Macrophage and Foam Cell Release of Matrix-bound Growth Factors

Vol. 268, No, 16, Issue of June 5, pp. 11951-11958,1993 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Q 1993 by The American Society for Biochemistry and Molecular Biology, Inc

Macrophage and Foam Cell Release of Matrix-bound Growth Factors ROLE OFPLASMINOGEN

ACTIVATION* (Received for publication, December 11, 1992)

Domenick J. Falcone$#T,Timothy A. McCaffrey[[**,Adriana Haimovitz-Friedman$$, Jo-Anne Vergilio$#Q,and Andrew C.Nicholson$ From the Departments of $Pathology, $Cell Biology and Anatomy, and IIMedicine, Cornell Medical College and the $$Department of Radiation Oncology, Memorial Sloan-Kettering Cancer Center, NewYork,New York 10021

rophages and lymphocytes (1-3). It has been suggested that We have determined whether macrophage derivedfoam cells, a prominent component of the atheroscle- following their emigration from the blood and transformation rotic lesion, express more urokinase-type plasminogen into lipid-laden foam cells, macrophage release growth factors activator (uPA) and whether their ability to generatethat stimulate vascular smooth muscle cell proliferation and release of matrix-bound growth matrix synthesis (4).In this regard, macrophage do not acplasmin stimulates the factors. Steady state levels of uPA mRNA and both cumulate lipid when incubated with LDL’ (5-7); however, membrane and intracellular uPA activities were sig- several chemical modifications of LDL have been described nificantly increased in foam cells. When cultured on that result in adramatic increase in theiruptake (5-7). cell-derived matrices containing bound ‘261-basi~ fi- Endocytosis of these modified lipoproteins is mediated by broblast growth factor (bFGF), both macrophage and what is now commonly referred to as the scavenger receptor. foam cells released intact l2’1-bFGF into their media. More recently it has been demonstrated that oxidation of The release of ”‘I-bFGF by either cell was signifi- LDL leads to modification of the lysine groups of apoB much cantly enhancedin thepresence of plasminogen. Howlike that observed with chemical derivatization of LDL (8). moremembrane ever, foam cells,whichexpressed However, exposure of macrophages or endothelial cells to uPA, released more “‘I-bFGF than control cells. The release of matrix-bound bFGF was independent of hep- these modified forms of LDL results in decreased expression aranase activity, since neither macrophage nor foam of platelet-derived growth factor, a potent mitogen for vascular smooth muscle cells (9-11). Therefore, it appears uncells degraded 36S04-labeledheparan sulfate proteolikely that modified LDL would enhance the macrophage’s glycans. In addition, media derived from foam cells cultured on cell-derived matrices in the presence of ability to stimulate smoothmuscle cell proliferation and lesion plasminogen hadincreasedlevels of transforming progression in thismanner. growth factor (TGF) /3 activity as compared to cells In addition to secreting growth factors, macrophage may grown in the absence of plasminogen. In contrast,plas- affect the growth of neighboring cells by altering the extraminogen had no effect on TGF-@ activity recovered in cellular matrix. Matrix is a complex reservoir of bioactive the media of foam cells grown on plastic. Moreover, substances including growth factors (12-15) and inhibitors when macrophage were culturedon matrices contain- (16), adhesion molecules (17, IS), and modulators of coaguing bound ‘a61-TGF-@, the release of labeled TGF-/3was lation andfibrinolysis (19-22). One of the best studied matrixincreased in the presence of plasminogen. This is the associated growth factors is basic fibroblast growth factor first demonstration that foam cells can release two (bFGF). bFGFis a highly cationic protein that when released important growth regulators, bFGF and TGF-8, from by cells, in an as yet undetermined manner, can bind to its a mechanism by high affinity receptor and/or heparan sulfate proteoglycan on the extracellular matrix, and provides which macrophage and foam cells can stimulate ath- the cell surface or in the matrix (reviewed in Ref. 23). In this erosclerotic lesion development. regard, bFGF has been identified in the extracellular matrix both in vitro (12,24, 25) and in vivo (26-29). Whereas receptor-bound bFGF is internalized and degraded, matrix-bound bFGF persists as a source of growth factor activity (30, 31). Atherosclerosis shares many properties with chronic in- Matrix-bound bFGF has been reported to be released by two flammatory diseases including the persistent presence of mac- mechanisms: heparanase-mediated degradation of the glycos* These studies were supported in part by Research Grants HL- aminoglycan side chains of heparan sulfate proteoglycan (32) 40819, HL-46403, and HL-35724 from the National Heart Lung and or the release of bFGF/heparan sulfate proteoglycan comBlood Institute, National Institutes of Health. The costs of publica- plexes from matrix by plasmin (33). Agents that affect either tion of this article were defrayed in part by the payment of page of these enzyme systems may regulate the release of matrixcharges. This article must therefore he hereby marked “aduertise- bound growth factors. ment” in accordance with 18 U.S.C. Section 1734 solely to indicate We have previously demonstrated that lipoprotein and nonthis fact. lipoprotein ligands of the scavenger receptor trigger macro1 Recipient of National Institutes of Health Research Career Development Award HL-01962. To whom correspondence should be phage secretion of urokinase-type plasminogen activator addressed Dept. of Pathology (C-440), Cornell Medical College, New (uPA) (34, 35). The effect of these highly charged polyanions York, NY 10021. Tel.: 212-746-6457; Fax: 212-746-8789. ** Recipient of National Institutes of Health FIRST Award HL42606. $3 Recipient of the Ross Reid Memorial Medical Student Fellowship from the New York City Affiliate of the American Heart Association.

The abbreviations used are: LDL, low density lipoprotein(s); uPA, urokinase-type plasminogen activator; bFGF, basic fibroblast growth factor; PAI, plasminogen activator inhibitor; DPBS, Dulbecco’sphosphate-buffered saline; BCEC, bovine corneal endothelial cell(s); BB unit, Berger-Broida unit; PAGE, polyacrylamide gel electrophoresis.

11951

11952

Macrophage Release

of Matrix-bound Growth

Factors

on macrophage expression of uPA activity is protein kinase C-dependent and requires protein and RNA synthesis (36). These data suggest that exposure of macrophage to modified lipoproteinswouldtrigger uPA secretion and plasmin-dependent releaseof matrix-bound growthfactors. However, the effect of macrophage uPA secretion isblunted by theirsecretion of plasminogen activator inhibitor (PAI) as well as PA1 released byothercells (37, 38). In order to localize uPA activity as well as minimize its inactivation,macrophage express a receptor for uPA (39-41). Although, recent experiments have demonstrated that receptor-bound u P A is accessible t o PA1 (42-44), functional assays for uPA activity indicate that receptor-bound uPA is relatively insensitive to inhibition by PAI, and its catalytic efficiency is increased (43,

to 6.70 2 0.02 ( n = 5). Preincubation (30 min a t 4 "C) of purified human uPA (0.1-100 millunits) in 100 mM NaC1, 50 mM glycine, pH 3.0, was determined to have no effect on uPA activity as determined by plasmin-mediated caseinolysis. Moreover, reducing the pH of the uPA assay mixture from 7.16 to 6.70 had no effect on our ability to measure uPA activity. Data are expressed as milliunits of uPA normalized for cellular protein. Lactate Dehydrogenase Activity-In order to determine whether the exposure of macrophages to the acid glycine dissociation buffer lead to the nonspecific release of intracellular uPA, the release of lactate dehydrogenase (LDH) was evaluated utilizing a colorimetric assay (Sigma Diagnostics) as per manufacturer's instructions (Procedure 500). Preincubation (30 min at 4 "C)of purified rabbit muscle LDH (75-200 milliunits; Calbiochem) with 100 mM NaCl, 50 mM glycine, pH 3.0, resulted in a significant reduction of LDH activity (BB units) asdetermined by the above colorimetric assay. Values for LDH release from cells exposed to NaCl/glycine dissociation buffer 45, 46). were corrected for this loss of activity by determining the relationship In experiments reported here, we have determined whether between BB units of LDH activity obtained with standard LDH and macrophage-derived foam cells express more membrane-as- LDH preincubated with the NaCl/glycine buffer (LDH activity = sociated uPA and whether their increased ability to generate (observed activity)(2.476) 597; r = 0.99; p < 0.009). Steady State uPA and TGF-Dl mRNA Levels-RNA was isolated plasmin is a mechanismbywhichfoam cells can release from J774A.1 cells (5 X 106/T-25 flask) incubated with 100 pg/ml matrix-bound growth factors. LDLor acetyl-LDL for 48 h as previously described (36). RNA samples were electrophoresed in agarose, transferred to a Zetaprobe MATERIALS ANDMETHODS membrane (Bio-Rad), and hybridized with a 32P-labeled murine Isolation of Low Density Lipoprotein-Low density lipoprotein cDNA for uPA (provided by Dr. J. L. Degen, Children's Hospital (LDL) (1.019 < d < 1,063 g/ml) was isolated by ultracentrifugation Research Foundation, Cincinnati, OH) or32P-labeledoligonucleotide and acetylated as previously described (34). Lipoproteins were di- probe for exon 7B of human TGF-Pl (26 base pairs; 97% homology alyzed against 140 mM NaCl, 10 mM phosphate buffer containing 0.5 with the murine gene) (R&D Systems). Hybridization was evaluated mM EDTA, pH 7.4, a t 4 "C for 24 h, filter sterilized, and stored under by autoradiography and expression was normalized to the constitunitrogen gas. Oxidation was monitored by measuring thiobarbituric tively expressed message for glyceraldehyde-3-phosphate dehydrogenacid reactive substances (47). ase. Cell Culture-Murine macrophage cell line J774A.1 (American Preparation of Extracellular Matrix-coated Disks-The isolation Type Tissue Culture, Rockville, MD) was mechanically harvested in and culture of bovine corneal endothelial cells (BCEC) has been Roswell Park Memorial Institute medium (RPMI; without HEPES) described previously (48). BCEC were harvested by trypsinization supplemented with 10% fetal bovine serum, penicillin (100 IU/ml), and aliquoted into 35-mm dishes. Cells were maintained inDulbecco's streptomycin (100 pg/ml), and glutamine (4 mM) (Flow Laboratories, minimal essential medium containing 10% calfserum, 5% fetal bovine McLean, VA). Cells were aliquoted into T-25 flasks (3 X lo6) or 12- serum, penicillin, streptomycin, glutamine, and 5% dextranT-40. well plates (0.5 X 106/well) and left undisturbed in a C02 incubator Five to seven days after reaching confluence, the cell layer was overnight at 37 "C. The next day media was changed and either 100 removed by sequential exposure to 0.5% Triton X-100 in PBS (10 pg/ml LDL or acetyl-LDL was added. Media was replaced on day 2. min a t room temperature) and 0.20 mM NH40H in PBS (3rnin). Iodination of Growth Factors-Recombinant human bFGF (Amgen On day 3, control cells and foam cells were washed 3 times with Dulbecco's phosphate-buffered saline (DPBS) andeither assayed for Biologicals, Thousand Oaks, CA)was iodinated as described by Neufeld and Gospodarowicz (49). Following iodination the labeled membrane-associated and intracellular uPA or plated onto intact matrices derived from bovine corneal endothelial cells (described bFGF was applied to a heparin-Sepharose column equilibrated with 20 mM phosphate buffer, pH 7.2, containing 0.2% gelatin. Free lZ5I below). was washed from the column with 0.6 M NaCl in phosphate buffer Isolation of Activated Mouse Peritoneal Macrophage-Activated peritoneal macrophage were obtained from female Swiss-Webster containing gelatin. lZ6I-bFGF(34,593 cpm/ng) was eluted with phosmice (Crl:CFW(SW)BR, Charles River) 11 days following intraperi- phate buffer containing gelatin and 2 M NaCl. In addition, recombitoneal injection of 2.1 mg/ml Corynebacterium parvum (Wellcome nant human 'T-bFGF with higher specific activity (137,037 cpm/ng) Research Laboratories, Beckenham, United Kingdom) as described was purchased from Amersham Corp. Recombinant human "'1-TGF01 (228,660 cpm/ng) was obtained from Du Pont-New England previously (34). Determination of Membrane and Intracellular Plasminogen Acti- Nuclear. Release of lzSI-bFGF or -TGF-/3 from Extracellular Matricesvator Activity-Membrane-bound uPA was dissociated from cells by exposing them to 100 mM NaCI, 50 mM glycine, pH 3.0, for 30 min Dishes (35 mm) previously derivatized with BCEC matrix were preinat 4 "C (35). The acid glycine buffer was recovered and cells were cubated with 5 ng of '"1-bFGF (34,593 cpm/ng) or 1 ng of '=I-TGFwashed with DPBS. The monolayers were mechanically harvested in (31in 1 ml of DPBS containing 0.15% gelatin overnight at 4 "C. The DPBS andcellular lysates prepared by brief sonication. uPA activity next day, the DPBS containing unbound '251-growth factors were in the acid glycine dissociation buffer or the cell lysates was quanti- removed and thedishes were washed once with DPBS/gelatin. Monotated utilizing a functional assay for plasmin as previously described layers of macrophage or macrophage-derived foam cells were washed (34, 35). Aliquots (20 p l ) of acid glycine buffer or cell lysates were 3 times with DPBSand mechanically harvested in RPMI-ITS+ added to 180 pl of DPBS containing 1 pg of plasminogen and 10 pg (Collaborative Research (Bedford, MA). Cells (1.6 X 106/dish) were of [3H]acetyl casein prepared as previously described (34). Samples plated directly on the BCEC matrixcontaining sequestered lZ5Iwere incubated for 2.5 h at 37 'C following which600pl of 2% growth factors. Cells were incubated on the matrices for 2 h to allow albumin and 200plof 50% trichloroacetic acid were added. Acid- for adherence. The media was then changed and supplemented with soluble 3H-labeledpeptides were recovered and quantitatedin a LKB 100pg/ml acetyl-LDL or LDL and 2 pg of glu-plasminogen (American liquid scintillation counter (model 1209). Plasminogen activator ac- Diagnostica, New York, NY). Following an overnight incubation, tivity expressed by J774A.1 macrophage was inhibited >90% when media was recovered and assayed for acid-precipitable "'I-bFGF or incubated overnight (4 "C) with a neutralizing anti-uPA IgG as pre- -TGF-Pl and acid-soluble lZSI-peptides. Conditioned media (500 pl) was transferred into a microcentrifuge viously described (36). uPA concentration in the samples was determined by calculating plasmin-dependent hydrolysis of [3H]acetyl tube and 300 pl of2% bovine serum albumin added. Proteins were casein (i.e. the difference in hydrolysis of substrate in the presence precipitated with the addition of 200 pl of 50% trichloroacetic acid (1 and absence of exogenous plasminogen) and extrapolating this value h at 4 "C). The precipitated proteins were pelleted by centrifugation, to a standard curve generated with known concentrations of uPA. and the supernatantcontaining degraded 1261-groWthfactor and free The standard uPA assay reaction mixture was determined to have a radioactive iodide was recovered. Free iodide in the supernatant was pH of 7.16 +. 0.04 (-+ S.D.; n = 5). The addition of 20 pl of the acid extracted by adding H20zin the presence of KI and extracting the glycine dissociation buffer to the uPA assay mixture reduced the pH free iodine with chloroform (50).

+

Macrophage Release

of Matrix-bound Growth Factors

The remaining monolayers and matrices were dissolved in 0.5 NaOH, 0.1% Triton X-100 and assayed for residual iodinated growth factor and protein. Cellular protein was determined as thedifference between total protein/dish and theprotein content of matrices incubated with media alone. The protein content of the matrices utilized for the bFGF and TGF-B studies were 141.6 -t 12.7 (+ S.D.) and 86 rf: 10 pgldish, respectively. SDS-PAGE Analvsis of '"I-bFGF Released from Matrix-Recombinant human '*'I-bFGF {4 ng; 137,037cpm/ng) in DPBS containing 0.15% gelatin was aliquoted into dishes previously derivatized with BCEC matrix. Dishes were incubated overnight at 4 "C and washed with DPBS-gelatin followedby DPBS. The amount of Y - b F G F bound to the dishes was 1.46 f 0.05 ng/dish (-t S.D.). Macrophage and macrophage-derived foam cells were plated onto the '"I-bFGFlabeled matrices as described above. Following an overnight incubation media were recovered, equal volumes were mixed with sample buffer containing @-mercaptoethanoland applied to 4-15% gradient gels. After electrophoresis gels were fixed in 20% trichloroacetic acid, stained with 0.1% Coomassie Blue, and dried. activity was visualized by autoradiography. Release and Degradation of 3SS04-lubeledHeparan Sulfate Proteoglycan from BCEC Matrix-BCEC matrix was prepared as described above except that cells were aliquoted into four-well plates (5 X lo' cells/l6-mm well) and 40 pCi/ml Na25S0, (250-1000 mCi/mmol; Du Pont-New England Nuclear) was added 3 and 7 days after plating. Previous studies demonstrated that 70-80% of the matrix-incorporated [Y3]sulfate was associated with heparan sulfate glycosaminoglycan (32, 51). Cell-mediated degradation of [36S]sulfate-labeled matrix was determined as described previously (48). Macrophages or foam cells were mechanically harvested in DPBS, washed and resuspended in RPMI-ITS+. Cells (0.5 X 10') were plated into 16-mm wells previously derivatized with [36S]sulfate-labeledmatrix and allowed to adhere for 2 h. The media was removed and replaced with fresh RPMI-ITS+ containing either 2 pg of glu-plasminogen, 100 pg of heparin (bovine mucosal; Calbiochem), or both plasminogen and heparin. The next day culture media were collected and centrifuged (10,000 X g, 5 min) and the supernatants stored at -20 "C until analyzed by gel chromatography. Supernatnants were fractionated on Sepharose CL-GB (Pharmacia LKB Biotechnology Inc.) (0.7 X 35 cm) previously equilibrated with PBS containing 0.1% sodium azide. Fractions (200 pl) were monitored for 3sSactivity by scintillation counting. Soluble proteoglycans eluted from the column with a K , < 0.2 (-500 kDa) and is defined as peak I material. Degraded glycosaminoglycan side chains were eluted with a K., between 0.50 and 0.75 (-5-10 kDa) and were defined as peak I1 material. Free [%]sulfate emerged with the V,. Recovery of the label ranged from 85 to 95% in different experiments. Bioassay for TGF-&TGF-b activity in macrophage supernatants was assayed by determining the inhibition of [3H]thymidine incorporation by CCL64 mink lung fibroblasts (American Type Culture Collection, Rockville, MD) as previously described (52). Aliquots (100 pl) of macrophage conditioned media were added to 300 plof Medium 199 plus 13% fetal bovine serum with 50 pg/ml gentamicin sulfate (Whittaker Bioproducts, Walkersville, MD) and thenadded to CCL64 cells plated 24 h previously at 1 X lo' cells/well of 96-well plate (100 pl/well X 3 wells). Activation of latent TGF-@ was achieved by heating the 100-pl samples at 100 'C for 10 minutes prior to dilution with culture media (53). After 18 h at 37 "C, the media was carefully decanted and 1 pCi/ml [3H]thymidine (20 Ci/mmol; Du Pont-New England Nuclear) in Medium 199 was applied for an additional 4 h. The reduction of [3H]thymidine incorporationinto CCL64 DNA was compared to that obtained from known concentrations of TGF-fil (0.001-1.0 ng/ml; R & D Systems). Control samples containing the heated J774A.1 culture media were treated identically to the cell supernatants and caused no reduction in thymidine incorporation. The identity of the inhibitory factorwas confirmed by prior treatment of selected samples with 100 pg/ml rabbit anti-porcine TGF-61 (R & D Systems) for 4 h prior to addition to thecells. RESULTS

Foam CelluPAmRNALevels and Membrane-bound and Intracellular uPA Activities Are Increased-Macrophage incubated with acetyl-LDL for 48 h contained numerous lipid dropletsas revealed byOil Red 0 staining (Fig. lA). In contrast, thecytoplasm of cells incubated with LDL was free of Oil Red 0-positive material (Fig. 1B). The steady state

11953

level ofuPA mRNA in foam and controlcells was determined by Northern blot hybridization and quantitated densitometrically. As seen in Fig. 2, there was a modest increase (2.5fold) in uPA mRNA in foam cells as compared to control cells. Membrane uPA derived from either foam or control cells was dissociated from its receptor by a brief treatment with an acid glycine buffer (35). Cell monolayers were then harvested mechanically and lysed by sonication. As seen in Fig. 3, membrane-associated uPA recovered from foam cells was increased 5.1-fold ( p < 0.001)over membrane-bound uPA recovered from cells incubated with LDL. Likewise, intracellular uPA derived from foam cells was 2.3-fold greater ( p < 0.001) than that observed in control cells. Overall, intracellularuPA activity exceeded membranebound uPA by approximately 4.5-fold ( p < 0.001). It is unlikely that exposure of macrophages to theacid glycine buffer lead to the nonspecific release of intracellular uPA, since the release of lactate dehydrogenase activity by cells incubated with the theacid glycine buffer was 5150 f 496 BB units/5 X lo6 cells (k S.D.; n = 3) or 1.65% of the cellular lactate dehydrogenase, whereas cells incubated with DPBS released 4796 k 358 BB units or 3.56%. Moreover, in previous studies we could not demonstrate a reduction in intracellular uPA activity in cells treated with the dissociation buffer (35). Plasminogen-dependent Release of '"I-bFGF From Extracellular Matrix-Cell-derived matrices contain bFGF bound to heparansulfate proteoglycan (33, 54, 55). Sequestered growth factorcan be released from the insoluble matrix through the action of plasmin (33). We therefore determined whether the expression of uPA activity by either macrophage or foam cells would release matrix-bound bFGF. For this purpose, we monitored the ability of macrophage and foam cells to release '"I-bFGF from BCEC matrices. When matrices were incubated with media alone 68 -C 9 (& S.D.) pg of acid-precipitable "'I-bFGF was released into media. The addition of plasminogen to the media controls had no effect on the release of Ia51-bFGF(79 k 11 pg). As seen in Fig. 4, the release of acid-precipitable "'I-bFGF from BCEC matrices by macrophage was enhanced 11-fold when plasminogen was added to theculture media. Moreover, the release of "'1-bFGF by foam cells, in the presence of plasminogen, was 70% ( p < 0.001) greater than thatreleased by control cells or nearly 18fold over cells without plasminogen. The total release of lZ5IbFGF by foam cells was 56% of the "'I-bFGF bound to the matrix. This appears to be the entire plasmin-releasable pool of'"I-bFGF in these studies, since the addition of more plasminogen or plasmin to these cultures did not lead to further release of '=I-bFGF (data not shown). Therefore, the observed difference in the release of matrix-bound 1251-bFGF by control and foam cells is likely underestimated in these studies. The dramatic release of matrix-bound "'I-bFGF observed when macrophages and foam cells were cultured inthe presence of plasminogen could not be due to activation of gluplasminogen by the matrix, since the addition of plasminogen to media controls had no effect on the release of lZ5I-bFGF. Macrophage degradation of "'I-bFGF was evaluated as the appearance of noniodine, acid-soluble Iz6I-peptides in the media. Acid-soluble '1 activity was not detected in the media controls. In the absence of plasminogen, macrophage and foam cells released nearly equivalent amounts of degraded and intact "'I-bFGF (Fig. 4). Although the amount of degraded "'I-bFGF was significantly bp < 0.001) increased when cells were incubated in the presence of plasminogen, the proportion of total bFGFreleased that was degraded dropped to 19% (Fig. 4).

Macrophage Release of Matrix-bound Growth Factors

11954

h G . 1. Oil Red 0 staining of macrophage and foam cells. J774A.1 cells (0.25 X 106/ml) were plated on glass coverslips (12 mm) in RPMI supplemented with fetal bovine serum. On day 1 and 2 media was changed and 100 pg/ml acetyl-LDL ( A ) or native LDL ( B ) was added. On day 3, cells were washed with DPBS, fixed in 0.1% glutaraldehyde, stained with Oil Red 0, and counterstained with Giemsa.

4000 3500

2.5 0

5 2.0

uPA

2.3 kb

GAPDH

1.3 kb

10

Membrane

T

2 2500 n = 2000 .-tcn 1500 3000

E

c

3

E

1000500

Ctrl Cell

Foam Cell

FIG.3. Enhanced expressionof membrane and intracellular

Ctrl Cell

Foam Cell

uPA activity by foam cells. Cells were aliquoted into 12-well plates (0.5X 106/well)and incubated with LDLor acetyl-LDL. uPA activity was determined by monitoring plasmin-dependent release of labeled peptides from [3H]acetyl casein. The datarepresent the mean S.E. from six separate wells.

*

Treatment FIG. 2. uPA mRNA levels are increased in foam cells. Cells (5 X 106/flask) were incubated with 100 pg/ml acetyl-LDL or LDL for 48 h in RPMI 1640, 10%fetal calf serum. Media was changed after 24 h. Total RNA was isolated, and uPA mRNA levels were determined by Northern blot hybridization utilizing a murine uPA cDNA probe. Expression is normalized to constitutively expressed glyceraldehyde-3-phosphate dehydrogenase (GAPDH).

In order to verify that acid-precipitable "'I activity presented in Fig. 4 was actually intact bFGF, we analyzed the released lZ5I-bFGFby SDS-PAGE (Fig. 5). Authentic '"IbFGF migrated asa single band with a molecular massof -16 kDa in a 4-16% gradient gel. Media derived from control or foam cells plated on matrices to which "'1-bFGF was previously bound contained intact bFGF (Fig. 5). When plasminogen was addedto macrophage or foam cell cultures, we

Macrophage Release of Matrix-bound Growth Factors 6oo 500 400

1

ActivatedMacrophage

Degraded

1

T

+

11955

I

'Ig

0

Media + Heparin

-

300 -

200

loo Foam

-

1

400

0 Ctrl Ctrl

FIG. 4. Macrophage and foam cell release of matrix-bound lS6I-bFGFis enhanced in the presence of plasminogen. Control and foam cells were harvested in RPMI-ITS+ and aliquoted into35mm dishes containing an adherentlayer of BCEC extracellular matrix to which lZ5I-bFGF(34,593 cpm/ng) was bound.The release of matrixbound bFGF into conditioned media, in the presence and absenceof plasminogen, was determined as acid-precipitable "'1 activity. Degradation of released '"I-bFGF was determined as the appearance of acid-soluble "'I activity. Cellular protein was 278 f 17 pg/dish (f S.E.) andtotalboundI-bFGF was 1.15 f 0.02 ng/dish (fS.E.). Neither cell protein nor bound bFGFvaried significantly with experimental conditions. '"I-bFGF released by media controls has been subtracted from the observed release of '"I-bFGF by cells. Data represent the mean S.E. of four dishes.

Cnn nn

5

0

Foam

10

15

25

20

30

Fraction Number 2ooo,

1600

Control Cells

i

0

0

Media + Plg + Heparin + Plg. Hep

0 0

5

10

15

25

20

30

Fraction Number

16.2 kD e Media

-+-+ -+-+ Plasminogen FIG.5. SDS-PAGE analysis of '%bFGF released from BCEC matrix. Aliquots of conditioned media derived from control and foam cells incubated on BCEC matrix (24 h) derivatized with "'1-bFGF (137,037 cpm/ng) were electrophoresed through a 4-15% polyacrylamide gel. 12sI-bFGFreleased from the matrixwas visualized by autoradiography and compared to authentic "'1-bFGF. On the right side,the gel was intentionally overloaded with conditionedmedia in order tovisualize fragments of lz5I-bFGF.

l6O01

R

0

+ + +

Plg Heparin Plg, Hep

400

0 0 5 10 15 20 25 30 observed a large increase in the release of intact "'1-bFGF. Fraction Number Foam cells released more matrix-bound bFGF than control cells bothinthe presence orabsence of plasminogen as FIG. 6. Foam cells do not express heparanase activity. Aliquots of conditioned media derived from cultures of activated peridetermined by a densitometric scan of the autoradiogram. Release and Degradation of Matrix Heparan Sulfate Proteo-toneal macrophage (top),J774A.1 macrophage (middle),and J774A.1derived foam cells (bottom) grown on SsS04-labeledmatrices for 24 h glycan-Another mechanism by which cells can mobilize were fractionated by gel sieve chromatography and fractionsassayed bFGF bound to heparan sulfate proteoglycan is via the expres- for 35S activity. The emergence of activity near the column void sion of heparanase activity (32, 54). Therefore, the abilityof volume (peak I) is associated with intact proteoglycan, whereas the macrophage and foam cells to release and degrade matrix- second included peak of activity (peak II) is associated with degraded glycosaminoglycan side chains. associated3sS04-labeled heparansulfate proteoglycanwas determined. When the media derived from C. parvum-activated macrophage cultured on 35S04-labeled matrices were can was inhibited, as evidenced by the disappearance of the fractionated on SepharoseCL-GB, intact proteoglycans (-500 second peak anda corresponding increase in the firstpeak. Bothcontroland foam cells releasedsmall amounts of kDa) emerge as a peak at or near the columnVo (Fig. 6). The smaller partially degraded sulfate-labeled glycosaminoglycan proteoglycan into their media (Fig. 6). The release of labeled When proteoglycan from matrix by foam cells was increased severside chains (-5-10 kDa) emerge as asecondpeak. heparin (an inhibitor of heparanase activity)was added to the alfold in the presence of plasminogen, whereas plasminogen culture medium, the degradation of 3sS04-labeled proteogly- had no effect on the release of proteoglycan by control cells

11956

Macrophage Release of Matrix-bound Growth

(Fig. 6) or when matrices were incubated with media controls (data not shown). These data corroborate our observations that foam cells express more uPA thancontrol cells. In addition to heparin’s ability to inhibit heparanaseactivity, we observed that the release of labeled proteoglycan (i.e. peak I) from cell-derived matrices was increased in the presence of heparin. This effect of heparin was not cell-dependent, since the same phenomenon was observed in the absence of cells (data not shown). There was a synergistic effect (10-fold) on the release of 35S04-proteoglycanfrom the BCEC matrix when foam cells were cultured in the presence of both heparin and plasminogen. Therefore, although foam cell release of 35S04proteoglycan was increased in the presence of plasminogen (Fig. 6), it appears to be a relatively small fraction of the releasable pool of labeled proteoglycan. In dramatic contrast to C. paruum-activated macrophage, neither control nor foam cells expressed heparanase activity, as evidenced by their failure to generate peak I1 material. These data demonstrate that the release of bFGF from the extracellular matrix by these cells is independent of heparanase activity. The Release of Transforming Growth Factor P by Macrophage and Foam Cells Is Increased by Plasminogen-Intact and partiallydegraded components of the extracellular matrix released by macrophage and foam cells may exert diverse effects on cellular proliferation. Since we demonstrated that macrophage can release exogenous ‘“I-bFGF bound to matrix, we next determined whether conditioned media derived from macrophage or foam cells grown on matrix (in the absence of exogenous ‘“1-bFGF) possessed bFGF activity. For this purpose BCEC, which respond mitogenically to bFGF, were utilized as target cells, Unexpectedly, conditioned media derived from macrophage and foam cells inhibited the uptake of thymidine by BCEC cells (data notshown). In addition,when exogenous bFGF was added to the conditioned media, its effect on thymidine uptake by BCEC cells was also inhibited, thus suggesting the presence of an inhibitor in the macrophage conditioned media. The ability of TGF-P to inhibit the proliferation of endothelial cells has been reported (56,57). Therefore, we assayed for the presence of TGF-P inmedia derived from macrophage and foam cells grown on extracellular matrix. The addition of TGF-/3 tothe media of CcL64 cells resulted in a dosedependent inhibition in their uptake of [3H]thymidine (data not shown). Utilizing this cell line as a target cell for TGF-8 (52), we observed that macrophage conditioned media contained both active and latent TGF-P;however, no more than that observed when matrices were incubated with media alone (Fig. 7). Overall, foam cells released more TGF-P than control cells ( p < 0.001). When control and foam cells were considered together, there was an increase in the release of TGF-@ inthe presence of plasminogen ( p < 0.05); however, the most striking increase (>$fold) was observed with foam cells. When conditioned medium derived from foam cells grown on matrix in the presence of plasminogen was heat-activated (53) and then preincubated with an anti-TGF-P IgG, TGF-P activity was inhibited >go%. In order to assess the relative contributions of cell-derived TGF-Pand matrix-derived TGF-P observed in Fig.7,we compared the amount of TGF-/3 activity released by foam cells grown on plastic and matrix. Conditioned media derived from cells cultured on matrixor plastic contained nearly identical amounts of TGF-P activity (Fig. 8). Conditioned media derived from foam cells cultured on matrix in the presence of plasminogen had -200 pg/ml more TGF-P than identical cells grown on matrix inthe absence of plasminogen.

Factors

0.35 7

FIG. 7. Release of TGF-j3 by macrophage and foam cells is enhanced in the presence of plasminogen. Conditioned media derived from control or foam cells incubated 24 h on BCEC matrix in the presence or absence of plasminogen were assayed for active and latent TGF-Pactivity as described under “Materials and Methods.’’ Data represent mean k S.E. of three separate cultures. 0.5

-E

\

-

Plasfic

0.4

0.3

c

CD

a

Matrix

0.2

I-

o. 1 0

1 1I Foam

+

Plg

Foam

+

Plg

FIG. 8. Release of TGF-j3 by foam cells cultured on plastic or BCEC matrix. Conditioned media from foam cells incubated in the presence or absence of plasminogen were assayed for total TGFactivity as described under “Materials and Methods.”

In contrast, the addition of plasminogen to foam cells cultured on plastic had no effect on the concentration of TGF-/3 recovered. Therefore, in this experiment nearly 50% of the 423 pg/ml TGF-fi released by foam cells cultured on matrix is matrix-derived and released in a plasminogen-dependent manner. The above data suggest that plasmin generated by foam cells is releasing matrix-bound TGF-8 and/or that matrix enhances the expression of cell-associated TGF-P, which can be released by plasmin. Therefore, we determined whether macrophage are able to release matrix-bound TGF-P. For this purpose we utilized ‘251-TGF-pas a tracer. When matrices containing bound lZ5I-TGF-pwere incubated with media alone, 9.7 +- 0.25 (kS.D.) pg of intact (acid-precipitable) ‘‘‘1TGF-P was recovered. When 2 or 4 pg of plasminogen were added to themedia, the release of intact lZ5I-TGF-/3 remained unchanged (9.9 -t 0.87 and 9.5 .t 0.47, respectively). Little or no degraded (acid-soluble) 1251-TGF-Pwas detected in the media controls. When adjusted for media controls, macrophage plated on BCEC matrices released -2.0 pg of intact and -17 pg of degraded lZ5I-TGF-P(Fig. 9). When cells were cultured on matrices in the presence of 2 pg of plasminogen, the recovery of intact ’251-TGF-Pincreased $-fold, whereas the recovery of degraded ‘251-TGF-@increased only 30%.

Macrophage Release

of Matrix-bound Growth Factors

11957

(62), activation or release of growth factors (33, 63, 64), and activation of other proteases (65). We have reported previ30 Intact Degraded ously that acetyl-LDL and otherpolyanions recognized bythe scavenger receptor trigger macrophage secretion of uPA (3436). In experiments reported here, these observations have been extended by the demonstration that macrophage-derived foam cells contain increased levels of uPA mRNA and express more membrane-bound and intracellular uPA. Consequently, the exposure of macrophage to modified LDL will stimulate both fluid phase and cell surface plasminogen activation. In contrast to the fluid phase, uPA bound to the cell surface is relatively insensitive to inhibition by plasminogen activator inhibitor (43,45,46). Furthermore, variety a of cells including macrophage possess plasminogen/plasmin receptors (66-68). 0 2.0 4.0 0 2.0 4.0 When plasminogen is activated on the surface of cells it is Plasminogen Ipg) relatively insensitive to inhibition by a2 anti-plasmin and a2 FIG. 9. Macrophage release of matrix-bound '2KI-TGF-Bis macroglobulin (67-69). Therefore in uiuo, cells are able to enhanced in the presence of plasminogen. Macrophage were activate plasminogen despite the abundant presence of speharvestedin RPMI-ITS+and aliquoted into 35-mm dishes containing cific and high affinity inhibitors. The increase in membrane an adherent layer of BCEC matrix to which '251-TGF-p was bound. uPA observed in these studiesmay reflect increased synthesis The release of matrix-bound TGF-fl into conditionedmediawas determined as acid-precipitable lZ61 activity. Degradation of TGF-P and expression of uPA receptors on the surface of foam cells, was determinedas the appearance of acid-soluble '"I activity. Cellular or represent increased receptor occupancy (70). We are curprotein was 366 -C 5 (-c S.E.) jtg/dish, and total bound 'Z61-TGF-p was rently exploring these possibilities. 98 f 1 (* S.E.) pg/dish. Neither cell protein nor bound '251-TGF-p There is evidence that foam cell formation in vivo may lead varied significantly between groups. Data representthe mean f S.E. to increased uPA expression. Lipid-laden macrophage isolated of three dishes. from carrageenan-induced granulomas in hypercholesterolemic rabbits express more PA activity than granuloma macIncreasing the plasminogen concentration to 4 fig had no rophage derived from normocholesterolemic rabbits (71). effect on the release of intact or degraded TGF-@. When However, secretion of uPA by these cells was not demonanalyzed by SDS-PAGE under reducing conditions, monostrated, and no distinction between membrane and intracelmeric lZ6I-TGF-@ migrates as a single band of approximately 14 kDa. Matrix-bound lZ5I-TGF-@released by macrophage lular activities was determined. bFGF is a component of the extracellular matrix of a variety appeared as a single 14-kDa band. When plasminogen was of tissues (24-29). We demonstrate that macrophage release added to thecultures, the radiolabeled 14-kDa bandincreased in intensity (data not shown). Lower molecular weight lZ5I- matrix-bound bFGFby the limited proteolytic degradation of the matrix. Although bFGF is released by the action of plaslabeled fragments were not visible. Finally, in an effort to determine whether TGF-@ synthesis min, it is relatively protected from proteolytic degradation. was altered in foam cells, we examined the steady state levels The mechanism by which bFGF is protected is likely via its of TGF-01 mRNA in control and foam cells. In contrast to association with heparan sulfate proteoglycan. This concluthe observed increase in steady state levels of uPA mRNA sion is based on the observation that when foam cells are (Fig. 2), levels of TGF-/3 mRNA were equivalent in control culturedon bovine corneal endothelial cell matrices, they and foam cells (data not shown). These data must be inter- release more matrix heparan sulfate proteoglycan when plaspreted with caution, however, since the regulation of TGF-@ minogen is added to themedia. Moreover, it hasbeen reported expression has been reported to be post-transcriptional (58). that bFGF bound to heparan sulfate is protected from degradation by plasmin (72), and treatment of endothelial cell cultures with plasmin or plasminogen leads to the release of DISCUSSION A central question in the pathobiology of atherosclerosis is bFGF/heparan sulfate proteoglycan complexes (33). An unexpected finding of these studieswas the demonstrathe source of the mitogens responsible for the proliferation of tion that foam cells cultured on matrices in the presence of vascular smooth muscle cells. Although it hasbeen speculated plasminogen released more TGF-/3 than control cells. These that foam cells might provide a source of growth factors (reviewed in Ref. 4),it remains to be demonstrated that the data suggest that TGF-@is sequestered in the matrix and its release of growth factors by macrophage is increased when release is regulated by macrophage expression of uPA. This interpretation is corroborated by three observations: steady they are transformed into foam cells. Paradoxically, it has been reported that exposure of cells to modified lipoproteins, state levels of TGF-Pl mRNA are unchanged in foam cells capable of inducing a foam cell phenotype in macrophage, relative to control cell; the addition of plasminogen to cultures appears to decrease their expression of platelet-derived growth of foam cells grown on plastic did not affect the level of TGFfactor, a potent smoothmuscle cell mitogen (9-11). In studies @ activity recovered in their media; and therelease of matrixby macrophage was increased in the presreported here, we demonstrate that macrophage expression of bound 1251-TGF-@ uPA regulates theirplasmin-dependent release of matrix- ence of plasminogen. In these studies, the extracellular matrix components to bound bFGF andTGF-8. Moreover, we demonstrate that which TGF-@was bound were not identified. However, it is foam cell uPA expression is enhanced, which leads to an increase in their release of these growth factors. These studies likely that TGF-@ binds to multiple sites in the matrix. pprovide a mechanism by which foam cells may contribute to Glycan, a cell-associated chondroitin sulfate/heparan sulfate smooth muscle cell proliferation in theatherosclerotic lesion: proteoglycan, may act as reservoir a of TGF-@(73). In contrast release of growth factors from the extracellular matrix. to thebinding of bFGF to heparan sulfate proteoglycan, TGFCellular uPA activity can regulate a variety of important 3'/ appears to bind the protein core of @-glycan(74). In addition processes including cell migration (59-61), tissue remodeling to the cell surface form, @-glycanhas been reported in tissue

1

I

Macrophage Release of Matrix-bound GrowthFactors

latent TGF-p, plasmin may release TGF-/3 bound to proteoglycans or other extracellular matrix components. Several recent in vivo studies have implicated both bFGF and TGF-8 as important mediators of the blood vessel response to injury. Following arterial injury, levels of TGF-6 mRNA and immunoreactivity are increased (79). In addition, systemically administered TGF-8 stimulates smooth muscle cell proliferation following vascular injury (79). Likewise, following vascular injury, administration of bFGF results in increased smooth muscle proliferation andantibodies to bFGF significantly inhibits smooth muscle cell proliferation (80, 81). Data reported here clearly demonstrate that macrophage-derived foam cells release matrix-associated bFGF and TGF-(3. An important mechanism for this release appears to be the enhanced expression of uPA. This is the first demonstration of a mechanism by which foam cells could contribute to therelease of growth factors involved in progression of the atherosclerotic lesion. REFERENCES 1. Gown, A. M., Tsukada, T., and Ross, R. (1986) Am. J. Pathol. 1 2 5 , 191207 2. Jonasson, L., Holm, J., and Skalli,0.(1986) Arterioscler. Thromb. 6 , 131138 3. Hansson, G. K., Seifert, P. S., Olsson, G., and Bondjers, G. (1991) Arterioscler. Thromb. 1 1 , 745-750 4. Ross, R. (1986) N. Engl. J. Med. 314,488-500 5. Golstein, J. L., Ho, Y.K., Basu, S., and Brown, M. S. (1979) Proc. Natl. Acad. Sci. U. S. A. 76,333-337 6. Shechter, L.A., Fogelman, A. M., Haberland, M. E., Seager, J., Hokom, M., and Edwards, P. A. (1981) J. Lipid Res. 22,63-71 7. Mahley, R. W., Innerarity, T. L., Weisgraber, K. H., and Oh, S. Y. (1979) J. Clin. Inuest. 6 4 , 743-750 8. Steinbrecher, U. P. (1989) J. Biol. Chem. 262,3603-3608 9. Fox, P. L., and DiCorleto, P. E. (1986) Proc. Natl. Acad. Sci. U. S. A. 8 3 , 4774-4778 10. Fox. P. L..Chisolm. G. M.. and DiCorleto. P. E. (1987) . . J. Biol. Chern. 262. 6046-6054 11. Malden, L. T., Chait, A,, Raines, E., and Ross, R. (1991) J. Biol.Chem. 2 6 6 , 13901-13907 12. Vlodavsky, I., Folkman, J., Sullivan, R., Fridman, R., Ishai-Michaeli, R., Sasse. J.. and Klaesbrun.. M. (1987) . , Proc. Natl. Acad. Sci. U. S.A. 84. 2292-2296 13. Roberts, R., Gallagher, E., Spooncer, T. D., Allen, F., Bloomfield, F., and Dexter, T. M. (1988) Nature 332,376-378 14. Andres, J. L., Stanley, K., Cheifetz, S., and Massague, J. (1989) J. Cell Biol. 109,3137-3145 15. Hauschka. P. V., Chen, T. L., and Mavrakos, A. E. (1988) Ciba Found. Symp. 136,207-225 16. Castellot, J. J., Wright, T. C., and Karnovsky, M. J. (1987) Semin. Thromb. Hemostasis 13,489-503 17. Kleinman. H. K.. McGarvev. M. L.. Hassell. J. R.. Star. V. L.. Cannon. F. B., La&, G. W., and Mikin,G.' (1986) Biochemist625,312-318 18. Yurchenco, P. D., and Schittny, J. C. (1990) FASEB J. 4,1577-1590 19. Bar-Shavit. R.. Benezra. M.. Eldor, A.. Hv-Am. E.. Fenton, J. W., 11, Wilner, G . D:, and Vlodavsky, I. (1990) Cill Regul. 1,453-463 20. Knudsen. B. . . J. Clin. Znuest. - S.. Hamel. P. C.. and Nachman. R. L. (1987) 80, 1082-1089" ~' ' 21. Seiffert, D., Mimuro, J., Schleef, R. R., and Loskutoff, D. J. (1990) Cell D i p De: 3 2 , 287-292 22. Knuen, B S., Sllverstem, R.L., Leung, L. L. K., Harpel, P. C., and Nachman, R. L. (1986) J. Bwl. Chem. 261,10765-10771 23. Rifkin, D. B., and Moscatelli, D. (1989) J. Cell Biol. 1 0 9 , 1-6 24. Globus, R. K., Plouet, J., andGospodarowicz,D. (1989) Endocrinology 1 2 4 ,

-

~

~

x~

ma-1"_. "-" 25. Ro elj, S., Klagsbrun, M., Atzmon, R., Kurokawa, M., Haimovitz, A., Fuks, I

k,

547

and Vlodavsky, 1. (1989) J. Cell Biol. 109,823-831 26. Morton, K., Hutchinson, C., Jeanny, J. C., Karpouzas, I., Pouliquen, Y., and Courtois, Y. (1989) Curr. Eye Res. 8,975-987 27. Gonzalez, A. M., Buscaglia, M., Ong, M., and Baud, A. (1990) J. Cell Biol. 110,753-765 28. Yamada, S., Buffinger, N., DiMario, J., and Strohman,R. C. (1989) J. Med. Sci. Sports Exercise 2 1 , S173-180

37. Chapman, H. A,, and Stone, 0. L. (1985) Biochem. J. 2 3 0 , 109-116 38. Genton, C., Kruithof, E. K. O., and Schleuning, W.-D. (1987) J. Cell Biol. 1 0 4 , 705-712 39. Vassalli, J.-D., Baccino, D., and Belin, D. (1985) J. Cell Biol. 1 0 0 , 86-92 40. Plow, E. F., Freany, D. E., Plescia, J., and Miles, L. A. (1986) J. Cell Biol. 103,2411-2420 41. Blasi, F., Vassalli, J.-D., and Dano, K. (1987) J. Cell Bwl. 1 0 4 , 801-804 42. Cubellis, M. V., Andreasen, E. A., Ragno, P., Mayer, M., Dano, K., and Blasi, F. (1989) Proc. Natl. Acad. Sci. U. S.A. 86,4828-4832 43. Kirchheimer, J. C., and Remold, H. G. (1989) Blood 74,1396-1402 44. Cubellis, M. V., Wun, T.-C., and Blasi, F. (1990) EMBO J. 9 , 1079-1085 45. Chapman, H. A., Jr., Vavrin, Z., and Hihbs, J. B., Jr. (1982) Cell 28,653662 46. Saksela, O., Hovi, T. and Vaheri, A. J. (1985) J. Cell. Physiol. 1 2 2 , 125132 47. Hessler. J. R., Morel, D.W.,Lewis, L. J.. and Chisolm, G.M. (1983) Arterwsclemsis 3,215-222 48. Haimovitz-Friedman, A., Falcone, D. J., Vlodavsky, I., Schirrmacher, V., and Fuks, 2.(1991) Blood 78,789-796 49. Neufeld, G., and Gospodarowicz, D. (1985) J. Biol. Chem. 2 6 0 , 1386013868 50. Falcone, D. J., Mateo, N., Shio, H., Minick, C. R., and Fowler, S. D. (1984) J. Cell Biol. 99.1266-1274 51. Vlodavsky, I., Fuks, Z., Bar-Ner, M., Ariav, Y., and Schirrmacher, V. (1983) Cancer Res. 43,2704-2711 52. McCaffrey, T. A,, Falcone, D. J., and Du, B. (1992) J. Cell. Physiol. 1 6 2 , 430-440 53. Lawrence, D. A., Pircher, P., and Jullien, P. (1985) Biochem. Biophys. Res. Commun. 133,1026-1034 54. Bashkin, P., Doctrow, S., Klagsbrun, M., Svahn, C. M., Folkman, J., and Vlodavskv. I. (1989) Biochemistry 28,1737-1743 55. Vigny, M., Oilier-Hakmann, M. P.; Lavigne, M., Fayein, N., Jeanny, J. C., Laurent, M., and Courtois, Y. (1988) J. Cell. Physiol. 137,3217328 56. Madri, J. A,, Pratt, B.M., and Tucker, A.M. (1988) J. CellBtol. 106, 1375-1384 57. Muller, G., Behrens, J., Nussbaumer, U., Bohlen, P., and Birchmeier, W. (1987) Proc. Natl. Acad. Sci. U.S.A. 84,5600-5604 58. Assoian, R. K., Fleurdelys, B. E., Stevenson, H. C., Miller, P. J., Madtes, D., Raines, E.,Ross, R., and Sporn, M. B. (1987) Proc. Natl. Acad. Sci. U. S.A. 84,6020-6024 59. Pepper, M. S., Vassalli, J. D., Montesano, R., and Orci, L. (1987) J. Cell Biol. 106,2535-2541 60. Pollanen, J., Hedman, K., Nielsen, L. S., Dano, K., and Vaheri, A. (1988) J. Cell Bwl. 106.87-95 61. Meissauer, A,, Kramer, M. D., Hofman, M., Erkell, L. J., Jacob, E., Schirrmacher, V., and Brunner, G. (1991) Ex Cell Res. 192,453-459 62. Werb, 2. M., Banda, J., and Jones, P. A. (19aOP.J.Exp. Med. 1 6 2 , 13401157

Keski-Oja, J., and Moses, H. L. (1988) J. CellBwl. 1 0 6 , 63. Ly&i;R.M., 1659-1665 64. Sato, Y., Tsuboi, R., Lyons, R., Moses, H. L., and Rifkin, D. B. (1990) J. Cell Bwl. 11 1,757-763 65. Werb, Z., Mainardi, C. L., Vater, C.A., and Harris, E. D. (1977) N. Engl. J. Med. 2 9 6 , 1017-1023 66. Ha"ar K. A. Harpel P. C., Jaffe, E. A,, and Nachman, R. L. (1986) J. &ol: C h e d 2 6 1 , 1i656-11662 67. Plow, E. F., Freany, D. E., Plescla, J., and Miles, L. A. (1986) J. Cell Biol. 103,2411-2420 68. Silverstein, R. L., Friedlander, R. J., Nicholas, R. L., and Nachman, R. L. (1988) J. Clin. Znuest. 8 2 , 1948-1955 69. Hajjar, K. A., and Nachman, R. L. (1988) J. Clin. Invest. 8 2 , 1769-1778 70. Kirchheimer, J. C., Nong, Y.-H., and Remold, H. G. (1988) J. Zmmunol. 141,4229-4234 71. Kelly, J. L., Suenram, C. A,, Rogek, M. M., and Schwartz, C. J. (1988) Am. J. Pathol. 131,539-546 72. Saksela, O., Moscatelli, D., Sommer, A., and Rifkind, D. B. (1988) J. Cell Biol. 107,743-751 73. Andres, J. L., Stanley, K., Cheifetz, S., and Massague, J. (1989) J . Cell Biol. 109,3137-3145 74. Cheifetz, S . , and Massa e, J (1989) J. Biol. Chem. 2 6 4 , 12025-12028 M.[ and Ruoslahti, E. (1990) Nature 3 4 6 , 28175. Yamaguchl, Y., Mann, 284 76. McCaffrey, T. A., Falcone, D. J., Brayton, C. F., Agarwal, L. A., Welt, F. G. P., and Weksler, B. B. (1989) J. Cell Biol. 109,441-448 77. Paralkar. V. M..Vukicevlc. , S... and Reddi. A. H. (1991) Deu. Biol. 1 4 3 , 303-308 78. Mooradian D. L., Lucas, R. C., Weatherbee, J. A., and Furcht, L. (1989) J. Cell. Biochem. 4 1 , 189-200 79. Majesky, M. W., Lindner, V., Twardzik, D. R., Schwartz, S. M., and Reidy, M. A. (1991) J. Clin. Znuest. 88, 904-910 80. Lindner, V., Lappi, D. A,, Baird, A,, Majack, R. A., and Reidy, M. A. (1991) Circ. Res. 6 8 , 106-113 81. Lindner, V., and Reidy, M.A. (1991) Proc. Natl. Acad. Sci. U. S. A. 88, 3739-3743

F

~

~~

~~~