THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 267, NO. 4, Issue of February 5, pp. 2451-2458,1992 Printed in U. S. A.

A 130-kDa Protein on Endothelial Cells Binds to Amino Acids 15-42 of the B@ Chain of Fibrinogen*

(Received for publication, September 3, 1991)

John K. ErbanSg and Denisa D. WagnerSllII From the $Center for Hemostasis and Thrombosis Research, Division of Hematology/Oncology, New England Medical Center and the llDepartment of Anatomy and Cellular Biology, Tufts University School of Medicine, Boston, Massachusetts 021 1 I

Factors which stimulate the release of von Wille- brand factor (vWf) from endothelial cell Weibel-Palade bodies and which induce the expression of the leuko- cyte-binding adhesion molecule P-selectin (PADGEM, GMP-140, CD62) on the endothelial cell surface re- main incompletely characterized. Fibrin but not fibrin- ogen is a potent stimulus for the release of stored von Willebrand factor from endothelial cells. Removal of fibrinopeptides A and B from fibrinogen occurs during the formation of fibrin, and the removal of fibrinopep- tide B is a requirement for fibrin to induce vWf secre- tion. The cleavage of fibrinopeptide A by reptilase enzyme forms a fibrin gel yet it is incapable of stimu- lating Weibel-Palade body degranulation. As a conse- quence of removing fibrinopeptide B, BB15-42 be- comes the new NH2 terminus of the /3 chain of fibrin. We have shown that the peptide B/315-42 in solution inhibits the release of vWf stimulated by fibrin. In addition, B/315-42 coupled to ovalbumin supports the binding and spreading of endothelial cells, while a scrambled form of this peptide coupled to the same carrier does not. We investigated whether these deter- minants near the amino terminus of the /3 chain of fibrin bind to a specific protein on the surface of endo- thelial cells. A 130-kDa protein was isolated from surface-labeled human umbilical vein endothelial cells by specific binding to Bj315-42 immobilized on Seph- arose. This glycoprotein was eluted with the B/315-42 peptide in solution but not with the scrambled form of this peptide. The fibrin-derived peptides B/319-26 and B837-56-cysteine were also incapable of eluting the 130-kDa protein bound to immobilized Bj315-42 as were the arginine-glycine-aspartic acid-serine RGDS tetrapeptide and EDTA. The 130-kDa protein is rec- ognized neither by antibodies to the known integrins found on endothelial cells nor by antibodies to CD3l(endoCAM, PECAM-l), a member of the immu- noglobulin family of receptors found on endothelial cells. The j 3 chain of fibrin thus contains a sequence near its amino terminus which specifically binds to what is likely a novel endothelial cell surface protein.

* This investigation was supported by National Heart, Lung and Blood Institute Grant PO1 HL42443 and Individual National Re- search Service Award HL08188. An abstract of this work was pre- sented at the 31st annual meeting of the American Society of He- matology, December 1989. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Recipient of a National Institutes of Health Clinical Investigator Award from the National Heart, Lung and Blood Institute. To whom reprint requests should be addressed Div. of Hematology/Oncology, New England Medical Center, 750 Washington St., Box 832, Boston, MA 02111. Tel.: 617-956-5650.

11 Established Investigator of the American Heart Association.

This glycoprotein may promote endothelial cell adhe- sion to fibrin during the wound healing process and is a candidate for a receptor involved in fibrin-mediated release of Weibel-Palade bodies from endothelial cells.

When a blood vessel is injured, the process of vascular healing begins immediately. Fibrin plays a critical role in the acute response to injury through stabilization of platelet ag- gregates and formation of mural thrombi. Fibrin also serves to stimulate a number of cellular processes which are neces- sary for wound healing and vascular repair. Upon stimulation with fibrin, endothelial cells rapidly release von Willebrand factor (vWf)’ from storage granules known as Weibel-Palade bodies (1). This response can also be induced by thrombin, histamine, and components of complement (2-5). The storage pool of vWf is mobilized within minutes and becomes acces- sible to platelets for adhesion to the vessel wall (6). In addi- tion, an adhesion molecule for leukocytes known as P-selectin (PADGEM, GMP-140, CD62), a membrane component of Weibel-Palade bodies (7, 8) is rapidly translocated to the cell surface, presumably to participate in the inflammatory re- sponse.

The fibrinogen molecule is composed of pairs of Aa, BO, and y chains linked by disulfide bonds (9). Cleavage of fibri- nopeptide A from Aa chains produces a net change in charge which allows individual fibrin monomers to polymerize and form a fibrin network (10,l l) . While the snake venom enzyme reptilase cleaves only fibrinopeptide A, thrombin cleaves both fibrinopeptides A and B from fibrinogen (12, 13). The addi- tional cleavage of fibrinopeptide B, while not critical for gel formation, allows the amino terminus of the p chain to asso- ciate with the D domain of a second fibrin monomer. This cleavage has been associated with several responses in endo- thelial cells including the rapid and complete release of vWf (14) and the increased production of tissue plasminogen ac- tivator and prostacyclin (15, 16). A receptor for determinants near the amino terminus of fibrin on endothelial or hemato- poietic cells has not yet been demonstrated. Shainoff et al. (17) have proposed, however, that this region of the molecule is involved in the binding of fibrin to mononuclear cells in experiments using the soluble NHp-terminal disulfide knot prepared after cyanogen bromide cleavage of fibrin.

Since previous studies of Ribes et al. (1) demonstrated that fibrin but not fibrinogen induced a rapid release of stored von

The abbreviations used are: vWf, von Willebrand factor; RGDS, arginine-glycine-aspartic acid-serine tetrapeptide; PBS, phosphate- buffered saline; PBS’’, phosphate-buffered saline with 100 mg/liter of both calcium chloride and magnesium chloride; SDS, sodium dodecyl sulfate; TPCK, L-1-tosyl-amide-2-phenylethyl chloromethyl ketone; WGA-Sepharose, wheat germ agglutinin-Sepharose.

2451

2452 A Receptor for Fibrin Bp15-42 on Endothelial Cells

Willebrand factor from cultured human umbilical vein endo- thelial cells, we were interested in investigating whether the newly exposed determinants near the amino terminus of the p chain bind to a specific protein on the surface of endothelial cells. We have now identified a 130-kDa protein which binds specifically to the fibrin derived peptide Bp15-42.

EXPERIMENTAL PROCEDURES

Cell Culture-Human umbilical cords less than 48 h old were cannulated and endothelial cells detached from basement membrane using 0.5% Pronase (Calbiochem Corp., San Diego, CA) as described (18,19). Cells were passaged no more than twice before use. Passaged cells were plated in flasks coated with 0.2% gelatin, and medium was supplemented with 100 pg/ml heparin (Sigma) and 50 pg/ml endo- thelial cell mitogen (Biomedical Technologies Inc., Stoughton, MA). Multiply passaged human foreskin fibroblasts and mouse 3T3 cells were grown in Dulbecco's modified Eagle's medium with 10% fetal calf serum.

'''I Surface Lubeling-Confluent 75-cm2 flasks were washed three times with phosphate-buffered saline containing 100 mg/liter CaC12 and 100 mg/liter MgC1, (PBS2+). Cells were labeled using a modifi- cation of the lactoperoxidase method (20). To each 75-cm2 flask, 1.2 mCi of lZ5I (629 GBq/mg, 100 mCi/ml, Du Pont-New England Nu- clear) in 3 ml of PBS" with 5 mM (3-D-glucose was added. Iodination was initiated by adding 150 p1 of a mixture of lactoperoxidase (Cal- biochem Corp.), l mg/ml, and glucose oxidase (Type VII-S, Sigma), 10 units/ml, in PBS". The reaction continued for 10 min at room temperature. Flasks were washed twice with phosphate-buffered io- dide and cells were lysed in lysing buffer (0.025 M Tris-HC1, 0.14 M NaCl, 0.5 mM MnC12, 0.5 mM CaC12, 200 mM (3-D-octylglucopyrano- side (Calbiochem), 3 mM phenylmethylsulfonyl fluoride). Lysate from 3 to 5 flasks was collected and centrifuged at 12,000 X g for 30 min at 4 "C and then at 40,000 X g for 30 min. 10 pl of cell lysate typically contained 250,000-800,000 cpm.

Peptide Synthesis-Sequences of B(319-26 and Bf315-42 were ob- tained from published data (21). Bp15-42S sequence was generated by randomly reassigning the order of amino acids in B(315-42. Syn- thetic peptides were synthesized on an Applied Biosystems 430A Peptide Synthesizer. Peptides were purified on Waters reverse-phase high performance liquid chromatography using a Bio-Rad C18 pre- parative column. Gradients were generated using 0.1% trifluoroacetic acid in water and 0.1% trifluoroacetic acid in acetonitrile, and peptide elution was monitored at 214 nm. Peak fractions were collected and lyophilized. Purified peptide was sequenced on an Applied Biosystems 470A Protein Sequencer to confirm sequence and determine purity.

Preparation of Peptide Conjugates-Peptide conjugates were pre- pared using the following method. Five mg of ovalbumin (grade V, Sigma) was dissolved in 5 ml of PBS for each conjugate to be prepared. B(315-42 or scrambled peptide was added in a final molar ratio of 201 peptide/ovalbumin. The solutions were cooled to 4 "C. Two % glutaraldehyde (G 5882, Sigma) was prepared fresh by diluting the 25% stock with distilled water, and 5 ml was added dropwise with constant stirring. The reaction was continued for 1 h at 4 "C and then stopped by adding sodium borohydride to a final concentration of 10 mg/ml. The mixtures were kept at 4 "C for an additional hour and then dialyzed against four changes of PBS over 48 h. Conjugates were analyzed by SDS-polyacrylamide gel electrophoresis and stored at -80 "C.

Endothelial Cell Binding Assays-96-well plates (non-tissue culture treated, Costar Corp., Cambridge, MA) were coated with 50 pl/well of peptide conjugate solutions or ovalbumin in PBS for 24 h at 4 "C. Ovalbumin was used as the control to determine background binding. After 24 h wells were aspirated dry and 200 pl of PBS with 0.5% ovalbumin were added for an additional 24 h. Wells were then aspirated and endothelial cell medium was added for 15 min at 37 "C in 5% C02 atmosphere. First or second passage endothelial cells in a single 75-cm2 flask were labeled for 48 h with 5 pCi/ml meth~l-[~H] thymidine (2.7 TBq/mMol, 185 MBq/5 ml, Du Pont-New England Nuclear). The cells were then detached using trypsin-EDTA, pelleted, and resuspended in 50 ml of fresh medium. 200 pl of cell suspension were added to each well and incubated for 2 h a t 37 "C The wells were then washed four times with medium, inspected for cell attach- ment and spreading, and attached cells were lysed with 2% SDS. The contents of each well was transferred to scintillation vials and bound cells were quantitated by counting 3H in a Beckman LS1801 scintil- lation counter. Background binding represented the number of cells

bound to ovalbumin. The mean of four determinations for each concentration of conjugate or ovalbumin was calculated and % bind- ing was determinated by the formula (mean sample counts-mean background)/total counts added X 100.

Affinity Matrix Preparation-Cyanogen bromide-activated Sepha- rose 4B (Pharmacia LKB Biotechnology Inc.) was washed per the manufacturer's instructions. Peptides were dissolved in coupling buffer (0.1 M NaHC03, 0.5 M NaCl, pH 8.3) at a concentration of 10 mg/5 ml and mixed with 1 ml of washed Sepharose overnight at 4 "C. Unreacted sites were blocked with 0.1 M Tris-HC1, pH 8.0, and gel was washed with alternating high and low pH buffers. Typically 95- 98% of peptide was covalently bound to Sepharose as determined by optical density.

Affinity Chromatography-1-ml columns were packed and washed in column buffer (0.025 M Tris-HC1, 0.14 M NaCl, 0.5 mM MnC12, 0.5 mM Cacl2, 50 mM octylglucopyranoside, 1 mM phenylmethylsulfonyl fluoride). Cell lysate (2-4 ml) was applied to the column over 2-3 h at room temperature. Columns were washed with 10-15 volumes of column buffer. Elution with peptides was carried out by dissolving each peptide in column buffer a t 1 mM concentration and passing 1 ml of peptide containing buffer over the column at 1 ml/h, followed by 4 ml of column buffer. Fractions of 700 pl were collected and counted in a Beckman 9000 series y-counter. After eluting with peptides, columns were washed with either 4 M guanidine hydrochlo- ride or 6 M urea in column buffer.

Analysis of Column Fractions-70-pl aliquots of fractions from columns were mixed with sample buffer and analyzed under reducing or nonreducing conditions on 6% polyacrylamide gels using the method of Laemmli (22). Gels were fixed, dried, and exposed for 3- 14 days to X-OMAT RP film (Kodak) at -80 "C.

Gel Densitometry-Relative amounts of 130- and 120-kDa protein were assessed by scanning autoradiographs on an LKB Ultroscan XL laser densitometer.

Fibrin-stimulated Release of uon Willebrand Factor from Endothe- lial Cells-Primary human umbilical vein endothelial cells were plated onto 1-cm2 glass coverslips. Cells were grown for 72-96 h before use. Fibrin was prepared by dissolving lyophilized fibrinogen (grade L, KabiVitrum, Stockholm, Sweden) to a concentration of 2.8 mg/ml in phosphate-buffered saline containing 10 mM calcium chloride. Clot- ting was initiated by adding 5 X units/ml thrombin (Sigma). 750-p1 aliquots were placed onto inverted glass scintillation vials and clots allowed to form in a humidity chamber overnight at 20 "C. Clots were then incubated in endothelial cell medium supplemented with 2.5 units/ml hirudin (Sigma) with three changes of medium over >12 h before use.

Coverslips with endothelial cells were placed singly in microtiter wells (Costar Corp., Cambridge MA) and preincubated with medium containing 2.5 units/ml hirudin for 30 min. Medium was removed and 500 pl of medium alone or medium containing peptides at various concentration were added to the wells for 30 min. Fibrin clots were then layered gently onto the coverslips for an additional 10 min in the presence of medium or medium with peptides. To determine release of von Willebrand factor from cells induced by the peptide alone, coverslips were incubated for 40 min in medium with peptide without fibrin addition. Coverslips were gently removed and washed three times in PBS" and fixed in 3.7% formaldehyde in PBS2+. To assess the percentage of cells demonstrating release, fixed coverslips were permeabilized with 0.5% Triton X-100 in PBSZ+ for 15 min and then incubated with polyclonal antisera to von Willebrand factor (Assera, Diagnostica Stago, Asnieres, France) 1:50 dilution followed by fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (ICN Immunobiological, Costa Mesa, CA) diluted 1:lOO. Coverslips were mounted and examined by fluorescent microscopy for release of Weibel-Palade bodies. A cell that released was defined as one in which no Weibel-Palade bodies could be identified. The percentage of released cells was calculated after examining at least 200 cells/ coverslip.

Immunoprecipitation with Antibodies to Integrins and CD31-Pro- tein A-Sepharose (Pierce Chemical Co.) was washed twice with 0.1 M Tris, pH 8.3. Aliquots of 30 mg were incubated with antiserum to the cytoplasmic domain of the PI subunit of integrin ((23) kindly provided by Dr. Richard Hynes, Center for Cancer Research, Massachusetts Institute of Technology, purified polyclonal antibody to gpIIb/III, (kindly provided by Dr. David Phillips, COR Therapeutics, San Francisco, CA), polyclonal antiserum to the p 5 subunit of integrin (kindly provided by Dr. Martin Hemler, Dana Farber Cancer Insti- tute, Boston, MA), or polyclonal antiserum to endoCAM ((33), kindly provided by Dr. Steven Albelda, Wistar Institute) for 2 h, then washed

A Receptor for Fibrin Bp15-42 on Endothelial Cells 2453

TABLE I Peptides used for elution of Bp15-42 columns

The final concentration of petpides was 1 mM. BO peptides are all derived from the published sequence of the BP chain of human fibrinogen. Bp15-42s refers to a synthetic pept,ide composed of the same amino acids as Bp15- 42 but in random order.

LDKKREEA Bp19-26 GHRPLDKKREEAPSLRPAPPPISGGGYR Bp15-42

DRGAPAHRPPRGPISGRSTPEKEKLLPG B(315-425' RGDS Arg-Gly-Asp-Ser

SGGGYRARPAKAAATQKKVEC Bj337-56-c~~

twice with 0.1 M Tris-HC1, pH 8.3, to remove unbound antibody. Monoclonal antibody to CD31 antigen (Clone 5.6E, IgG, isotype, AMAC Corp., Westbrook, ME) was covalently coupled to Sepharose for use in immunoprecipitation. Beads with bound antibody were washed with 0.025 M Tris-HC1, 0.14 M NaCl, 0.5 mM MnCl,, 0.5 mM CaC12, 0.5% Nonidet P-40 and then incubated either with 250 pl of 1251-labeled endothelial cell lysate or with 300-600 pl of sample containing labeled 130-kDa protein (2 h, 25 "C or overnight at 4 "C). Beads were washed five times with detergent containing buffer and boiled in the presence of eletrophoresis sample buffer (23).

Trypsin Cleavage of CD31 and 130-kDa Proteins-In order to prepare CD31 and 130-kDa protein for digestion under identical conditions, CD31 immunoprecipitated from "'1-labeled endothelial cell lysate was eluted from beads with 500 p1 of 0.05 M diethylamine, 0.5% Nonidet P-40, pH 11.5. Eluates were immediately neutralized with 50 p1 of 1 M Tris-HC1, pH 6.8. Eluates containing CD31 were then incubated with 250 pl of Sepharose beads to which wheat germ agglutinin ((WGA) Triticum vulgaris, Sigma) had been covalently linked at a concentration of 5 mg/ml gel. In parallel, fractions from affinity column experiments containing 130-kDa protein were incu- bated with WGA-Sepharose. Incubations were carried out overnight at 4 "C The beads were then washed two times with 0.025 M Tris- HC1, 0.14 M NaCl, 0.5 mM MnCl,, 0.5 mM CaCl,, 0.5% Nonidet P-40, p H 7.4. CD31 or 130-kDa protein were then eluted from WGA- Sepharose into 300 pl of the same buffer containing 300 mM N - acetylglucosamine. Each sample was split into two equal aliquots and half was digested by adding trypsin (Type XIII, TPCK treated, Sigma) to a final concentration of 5 pg/ml for 15 min at 37 "C. The reactions were stopped by adding 1/10 volume 20% SDS, electropho- resis reducing sample buffer and boiling samples for 8 min.

Determination of Lectin Affinity-WGA-Sepharose or concana- valin A-Sepharose (Sigma) were washed with Tris-buffered saline containing 50 mM P-D-octylglucopyranoside. Aliquots of fractions containing 130-kDa protein were incubated for 2 h with 200-pl samples of each lectin resin and then washed with the buffer described above. Retained proteins were analyzed on gels. In separate experi- ments, I-ml columns of each lectin-Sepharose were prepared. 2-ml samples containing '2sII-labeled 130-kDa protein were passed over each column. WGA-Sepharose was first eluted with 100 mM a- methylmannoside followed by 100 mM N-acetylglucosamine. Concan- avalin A-Sepharose was eluted with the same sugars in reverse order. Elution of bound 130-kDa protein was determined by autoradiography of SDS-polyacrylamide gel of fractions from each column.

RESULTS

Inhibition of Fibrin-mediated Release of uon Willebrand Factor by Bp15-42 Peptide-When fibrin formed by reptilase is layered onto endothelial cells in culture, no release of von Willebrand factor is observed (14). When both fibrinopeptides A and B are cleaved, the fibrin formed becomes a potent and rapid secretagogue of von Willebrand factor (1, 14). In con- trast, Bp15-42 appears to have weak agonist activity only after 3 h of incubation with endothelial cells (14). To deter- mine whether the sequence Bp15-42 might indeed be directly involved in fibrin-stimulated release, endothelial cells were incubated with 1 mM Bp15-42 or scrambled Bp15-42 (Bp15- 42S, Table I) in medium or with medium alone for 30 min and then stimulated with fibrin for 10 min in the presence or absence of the peptide. As shown in Fig. 1, free Bp15-42 had no significant secretory capacity after a 40-min incubation. However, preincubation of cells with 1 mM Bp15-42 did

Ffbrln

Flbrln + 8015-42

Flbrm + B015-42S

0 20 40 60 EO 3

PERCENT OF CELLS RELEASING vWf

FIG. 1. Release of von Willebrand factor from human um- bilical vein endothelial cells. Cells were grown on coverslips and incubated with medium (control) or medium containing 1 mM peptide for 30 min. Coverslips were then incubated for an additional 10 min in the absence or presence of fibrin clot. Cells were fixed, permeabil- ized, stained with anti-vWf antibody and scored for release by count- ing the percentage of cells without residual Weibel-Palade bodies. Each result is the mean t S.D. of three or more determinations (2200 cells scored/determination).

I I

0.5 1 10 100

Peptide conju~ale Concentration (uplml)

FIG. 2 . Binding of endothelial cells to either Bp15-42-oval- bumin or B@15-42S-ovalburnin. Five X lofi human umbilical vein endothelial cells labeled with 'H were detached and resuspended in 50 ml of media. 200-pl aliquots were added to each well of a 96-well plate coated with increasing concentrations of either B(315-42-oval- bumin, B/315-42S-ovalbumin, or ovalbumin alone. Wells were incu- bated for 2 h at 37 "C and then washed with media. Attached cells were lysed and percent binding for each well was determined by subtracting counts bound to ovalbumin alone and dividing by the total counts in 200 p1 of cell suspension. Each point represents the mean t S.D. of four determinations in a representative experiment. Closed circles, Bpl5-42-ovalbumin. Open circles, Bo15-425'-ovalbu- min.

inhibit fibrin-mediated release of vWf from endothelial cells by approximately 50% (t test p < 0.0001). Incubation with Bp15-42 at 0.1 mM concentration did not significantly inhibit fibrin-mediated release (not shown). In contrast, preincuba- tion with peptide Bp15-42S, composed of the same amino acids as Bp15-42 in random order, did not inhibit release of vWf stimdated by fibrin a t a concentration of 1 mM. These

2454 A Receptor for Fibrin Bp15-42 on Endothelial Cells

results are consistent with the hypothesis that the BP15-42 sequence of fibrin interacts directly with the endothelial cell, since free BP15-42 at 1 mM final concentration partially inhibits this interaction.

Binding of Endothelial Cells to Immobilized BP15-42"To determine whether BP15-42 would support the attachment and spreading of endothelial cells, BPl5-42-ovalbumin or the scrambled form of this peptide (BP15-42s) were coupled to ovalbumin and bound to plastic wells. By SDS-polyacrylamide gel electrophoresis comparison of derivatized ovalbumin to native ovalbumin, it was estimated that each mole of conju- gate contained an average of 4 mol of peptide for both re- agents. Increasing concentrations of BP15-42 peptide conju- gate led to an increase in the binding of endothelial cells (Fig. 2). This binding was saturable and was accompanied by spreading of endothelial cells as observed by inverted light microscopy (not shown). In contrast no saturable binding to the conjugate containing BP15-42s was seen (Fig. 2) and cells which did attach remained rounded (not shown). These re- sults suggested the presence of a specific binding site on the surface of endothelial cells which interacts with the sequence BP15-42 of fibrinogen, and we decided to characterize this putative BP15-42 receptor.

Identification of an Endothelial Protein Specifically Binding to the BP15-42 Sequence of Fibrinogen-To determine whether BB15-42 binds to a protein on the surface of endo- thelial cells, human umbilical vein endothelial cells were surface labeled with '''1 using the lactoperoxidase method (20). Labeled endothelial cells were lysed in octylglucopyran- oside-containing buffer and the lysate was applied to a column of BP15-42 coupled to Sepharose 4B. The column was washed with column buffer and eluted sequentially with one of the peptides shown in Table I, followed by BP15-42, and finally with guanidine hydrochloride or urea. Fractions were collected and aliquots of each fraction were analyzed by SDS-polyacryl- amide gel electrophoresis under reducing conditions. Analysis of fractions collected in one such experiment after eluting the column with BP37-56-cysteine followed by BP15-42 is shown in Fig. 3. No labeled proteins were eluted with the first peptide but two proteins of 130 and 120 kDa were eluted by BP15-42. The relative amount of the less prominent 120-kDa protein varied from 3 to 20% of the major band as determined by densitometric analysis of autoradiographs among different experiments, and only the 130-kDa protein was seen on silver- stained gel (not shown). The variable amount of the 120-kDa

*0°- 1 97-

69 -

I 11 111

FIG. 3. Binding of 130- and 120-kDa proteins to BB15-42- Sepharose column. Approximately 1.2 X 10' human umbilical vein endothelial cells were surface labeled with '*'I and lysed. Cell lysate was applied to a Bpl5-42-Sepharose column, and the column was washed with 10 column volumes of buffer. After washing, column was eluted with 1 mM BB37-56-cysteine (single arrow), 1 mM BB15-42 (double arrow), or guanidine hydrochloride (triple arrow). Aliquots of each column fraction were reduced and loaded onto 6% SDS gels. Gels were fixed and exposed to autoradiography for 72 h with single enhancing screen. Arrowheads depict positions of eluted 130- and 120-kDa proteins. Molecular size standards are shown in the left lane.

protein indicated to us that it may represent a degradation product of the 130-kDa protein. The amount of 130-kDa protein recovered was estimated a t 50-150 ng/107 cells as judged by silver stain. Guanidine hydrochloride eluted other proteins which bound to the column but were not eluted by BP15-42 (Fig. 3), in addition to a small amount of residual 130-kDa protein.

When the column was eluted with 1 mM BP15-42S, trace elution of the 130-kDA protein was seen compared to what was observed when BP15-42 was subsequently applied to the column (Fig. 4A). BP19-26, a segment of BP15-42 which contains six consecutive charged amino acids (Table I) failed to elute the 130-kDa protein which was then eluted by BB15- 42 (Fig. 4B), indicating that this charged sequence had little or no affinity for the 130-kDa protein as compared to Bj315- 42. The arginine-glycine-aspartic acid-serine (RGDS) tetra- peptide mediates the binding of various adhesive proteins such as fibrinogen, fibronectin, and vitronectin to their re- spective integrin receptors (24). When 1 mM RGDS was added to the column, no elution of either band from the BP15-42 column was detected (Fig. 4C), nor did EDTA elute residual 130-kDa protein which remained bound to the column (not shown). The 130- and 120-kDa proteins did not bind to a bovine serum albumin-Sepharose column to which labeled cell lysate was applied (not shown). To further establish the specificity of binding, BP15-42S (Table I) was coupled to Sepharose, and cell lysate was applied to this column. The flow through from this column was then applied to a BB15- 42-Sepharose column. Both columns were washed and eluted with 1 mM BP15-42. Neither the 130- nor 120-kDa proteins were eluted from the scrambled peptide column (Fig. 5A). In

RGDS

FIG. 4. Elution of BB15-42-Sepharose with control pep- tides or BB15-42. In panels A-C, Bpl5-42-Sepharose was first eluted with 1 mM control peptide as labeled (top) followed by 1 mM Bp15-42 (bottom). Only elution lanes are shown. In all cases, BB15- 42 eluted significantly more 130-kDa protein from the column than did control peptides. Arrowhead indicates expected migration for a protein of 130 kDa size.

FIG. 5. Elution of BB15-42s and BB15-42 columns with B815-42. Labeled endothelial cell lysate was applied to BB15-42S- Sepharose. Flow through from this column was then applied to Bj315- 42-Sepharose and both columns were eluted with 1 mM BB15-42. Elution lanes are shown. A, BB15-42S column. B, BB15-42 column. The 130-kDa protein bound to and was eluted only from the native B615-42-Sepharose column.

A Receptor for Fibrin Bp15-42 on Endothelial Cells 2455

NR R 1 1

200-

I

I

97- I

69- I I i

43-

FIG. 6. Analysis of 130- and 120-kDa proteins under re- ducing and nonreducing conditions. Radiolabeled proteins were analyzed on a 6% SDS polyacrylamide gel under reducing ( R ) or nonreducing ( N R ) conditions. The autoradiograph of the gel is shown with molecular size standards to the left.

contrast, these proteins clearly bound to and were eluted from the Bj315-42-Sepharose column (Fig. 5B) . The results of these experiments demonstrate that binding of the 130- and 120- kDa proteins to Bj315-42 is specific and that the determinants sufficient for elution were not found in other peptides tested, including a partially overlapping sequence of the Bj3 chain (BP37-56-cysteine), or an internal sequence (Bj319-26).

Further Characterization of the Bj315-42-binding Proteins- SDS-polyacrylamide gel electrophoresis of the eluted proteins under reducing and nonreducing conditions is shown in Fig. 6. Under reducing conditions the bands migrated more slowly, compatible with larger molecular radii on reduction and in- trachain disulfide bonding. No light chains were identified after reduction when examined on 12.5% SDS-polyacrylamide gel (not shown). The 130-kDa species bound to the WGA- Sepharose column (not shown; WGA-Sepharose binding shown in Fig. 8 A ) and was eluted with 100 mM N-acetylglu- cosamine but not 100 mM a-methylmannoside, establishing that the 130-kDa protein is a glycoprotein. Binding to con- canavalin A-Sepharose under the same conditions was mini- mal (not shown). Since wheat germ lectin is known to bind to tri- and tetra-antennary complex oligosaccharides contain- ing N-acetylglucosamine, we assume that the 130-kDa protein is modified by oligosaccharides of this type.

To determine whether fibroblast contained the 130-kDa protein, human foreskin fibroblasts and mouse 3T3 cells were labeled and lysates applied to Bj315-42 columns. No elution of the 130-kDa protein with Bp15-42 was seen with either cell (not shown), suggesting either a tissue-specific distribu- tion of the protein or a difference in quantity between endo- thelial cells and the fibroblasts.

The receptors for fibrinogen which have been described include the a&/& integrin receptor on platelets (25-27), the

receptor on endothelial cells (28, 29), Mac-1 (CDllb/ CD18), an integrin of the B2 subclass on leukocytes (30), and a 90-kDa mitogenic receptor for fibrinogen on leukocytes (31). Endothelial cells possess integrins of the and j33 subclass (32) which are active in mediating the interaction of these cells with components of the basement membrane as well as small numbers of the p5 subclass.* To determine if the Bp15- 42-binding proteins were related to either or 03, surface- labeled endothelial cells were lysed and aliquots were incu- bated with polyclonal anti-P1 and anti-gpIIb/IIIa antibodies bound to protein A-Sepharose. The immunoprecipitated pro- teins were analyzed by SDS-polyacrylamide gels under reduc- ing conditions. In parallel, the 130- and 120-kDa proteins eluted from the Bj315-42 column were incubated with both

Martin Hemler, personal communication.

antibodies under the same conditions. As shown in Fig. 7, anti-& as well as anti-IIb/IIIa antibodies, which recognize the p3 subunit of integrin, bound heterodimeric receptors from the endothelial cell lysate. In contrast, neither antiserum recognized the proteins eluted from the Bj315-42 column. In addition, Western blots of 130- and 120-kDa proteins with anti-& antiserum were negative (not shown). Upon reduction, the 130-kDa protein can be distinguished from the av subunit of the vitronectin receptor by the absence of a light chain and by slightly slower migration on gels. When compared on nonreduced gels the av subunit is larger by approximately 20,000 daltons, reflecting the additional mass of the disulfide- linked light chain. None of the other integrin subunits im- munoprecipitated from endothelial lysate with either anti+, or anti-& antisera comigrated with the Bj315-42-binding pro- teins on reduced or nonreduced gels. When radioiodinated endothelial cell lysate was incubated with anti-p5 antiserum, no immunoprecipitated bands were detected (not shown). Taken together, these results indicate that the proteins eluted from the Bp15-42 column are not immunologically related to known integrins on endothelial cells, including the avj33 fi- brinogen receptor.

Several groups have recently characterized a 130-135-kDa glycoprotein on endothelial cells known as endoCAM (33), PECAM-1 or CD31(34,35), a member of the immunoglobulin family of cell surface receptors which is thought to function in cell-cell adhesion. Although polyclonal antiserum to CD31 neither stimulated release nor inhibited fibrin-mediated re- lease of Weibel-Palade bodies from endothelial cells (not shown), we wished to demonstrate that the CD31 protein on endothelial cells and our 130-kDa protein were not identical. CD31 and Bpl5-42-binding protein comigrate on both re- duced and nonreduced polyacrylamide gels (not shown). We have attempted to immunoprecipitate the Bj315-42-binding protein with both polyclonal (Fig. 8A) and monoclonal anti- CD31 antibodies (not shown). Although Both recognized a 130-kDa protein in iodinated endothelial cell lysate, neither immunoprecipitated the 130-kDa protein eluted from the Bj315-42 column, indicating that these molecules are immu- nologically distinct. To further establish that CD31 and the 130-kDa proteins were different molecules, both were bound to WGA-Sepharose, eluted with N-acetylglucosamine into identical buffers, and cleaved with 5 pg/ml trypsin. CD31 cleavage under these conditions resulted in the appearance of a major band at approximately 110 kDa as determined by SDS-polyacrylamide gel electrophoresis in conjunction with

i 2 3 4 5 6

97-

69-

43-

FIG. I. Reactivity of antibodies to 8, or /3~ subunit of inte- grin with cell lysate and eluate from BB15-42 column. lz5I-

Labeled cell lysate (lanes 1-3) or 130-kDa protein eluted from a B815-42-Sepharose column (lanes 4-6) were incubated with protein A-Sepharose to which antibodies to 8, (lanes 2 and 5) or 83 (lanes 3 and 6 ) subunit of integrin were bound. Lanes 1 and 4 represent the starting material prior to immunopurification. Numbers on the left indicate migration of molecular size standards.

2456

A

A Receptor for Fibrin BD15-42 on Endothelial Cells

FIG. 8. Comparison of 130-kDa protein and CD31. A, pooled column fractions containing iodinated 130-kDa protein eluted from BP15-42 Sepharose were divided into two portions and incubated either with polyclonal anti-CD31 antiserum adsorbed onto protein- A-Sepharose beads or with WGA-Sepharose. Supernatants were saved and beads washed as under “Experimental Procedures.” Beads and aliquots of supernatants were boiled with SDS-polyacrylamide gel sample buffer and all samples were analyzed on 8% SDS-poly- acrylamide gel electrophoresis under reducing conditions. Lanes I and 2, supernatant after incubation with CD31-Sepharose and WGA- Sepharose beads, respectively. Lanes 3 and 4, iodinated proteins bound to CD31-Sepharose and WGA-Sepharose, respectively. B, io- dinated CD31 or 130-kDa protein eluted from WGA-Sepharose with N-acetylglucosamine were digested with trypsin (5 pg/ml) for 15 min at 37 “C and reaction stopped by adding SDS and reducing sample buffer before boiling for 8 min. Samples were loaded onto 8% SDS gel and compared with starting material. Lanes 1 and 2, 130-kDa protein before and after trypsin digestion. Lunes 3 and 4, CD31 antigen before and after trypsin digestion. Lane 5, molecular size standards.

complete disappearance of the starting material (Fig. 8B). In contrast, 130-kDa protein cleaved under identical conditions generated a different set of fragments at approximately 120 and 115 kDa with residual 130-kDa protein remaining (Fig. 8B). This strongly suggests that the primary structures of the two proteins are different, and coupled with the immunopre- cipitation experiments described above it can be concluded that the 130-kDa protein is distinct from CD31 on endothelial cells.

DISCUSSION

Deposition of fibrin at the site of vascular injury results from a cascade of events involving soluble blood coagulation factors and cofactors which interact with the endothelium, platelets, and the subendothelial basement membrane. Fibrin formation and degradation stimulate cellular responses im- portant in wound healing. In addition to stimulating the release of stored vWf (l), fibrin has been shown to induce endothelial cells to retract and migrate (36, 37) and to stim- ulate the synthesis of tissue plasminogen activator and pros- tacyclin (15, 16). Fibrinogen in contrast does not initiate these responses. Early in the process of fibrinolysis the car- boxyl terminus of the a chain as well as Bp15-42 are cleaved from fibrin, followed by more complete enzymic release of fragments E and D and, in the case of cross-linked fibrin, D dimer (38, 39). B/315-42 is chemotactic for neutrophils (40) and Bpl-42, released during fibrinogenolysis, has been shown to induce endothelial cell retraction and fibroblast migration (38). Binding of the soluble NH2-terminal disulfide knot of fibrin to U937 cells and rabbit monocyte/macrophage-derived cell lines has been demonstrated, although the determinants involved in binding have not been established (17). Chen et al. (41) have shown that BO15-42 inhibits platelet aggregation, and although they propose that B@5-42 binds to fibrinogen and inhibits it from binding to gpIIb/IIIa, the interaction of Bp15-42 with a specific platelet receptor is not entirely ruled out. Therefore, the dual processes of fibrin deposition and

degradation not only serve a hemostatic function but influ- ence the inflammatory response and subsequent wound heal- ing.

The release of von Willebrand factor by endothelial cells in response to fibrin is rapid and dependent on contact with cells (1). In the process of release of von Willebrand factor, Weibel-Palade bodies fuse with the cell surface, and the granule membrane adhesion molecule P-selectin (PADGEM, GMP-140, CD62) is expressed (42, 43, 44). The physiologic role for P-selectin remains unknown although it is reasonable to speculate that it plays a role in modulating inflammation through the binding of leukocytes, Release of von Willebrand factor occurs in response to fibrin, thrombin, complement, and histamine (1-5). The release in response to fibrin has been correlated with cleavage of fibrinopeptide B (14). Kaplan et al. (16) have demonstrated a similar effect on prostacyclin and tissue plasminogen activator synthesis, noting that fibrin lacking only fibrinopeptide A is inactive. To date no receptors which recognize fibrinogen have been shown to bind to deter- minants on the BP chain. Using purified S-carboxymethylated fibrinogen Aa, BP, and y chains, Hawiger et al. (45,46) found that the a chain and the carboxyl terminus of the y chain contain sites which bound to platelet fibrinogen receptors. Binding of the intact BP chain to platelet receptors could not be demonstrated (45). Distinct binding sites on the a and y chains of fibrinogen have been demonstrated for the fibrino- gen receptor on platelets (gpIIb/IIIa) and the fibrinogen receptor on endothelial cells (a,/P3) (47). In this study, we now show that Bp15-42 bound to an endothelial surface protein of 130 kDa not found on human foreskin fibroblasts or mouse 3T3 cells under similar conditions. Elution of this protein from an affinity column required intact BO15-42 and was not accomplished by using a highly charged region of the peptide near its amino terminus nor by using a partially overlapping peptide which spans the plasmin cleavage site between amino acids 42 and 43 of the B/3 chain (Fig. 3). Free BB15-42 inhibited fibrin-mediated release of von Willebrand factor from endothelial cells (Fig. 1). This was compatible with the hypothesis that the amino terminus of the BP chain exposed after fibrinopeptide B cleavage interacts alone or in concert with other determinants on fibrin with an endothelial cell receptor. Free BP15-42 at 1 mM concentration is a poor stimulus of vWf release relative to fibrin (14). As part of a fibrin network the amino terminus of the B chain may bind and cross-link receptors signaling a response or may interact with other determinants in the fibrin molecule to provide the entire ligand needed to signal vWf release. In this model free BD15-42 may partially inhibit secretion by binding to the receptor, yet by itself be incapable of rapidly signaling the cell to degranulate. During fibrinolysis BB15-42 is cleaved from fibrin by plasmin and can be measured in plasma as an early fibrin degradation product (48). Since this cleavage does not contribute significantly to mechanical disruption of a fibrin gel (49), we consider whether removal of BP15-42 is instead important in terminating an endothelial cell response by releasing this potential ligand. We cannot rule out that the 130-kDa protein does not bind either to intact fibrinogen or to BB1-42 by the present studies, although in column exper- iments using fibrinogen-Sepharose as the affinity matrix, no binding of the 130-kDa protein to fibrinogen was identified, while the a& receptor was clearly retained and eluted with either EDTA or the RGDS tetrapeptide.“ Intact B/3 chain which we purified after reduction and alkylation of fibrinogen using published methods (50) proved unsatisfactory for use to synthesize an affinity matrix due to its virtual insolubility at

‘ J. K. Erban and D. D. Wagner, unpublished observations.

A Receptor for Fibrin BB15-42 on Endothelial Cells 2457

neutral or alkaline pH. In contrast, synthetic B(315-42 is highly soluble in aqueous solution and could be easily coupled to Sepharose to provide a suitable affinity matrix.

Prior studies examining endothelial response to fibrin have demonstrated that cell migration and retraction occur on prolonged exposure (36-38)3 although some controversy re- mains about the fibrin determinants necessary to induce these responses. Dang et al. (51) have suggested that fragment D can induce many of the same responses while fragment E is inactive, although Delvos et al. (52) were unable to show an effect of fragment D on endothelial cells even after prolonged exposure. In the process of fibrin assembly the E domain of one fibrin monomer and the D domains of others become closely opposed in a half-staggered configuration (11). When fibrin is incubated with antibodies to either the E or D domains, the ability of fibrin to induce release is a t ten~ated .~ Since B@15-42 bridges the E and D domains during fibrin gel formation, this region of the molecule is closely related to both domains and would potentially be masked by antibodies to either. Bunce et al. (53) have recently reported that while fibrin supports the spreading of endothelial cells as deter- mined by rhodamine phylloidin staining of F-actin filaments, des-Bpl-42-fibrin does not facilitate spreading nearly as well. They propose that the B(315-42 sequence in the fibrin mole- cule may be important for endothelial cell spreading on fibrin gels. We now show that B(315-42 conjugated to ovalbumin supports the attachment (Fig. 2) and spreading (not shown) of endothelial cells on non-tissue culture-treated plastic.

Although several column elution experiments appeared to show a second band at 120 kDa in addition to the major 130 kDa species (Figs. 3-5), its presence was variable. It therefore appeared that binding to B(315-42 was dependent only upon a single polypeptide chain and not upon a complex of subunits as is the case with integrins (54). The argument that the 130- and 120-kDa proteins were not an integrin complex is further strengthened by the observations that EDTA (not shown) and RGDS (Fig. 4C) were ineffective in eluting the protein and that neither PI- nor &-specific antibodies recognized the protein (Fig. 7). The appearance of a minor satellite band was similar to what has been described for CD31, a transmem- brane protein of similar size in the immunoglobulin family of receptors. We demonstrated that CD31 and the 130-kDa protein were distinct since antibodies to the CD31 antigen did not recognize the 130-kDa protein (Fig. &I), and partial tryptic digestion of the two proteins yielded different cleavage products under identical conditions (Fig. 8B). In addition polyclonal antiserum to CD31 did not stimulate nor did it inhibit fibrin-mediated release of Weibel-Palade bodies from endothelial cells (not shown). It is of interest that the major tryptic cleavage fragment seen after digestion of CD31 under these conditions corresponded in size in our hands to the satellite band of reported by Albelda and others (33). We cannot at this time rule out the possibility that the 130- and 120-kDa bands which elute from the B(315-42 column repre- sent the translated products of alternatively spliced mRNA as has been described for VCAM-1 and other members of the immunoglobulin family of receptors (55-57). It is tempting to speculate rather that the smaller band represents a degrada- tion product of the larger since one band seen after tryptic digestion corresponds in size to the 120-kDa protein which elutes from the B(315-42 column. The slower relative migra- tion of the 130-kDa protein after reduction suggests intra- chain disulfide bonding characteristic of membrane receptor proteins. It is likely that the 130-kDa B(315-42-binding pro- tein is a new member of a class of single chain membrane adhesion receptors which on endothelial cells includes the

immunoglobulin family adhesion receptors ICAM-1 (58, 59), ICAM-2 (60), VCAM-1 (55, 61, 62), CD31 (33), calcium- dependent cell-cell adhesion molecules (63), and the adhesion proteins E-selectin (ELAM-1) (64) and P-selectin (PAD-

In summary, we have identified a 130-kDa endothelial cell surface glycoprotein which specifically binds to an amino- terminal sequence derived from the @ chain of fibrin. The 130-kDa glycoprotein is a likely candidate for a receptor involved in fibrin-mediated release of Weibel-Palade bodies and in the endothelial cell response to fibrin deposited at sites of vascular injury.

GEM, GMP-140, CD62) (7,65).

Acknowledgments-We would like to thank Margaret Jacobs for peptide synthesis, Drs. Bruce and Barbara Furie for many helpful discussions, Dr. Richard Hynes for providing anti-& antisera, Dr. Martin Hemler for providing anti-p5 antiserum, Dr. David Phillips for anti-IIb/IIIa antibodies, Dr. Steven Albelda for polyclonal anti- serum to endoCAM and for helpful discussion, Dr. Tanya Mayadas Norton for critical reading of the manuscript, Dr. Joost Koedam for helpful discussions, Dr. Sheldon Wagner for statistical analysis, and Tony Korosi, Paula Pellerin, and Kathy Reebenacker for expert technical assistance.

1.

2.

3.

4. 5.

6.

7.

8.

9. 10. 11. 12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22. 23.

24.

25.

26.

27.

REFERENCES

Ribes, J., Francis, C., and Wagner, D. (1987) J. Clin. Znuest. 79 ,

Levine, J., Harlan, H., Harker, L., Joseph, M., and Counts, R.

Reinders, J., Vervoorn, R., Venveij, C., van Mourik. J., and

Hamilton, K., and Sims, P. (1987) J. Clin. Znuest. 79,600-608 Hattori, R., Hamilton, K., McEver, R., and Sims, P. (1989) J.

Sakariassen, K., Bolhius, P., and Sixma, J . (1979) Nature 279 ,

Bonfanti, R., Furie, B. C., Furie, B., and Wagner, D. (1989) Blood

McEver, R., Beckstead, J., Moore, K., Marshall, C. L, and Bain- ton, D. (1989) J . Clin. Znuest. 8 4 , 92-99

Doolittle, R. (1984) Annu. Reu. Biochem. 5 3 , 195-229 Blomback, B. (1958) Ark. Kemi. 12, 321-335 Doolittle, R. (1973) Adu. Protein Chem. 2 7 , 1-109 Shen, L., Hermans, J., McDonagh, J., and McDonagh, R. (1977)

Kwann, H., and Barlow, G. (1971) Thromb. Diath. Haemorrh.

Ribes, J., Wagner, D., and Francis, C., (1989) J. Clin. Znuest. 8 4 ,

Kaplan, K., Mather, T., Demarco, L., and Solomon, S. (1987)

Kaplan, K., Mather, T., DeMarco, L., and Solomon, S. (1989)

Shainoff, J., Stearns, D., Dibello, P., and Hishikawa, I. (1990)

Wagner, D., Olmsted, J., and Marder, V. (1982) J . Cell Biol. 9 9 ,

Gimbrone, M., Cotran, R., and Folkman, J . (1984) J. Cell Biol.

Lebien, T., Boue, D., Bradley, J., and Kersey, J. (1982) J. Zm- munol. 129,2287-2292

Dayhoff, M. (ed) (1973) Atlas of Protein Sequence and Structure B(supp1. I) S-29, National Biomedical Research Foundation, Silver Spring, MD

117-123

(1982) Blood 6 0 , 531-533

DeGroot, P. (1987) J. Cell Physiol. 133, 79-87

Biol. Chem. 264,9053-9060

636-638

73,1109-1112

Am. J . Physiol. 232,629-633

~ ~ ( s u P P ~ ) , 361-369

435-442

Thromb. Haemostusis 58,939-946

Arteriosclerosis 9 , 43-49

Thromb. Haemostusis 63,193-203

2123-2130

6 0 , 673-684

Laemmli, U., (1970) Nature 237,680-685 Marcantonio, E., and Hynes, R. (1988) J. Cell Biol. 106 , 1765-

Hautanen, A., Gailit, J., Mann, D., and Ruoslahti, E. (1989) J.

Marguerie, G., Plow, E., and Edgington, T. (1979) J. Biol. Chem.

Plow, E., Srouji, A., Meyer, D., Marguerie, G., and Ginsburg, M.

George, J., Nurden, A., and Phillips, D. (1984) N. Engl. J . Med.

1772

Biol. Chem. 2 6 4 , 1437-1442

254,5357-5363

(1984) J . Biol. Chem. 259,5388-5391

311,1084-1098

2458 A Receptor for Fibrin Bp15-42 on Endothelial Cells 28. Dejana, E., Languino, L., Polentarutti, N., Balconi, G., Ryche- 46. Kloczewiak, M., Timmons, S., and Hawiger, J. (1982) Biochem.

47. Cheresh, D., Berliner, S., Vincente, V., and Ruggeri, Z. (1989)

48. Kudryk, B., Robinson, D., Metre, C., Hessel, B., Blomback, M.,

Luike, J. (1988) Proc. Nutl. Acad. Sci. U. S. A. 80,5699-5703 49. Nuewuariwsju, S., Stewart, G., and Marder, V. (1970) Biochim.

waert, J., Larrieu, M., Donati, M., Mantovani, A,, and Mar- Biophys. Res. Commun. 107 , 181-187 guerie, G. (1985) J . Clin. Invest. 75, 11-18

29. Fitzgerald, L., Charo, I., and Phillips, D. (1985) J . Biol. Chem. Cell 58,945-953

30. Wright, S., Weitz, J., Huang, A., Levin, S., Silverstein, S., and and Blomback, B. (1982) Thromb. Res. 25, 277-291

31. Levesque, J., Hatzfeld, J., and Hatzfeld, A. (1990) J. Biol. Chem. Biophy~. Acta 22 1 , 326-332

32. Conforti, G., Zanetti, A,, Colella, S., Abbadini, M., Marchisio, P., FEBS Lett. 14,37-41

260, 10893-10896

265,328-333 50. Murano, G., Wiman, B., Blomback, M., and Blomback, B. (1971)

E. (1989) Blood 73, 1576-1585 Pytela, R., Gioancotti, F., Tarone, G., Languino, L., and Dejana, 51. Dang, C., Bell, W., Kaiser, D., and Wong, A. (1985) Science 227,

33. Albelda, S., Oliver, P., Romer, L., and Buck, C. (1990) J . Cell 52. Delvos, U., and Muller-Berghaus, G. (1988) Huemostasis 18,99-

34. Newman, P., Berndt, M., Gorski, J., White, G., Lyman, S., 53. Bunce, L., Sporn, L., andFrancis, C. (1991) Thromb. Huemostusis

35. Muller, W., Ratti, C., McDonnell, S., and Cohn, Z. (1989) J. Exp. 54. Hynes, R. (1987) Cell 48, 549-554

,36. Kadish, J., Butterfield, C., and Folkman, J . (1979) Tissue & Cell 18,5901

37. Weimar, B., and Delvos, V. (1985) Arteriosclerosis 6, 139-145 Walsh, F. (1989) J. Cell Biol. 109, 3465-3476 38. Rowland, F., Donovan, M., Picciano, P., Wilner, G., and Kruetzer, 57. Jessell, T. (1988) Neuron 1, 3-13

39. Kopel, M., and Latallu, Z. (1978) Hundb. Exp. Phurmacol. 46, Immunol. 137, 1270-1274

40. Skogen, W., Senior, R., Griffin, G., and Wilner, G. (1988) Blood T. (1988) Cell 52,925-933

41. Chen, C., Chou, S., and Thiagarajan, P., (1988) Blood 27, 6121-

42. Larsen, E., Celi, A., Gilbert, G., Furie, B. C., Erban, J., Bonfanti. Chi-Rosso, G., and Lobb, R. (1989) Cell 59, 1203-1221

43. Johnson, G., Cook, R., and McEver, R. (1989) Cell 56, 1033-44 63. Takeichi, M. (1990) Annu. Reu. Biochem. 59,237-252 44. Berrnan, C., Yeo, E., Wencel-Drake, J., Furie, B. C., Ginsberg, 64. Bevilaqua, M., Stengalin, S., Gimbrone, M., Jr., and Seed, B.

45. Hawiger, J., Timmons, S., Kloczewiak, M., Strong, D., and Doo- 65. Hattori, R., Hamilton, K., Fugate, R., McEver, R., and Sims, P.

1487-1490

Biol. 220, 1227-1237 105

Paddock, C., and Muller, W. (1990) Science 247, 1219-1222 65, 1225 (abstr.)

Med. 170,399-414 55. Polte, G., Newman, W., and Gopel, T. (1990) Nucleic Acids Res.

11,99-I08 56. Pizzey, J., Rowett, L., Howard Barton, C., Dickson, G., and

D. (1984) Am. J. Pathol. 117,418-428 58. Rothlein, R., Dustin, M., Marlin, S., and Springer, T. (1986) J.

59. Stauton, D., Marlin, S., Stratowa, C., Dustin, M., and Springer, 81-105

71, 1475-1479 60. Stauton, D., Dustin, M., and Springer, T. (1989) Nature 339,

6126

R., Wagner, D., and Furie, B. (1989) Cell 59, 305-312 62. Rice, E., and Bevilacqua, M. (1989) Science 246, 1303-1305

61-64 61. Osborn, L., Hession, C., Tizard, R., Vassallo, C., Luhowsky, S.,

M., Furie, B. (1986) J. Clin. Inuest. 78 , 130-137 (1989) Science 243, 1160-1165

little, R. (1987) Proc. Nutl. Acad. Sci. U. S. A. 79, 2068-2071 (1989) J . Biol. Chem. 264,7768-7771