50 ng hraFGF, hrbFGF, or prestained Mr standards (Enprotech), were subjected to NaDodS0,-SDS-PAGE, using 10-20% gradient minigels (Enprotech) at 35 mA for 3 hours. After transfer of electrophoresed proteins to 0.05 pm nitrocellulose (Schleicher and Schuell) in a Polyblot (American Bionetics) for 60 minutes, the antigens were visualized using the following peptide antisera: anti-aFGF (50-82) (SIG . . . NEE), antibFGF (3343) (LRI . . . DGV) or IgG-purified antibFGF (1-24) (PAL . . . RLY), 967 (raised against human recombinant bFGF), gifts from Dr. A. Baird, and anti-bFGF monoclonal 78 (a gift from Takeda), followed by biotinylated goat anti-rabbit IgG. Streptavidin biotinylated horseradish peroxidase complex and 4-chloro1-naphthol substrate was used as described (Klagsbrun et al., 1986). Immunohistochemistry Competitive Radioimmunoassays (RIA) For quantification of bFGF we used [lZ51]human Freshly scraped BAEC or enzyme-dispersed SMC recombinant bFGF as a ligand and anti-bFGF IgG "cat" (1-3 X lo3) in a 5-10 pl PBS drop were allowed to (a gift from Dr. D. Gospodarowicz)as the binder accord- sediment for 1 hour on a glass microscope slide which ing to the published method (Schweigerer et al., 1987~). had been coated with 1%3-aminopropyl-triethoxysilan The RIA for aFGF was developed in our laboratory (TESPA, Fluka Co.) in toluene as described by Berger (Shrivastav et al., in preparation) with antisera raised (1986). Cells attached by Cytospin to glass alone against a synthetic peptide corresponding to aFGF washed off, whereas those on TESPA did not. The glass residues 50-82 (SIG . . . NEE) conjugated to keyhole slides with cells attached were immersed into 4% limpet hernocyanin, (J. Sasse et al., in preparation). paraformaldehyde in PBS for 20 minutes at room Four micrograms of aFGF in 8 p1 of 10 mM Tris pH 7, temperature (RT), then washed 3 x 5 minutes with 20 pg of heparin sodium (Upjohn) and 30 pl of water PBS. After dehydrating the cells in (5 minutes each) were added to a glass test tube coated with 2 pg of 30%, 60%, SO%, 95%, and 100% ethanol at RT, the Iodogen (Pierce) in methylene chloride. After that, the slides were air dried, stored at 4°C for 1-2 days, and procedure follows the one published for the bFGF RIA processed as described below for immunostaining. (see above). RIAs were performed in solution, at equiSparse BAEC or SMC (passage 3-10) rown in librium, carried out in duplicate. Cross-reactivity with 2-chamber culture dishes (LabTek) were was ed 2 x 5 hrbFGF in the aFGF RIA or with hraFGF in the bFGF minutes with PBS at RT, washed 3 x 5 minutes with RIA was 1%in each case. Binding of tracer in the 0.1 mM CaClz and 1 mM MgClz in PBS" (hereafter absence of primary antisera was 1%.Sample volumes referred to as PBS") at RT, then ermeabilized in 0.2% were P l o p 1 of peak heparin-sepharose fractions. In saponin/PBS* for 5 minutes at R . Following 3 washes some cases, all fractions from the HS column were with PBS" for 5 minutes each, cells were incubated tested by RIA. Peak RIA activity coincided with peak with 1%HzOziwater for 30 minutes at RT, washed 1 x 5 minutes with water and 2 x 5 minutes with 3T3 mitogenicity. PBS*, then incubated with 5% normal goat serum Immuno-neutralizationwith anti-aFGF or (NGSIIPBS for 30 minutes at RT. After a 5-minute anti-bFGFAntibodies (Ab) wash with PBS", cells were incubated overnight with Forty micrograms of bovine anti-aFGF Ab (R&D)or primary antibodies (Ab)diluted in 2% NGS, 1%normal 4 pg of anti-bFGF "dog" Ab (a gift from Dr. D. Gospo- rat or bovine serum. The following antibodies were darowicz) selectively blocked bioactivity of 0.5 ng used: R119 anti-aFGF IgG, residues (50-821, or resihraFGF or O.ln of hrbFGF, respectively, in Balbic3T3 dues (7-161, raised by J. Sasse; monoclonal anti-aFGF cells by 90%, w ile Ab alone had no effect. The same which recognizes the region 111-126 of bovine aFGF amounts of anti-aFGF or anti-bFGF IgG neutralized (purchased from Upstate Biotechnology Inc., Lake the mitogenic activities of 1 p1 of 1.1 M or 1.5 M NaCl Placid, NY). For immunolocalization of bFGF we used HS column eluates from lysates of smooth muscle cells anti-bFGF (1-24) and 967 (the latter raised against or endothelial cells. Neutralizing anti-bFGF or anti- human recombinant bFGF, both gifts from Dr. Baird), aFGF was previously shown to block the activity of and anti-bFGF monoclonal 78 which recognizes the exogenously added bFGF (Gospodarowicz et al., 1988) NH2-terminus region AA 1-9. After incubation with appropriate secondary antibodies, immunoreactivity or aFGF. was visualized using Vector AB complex for 10 minutes Western blot analysis at RT, followed by 3 washes with PBS and staining with Aliquots of 0.5-1 ml of peak fractions eluted from DAB for 10 minutes at RT. Finally, the stained cells heparin-sepharose were desalted and concentrated 5- were washed for several minutes with tap water. The

a

IF

R

FGFs IN VASCULAR EC AND SMC EX VIVO VS. IN VITRO

chambers were removed and the cultured or freshly isolated cells on slides were dehydrated and mounted with coverslips. Cross-sections (6 pm) of rat aorta from erfusionfixed animals (10% formalin) were prepare $ and immunostained as described for tissue sections (Casscells et al., 1990). To test for nonspecific staining the following controls were used. A: Immune IgGs (2 pgiml) readsorbed with human recombinant bFGF (30 pg/m) in the case of antibodies to the whole molecule (967, “dog”, “cat”),or with excess peptide for (7-161, and (1-24); antisera (1-10 pgiml) were incubated for 24 hours at 4°C in presence of 1%crystalline BSA in PBS, then centrifuged at 16,OOOg for 30 minutes and passaged through a small heparin-sepharose column (0.1 ml bed volume for 500 ~1 incubation mixture). B: non-immune rabbit antisera or mouse IgG (Cappel) at identical protein concentrations to the immune I Gs. C: Omission of antibodies; D: [l-241 b#GF or 150-821 antiaFGF epleted antibodies from the anti [1-24lbFGF or anti[50-82] aFGF IgG purification on Affigel (Pierce Chemical Go), obtained from Drs. A. Baird and J. Sasse.

P

365

were trypsinized and the number of cells per culture was determined with a hemocytometer. BAEC were grown from isolated colonies (Gospodarowicz et al., 1976) to near-confluence, kept in serum-free DMEM, 1%ITS prior to the binding assay, then processed as described for SMC. In some experiments, cells were removed from the dishes by incubation in Pucks saline for 15 minutes (8 g NaCl, 0.4 g KC1, 1 g glucose, 0.35 g NaHC03, EDTA 0.2 gb). The cells were pelleted by centrifugation ( 7 0 0 10 ~ minutes, VC), resuspended in binding buffer at 10 cellsiml, and used for binding assays as described for freshly isolated cells.

Single cell autoradiography Sparse cultures in Nunc 4-chamber slides were rinsed twice with binding buffer (M199,25 mM Hepes, 0.15% gelatin) and incubated for 1 hour at 22°C with 200 pM ‘“1-aFGF in the presence or absence of 200-fold excess aFGF. Some dishes were incubated with C3Hlthymidine (American Radiochemical Co.) 6.7 Ciimmol, 2 uCi/ml for 8 hours. The chambers were rinsed four times with ice-cold PBS, fixed with 2% glutaraldehyde in PBS at 4°C for 2 minutes, then rinsed 10 times with distilled HzO, dried, coated with Kodak NTB2 emulBinding of IZ5I-aFGFto intact BAEC or SMC sion, and exposed for 10 days at 4°C (Olwin and Freshly isolated BAEC or SMC were washed in cold Hauschka, 1988). PBS and pelleted by centrifugation (700g, 10 min, 5°C). This wash was repeated twice. The final pellets were Cross-linking of 1251-aFGFto cells resuspended in binding buffer (M199, 25 mM, Hepesd Subconfluent BAEC or SMC in monolayers were 0.15% gelatin; Moscatelli and Quarto, 1989) at 2 X 10 cellsiml. The rat SM cells were incubated with Iz5I- prepared as described by Moscatelli and Quarto (1990). aFGF at saturating concentrations (0.8 nM, Hoshi Briefly, 1-2 x lo6 cells were kept serum-free for 24 et al., 1988) in the presence or absence of 500-fold hours, washed twice with cold 2 M NaCl in 20 mM excess of nonradioactive bFGF for 2 hours in a shaking sodium acetate, incubated in DMEM or M199 (25 mM ice bath. The cells were then pelleted by centrifugation Hepes, 0.15% gelatin) and 10 ngiml Iz5I-aFGF.After 2 (1minute) in a Beckman microfuge, the supernatant hours at 4”C, the medium was replaced with PBS condecanted, and the pellet rinsed three times with ice-cold taining 1 mM bis (sulfosuccinimidyl) substrate (Pierce PBS, then counted in a gamma counter (Orlow et al., Chemical Co.) and 10% DMSO. After a 15-minute 1990). Pellets were resuspended in 2 M NaC1, centri- incubation at 22”C, cells were extracted with Laemmli fuged, and the supernatant decanted. This was re- buffer, and the extract was heated in a boiling water peated once. The pellets were then resuspended in 2 M bath for 2 minutes. Samples were subjected to SDSNaCl in 20 mM sodium acetate, pH 4, centrifuged, the PAGE on a 7.5% resolving gel (Enprotech). The gels supernatant removed, and this step was repeated once. were dried and exposed to Kodak X-MAR (Eastman All the supernatants as well as the pellets were counted Kodak) film. (final volume 1 ml). Primary SMC (lo4 cells/cm2)were RESULTS seeded in 6 cm dishes (Falcon) in M199, 10% fetal Characterization of BAEC and SMC in Culture bovine serum, 1%antibiotic-antimycotic, (Biofluids) at 37”C, 5%C02.After 48 hours the cells were given fresh The isolation of BAEC was done according to Gospomedium, which was replaced after 24 hours with se- darowicz et al. (1976). When confluent, BAEC always rum-free medium (M199 with 1%insulin, transferrin, exhibited characteristic “cobblestone” morphology. selenous acid, from Collaborative Research, Bedford (Fig. 5k). BAEC were pro agated by passaging confluMA)). Seventy-two hours after seeding, the cells were ent flasks weekly at a Iilution of 1:lOO. bFGF was washed twice with cold PBS, once with binding buffer, added every other day (1 ngiml); cells were grown in and incubated in monolayer with 0.8 nM Iz5I-aFGF in 24 x 24 cm culture dishes. binding buffer, as described above. Each dish contained In a separate experiment BAEC were grown without approximately 2 x lo6 cells. Cells were harvested with FGF and passaged at 1:3, then harvested and processed 2 ml Pucks saline per dish and gentle scraping. as described in the Methods section, There was no Secondary SMC (passage 8-12) were plated at lo4 difference (in subsequent assays for FGF content) becellskm’ for 48 hours in growth medium, then kept tween results from BAEC grown with or without FGF. serum-free in M199, 1%ITS for 24 hours prior to the In another experiment we added lZ5I-aFGFor bFGF to binding assay. Cultures were then washed twice with cells 3 days before confluence and harvesting, which cold PBS, equilibrated in binding buffer for 15 minutes, was the time of the last FGF feeding for the cells then incubated with 0.8 nM Iz5I-aFGFin binding buffer maintained with growth factor. In crude cell lysates as described for primary cultures. Parallel cultures and ECM we recovered less than 10%of labeled FGFs.

primardr

366

SPEIR ET AL

W I

z X

5 a

u

Fractions

A.

Rat Aorta Smooth Muscle Cells (7 x lo7)

Fig. 1. Heparin-sepharose affinity chromatography IHS) of freshly isolated rat aorta smooth muscle cells (A) or bovine aorta endothelial cells (B),analyzed for content of aFGF (1.1M NaCl peak) or bFGF (1.5 M NaCl peak) by 3T3 mitogen assay. (A) 7 x lo7 enzyme-dispersedrat aorta SMC were batch-absorbed overnight on HS, transferred to a disposable 10 ml BioRad column, and eluted with a step-gradient of

We conclude that yields of FGFs from cells including nuclei or ECM were not significantly “contaminated” with exogenous FGF (not shown). SMC grew t o the typical hill and valley pattern at confluence (Fig. 6i). Immunoperoxidase reaction product to anti-a-smooth muscle-1-antibodyHHF-35 (1:8000) is clearly seen in SMC but not in endothelial cells from rat aorta. Omission of primary antibody or incubation with mouse IgG (1:8000) gave negative results (not shown).

6.

Fractions Bovine Aorta Endothelial Cells ( 1 x 107)

NaCl as shown. (B) 1 X lo7 freshly scraped bovine aortic endothelial cells (BAEC) were gently scraped from 5 bovine aortas, lysed, and processed, as described in Methods. BAEC lysates contain roughly 10-fold more activity than SMC lysates. Aliquots of column fractions (0.5-1 ~ 1 were ) added in triplicate to quiescent BalbicJT3 cells in 96-well plates as described.

The mitogen peaks were well separated and RIA and Western blotting indicate little contamination of the 1.1 M fractions with bFGF or of 1.5M fractions with aFGF. Thus, it was possible to estimate the amounts of aFGF and bFGF. Further assurance is given by the better than 80% inhibition of mitogenicity in the 1.1M fractions by anti-aFGF blocking antibody (R&D, #P,D,j and of the 1.5 M fractions by anti-bFGF “dog” (Fig. 3). Extracellular matrix from cultured cells (BAEC or SMC) was extracted with collagenase and the soluble portion processed as described under Methods. Elution Growth factor activity in cell and nuclei lysates profiles are shown in Figure 2a,b. Heparin affinity chromatography (HS) and 3T3 miWhen isolated nuclei from BAEC were lysed and togen assays were used to analyze the growth factor subjected to heparin-Sepharose chromatogra hy, they content of freshly isolated rat aorta SMC, or bovine contained approximately 3 ng of aFGF an(P 2 ng of aorta EC (BAEC), of monolayers of SMC or BAEC bFGF-like activity (6% and 5% of the cell-associated grown in culture and of 2 x 10’ isolated nuclei (chro- activity). The values for cultured SMC nuclei are matographic profile and the high bioactivity of nuclei approximately 2 ng of aFGF and 2 ng of bFGF or 16% extracts are not shown). The monolayers were sepa- and 25% of the cell-associated activity (Table 1).The rated into cell lysates and extracellular matrix, as nuclei were intact and not contaminated with memdescribed in Methods. Mitogenic activity for 3T3 cells brane. This was verified by ultrathin-section electron eluted from the columns as two peaks, one at 1.1M, the microscopy (not shown). other at 1.5 M NaC1. The 3T3 assays of freshly isolated Radioimmunoassays (RIAs) SMC and BAEC are shown in Figure la,b; those of cultured SMC or BAEC are shown in Figure 2a,b. The RIAs were performed using 2-10 ~1of peak fractions panels depict mitogenicity of a) cell lysates, and b) after heparin-affinity chromatography (HS). In some extracellular matrix extracts. The number of fractions cases, all eluates from HS were subjected to RIAs for is shown on the abscissa (0.5 ml each). RIA results aFGF (1.1 M) or bFGF (1.5 M NaC1) (Fig. 2a,b). Peak ex ressed as fmole/pl are on the ordinate to the right RIA activity coincided with peak 3T3 mitogenicity. si e of the profiles. Reproducibility of the RIA in Results are compiled in Table 1. Cultured BAEC intra-assay experiments is within 2 10%; inter-assay yielded 2-3 times as much aFGF and 4 times as much variations are ? 15%.Values shown are means of 5-6 bFGF as fresh BAEC. The ECM alone contained almost determinations. Details of the RIA procedure will be as much aFGF as the lysates from freshly isolated cells published separately (Shrivastav et al., in preparation). and about half as much bFGF. Clearly, there is more

B

367

FGFs IN VASCULAR EC AND SMC EX V N O VS. IN VITRO 0.61.11.5

3M 1.61.1 1.5 3M

c+

+

4

ECM 5 pl

Cell Lvsate 1 PI sam

10 I

30

20

10

b

\

1

5

5

a

b

-

5

5

a

A Rat Aorta Smooth Muscle Cells P, (3x lo7)

b

B Bovine Aorta Endothelial Cells (5 x lo7 cells) a1 Cell lysate (1 pl/well) b ) Extracellular Matrix Extract (2 pl/well) RIA for bFGF: lmmunoreactivity coincides with Mitogenicity

Fig. 2. A Heparin-sepharose chromatography and mitogen assays of lysates of confluent cultures of SMC and extra-cellular matrix (ECM). (a) 3 x loT cells were harvested, and cell and matrix extracts separated, as described in Methods. The column was eluted stepwise with NaCl/O.OlM Tris as shown. Column fractions (1FUwell) were tested in triplicate using the 3T3 bioassay (0-0).(b) ECM extracts (2 pl/well) were plated in triplicate; mitogen activity fin cpm x and immunoreactivity (ir) in fmoleipl are plotted in the same graph (ordinate); aFGFir (A-A) and bFGF in ( e 4 ) peaks coincide with peak mito-

genicity. Fifteen (0.5 ml) fractions were collected (abscissa). Intra- and inter-assay varation were 5 10% and 2 15%, respectively. B Heparin-sepharose chromatography (HS) and mitogen assays of confluent bovine aorta endothelial cells and their ECM. (a) 5 x lo7 cells were processed as described for SMC; 1 ~1 aliquots were plated in the 3T3 bioassay, (0-0). (b)activity of ECM extracts (0-0).RIA for both (a) and (b) was measured in 6 pl aliquots. Assays based on mitogenicity and immunoreactivity for both aFGF and bFGF are in close agreement.

anti-bFGF Ab “dog”

anti-aFGF Ab R&D

m

4c

I

z X

E

6

2c

At:

1 - +

-

Control

aFGF 0.5 ng

b) Extracellular Matrix Extract

a) Cell Lysate

+

bFGF 0.1 ng

n + 1.1M BAEC

- + 1.1M SMC

bFGF 0.1 ng

aFGF 0.5 ng

n. -

+

1.5M BAEC

-

+

1.5M SMC

Fig. 3. Immuno-neutralization of mitogenic activity for Balbic 3T3 cells. Bovine anti-aFGF IgG (40 Fg) or human anti-bFGF IgG “dog” (4 pg! block 0.5 ng recombinant aFGF or 0.1 ng rbFGF, respectively. Furthermore, 40 ug of the same anti-aFGF IgG or 4 pg ofthe anti-bFGF IgG inhibit the mitogenic activity of 1 p1 (HS) aliquots from aortic endothelial cell lysates (BAEC) or smooth muscle cell lysates (SMC).

368

SPEIR ET AL. TABLE 1. Cells in vitro contain more PGFs than in vivol Bovine aortic EC

aFGF RIA Fresh cells Culture ECM Nuclei

bFGF

3T3

RIA

24 45 15

3

3173

RIA

15

10

44

42 8 2

12 40 9

19

Rat aortic SMC aFGF bFGF 3T3 RIA 3T3 2 12 5 2

2

9 6

1 8 5 4

1

7 3

‘TheamountsofaFGFandbFGFareestimatedfrom3T3cellgrowthfactorunits. Oneunitis theamountofmitogenin0.2ml required t o stimulate halfmaximal DNA synthesis (about50,000-70,000cpm/well in a 96-well plate). The sensitivity of 3T3 cells is approximately 2 n g h l hraFGF per 3T3 unit and 0.5-1 ng/ml hrbFGF per unit. RIA values are in close agreement. Values are given as ng/106 cells or nuclei.

cell-associated aFGF and bFGF in the cultured endothelial and smooth muscle cells than in the fresh cells. Both fresh and cultured endothelial cells do contain aFGF, and aFGF is found in the ECM. The ratio of extracellular to intracellular aFGF content is even higher for aFGF than for bFGF. Finally, these data indicate that endothelial cells contain more aFGF and bFGF than smooth muscle cells in vivo, and also in vitro, at least under these conditions (Figs. 1, 2). However, in this study we can not directly compare in vivo vs. in vitro amounts of matrix, as there are surely differences in the amounts of matrix in vivo vs. in vitro; furthermore, scraping may not remove all the intimal matrix and some FGFs may be solubilized by the long collagenase and elastase incubation necessary to free the smooth muscle cells.

Immunodetection of BAEC, SMC, their ECM and nuclei-derived mitogens (Western blots) When heparin-sepharose purified cell-, ECM-, or tissue-derived growth factors were subjected to SDS 10-20%-PAGE and subsequent electrophoretic transfer to nitrocellulose, aliquots of the 1.1M or 1.5 M fractions showed bands comigrating with human recombinant aFGF or human recombinant bFGF (155 amino acids each) with a molecular weight of approximately 18,000 Mr. Antibodies to synthetic peptides corresponding to residues (50-82) and (7-16, not shown) for aFGF and (1-24) or (33-43) for bFGF reacted with Mr 18000 bands obtained from PAGE-transfers of 1.1 M NaCl (aFGF) or the 1.5 M NaCl (bFGF) fractions from lysates of freshly isolated cells or cultured cells and their ECM to nitrocellulose (Fig. 4a-d). Similar results were obtained with anti-bFGF “dog”, anti-bFGF 967, and monoclonal anti-bFGF 78 (not shown). Cross-reactivity with bFGF was 1% when using anti-aFGF (50-82), 2% in the case of anti-aFGF (7-16). Monoclonal anti-aFGF had no detectable binding to bFGF. Anti-bFGF (1-24) at a dilution of 1:10,000 had less than 1%cross-reactivity with aFGF; antibFGF (3343) had 2%. Anti-bFGFs “dog” and 967 both showed 1%cross-reactivity. Nuclear extracts reacted with anti-aFGF (50-82), showing a doublet at approximately 16 and 18 kD and co-migrating with the human recombinant aFGF standard. There is a faint band at about 36 kD which could be a dimer. This band appears also in Western blots of cell extracts, even though the gel is run under reducing conditions. The 1.5 M fractions showed bands co-mi-

grating with the human bFGF standard. Cell extracts were loaded with 0.5-1 pg protein (maximal sample size 15 11.1).Nuclear samples had 2-4 pg protein/l5 ~1 concentrated sample (not shown). When 10’ isolated nuclei of BAEC or SMC were extracted with SDS and analyzed by Western blotting (without heparin-sepharose chromatography), anti-aFGF (50-82) and anti-bFGF “78” gave clear bands comigrating with the FGF Standards at around 17 kD, with faint bands at 36 kD. In addition to the 17 kD band, anti-bFGF (1-24) had high molecular weight cross-reacting bands, but we used immunopurified (1-24 j for the immunostaining and crude antisera for the Western. Anti-aFGF (7-161 also had cross-reacting high molecular weight bands and the 17 kD band. Subcellular immunolocalization of aFGF and bFGF in BAEC and SMC When freshly isolated BAEC or SMC (Figs. 5a-c; 6a-b) were examined by immuno eroxidase, little reaction product was detected in BA C and in SMC. This confirmed the biochemical data from fresh cell lysates. In contrast, BAEC grown in culture for 24 hours to 48 hours showed dramatic staining in the nucleo lasm and weak staining in the cytoplasm for !FGF (Fig. 5e,h,i). The intensity of the stain was dependent on cell density: peroxidase reaction product consistently was localized more intensely within nuclei and cytoplasm of sparse cultures and was much weaker in confluent cultures, despite the latter’s thicker (less flattened) cytoplasm (not shown). Senescent BAEC, which did not incorporate [3H]thymidine (by autoradiography with NTB2 emulsion, not shown), consistently showed lack of nuclear staining (Fig. 5h, cell in upper right). Staining for aFGF showed nucleolar patterns and an even distribution in the cytoplasm (Fig. 6e,h). Both acidic and basic FGF were present in lysates of isolated nuclei of BAEC and SMC, as shown by Western blot (Fig. 4e,f ). In other cell types, evidence of nuclear FGF has recently been obtained by immunocytochemistry and by Western blotting (Renko et al., 1990; Speir et al., in preparation). SMC in secondary culture (passage 8-12) were in a state of rapid growth after 36 hours in M199,10% fetal bovine serum; they continued to synthesize DNA during the subsequent 24 hours in serum-free medium. Immunoperoxidase staining was intense, and diffuse throughout the cytoplasm and nucleoplasm; but some cells did not have nuclear staining (Fig. 6e). The

E

FGFs IN VASCULAR EC AND SMC EX VIVO VS. IN VITRO

BAEC OR SMC anti aFGF !D@ ' 1 2 3 4 5 6 7 '

anti bFGF 33-43

' 1 2 3 4 5 6 7 8

55-

433629-

-

18.4 12.4 -

B

A anti-aFGF (50-82)

ri-i-3

18.4

anti-bFGF (1-24)

m

-

C

D

Fig. 4. Western blot analysis of confluent BAEC or SMC lysates and extracellular matrix (ECM). 1.1 M or 1.5 M column fractions were concentrated 5-fold and subjected to PAGE, as described. Lanes A1 or B1 contain 10 ng of human recombinant aFGF or bFGF. SMC lysate (A2)or ECM extracts (A3, A4), as well as BAEC lysates (A5, A6) and ECM (A7) gave bands in the expected molecular weight range when blotted with anti-aFGF (50-82) or (lanes B1-8) with anti-bFGF (33-43), both at 1:lOOO dilutions. Samples were applied in duplicate; SMC lysate: B2, B3; SMC ECM: B4,B5;BAEC lysate B6, B7; BAEC ECM: B8. Identification of aFGF In 1.1M NaCl (C2, C3) and bFGF in 1.5 M NaCl (Dl,D2) HS eluates of isolated BAEC or SMC nuclei. C1 or D3: 20 ng of aFGF or bFGF. C2 or D2: BAEC (2 pg) C3 or D1:SMC (4pg protein). Western blot analysis of freshly isolated BAEC (lo7!or SMC (7 x lo7)yielded barely detectable levels of a FGF and basic FGF (not shown).

pattern was similar for both aFGF and bFGF distribution (Fig. Gd,e,g,h). SMC in primary culture were kept in growth medium for 48 hours, then placed in serum-free medium for 24 hours. These cells did not synthesize DNA without serum as evidenced by lack of [3H]thymidine incorporation (autoradiography with NTB2 emulsion, Fig. 7a). Durin the first 36 hours after plating, primary cells forme small colonies of a few rounded cells, a fairly homogeneous cell population with slow growth, unlike the thin, flat, morphologically heterogeneous cells of the passaged SMC which grow rapidly. In the primary

%

369

SMC, FGFs were sometimes localized in the cytoplasms, other times in the nuclei or in the nucleoli (Fig. 6d,e), whereas in the passaged SMC, FGFs were always seen in the cytoplasms and in the nuclei (Fig. 6g,hj. In control experiments, cells were incubated with anti-FGF antibodies preabsorbed with FGF or with the peptides used to generate the antisera. We also used non-immune IgGs as well as anti-bFGF or antiaFGF-depleted antibodies (the column flow-through from affinity purification of the anti-FGF IgGsj. Some of these negative controls are shown: BAEC and SMC treated with antisera that had been preadsorbed with antigen (Fig. 5a,d,g) and (Fig. 6c,f 1. The growth phase of primary and secondary SMC during the 60 hours following plating was determined by f3H1thymidine incorporation and visualized by NTB2 photographic emulsion autoradiography. Primary cultures were slow to begin proliferating and by 60 hours still have a low labeling index (< 10%). In contrast, 50% of the cells in secondary cultures have labeled nuclei (Fig. 7a,b) after a single &hour exposure, FGF Receptor Expression To analyze the ability of freshly isolated SMC as well as primary and secondary SMC to interact with FGF, receptor-binding experiments with 1251-aFGFwere performed. The possibility that receptors were removed by extensive collagenase incubation of aortic tissue during the cell isolation procedure cannot be excluded. We wanted to compare primary cultures to passaged cultures. Explant cultures contain tissue clumps, as well as cell clumps concentrated around the explants which makes these primary cultures unsuitable for binding studies. However, binding studies were performed with normal intact rat aortic tissue: there is no specific binding of 1251-aFGFto normal rat aorta (Casscells et al., submitted). Moreover, results were similar in freshly isolated BAEC, which were not exposed to enzymes but harvested by scraping the aortas. lZ5IaFGF binding was negligible in both types. In contrast, we found high affinity binding of FGF to BAEC or SMC in secondary culture for 48-60 hours (approximately 10 or 12 fmoli106 cells); primary SMC bound 3 fmol/106 cells (Fig. 8a, bar 2). Binding sites for 1251-aFGFin secondary SMC were visualized by single cell autoradiography (Fig. 7c,d), whereas primary SMC showed background signal only (not shown). Receptors were cross-linked by incubation of SMC with 10 ng/ml 1251-aFGFfor 2 hours at 4°C in presence (cold competitor) or absence of 200-fold excess unlabeled FGF, and then the cross-linking agent bis(sulfosuccinimidy1) suberate was added for 15 minutes at 22°C. When extracts of SMC were analyzed by electrophoresis followed by autoradiography, only secondary SMC showed high molecular weight bands at approx 150 kD and 170 kD (not shown). This is in agreement with the molecular weight of these complexes published by Neufeld and Gospodarowicz (1985). Results of 1251-aFGFbinding to cells in suspension were essentially the same (not shown). DISCUSSION The results of this study indicate that cells in normal, non roliferatin adult tissues such as aortic intima and me ia contain ow levels of aFGF and bFGF. This is

cf

Y

370

SPEIR ET AL.

Fig. 5. Localization of aFGF with anti-aFGF (50-82) in cultured bovine aortic endothelial cells to nuclei and cytoplasm (F).Identification of bFGF in cultured BAEC with anti-bFGF (1-24) (E) or 967 (H, I): there is intense nuclear and pale cytoplasmic staining: little stain is seen in freshly isolated cells with 967 (B) or (1-24) (not shown). Immunostaining with anti-aFGF (50-82) gave negative results in freshly isolated BAEC (0.Anti-bFGF 967 IgG preadsorbed with human recombinant bFGF (A, G) or anti-bFGF (1-24) bFGF-depleted IgGs did not stain freshly isolated (not shown) or cultured BAEC (D). Senescent “giant” or binucleated BAEC consistently showed lack of staining for both FGFs (shown is one sample: (H) cell in upper right corner). Sparse BAEC were identified with uptake of DiI-Ac-LDL (J) and cobblestone mor hology at confluence. (K).Magnification: (A-C and G-I) x 600; (D-f x 300; J or K: x 250 or x 50).

Fig. 6. Localization of aFGF or bFGF in SMC (primary culture: C-E; passage 12 SMC, F-HI. Shown are results using anti-aFGF (50-82) IgG (E,H) and anti-bFGF (1-24) IgG, (D,G) 2 pg/ml, and their respective controls (C,F) (immune IgGs preadsorbed with excess peptide used to generate the Abs): C (50-82) F (1-24). SMC in primary culture showed mostly nuclear and cytoplasmic staining (not shown) for aFGF, but occasionally intense staining was seen in cytoplasm and nucleoli (E). Passaged SMC consistently showed intense staining for aFGF in cytoplasm and inconsistently in the nuclei (H). bFGF was localized in nuclei and cytoplasm (GI. At confluence, SMC showed the typical “hill-and-valley” morphology (I).Cross-sections (6 p.m) of rat aorta were reacted with anti-aFGF (50-82): J; or anti-bFGF (1-24): K. Freshly isolated SMC show lack of immuno-reactivity with both anti-aFGF and anti-bFGF IeGs: Ail-24): R.i50-821. whereas faint color is seen in the media-of tissue sections of rat aorta (J, HI. Magnification: (A, B, E) x 600; (D, F, G) x 160x; (H) x 300; (I) x 50 (J,K) x 133.

based on extraction of freshly isolated aortic endothelial cells (BAEC) and smooth muscle cells (SMC) by established methods (Vlodavsky et al., 1987) and analysis by 3T3 mitogenicity, RIA, and Western blotting. By contrast, using the same techniques, much higher amounts of FGFs were detected in aliquots of the same BAEC and SMC population grown to confluence in cell

culture. These findings are supported by immunocytochemistry of freshly isolated and cultured BAEC and SMC, which shows little stain in freshly isolated cells, but intense stain in cultured cells. In fact, amounts similar to those present in fresh cell lysates were deposited in the extracellular matrix alone of the cultured cells. An even larger proportion of growth

371

FGFs IN VASCULAR EC AND SMC EX VIVO VS. IN VITRO

Fig. 8. Binding of '"I-aFGF to intact BAEC or SMC. High affinity binding to freshly isolated BAEC or SMC (A1 or B1)was negligible. Primary SMC bound 3.8 fmoli106 cells (A2) while secondary SMC (passage 12i or BAEC (passage 4) bound 9 or 12 fmol!106 cells.

Fig. 7. ["Hlthymidine autoradiography with Kodak NTB2 emulsion. After 60 hours in culture, primary SMC synthesized little DNA in serum-free medium (A) whereas secondary SMC (passage 12) showed extensive labeling (B). '"I-aFGF autoradiography of secondary SMC. Cells were plated a t 5000 per cm2 in Nunc 4-chamber glass slides. After 24 hours, the medium was changed to serum-free M199 for an additional 24 hours. SMC were incubated for 1 hour with 10 ng!ml '251-aFGF a t 22"C, rinsed with PBS, fixed with 2%'glutaraldehyde at 4"C, dried, coated with NTB2 emulsion and exposed for 10 days at 4°C. Brightfield photography x 300 shows SMC colonies (C); with darkfield photography labeling of receptors is apparent (D).

factors might be sequestered in the matrix found in vivo. It has been shown that BAEC synthesize and deposit bFGF in culture, but aFGF has been localized to ECM in some studies (Baird and Ling, 1987; Weiner and Swain, 1989) but not in others (Vlodavsky et al., 1987). No explanation for these contrasting findings is apparent, since Vlodavsky et al. also used BAEC grown to confluence without bFGF and extracted and tested in similar fashion. It is possible that some of the mitogenicity eluting in a single broad peak between l M and 2M NaCl contained aFGF, or that anti-aFGF antisera with the necessary sensitivity were not widely available, or that their clone of 3T3 cells was unusually unresponsive t o aFGF. Also, aFGFs greater sensitivity to proteolysis (at least by thrombin, Lobb, 1988) and its

reater requirement for heparin for protection against ienaturation by heat or acid (Gospodarowicz et al., 1987) may explain previous reports of aFGF's absence. In normal tissues rich in FGF activity, FGF mRNA levels are very low and have only been reported in bovine hypothalamus (Abraham et al., 1986)for bFGF, and in rat aorta for aFGF and bFGF mRNAs (Sarzani et al., 1989; see also Shimasaki et al., 1988). aFGF and bFGF activity has been found in many tissues and cultured cells, but to date these mitogens have not been localized to endothelial or smooth muscle cells in vivo or to the nuclei of both cell types. In vivo interaction between FGFs and other growth factors could be important in regulation of BAEC and SMC growth and function through autocrine and paracrine mechanisms and in response to cell trauma and wound repair. This is the first study to localize aFGF and bFGF to freshly isolated bovine endothelial cells. Previous studies of cultured bovine endothelial cells have clearly shown that they can make both bFGF and aFGF, according to one study (Baird and Ling, 19871, and bFGF according to others (Vlodavsky, 1987; Schweigerer, 1987a). Since bFGF and aFGF are endothelial mitogens, and the addition of one or the other is an absolute requirement for their long-term proliferation when seeded sparsely in culture, (but not when seeded at high density), it might be expected that intracellular bFGF or aFGF would be found in proliferatin cells as we indeed found in this study. However, we a so found large amounts of both peptides in post-confluent endothelial cells and in nearly confluent smooth muscle cells, which were already depositing large amounts of these peptides in their ECM. In contrast, confluent cells in vivo, scraped from the same animal, contained

f

372

SPEIR ET AL.

substantially smaller amounts. Cultured endothelial and smooth muscle cells may require bFGF and aFGF for other purposes in addition to proliferation. Indeed, Gospodarowicz has reported (1987) and we have also noted that viability, polarization, and contact-inhibition of post-confluent BAEC can be maintained for several weeks if the serum is supplemented with bFGF, but only for a few days in serum without FGF. The presence of more aFGF and bFGF in cultured endothelial cells than smooth muscle cells could simply be due to the fact that the serum contains PDGF and TGF-beta (Childs et al., 1982), which are not mitogenic for BAEC, but which are mitogenic for SMC and may make synthesis of aFGF or bFGF unnecessary. However, the same difference was noted in the ex vivo cells and no explanation is immediately apparent. The mechanism of regulation of FGF synthesis is not known, but we speculate that factors in the growth media, loss of contact inhibition, and loss of matrix may be important. Murphy et al. (1988) reported increased bFGF mRNA transcripts after stimulation of astrocytoma cells with PDGF or other protein kinase C activators. Large vessel endothelial cells lack receptors for PDGF (Bar et al., 1989) but SMC express the PDGF receptor in a phenotype- and growth-state dependent manner. Sjolund et al. (1988) reported that freshly isolated SMC from adult rat aorta expressed barely detectable PDGF transcripts, did not release PDGFlike mitogens, and had no detectable receptor binding or mitogenic activity. Primary SMC secreted PDGF only after exposure to PDGF or serum and had receptor activity in both confluent and subconfluent cultures. When secondary SMC were grown in the presence of serum, PDGF receptor activity was no longer detectable in either confluent or subconfluent cultures, but subconfluent cells still secreted PDGF (Sjolund et al., 1988). In confluent SMC, bFGF has been shown to interact with PDGF and IGF-1 through their respective receptors, a phenomenon termed receptor-cross talk (Pfeifle et al., 1987). It is possible that the low FGF receptor activity we found in our primary SMC (grown in the presence of serum) is due to down-regulation in the presence of the other competence factor, PDGF. Secondary SMC nearing confluence show dramatic loss of PDGF receptors and secrete little PDGF, but large amounts of extracellular matrix (ECM)are synthesized and deposited by SMC and BAEC. This ECM could well be the bioavailable FGF depot which could modify the proliferative response of both cell types through interaction with rtroteolvtic enzymes and heparin-like molecules. We isolated re1ati;ely large amounts of both FGFs from ECM (Table 1). The localization of both FGFs to nuclei of BAEC and SMC by Western blot of extracts of isolated nuclei is intriguing. Other growth factors, notably EGF, NGF, and insulin have been shown to accumulate in the nuclei (Burwen and Jones, 1987), but the roles of growth factors localized in the nucleus remain to be established. A role as transacting transcriptional regulators of genes relating to cell division seems possible. Since bFGF is highly basic, it could bind to nuclear chromatin through its net positive change at physiologic pH; although this is not true of aFGF, which has a net negative charge and is also found in the nucleus,

the two peptides have some homologous cationic regions. Delivery of endocytosed polypeptide to the nucleus could also explain the slow degradation of FGFs in cultured cells. In an elegant recent study, Bouche et al. (1987) reported translocation of exogenous bFGF to the nucleus and nucleolus of cultured BAEC where it directly stimulated the transcription of ribosomal genes by enhancing RNA polymerase I activity. However, the results of the present study suggest that endogenous FGFs are also transported to the nucleus. Several cationic nuclear translocation signals are known, and although both FGFs lack classic hydrophobic sequences signalling export from the cell, each has a cationic sequence near the NH,-terminus which could specify nuclear translocation. BAEC in vitro are the victims of “culture shock’: isolation from tissue matrix, migration and proliferation, exposure to serum and other wound-associated factors, imposition of a plastic substratum and cytoskeletal rearrangement could mimic pathologic stress in vivo (Sage, 1986). The in vitro changes of SMC are even more dramatic; growth-related transitions in actin and myosin isoforms have been reported. Owens et al. (1986)found that cultured SMC lost SMC specific cu-actin, while non-muscle P-actin was high in proliferating cells. Similarly, SMC-specific myosin is lost in log-phase cultured SMC and non-muscle myosin is prevalent (Kawamoto and Adelstein, 1987). Finally, post-confluent cells in culture have the morphology and sustained proliferation of transformed cells and form tumor-like structures, reminiscent of lesions caused by excessive SMC proliferation after balloon injury of the vessel in patients after angioplasty.

LITERATURE CITED Abraham, J.A., Mergia, A.? Whang, J.L., Tumolo, A., Friedman, J., Hjerrild, K.A., Gospodarowicz, D., and Fiddes, J.C. (1986) Nucleotide sequence of a bovine clone encoding the angiogenic protein, basic fibroblast growth factor. Science, 233:545-548. Baird, A,, and Ling. N. (1987)Fibroblast growth factors are present in the extracellular matrix produced by endothelial cells in vitro: Implications for a role of heparinase-like enzymes in the neovascular response. Biochem. Biophys. Res. Commun., 142:428-435. Bar, R.S., Boes, M., Booth, B.A., Dake, B.L., Henley, S., and Hart, M.N. (1989)The effects of platelet-derived growth factor in cultured microvessel endothelial cells. Encocrinology, 124:1841-1848. Berger, C.N. (1986) In situ hybridization of immunoglobulin-specific RNA in single cells of the B lymphocyte lineage with radiolabelled DNA probes. EMBO J., 5%-93. Bohlen, P., Esch, F., Baird, A,, and Gospodarowicz, D. (1985) Acidic fibroblast growth factor from bovine brain. Amino terminal sequence and comparison to basic fibroblast growth factor. EMBO J., 4:1951-1956. Bouche, G., Gas, N., Prats, H., Baldin, V., Tauber. J.P.,Teissie, J., and Almaric, F. (1987) Basic fibroblast growth factor enters the nucleolus and stimulates the transcription of ribosomal genes in ABAE cells undergoing Go-G1 transition. PIoc. Natl. Acad. Sci. U.S.A. 84:6770-67?4. Burwen, S.J., and Jones, A.L. (1987) The association of polypeptide hormones and growth factor with the nuclei of target cells. TIBS, 12:159-162. Casscells, W., Speir, E., Sasse, J., Klagsbrun, M., Allen, P., Lee, M., Calvo, B., Chiba, M., Haggroth, L., Folkman, J., and Epstein, S.E. (1990) Isolation, characterization and localization of heparinbinding growth factors in the heart. J. Clin. Invest.. 85:43341. Casscells, W., Bazoberry, F., Speir, E., Thompson, N., Flanders, K., Kondaiah, P., Ferrans, V., Epstein, S.E., and Sporn, M. (1990) Transforming growth factor B1 in normal heart and in myocardial infarction. In: Transforming Growth Factor B’s: Chemistry, Biology and Therapeutics. Ann. N.Y. Acad. Sci., 593:148-160.

E'GFs IN VASCULAR EC AND SMC EX VlVO VS. IN VITRO

Childs, C.B., Proper, J.A., Tucker, A.F., andMoses, H.L. (1982)Serum contains a platelet-derived transforming growth factor. Proc. Natl. Acad. Sci. U.S.A., 79:5312-5316. Gospodarowicz, D., Moran, J., Braun, D., and Birdwell, C. (1976) Clonal growth of bovine endothelial cells: Fibroblast growth factor as a survival agent. Proc. Natl. Acad. Sci. U.S.A., 73~4120-4124. Gospodarowicz, D., Ferrara, N., Schweigerer, L., and Neufeld, G. (1987) Structural characterization and biological functions of fibroblast growth factor. Endocr. Rev., 8~95-109. Gospodarowicz, D., Ferrara, N., Haaparanta, T., and Neufeld, G. 11988) Basic fibroblast growth factor: Expression in cultured bovine vascular smooth muscle cells. Eur. J. Cell Biol., 46:144-151. Hauschka, P.V., Mavrakos, A.E., Iafrati, M.D., Doleman, S.E., and Klagsbrun, M. (1986) Growth factors in bone matrix. J. Biol. Chem., 261~12665-12674. Hoshi, H., Kan, M., Chen, J.K., and McKeehan, W.L. (1988) Comparative endocrinology-paracrinology-autocrinologyof human adult large vessel endothelial and smooth muscle cells. In Vitro, 24:309320. Kawamoto, S., and Adelstein, R.S. (1987) Characterization of myosin heavy chains in cultured aorta smooth muscle cells. J. Biol. Chem., 212:7282-7286. Lindner, V., Reidy, M.A., and Fingerle, J. (19891Regrowth of arterial endothelium. Denudation with minimal trauma leads to complete endothelial cell regrowth. Lab. Invest. 61:556-563. Lobb. R.R. (1988)Thrombin inactivates acidic fibroblast growth factor but not basic fibroblast growth factor. Biochemistry, 27:2572-2578. Mansson, P.E., Marlak, M., Sawada, H., Kan, M., and McKeehan, W.L. (1990) Heparin-binding (fibroblast) growth factors type one and two genes are co-expressed in proliferating normal human vascular endothelial and smooth muscle cells in culture. In Vitro Cell. Dev. Biol., 26~20%-212. Moscatelli, D., and Quarto, N. (1989) Transformation ofNIH 3T3 cells with basic fibroblast growth factor or the hstik-fgf oncogene causes down regulation of the fibroblast growth factor receptor: reversal of morphological transformation and restoration of receptor numbers by Suramin. J. Cell Biol., 109:2519-2527. Murphy, P.R., Sato, Y., Sato, R., and Friesen, H.G. (1988) Fibroblast growth factor messenger ribonucleic acid exoression in a human astrocytoma cell line:-regulation by serum ahd cell density. Mol. Endocrinol., 2591-598. Neufeld. G., and Gosoodarowicz, D. (1985) The identification and partial characterizaiion of the fibroblast growth factor receptor of baby hamster kidney cells. J. Biol. Chem., 260:13860-13868. Olwin, B.B., and Hauschka, S.D. (1988) Cell surface fibroblast growth factor and eoidermal growth factor receotors are oermanentlv lost during skelital muscg terminal differehiation i; culture. J: Cell Biol., 107:761-769. Orlow, S.J., Hotchkiss, S., and Pawelek, J.M. (1990)Internal binding sites for MSH: Analvsis in wild-tvoe and variant Cloudman melanoma cells.