identification and characterization of human endothelial cell

THE JOURNAL OF B~OLOGWAL CHEMISTRY 0 1990 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 265, No. 5, Issue of February 15, pp. 2908-2916, 1990 Printed in L? S. A.

Identification and Characterization of Human Endothelial Cell Membrane Binding Sites for Tissue Plasminogen Activator and Urokinase*

(Received for publication, June 21, 1989)

Katherine A. Hajjar$g and Nancy M. Hamel From the Divisions of Hematology-Oncology, Departments of Pediatrics and Medicine, Cornell University Medical College, New York, New York 10021

Cultured human endothelial cells synthesize and secrete two types of plasminogen activator, tissue plasminogen activator (t-PA) and urokinase (u-PA). Pre- vious work from this laboratory (Hajjar, K. A., Hamel, N. M., Harpel, P. C., and Nachman, R. L. (1987) J. Clin. Invest. 80, 1712-1719) has demonstrated dose- dependent, saturable, and high affinity binding of t- PA to two sites associated with cultural endothelial cell monolayers. We now report that an isolated plasma membrane-enriched endothelial cell fraction specifically binds “‘I-t-PA at a single saturable site (& 9.1 nM; Bmax 3.1 pmol/mg membrane protein). Ligand blotting experiments demonstrated that both single and double-chain t-PA specifically bound to a M, 40,000 membrane protein present in detergent extracts of isolated membranes, while high molecular weight, low molecular weight, and single-chain u-PA associated with a M, 48,000 protein. Both binding interactions were reversible and cell-specific and were inhibitable by pretreatment of intact cells with nanomolar concentrations of trypsin. The relevant binding proteins were not found in subendothelial cell matrix, failed to react with antibodies to plasminogen activator inhibitor type 1 and interacted with their respective ligands in an active site-independent manner. The isolated t-PA binding site was resistant to reduction and preserved the capacity for plasmin generation. In contrast, the isolated u-PA binding protein was sensitive to reduction, and did not maintain the catalytic activity of the ligand on the blot. The results suggest that in addition to sharing a matrix-associated binding site (plasminogen activator inhibitor type l), both t-PA and u-PA have unique membrane binding sites which may regu- late their function. The results also provide further support for the hypothesis that plasminogen and t-PA can assemble on the endothelial cell surface in a manner which enhances cell surface generation of plasmin.

Vascular endothelial cells play a central role in f’ibrinolysis.

* This work was supported by National Institutes of Health Grant HL 42493, The Council for Tobacco Research, Inc., U. S. A. Grant 2169, and the American Heart Association, New York Affiliate. The costs of publication of this article were defrayed in part by the payment -of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 USC. Section 1734 solely to indicate this fact.

$ Established Investigator of the American Heart Association, and a Syntex Scholar (1989).

§ To whom correspondence should be addressed: Dept. of Medicine, Cornell University Medical College, 1300 York Ave., New York, NY 10021.

They synthesize and secrete tissue type plasminogen activator (t-PA)’ (1, 2), urokinase-like plasminogen activator (u-PA) (3, 4), and plasminogen activator inhibitor type 1 (PAI-1) (5, 6). Furthermore, endothelial cells provide binding sites which promote the assembly of plasminogen and its activators (7- 12). For example, the interaction of Glu-plasminogen with the cell surface is associated with a 12-fold increase in the catalytic efficiency of its activation by t-PA (7). The binding process involves conversion of the ligand to its truncated form, Lys-plasminogen, which indicates the presence of an active “plasmin-like” serine protease on the cell surface (9). Both plasminogen binding to endothelial cells and cell surface plasmin generation have recently been shown to be modulated by the multiple kringle-containing low density lipoprotein- like particle, lipoprotein(a) (10-E’).

Although both t-PA and u-PA interact specifically with endothelial cell monolayers (8, 13-16), relatively little is known of the identity of membrane-associated binding sites directed toward these activators. For urokinase, putative membrane “receptor proteins” ranging in M, from 40,000 to 60,000 have been reported on monocytoid U937 and HeLa cells (17,18). A receptor for u-PA has, however, not yet been identified on human endothelial cells. For t-PA, previous reports have suggested the presence of at least two binding sites on human endothelial cell monolayers (14, 15). The higher affinity site is PAI-related since its interaction with t- PA is inhibitable by excess quantities of u-PA (14), and since it formed SDS-stable 105,000-dalton complexes which im- munoprecipitated with antibody to PAI- (15). However, it is not clear whether this PAI- was derived from the endothelial cell membrane itself or from exposed portions of PAI-rich subendothelial cell matrix. The identity of the second, lower affinity t-PA binding site, which preserves its function, has not been well defined to date.

For a protease “receptor” to have functional significance, it should be capable of enhancing overall catalytic function. This could be accomplished by 1) promoting the local concentration of the protease, 2) permitting conformational changes or cleavage events which favorably affect lz,., or K,, or 3) protecting the protease from circulating inhibitors. Studies described herein have identified and characterized discrete

’ The abbreviations used are: t-PA, tissue plasminogen activator; u-PA, urokinase plasminogen activator; BSA, bovine serum albumin; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-l-propanesulfo- nate; DFP, diisopropyl fluorophosphate; Hepes, 4-(2-hydroxyethylj- l-piperazineethanesulfonic acid, HBS, Hepes-buffered saline; HUVEC. human umbilical vein endothelial cells; PMSF, phenyl- methylsulfonyl fluoride; PVDF, polyvinylidene difluoride; RCM, reduced and carboxymethylated; TBS, Tris-buffered saline; PAI, plasminogen activator inhibitor; SDS, sodium dodecyl sulfate; WGA, wheat germ agglutinin; PAGE, polyacrylamide gel electrophoresis.

by guest on February 15, 2018http://w

ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

Plasminogen Activator Binding Sites on Endothelial Cells 2909

binding sites for t-PA and u-PA in detergent extracts of isolated human vascular endothelial cell plasma membranes. While both proteins are externally oriented, active site-independent, and unrelated to matrix, they have clearly distinct properties including mobility in SDS gels, sensitivity to reduction, and cellular specificity. In addition, the t-PA binding protein, when immobilized on a solid support, promotes plasminogen activator activity, whereas the u-PA binding protein does not.

EXPERIMENTAL PROCEDURES

Materials-T-75 Nunc tissue culture flasks (Laboratory Disposable Products) were emuloved. DFP, PMSF. 1euneDtin. uenstatin A, BSA -- __ (essentially fatty alid- and globulin-free), CHAPS, and tissue culture- grade EDTA were from Sigma. Polyvinylidene difluoride (Immobilon) transfer membranes were purchased from Millipore. p-Nitro blue tetrazolium chloride, 5-bromo-4-chloro-3-indolylphosphate p-toluidine, and dithiothreitol were obtained from Bio-Rad. Iodoacetic acid was purchased from Eastman and trypsin from Difco.

P&$ed Proteins-Human recoibinant t-PA, 80-90% single- chain as iudeed bv SDS-PAGE. was nrovided bv Genentech. Sinele- chain an& high molecular weight human urinary urokinase, as well as two-chain Bowes melanoma human t-PA were purchased from American Diagnostica. Preparations of low molecular weight urokinase were obtained from American Diagnostica, Calbiochem, and Abbott Laboratories. Human Lys-plasminogen was supplied by Im- muno, Vienna, Austria. Purified C, esterase inhibitor, and affinity- purified rabbit anti C1 inhibitor, alkaline phosphatase-conjugated rabbit anti-human t-PA, alkaline phosphatase-conjugatedrabbit anti- human urokinase, and rabbit anti-human PAI- were provided bv Dr. Peter Harpel, Cdrnell University Medical College, Gew York,-NY. Rabbit anti-human PAI- was orovided bv Dr. E. K. 0. Kruithof. Centre Hospitalier Universitaiie Vaudois: Lausanne, Switzerland: Alkaline phosphatase-conjugated goat anti-rabbit IgG was from Cooner Biomedical. Wheat eerm aealutinin was nurchased from Phar- mac’la LKB Biotechnology inc. ly

Cell Culture-Early passage human umbilical vein endothelial cells (HUVEC) were cultured and quantified as described previously (7,9, 14).

Subcellular Fractionation-HUVEC (PZ-P6) grown in monolayer (20-40 T-75 flasks/run) were washed twice with HBS (nH 7.4) containing 1 mM PMSF, harvested by scraping into HBi/PMSF; washed twice in HBS (150 X g, 5 min, 4 “C), treated with DFP (19) (2 mM final concentration, 5 min, 4 “C), washed again twice, and resuspended in homogenization buffer (HBS/0.25 M sucrose containing 21 pM leupeptin, 1 mM PMSF, and 15 gM pepstatin A). The resulting cell suspension (1.35 f 0.12 X lo6 cells/flask; SE.; n = 9) was sonicated in a polypropylene tube at 4 “C! for five 10-s intervals (Micro-Ultrasonic cell disruptor, Kontes; power setting 8). The post- nuclear supernatant was collected (150 x g, 5 min, 4 ‘C), and centrifuged at 100,000 X g, 60 min, 4 “C in a SW 50.1 Beckman rotor. The microsomal pellet was either extracted with detergent or resuspended in 1 ml of homogenization buffer and applied to one or two 15-45% continuous linear sucrose gradients in lb-mM Tris, 1 mM EDTA, pH 7.4 (20, 21). The gradients were develoned at 200.000 x P. 90 min. 4 “C (SW 50.1), and 0.25-ml fractions were collected manially fro& the top. Fractions were either assayed immediately, or stored at 4 “C. In some experiments, HUVEC were treated with “‘I-wheat germ agglutinin (22, 23) (specific activity 7.73 x lo6 cpm/pg), washed twice prior to DFP treatment, and then sonicated.

Radioisotope Labeling-Human recombinant t-PA and human u- PA were radiolabeled by the lactoperoxidase method as described previously (24). Wheat germ agglutinin (WGA) was iodinated according to the chloramine-? method as described (23).

Radio&and Binding Studies-Radioligand binding studies were carried out in a fluid phase assay (25). Aliquots (20 ~1; 5-50 fig of protein) of various subcellular fractions were incubated (30 min, 4 “C) in duplicate in a total volume of 120 ~1 with lz51-t-PA (3-5 X lo5 cpm/ pmol; 1.6-32 nM) in a buffer containing HBS, 2 mM CaCl*, 1 rnk M&l,, and 5 ma/ml BSA. Eiahtv-microliter aliauots of the incubation mixture were layered over a-20&,1 20% sucroie cushion in 5 x 15- mm polypropylene tubes. The tubes were centrifuged at 49,100 x g (Sorvall SS-34 rotor), 180 min, and snap-frozen in dry ice. Radioac- tivity in the pellets (amputated tips) and supernatants (upper portion) was counted using a Searle 1185 y counter (Searle Radiographics). Specific binding was defined as that inhibited by a 50-fold excess of

unlabeled t-PA. The assay system was calibrated bv counting pelleted radioactivity from plasma membrane-enriched fractions de&d from ‘251-WGA-labeled HUVEC. Usine this svstem. at least 80% of the input radioactivity could be recovired inthe pellet, whereas 0.6% or less of unbound lZ51-WGA sedimented under the same conditions. Binding data were analyzed using the “Ligand” program (26).

Marker Enzyme Assays-Specific activities for lactate dehydrogenase, a cytosolic enzyme, and Na’-K’ ATPase and 5’-nucleotidase, plasma membrane markers, were measured as described (27-30). Total protein was determined by the method of Lowry et al. (31).

Detergent Extraction-Crude microsomal membranes or purified plasma membranes were extracted with 15 mM CHAPS (critical micelle concentration 4-6 mM) (32, 33) in 20 mM Tris, 100 mM NaCl, 1 mM PMSF, 1 mM leupeptin, and 1 mM pepstatin A (30 min, 21 “C). Extracted proteins were harvested by centrifugation (100,000 X g; 60 min; Beckman TLA 100).

SDS-PAGE-Detergent extracts were dissolved in 5 x sample buffer (5% SDS, 50 mM Tris, 5 mM EDTA, 25% sucrose (w/v), and 50 pg/ml bromphenol blue), and applied to 9.0% Laemmli slab gels with 3.9% stacking gels. The gels were run under nonreducing conditions overnight at constant power with cooling. Gels were either stained with Coomassie Brilliant Blue (0.1% (w/v) in 50% methanol, 10% acetic acid) and destained in 10% methanol, 7% acetic acid, or used in ligand blotting procedures.

Zmmunoligand Blotting-Proteins resolved on SDS-polyacrylamide gels were electrophoretically blotted onto polyvinylidene difluoride (PVDF) transfer filters in a tank system using a buffer containing 25 mM Tris, 190 mM glycine, and 20% (v/v) methanol (90 min, 800 mA, 21 “C) (34, 35). Nontransferred strips from the same gel were stained with Coomassie Blue. The filters were blocked with 150 mM NaCl, 20 mM Tris (TBS), containing 5% BSA with or without 0.2% nonfat drv milk (Carnation). Following three washes with TBS containing 0.1% BSA, the filters were incubated with ligand diluted in TBS containing 1% BSA and 3 mM CaCl? (60 min. 21°C). washed three times, incubited with the alkaline ph&phata&onjugated antibody probe, washed three times again, and developed with p-nitro blue tetrazolium chloride (180 pg/ml) and 5-bromo-4-chloro-3-indolylphosphate p-toluidine (90 pg/inl), 15 min, 21 “C (36).

Phmirwgen Activator Function Studies-The plasminogen activating capacity of t-PA and u-PA bound to blotted membrane proteins was tested in a fluorogenic assay. Following incubation with ligand, PVDF strips were washed extensively (three changes, 1 h each) in TBS containing 0.1% BSA (DH 7.4). sliced horizontallv into 5-mm pieces, and immersed in 16-m; plastic wells containing l”0 XIIM Tris, 1 mM EDTA, 29 nM Lys-plasminogen, and 100 pM D-Val-Leu- Lys-aminofluoromethylcoumarin (Enzyme Systems Products). At various time points (0, 12, and 24 min), aliquots were removed and substrate hydrolysis measured at excitation 400 nm, emission 505 nm, and slit widths 2 nm in a model 650-10s Perkin-Elmer fluorescence spectrofluorometer as described previously (7, 14). Initial rates of plasmin generation were calculated using linear regression analysis of plots of relative fluorescence units versus t2.

RESULTS

Interaction of Tissue Plasminogen Activator with Endothe- lial Cell Plasma Membranes-WGA is a plant lectin which binds specifically to surface polysaccharides on a wide variety of cells (22), including endothelial cells (37). When endothelial cells were pretreated with lZ51-WGA labeled to high specific activity (23) and then subjected to subcellular fractionation, the major peak of radioactivity appeared in a visible band corresponding to 26-28% sucrose (dig 1.1101-1.1195) (Fig. 1). This result agreed well with published reports describing isolation of plasma membranes from bovine and porcine endothelial cells (20, 21). To further quantify the degree of plasma membrane enrichment in these peak fractions, Na+-K+ ATP- ase activity was measured (Fig. 2). Peak fractions contained 30-60 nmol/rg/h ATPase activity, representing a 15-30-fold increase in specific activity over the post-nuclear supernatant. Furthermore, 5’-nucleotidase assays yielded a similar profile of enrichment in specific activity while lactate dehydrogenase activity in these fractions was negligible. Subcellular fractions were also evaluated for the ability to bind lz51-t-PA. Peak specific binding of t-PA coincided with peak ATPase activity


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

Plasminogen Activator Binding Sites on Endothelial Cells

- Gradient A 30,000 O---O Gradient B

1 2 4 6 8 10 12 14 16 IS 20

Fraction number

FIG. 1. Wheat germ agglutinin profile of fractionated crude membranes from prelabeled endothelial cells. Endothelial cells (2.6 X 10’) were pretreated with iz51-wheat germ agglutinin (specific activity 7.73 X lo6 cpm/pg; 5 min, 4 “C) and DFP (2 mM; 5 min, 4 “C), washed three times, and fractionated in the presence of protease inhibitors as described under “Experimental Procedures.” The 100,000 x g microsomal pellet was resuspended in 2 ml of homogenization buffer, applied to two linear 15-45% sucrose gradients, and centrifuged (200,000 x g, 90 min, 4 “C; Beckman SW 50.1). Fractions (0.25 ml) were collected manually from above and counted (Searle 1185 y counter). Lectin profiles from gradients run in duplicate are shown.

50

Fraction

FIG. 2. Na+-K+ ATPase and ‘*?-t-PA binding profiles of endothelial subcellular fractions. Microsomal pellets were prepared as described under “Experimental Procedures,” and further fractionated by sucrose density gradient centrifugation. Specific Na+- K+ ATPase activity was measured according to the methods of Wallach and Kamat (28) and Ames (29), and total protein determined according to the method of Lowry et ~2. (31). Specific binding of lZ51- t-PA (16.5 nM) was quantified in a fluid phase sucrose barrier assay as described under “Experimental Procedures.” All assays were carried out for density gradient fractions as well as the post-nuclear supernatant (S,), the post-microsomal supernatant (S,), and the microsomal pellet resuspended in 1 ml of homogenization buffer (P,).

and wheat germ-containing fractions (Fig. 2), and represented approximately 2.80 pmol of t-PA/mg of membrane protein upon exposure to 16.5 nM t-PA. These data indicated that, when compared with adjacent gradient fractions, those enriched in plasma membrane markers interacted preferentially with radiolabeled t-PA.

Upon exposure of lz51-t-PA to isolated HUVEC plasma

membranes, binding took place in a dose-dependent and saturable fashion (Fig. 3A). This interaction plateaued at t-PA concentrations of 15-40 nM, at which time binding to each milligram of membrane protein approached 1.47 pmol of t- PA (-100 ng). Computer analysis of three separate isotherms using the “Ligand” program (26), yielded a Kd of 9.1 f 1.5 nM (S.E., n = 3), and B,,, of 3.1 f 0.7 pmol/mg membrane protein (S.E., n = 3). These values agreed well with previously determined binding constants (Kd 18.1 f 3.8 nM and B,,, 815,000 f 146,000 sites/cell) for the major t-PA binding site on endothelial cell monolayers (14). In addition, binding oft- PA to the membrane-enriched fraction was inhibited by excess amounts of unlabeled t-PA (Fig. 3B). At a 22-ill-fold molar excess, 99% of lz51-t-PA binding was inhibited, indicating a high degree of specificity. The 15,, for competition by unlabeled t-PA represented a 3.5-fold molar excess, indicating that the two ligands competed nearly equally for membrane binding sites. It was concluded that Y-t-PA could bind specifically to isolated HUVEC plasma membranes. Further- more, binding affinity was similar to that associated with the major binding site on intact HUVEC monolayers (14).

Identification of Membrane Binding Sites-For identification of potential binding sites for t-PA and u-PA on endothelial cell membranes, immunoligand blotting was employed. Crude membranes or purified plasma membranes from endothelial cells were extracted with the zwitterionic detergent, CHAPS, at three times its critical micelle concentration. This procedure routinely resulted in solubilization of 43.9 + 4.5% (S.E., n = 8) of total membrane protein. Following SDS- PAGE, approximately 30 bands ranging in apparent molecular weight from 15,000 to 250,000 were visualized upon staining with Coomassie Blue (Fig. 4A, lane 1). After transfer to PVDF membranes, the resolved proteins were recovered with 80-90% efficiency as judged by Amido Black staining of the blot (Fig. 4A, he 2). Of these proteins, a single band reacted with rabbit antibody to the integral membrane protein, angio- tensin converting enzyme. Reactivity of antibody to laminin, a matrix-related protein, was negative, even though this pro-

I I I I lb ‘. IO 20 30 40 20 40 60 so 100

%-PA, nM Fold molar excess t-PA

FIG. 3. Binding of ‘261-t-PA to endothelial cell plasma membranes. A, binding isotherm. Aliquots of the peak plasma membrane- enriched fractions from endothelial cells (dfi 1.1147; 6.3 fig/assay tube) were incubated with varying concentrations of radioiodinated t-PA (2.4-36 UM; specific activity 447,400 cpm/pmoi) in the presence of 5 mg/ml BSA, 2 mM CaC&, and 1 mM MgCl, in a total volume of 120 ~1 as described under “Experimental Procedures.” Following incubation (30 min, 4 “C), 80-~1 aliquots of the incubation mixture were centrifuged through a 200-~1 20% sucrose barrier (49,100 X g, 3 h) in polypropylene rocket tubes, and snap-frozen. Pellets (amputated tips) and supernatants (upper portion) were counted in a Searle 1185 y counter. Each point represents the average of duplicate determinations. B, inhibition of binding by excess unlabeled t-PA. Aliquots of endothelial cell plasma membranes (6.3 @g/tube) were incubated with ‘Y-t-PA (3.2 nM; specific activity 447,400) in the presence of varying excess concentrations of unlabeled t-PA in a total volume of 120 ~1. ‘*‘I-t-PA binding was quantified as described under A. Each point represents the average of duplicate determinations.


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/


B 1 2 c1 2 D1

FIG. 4. Ligand blotting of endothelial cell membrane extracts. A, SDS-PAGE and electrophoretic transfer to PVDF filters of resolved extracted proteins. Endothelial cell plasma membranes were extracted with CHAPS (15 mM) as described under “Experi- mental Procedures,” dissolved in sample buffer, and subjected to SDS-PAGE on 9% Laemmli slab gels which were either stained with Coomassie Blue (lane I) or transferred to PVDF membranes and stained with Amido Black (0.05% (w/v) in 40% ethanol, 10% acetic acid) and destained with 90% methanol, 5% acetic acid (lone 2). B, t-PA ligand blot. PVDF membranes bearing transferred, extracted polypeptides from endothelial cell plasma membranes (100 nug, lane i) b; crude membranes (100 rg, lane 2) were blocked with 5% BSA, 0.2% nonfat drv milk in TBS (1 h. 37 “0. washed three times (TBS/ 0.1% BSA), and incubated with t-PA (147 nM, 1 h, 21 “C). The blots were then washed three times (TBS/O.l% BSA; 5 min/wash), incubated with alkaline phosphatase-conjugated immunoaffinity purified rabbit anti-human t-PA at 1.1 rg/ml in TBS/l.O% BSA/B mM CaC12 (1 h, 21 “C), washed again three times and developed as described under “Experimental Procedures.” C, u-PA ligand blot. Detergent extracts of endothelial cell plasma membranes (100 pg, lane 1) or crude membranes (100 rg, lane 2) were subjected to SDS-PAGE, transferred to PVDF membranes which were blocked with 5% BSA in TBS (1 h, 37 “C), and washed as above. The blots were incubated with u-PA (97 nM, 1 h, 21 “C), rewashed, and probed with alkaline phosphatase-conjugated rabbit anti-human u-PA at 0.53 pg/ml. D, C1 inhibitor ligand blot. A PVDF blot bearing transferred proteins from 100 pg of extracted endothelial cell plasma membrane-proteins was blocked with 5% BSA/TBS. washed. incubated with C, inhibitor (94 nM), rewashed, and incubated with alkaline phosphatase-conjugated rabbit anti-human Cr esterase inhibitor at 0.6 rg/ml.

tein was efficiently recognized by the detecting antibody following a cycle of SDS-PAGE and PVDF transfer.

To identify polypeptides reacting with plasminogen activators, blocked PVDF membranes were exposed to either t- PA or u-PA followed by alkaline phosphatase-conjugated rabbit anti-human t-PA or u-PA. In this system, t-PA (147 nM) reacted with a single major protein of apparent M, 40,000 from extracts of either purified plasma membranes or crude membranes (Fig. 4B). When ligand blotting was performed with lZ51-t-PA, identical results were obtained, and the M, 40,000 signal was completely quenched upon coincubation of the blot with a 500-fold excess of unlabeled t-PA but not by unlabeled excess fibrinogen (Fig. 5A). When increasing concentrations of ligand were employed and the resulting ligand blots analyzed by laser densitometry, the intensity of staining of the M, 40,000 polypeptide plateaued at t-PA concentrations between 15 and 735 nM. Half-maximal binding was achieved at t-PA concentration of -6.5 nM, which agreed reasonably well with binding affinity estimates for the interaction of 1251- t-PA with purified endothelial cell plasma membranes (& 9.1 nM; Fig. 3). These results suggested that blotted extracts of endothelial cell membranes contained a protein which could bind t-PA with characteristics similar to those observed for the major site associated with intact cell monolayers.

A 12 3

B 123

: ,.

1

FIG. 5. Radioligand blotting. Detergent extracts of crude endothelial cell membranes were prepared as described in the legend to Fig. 4. Proteins (100 pg/lane) were resolved on 9% Laemmli SDS gels and transferred to PVDF membranes as described. A, ‘Y-t-PA ligand blot. PVDF strips were blocked with 5% BSA, 1% nonfat dry milk in TBS (1 h, 37 “C), washed, and then incubated (1 h, 21 “C) with lz51- t-PA alone (2.8 nM; 461,000 cpm/pmol; lane I), ‘251-t-PA in the presence of unlabeled t-PA (500-fold molar excess; lane 2), or rz51-t- PA in the presence of unlabeled bovine fibrinogen (500-fold molar excess; lane 3). The blots were then washed, dried, and exposed (Kodak X-Omat AR; 72 h; -70 “C). B, Y-u-PA ligand blot. PVDF membranes containing extracted membrane proteins were blocked as described in the legend to Fig. 4C, washed, and incubated (1 h, 21 “C) with iz51-U-PA (3.2 nM, 270,000 cpm/pmol; he I), ‘*?-u-PA in the presence of unlabeled low molecular weight u-PA (500-fold molar excess; he 2), or iZ51-U-PA in the presence of unlabeled bovine fibrinogen (500-fold molar excess; lane 3). The blots were washed, dried, and exposed as described above for 96 h.

When u-PA (97 nM) was employed as the ligand, it associated with a broad band of apparent M, 48,000 found in extracts of both crude membranes and purified plasma membranes (Fig. 4C). In some experiments, this band resolved into two components of equal intensity. When ligand blotting was performed with ‘251-~-PA, identical results were obtained, and the M, 48,000 signal was completely quenched by coincubation of the blot with a 500-fold excess of unlabeled u-PA but not by excess fibrinogen (Fig. 5B). By densitometry, binding plateaued at u-PA concentrations between 10 and 485 nM, and was half-maximal at a u-PA concentration of 4.5 nM. This value agreed well with previous estimates of the binding affinity for the interaction of u-PA with endothelial cell monolayers (& -1 nM, Ref. 14).

In data not shown, various plasmin-modified forms of t-PA and u-PA were tested as ligands. Interestingly, both single and double-chain t-PA associated with M, 40,000 polypeptide, while the M, 48,000 band interacted with both high and low molecular weight u-PA as well as its single-chain form. Blot- ted extracts which were not incubated with t-PA or u-PA showed no bands which reacted with anti-u-PA or anti-t-PA under the same conditions. In additional control experiments, endothelial cell plasma membrane extracts failed to react with purified C1 esterase inhibitor when probed with alkaline phosphatase-conjugated rabbit anti-human C, esterase inhibitor antibody (Fig. 40). These results suggested that the observed interactions between t-PA and u-PA were protein-specific.

To assess the possibility that the M, 48,000 u-PA binding protein and the M, 40,000 t-PA binding protein might have been derived from subendothelial matrix contaminating the membrane preparation, additional ligand blotting experiments were carried out (Fig. 6). Equal amounts, based on total protein content, of solubilized subendothelial matrix and


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

2912 Plasminogen Activator Binding Sites on Endothelial Cells

A B C 12 12 12

55- 4)

1 -48

FIG. 6. Ligand blotting of subendothelial matrix proteins. Matrix was isolated from fifth passage endothelial cells. The cells were removed from 75cm2 flasks by soaking in HBS containing 1 mM EDTA (30 min, 21 “C). Matrix proteins were solubilized in 5 x SDS sample buffer for ligand blotting. Matrix proteins from parallel flasks were solubilized in 0.5 N NaOH for protein determinations (31). The final preparation (2.16 mg/ml) was loaded on a 9% Laemmli SDS gel (100 &lane; lanes Al, BI, Cl) with endothelial cell membrane- extracts-(2.77 ‘mg/ml) (100. Kg/lane; lanes A2, B2, C2) in adjacent lanes. A, anti-PAI- immunoblot. The PVDF membrane was prepared as described in the legend to Fig. 4, and probed with rabbit anti-human PAI- (4.2 &ml, 1 h, 21 “C) followed by goat anti-rabbit IgG (0.32 pg/ml, 1 h, 21 “C). B, t-PA ligand blot The blot was incubated with t-PA (147 nM) followed bv alkaline nhosnhatase- conjugated rabbit anti-human t:PA as described in the iegenb to Fig. 4B. C, u-PA ligand blot. The blot was incubated with u-PA (97 nM) followed by alkaline phosphatase-conjugated anti-human u-PA as described in the legend to Fig. 4C.

detergent extracts of endothelial cell membranes were evaluated for the presence of plasminogen activator binding proteins (Fig. 6B). When exposed to solubilized matrix, t-PA reacted primarily with a M, 55,000 band which was distinctly different from the major (Mr 40,000) t-PA binding protein identified in endothelial cell membranes. A comigrating M, 55,000 matrix protein reacted intensely with anti-PAI-I antibody (Fig. 6A, he 1). Although trace amounts of comigrating immunoreactive material were identified, in some experiments, in an equivalent quantity of membrane extract (Fig. 6A, lane 2), the M, 55,000 protein unequivocally migrated more slowly than the membrane protein (M, 40,000) reacting with t-PA (Fig. 6B, lane 2). u-PA showed no interaction with the matrix extract under these conditions (Fig. 6C, lane 1). However, when greater than 200 pg of matrix extract were loaded per lane (not shown), u-PA interacted with a M, 55,000 protein which was clearly distinguishable from its M, 48,000 membrane binding site (Fig. 6C, lane 2). These experiments indicated that the t-PA and u-PA binding proteins identified in the endothelial membrane preparations did not originate in subendothelial cell matrix.

To determine the specificity of these putative binding sites for t-PA and u-PA, membrane preparations from six different cell types were studied as controls (Fig. 7). Detergent extracts of crude membrane preparations from bovine endothelial cells, human foreskin fibroblasts, murine myeloma cells (P3X63AG8.653), human brain, and human erythrocytes were evaluated by ligand blotting. With extracts of human fibro- blast membranes (Fig. 7A, lane 4), t-PA reacted intensely with a M, 40,000 protein which was indistinguishable from the M, 40,000 protein found in extracts of human endothelial cell membranes (lane 6). Extracts of low passage bovine endothelial cell membranes also displayed this t-PA reactive band, but this was on average about 50% less intense than

FIG. 7. Ligand blotting of membrane extracts from control cells. Crude membrane fractions were prepared from human umbilical vein endothelial cells (passage 3; lone 6), bovine aortic endothelial cells (passage 4; lane 5), human foreskin fihroblasts (passage 13; lane 4), a murine myeloma cell line (lane 3), human brain (lane 2) (44), and human erythrocytes (lane 1) (45) and extracted with 15 mM CHAPS. The extracts (100 pg/lane) were run on a 9% Laemmli gel, transferred to a PVDF membrane, blocked, and incubated with either 147 nM t-PA (A) or 97 nM u-PA (B). The blots were developed as described in the legend to Fig. 4.

the band derived from human endothelial cells (Fig. 7A, he 5). Of the proteins extracted from mouse myeloma cell membranes, t-PA reacted weakly with a M, 48,000 doublet and a single M, 32,000 band. In extracts of human brain, t-PA interacted weakly with a single band migrating at M, 48,000, and two additional bands at M, 35,000. The significance of these weakly reacting bands is currently under study. t-PA did not react at all with proteins found in an equal amount of CHAPS extract from human erythrocytes (Fig. 7A, lane 1). When the same membrane extracts were exposed to u-PA (Fig. 7B), only human endothelial cells and bovine endothelial cells displayed reactive bands, both migrating at an apparent M, 48,000. These results indicated that the putative binding sites identified for t-PA (Mr 40,000) and u-PA (Mr 48,000) were relatively specific for cell type.

It has been previously reported that interactions of t-PA and u-PA with endothelial cells were largely reversible (8,14). To determine whether this was also the case for the interaction of these ligands with their putative membrane binding proteins, PVDF filters bearing membrane extract were incubated with t-PA or u-PA. They were then either developed immediately or washed for various periods of time prior to development (Fig. 8). By densitometry scanning of stained blots, 52% of bound t-PA dissociated from its major M, 40,000 binding protein after a 60-min wash, while 78% dissociated after an overnight wash. Similarly, 59% of bound u-PA dissociated from its M, 48,000 binding protein after a 60-min wash, and 67% after an overnight wash. When suspended endothelial cells were incubated with t-PA or u-PA prior to fractionation, no immunoreactive complexes were detected in membrane extracts upon exposure to alkaline phosphatase- conjugated anti-u-PA or anti-t-PA probes. These data suggested that both ligand binding site interactions were reversible and did not represent formation of covalent complexes at the cell surface.

We then investigated whether the membrane-derived binding proteins for t-PA and u-PA were oriented toward the fluid phase environment of the intact cell. Thus, endothelial cells were subjected to controlled proteolysis prior to membrane preparation (Fig. 9). When equivalent amounts of membrane extract (100 pg) from trypsin-treated cells were tested by ligand blotting with t-PA (Fig. 9A) or u-PA (Fig. 9B), a


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/


A

wash time O ’ 0 1

0-d FIG. 8. Reversibility of ligand binding. Extracts of endothelial

cell membranes (100 pg/lane) were resolved on SDS gels and transferred to PVDF membranes. The blots were incubated with 147 nM t-PA (A) or 97 nM u-PA (B) for 1 h, 21 “C, and then either exposed to antibody probes and developed (AO, BO), or washed for 1 h prior to antibody exposure and development (AI, El) as described in the legend to Fig. 4.

A 6

40- %qy

Trypsin ’ ’ ’ ’ ’ ’ (nM)

loo 25 0 100 25 0 FIG. 9. Controlled proteolysis of intact endothelial cells.

Endothelial cells were harvested and incubated (10 min, 21 “C) with 0, 25, or 100 nM trypsin in protease inhibitor-free HBS as indicated. The cell suspensions were treated with DFP (25 mM, 5 min, 4 “C), and detergent extracts of crude membranes prepared and subjected to SDS-PAGE (100 rg/lane) as described in the legend to Fig. 4. Ligand blotting was carried out with both 147 nM t-PA (A) and 97 nM u-PA (B).

progressive decline in ligand binding to both M, 40,000 and 48,000 proteins was observed following treatment with increasing concentrations of trypsin. At the maximum tolerated trypsin concentration (100 nM), ligand staining intensity was reduced by 80-90% for both t-PA and u-PA. Furthermore, when ‘251-trypsin was added to HUVEC harvested under the

same conditions, Cl% of the original radioactivity was found to be associated with the cell pellet after four successive washes indicating that no substantial internalization of the protease had occurred. These results indicated that both putative binding sites were located on the external face of the cell membrane, in a location accessible to fluid phase ligands.

Characterization of Binding Sites-To determine the functional correlates of the t-PA or u-PA binding site interaction, several approaches were taken. First, t-PA and u-PA were subjected to two successive 24-h treatments with 25 mM DFP (4 “C). When evaluated in a fluorogenic assay of plasminogen activation, both samples displayed <O.l% of the original plasminogen activating activity. DFP-t-PA and DFP-u-PA were then compared with t-PA and u-PA in terms of their ability to associate with putative binding proteins. As shown in Fig. IOA, DFP-t-PA and untreated t-PA associated in an

140 120 L

- 80 cd .E 60 c

5 c E 2 4 6 8 IO 12 14 16 18 20 22

.- 2 Slice number ii z -u-PA CT

.- E

B - RCM-U-PA

z 60 -DFP-U-PA

40

2 4 6 8 IO I2 14 16 18 2022

Slice number FIG. 10. Functional ligand blotting. Extracts of endothelial cell

membranes were resolved on SDS gels and transferred to PVDF membranes. The blots were incubated with either unmodified ligand, DFP-conjugated ligand, or reduced and carboxymethylated ligand, and then either probed with specific antibodies or assayed for plasmin-generating activity. A, t-PA ligand blot. Duplicate PVDF membranes containing extracted membrane proteins were incubated with 147 nM unmodified t-PA (O), DFP-t-PA (0), or RCM-t-PA (A) as indicated. One strip from each pair was stained with alkaline phosphatase-conjugated anti-t-PA as described in the legend to Fig. 48. The second strip from each pair was washed in 0.1% BSA/TBS (3 h; 4 “C), sliced into-5 mm segments. Each segment was placed in 400 ~1 of Tris/EDTA containing Lys-plasminogen and the fluorogenic plasmin substrate D-Val-Leu-Lys-aminofluoromethylcoumarin as described under “Experimental Procedures.” Relative fluorescence units were recorded for each segment at 0, 12, and 24 min, and rates of plasmin generation calculated from initial points on the rate curve (relative fluorescence units (NW)/ min’) as previously described (7, 14). B, u-PA ligand blot. Duplicate PVDF strips were incubated with 97 nM unmodified u-PA (O), DFP-u-PA (0), or RCM-u-PA (A) as indicated. One strip from each pair was developed with alkaline phosphatase-conjugated rabbit anti-human u-PA as described in the legend to Fig. 4C. The second strip was assayed for plasmin generating capacity as described above.


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

2914 Plasminogen Activator Binding Sites on Endothelial Cells

identical fashion with the M, 40,000 binding protein. This indicated a lack of active site involvement in this interaction. Reduced and alkylated t-PA, on the other hand, failed to interact with this binding site, indicating that this interaction was conformation-dependent. Similarly, while DFP-u-PA and untreated u-PA complexed with the M, 48,000 binding protein in an identical fashion, reduced and alkylated u-PA was a totally ineffective ligand (Fig. 10B). These findings suggested that while the overall tertiary structure of both t-PA and u- PA was necessary for binding, functionally intact active sites were not required for these interactions.

Additional experiments supported the conclusion that the active sites of t-PA and u-PA were not involved in their interaction with membrane binding sites. In experiments not shown, inactivation of t-PA by pretreatment with plasminogen activator inhibitor type 1 (PAI-1) failed to block the interaction of the ligand with its putative (M, 40,000) binding site. Likewise, pretreatment of u-PA had no effect on its ability to bind to its (M, 48,000) membrane protein. Further- more, when the blotted membrane extract was preincubated with three different preparations of rabbit anti-human PAI- or rabbit anti-PAI-2, binding by t-PA or u-PA to their putative membrane binding sites was unimpaired. These results reiterated our previous finding that the active site was not required for the interaction of t-PA or u-PA with their respective membrane binding sites.

Since these experiments verified that both binding interactions were active site-independent, the functional integrity of membrane-associated t-PA and u-PA was investigated (Fig. 10). Thus, PVDF strips bearing blotted membrane extracts were blocked, exposed to unmodified or modified t-PA or u- PA, washed for several hours to eliminate nonspecific background, and then sliced into &O-mm segments. The strips were then assayed for plasmin-generating activity. When t- PA was used as the ligand (Fig. lOA), a single peak of plasmin- generating activity was noted. This peak represented a 5.2- fold increase over the background rate of plasmin generation and comigrated with the M, 40,000 t-PA-reactive band. In experiments where a trace amount of lz51-t-PA was included with unlabeled ligand, 38.8 + 4.8 ng of t-PA/100 pg membrane extract were found associated with the 1M, 40,000 band. When plasminogen activator activity associated with this band was compared with standard curves, the observed activity was equivalent to 46.7 f 7.2 ng of fluid phase t-PA. These data suggested that the putative M, 40,000 membrane binding site preserved the activity of t-PA. Furthermore, when fluid phase samples from control strips were tested for t-PA activity, no discrete peak was identified. This indicated that the observed focus of plasminogen activating activity did not represent dissociation of t-PA from the blot into the fluid phase. DFP- t-PA, as expected, showed no activity, even though the binding pattern obtained with this ligand was identical to that of the unmodified ligand. Reduced and carboxymethylated t-PA (RCM-t-PA) also showed no catalytic activity and failed to react with any proteins on the ligand blot. These data supported the previous finding that the interaction of t-PA with the M, 40,000 protein was active site-independent and suggested that the conformational structure of t-PA was important for binding. The data also suggested that t-PA associated with this protein retained its ability to activate plasminogen.

When blot-associated u-PA was tested by fluorogenic assay, no distinct peak of activity was observed on the PVDF strip. This occurred even though parallel strips stained with conjugated antibody displayed a broad band at M, 48,000 (Fig. 10B). DFP-u-PA associated with an n/l, 48,000 band in a manner indistinguishable from that of untreated u-PA. As

expected, DFP-u-PA was totally inactive on the blot. Like RCM-t-PA, RCM-u-PA showed no catalytic activity, and also failed to bind to the blot. Blots not exposed to either ligand were devoid of plasmin-generating activity. These data suggested that, like t-PA, u-PA binding to its putative cell surface protein was active site-independent, but conformation-dependent with respect to the ligand. However, unlike t-PA, when associated with its isolated, extracted binding protein, u-PA displayed no activity over base line. Trace-label experiments with ‘251-~-PA revealed that while 4.6 +. 1.0 ng of u- PA/100 pg of membrane extract were found associated with the M, 48,000 band, the associated catalytic activity was equivalent to only 0.18 f 0.03 ng of fluid phase u-PA. Thus, u-PA bound to the M, 48,000 lost most of its catalytic activity.

The data obtained in the above experiments suggested that the conformational structures of both t-PA and u-PA were essential for interaction with their respective binding sites. To determine whether the functional integrity of the two binding sites might depend upon intact disulfide-linked subunit structures, reduced and nonreduced endothelial cell membrane extracts were compared for their ability to interact with t-PA and u-PA (Fig. 11). When t-PA was exposed to reduced endothelial cell membrane extracts in a ligand blotting system, it interacted with a single band of apparent M, 42,000 (Fig. llA, lane 2). This band was clearly distinguishable from the M, 40,000 binding site identified in nonreduced extracts (lane 1). Interestingly, t-PA associated with this M, 42,000 polypeptide retained its ability to activate plasminogen (not shown). In contrast, the M, 48,000 u-PA-binding protein identified in nonreduced endothelial cell membrane extracts (Fig. llB, lane I), completely lost its ability to bind u-PA upon reduction with dithiothreitol (lane 2). Following incubation with u-PA, no discrete peak of plasminogen activating activity could be detected in a fluorogenic assay in blots of either reduced or nonreduced extracts. These data suggested

-42

40-

FIG. 11. Ligand blotting of reduced and nonreduced endothelial cell membrane extracts. Nonreduced (lanes AI, Bl) or reduced (530 mM dithiothreitol, 100 “C, 5 min) (lanes A2, 82) detergent extracts of endothelial cell membranes were subjected to ligand blotting with either 147 nM t-PA (A) or 97 nM u-PA (B). The blots were developed as described in the legend to Fig. 4.


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/


that the u-PA binding site may have a distinct disulfide- linked subunit structure which is critical for ligand binding. In contrast, the interaction of t-PA with its M, 40,000 binding site was unaffected by reduction of the membrane extract. This suggested a lack of dependence upon interchain disulfide bonds.

DISCUSSION

These experiments have identified, for the first time, endothelial cell membrane binding sites for tissue plasminogen activator and urokinase. Although previous reports in the literature indicate that t-PA and u-PA can bind to PAI- associated with endothelial cell monolayers (15, 16), our results indicate that these ligands can bind to sites which are plasma membrane-associated. These newly described membrane binding sites possess strikingly disparate structural and functional properties.

Studies of direct binding of radiolabeled t-PA to endothelial subcellular fractions indicated that t-PA preferentially interacted with a fraction enriched in plasma membranes (Fig. 2). Binding isotherms depicting the interaction of lz51-t-PA with this membrane fraction suggested the presence of a single, saturable site with a Kd (9.1 f 1.5 nM), similar to that observed for the major site on endothelial cell monolayers (18.1 f 3.8 nM) (14). Our study revealed a B,,, of 3.1 f 0.7 pmol/mg membrane protein. Based on an average protein yield in the plasma membrane-enriched fraction of 126 pg/l X lo6 cells, this value represents a binding capacity of approximately 235,000 sites/cell. Given the limits of accuracy associated with this type of calculation, this binding capacity agrees reasonably well with the B,,, of 815,000 sites/cell obtained in our previous report (14). The 235,000 value may, in fact, represent a minimum estimate for several reasons. First, some denatur- ation or dissociation of binding sites may have occurred during the membrane preparation procedure. Second, transmission electron microscopy of the membrane fraction suggested that as much as 15-20% of the larger membrane vesicles may have resealed around smaller vesicles. Third, a portion of the isolated membranes may have resealed with an inside-out orientation due to exposure to the relatively low ionic strength buffer employed in the sucrose gradients (38, 39). Under the latter two conditions, externally oriented binding sites might become relatively inaccessible to the fluid phase ligand.

Experiments reported herein suggest that the primary endothelial cell membrane binding site for t-PA has an apparent M, of approximately 40,000, and interacts reversibly with t- PA in a ligand blotting system. This binding protein is sus- ceptible to controlled trypsin treatment, and, therefore, is most likely oriented on the external face of the plasma membrane. In addition to human umbilical vein endothelial cell membrane extracts, comigrating ligand-reactive proteins were also detected in control membrane extracts of low passage bovine aortic endothelial cells and human foreskin fibroblasts, but not human erythrocytes. While the active site of t-PA was not involved in binding, the overall tertiary structure of t-PA appeared to be essential for interaction with the membrane-associated binding protein, since RCM-t-PA failed to interact with the M, 40,000 protein. Upon reduction, the blotted t-PA binding site migrated close to its unreduced position with an apparent M, of 42,000, and retained its ability to bind the ligand and support the efficient generation of plasmin.

The u-PA binding site, on the other hand, migrated as a broad band or doublet with an apparent M, 48,000. This protein interacted reversibly with u-PA and, like the t-PA binding site, lost its ligand binding capacity upon controlled

trypsin treatment of the intact cell. In addition to human endothelial cell membrane extracts, a comigrating u-PA-reactive protein was also detected in membrane extracts from bovine endothelial cells, but not human fibroblasts, mouse myeloma cells, human brain, or human erythrocytes. Human red cells have been previously reported to lack specific binding sites for u-PA (40). Although binding was active site-independent, the intact, disulfide-linked structure of u-PA appeared to be essential for binding to the M, 48,000 protein. In addition, u-PA bound to the detergent-extracted, blotted polypeptide showed no enhancement of activity. When reduced extracts of endothelial cell membranes were exposed to u-PA, no binding occurred. This suggested that binding site integrity was dependent upon interchain disulfide linkages.

Previous reports in the literature (15, 16) have suggested that plasminogen activator inhibitor type 1 can serve as a high affinity binding site for t-PA exposed to endothelial cells cultured in monolayer. Several published studies suggest that endothelial cells in monolayer may have substantial amounts of exposed subendothelial cell matrix, since subpassaged endothelial cells do not form the same types of junctional complexes found in uiuo (41-43). Although freshly isolated primary human umbilical vein endothelial cells may display gap and tight junctions and exhibit electrotonic transfer (41), passaged endothelial cells seem to cover the available culture surface incompletely (42). With repeated subpassage, furthermore, numbers of tight junctions have been observed to de- crease while intercellular fenestrations increased (43). Thus, depending on culture conditions, considerable amounts of PAI-rich matrix may, in fact, be available to fluid phase ligands such as t-PA and u-PA. Our studies, however, indicate the presence of membrane-associated, detergent-extractable binding sites for t-PA and u-PA which are not related to PAI- 1.

Several lines of evidence demonstrate that the major (Mr 40,000 and 48,000, respectively) membrane-associated binding sites identified here do not represent known immunoreactive or functional forms of plasminogen activator inhibitors. First, anti-PAI- immunoreactive polypeptides failed to comigrate with primary binding sites for either t-PA or u-PA. Second, neither binding site was active site-dependent, since DFP- treated and untreated activators interacted identically with these polypeptides. Third, neither pretreatment of blotted membrane proteins with three different preparations of anti- PA1 nor pretreatment of the activators with PA1 had any effect on ligand blotting of the binding sites by t-PA or u-PA. Fourth, both interactions were reversible, whereas plasminogen activators interact with their inhibitors in an irreversible fashion. Fifth, in the case of the t-PA binding site, catalytic activity of the ligand was preserved upon binding.

Cell surface binding proteins for urokinase have now been identified in a variety of cell types (17, 18). Nielsen et al. (17) identified a receptor protein for u-PA in seven neoplastic cell types. This receptor protein migrated at an apparent M, of 55,000-60,000 on SDS gels, was resistant to reduction, bound via the amino-terminal fragment of u-PA (46), and was up- regulated in the presence of phorbol myristate acetate. Es- treicher et al. (18) confirmed and extended these findings, reporting a cell surface protein of M, 45,000 f 5,000 which, as previously reported for U937 cells (47), preserved the catalytic activity of u-PA. Although it has a similar apparent molecular weight, the M, 48,000 u-PA binding protein identified in our study appears to differ from previously reported u-PA receptors in three important respects: 1) In contrast to the U937 cell protein (17), the endothelial cell protein we studied lost its ligand binding capacity upon reduction. 2) The


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/


isolated protein reported here did not promote the catalytic activity of u-PA on the blot. A previous report suggests that u-PA associated with intact U937 or HeLa cells retains its catalytic activity (18). 3) The M, 48,000 protein bound three different preparations of low molecular weight u-PA and high molecular weight u-PA equally well, suggesting that binding was not mediated by the amino-terminal portion of u-PA, unlike the other reported u-PA binding sites (17, 18). Thus, the binding site described here may constitute a novel, pos- sibly endothelial cell-specific, cell surface binding site for urokinase.

16. Sakata, Y., Okada, M., Noro, A., and Matsuda, M. (1988) J. Biol. Chem. 263.1960-1969

In summary, our data describe discrete, membrane-associated, plasminogen activator binding sites on the endothelial cell surface. These membrane proteins may play a role in regulating localized generation of plasmin at the surface of the vessel wall. In addition, the findings presented here support the hypothesis that the human fibrinolytic system, as represented by plasminogen and tissue activator, can assemble in a functionally significant manner on the endothelial cell surface.

Acknowledgments-We are indebted to Drs. Ralph L. Nachman, Aaron J. Marcus, and David P. Hajjar for helpful discussions.

5.

6.

7.

8.

9.

10.

11.

12.

13. 14.

15.

REFERENCES Bachmann, F., and Kruithof, E. K. 0. (1984) Semin. Thromb.

Hemostasis. 10,6-17 Levin, E. G., and Loskutoff, D. J. (1982) J. Cell Biol. 94, 631-

636 Booyse, F. M., Osikowicz, G., Feder, S., and Scheinbuks, J. (1984)

J. Biol. Chem. 259, 7198-7205 Van Hinsbergh, V. W. M., Binnema, D., Scheffer, M. A., Spren-

gers, E. D., Kooistra, T., and Rijken, D. C. (1987) Arterioscle- rosis 7, 389-400

van Mourik, J. A., Lawrence, D. A., and Loskutoff, D. J. (1984) J. Biol. Chem. 259, 14914114921.

Kruithof. E. K. 0.. Tran-Thane. C.. Ransiin. A.. and Bachmann. F. W. (1984) Blood 64, 907-%3 ’ * ’ ’

Hajjar, K. A., Harpel, P. C., Jaffe, E. A., and Nachman, R. L. (1986) J. Biol. Chem. 261, 11656-11662

Miles, L. A., Levin, E. G., Plescia, J., Collen, D., and Plow, E. F. (1988) Blood 72, 628-635

Haiiar. K. A.. and Nachman. R. L. (1988) J. Clin. Invest. 82, 1769-1778

Gonzalez-Gronow, M., Edelberg, J. M., and Pizzo, S. V. (1989) Biochemistrv 28.2374-2377

Miles, L. A., Fless,.G. M., Levin, E. G., Scanu, A. M., and Plow, E. F. (1989) Nature 339,301-303

Hajjar, K. A., Gavish, D., Breslow, J. L., and Nachman, R. L. (1989) Nature 339, 303-305

Beebe, D. B. (1987) Thromb. Res. 46, 241-254 Haiiar, K. A., Hamel, N. M., Harpel, P. C., and Nachman, R. L.

(19g7) J. Clin. Invest. 80,.171211719 Barnathan. E. S.. Kuo. A.. Van der Kevl. H.. McCrae, K. R..

Larsen, G. R., and Cines, D. B. (1988) ‘J. kol. Chen. 263; 7792-7799

17.

18.

19.

20.

21. 22.

23.

24.

25.

26.

27.

28.

29. 30.

31.

32.

33.

34.

35.

36. 37.

38.

39.

40.

41.

42.

43.

44. 45. 46.

47.

Nielsen, L. S., Kellerman, G. M., Behrendt, N., Picone, R., Dano, K., and Blasi, F. (1988) J. Biol. Chem. 263, 2358-2363

Estreicher, A., Wohlwend, A., Belin, D., Schleuning, W.-D., and Vassalli, J.-D. (1989) J. Biol. Chem. 264, 1180-1189

Marcus. A. J.. Safier. L. B.. Ullman. H. L.. Islam. N.. Broekman. M. J.; and van Schacky, k. (1987) J. Clk. Znue&. ?9,179-187

Magaral, W. W., Dickinson, E. S., and Slakey, L. L. (1978) J. Biol. Chem. 253,8311-8318

Teitel, J. M. (1986) J. Cell. Physiol. 128, 329-336 Bennett, V., and Cuatrecasas, P. (1973) Biochim. Biophys. Acta

3 11,362-380 Chang, K.-J., Bennett, V., and Cuatrecasas, P. (1975) J. Biol.

Chem. 250,488-500 Martin, B. M., Wasiewski, W. W., Fenton, J. W., and Detwiler,

T. C..(1976) Biochemist& 15, 4886-4893 Bennett. V.. and Branton. D. (1977) J. Biol. Chem. 252, 2753-

2763 Munson, P. J., and Rodbard, D. (1980) Anal. Biochem. 107,220-

239 Wroblewski, F., and LaDue, J. S. (1955) Proc. Sot. Exp. Biol.

Med. 90,210-213 Wallach, D. F., and Kamat, V. B. (1966) Methods Enzymol. 8,

164-172 Ames, B. (1962) Methods Enzymol. 8, 115-118 Avruch, J., and Wallach, D. F. H. (1971) Biochim. Biophys. Acta

233,334-347 Lowrv. 0. H., Rosebrouah. N. J., Farr. A. L., and Randall, R. J.

(1951) J. Biol. Chem. ig3, 265-275 Hielmeland. L. M. (1980) Proc. Natl. Acad. Sci. U. S. A. 77.

$368-6376 ’ Hjelmeland, L. M., and Chrambach, A. (1984) Methods Enzymol.

104,305-318 Daniel, T. 0.. Schneider, W. J., Goldstein, J. L., and Brown, M.

S. (1983) J; Biol. Chem. 258; 4606-4611 Beisiegel, U., Schneider, W. J., Brown, M. S., and Goldstein, J.

L. (1982) J. Biol. Chem. 257, 13150-13156 Turner, B. M. (1986) Methods Enzymol. 121, 848-855 Simionescu. M.. Simionescu. N., and Palade. G. (1982) J. Cell

Biol. 94,406:413 Steck, T. L., Weinstein, R. S., Straus, J. H., and Wallach, D. F.

H. (1970) Science 168, 255-257 Bennett, V., and Branton, D. (1977) J. Biol. Chem. 252, 2753-

2763 Miles, L. A., and Plow, E. F. (1987) Thromb. Haemostas. 58,

936-942 Larson, D. M., and Sheridan, J. D. (1982) J. Cell Biol. 92, 183-

191 Haudenschild, C. C., Cotran, R. S., Gimbrone, M. A., Jr., and

Folkman, J. (1975) J. Ultrastruct. Res. 50, 22-32 Haudenschild, C. C. (1984) in Biology of Endotheliul Cells (Jaffe,

E. A.. ed) DD. 129-140. Martinus Niihoff. Boston Davis, J., and-Bennett, V. (1983) J. Biol. &em. 258, 7757-7766 Bennett, V. (1983) Methods Enzymol. 96, 313-324 Appella, E., Robinson, E. A., Ullrich, S. J., Stoppelli, M. P., Corti,

A., Cassani, G., and Blasi, F. (1987) J. Biol. Chem. 262,4437- 4440

Vassalli, J.-D., Baccino, D., and Belin, D. (1985) J. Cell Biol. 100,86-92


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

K A Hajjar and N M Hamelsites for tissue plasminogen activator and urokinase.

Identification and characterization of human endothelial cell membrane binding

1990, 265:2908-2916.J. Biol. Chem.

http://www.jbc.org/content/265/5/2908Access the most updated version of this article at

Alerts:

When a correction for this article is posted•

When this article is cited•

to choose from all of JBC's e-mail alertsClick here

http://www.jbc.org/content/265/5/2908.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/content/265/5/2908

http://www.jbc.org/cgi/alerts?alertType=citedby&addAlert=cited_by&cited_by_criteria_resid=jbc;265/5/2908&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/265/5/2908

http://www.jbc.org/cgi/alerts?alertType=correction&addAlert=correction&correction_criteria_value=265/5/2908&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/265/5/2908

http://www.jbc.org/cgi/alerts/etoc

http://www.jbc.org/content/265/5/2908.full.html#ref-list-1

http://www.jbc.org/

identification and characterization of human endothelial cell

Documents