Human and bovine endothelial cells synthesize membrane proteins similar to human platelet glycoproteins IIb and IIIa.

Human umbilical vein endothelial (HUVE) and bovine aortic endothelial (BAE) cells in culture were examined to determine whether membrane proteins similar to human platelet glycoproteins (GP) IIb and IIIa were present. The HUVE and BAE cells were either 125I-surface labeled or metabolically labeled. Triton X-100 lysates of labeled cells were immunoprecipitated with polyclonal antibodies prepared against purified human platelet GP IIb-IIIa complex. Two membrane proteins were detected on both HUVE (Mr = 130,000 and 110,000) and BAE (Mr = 135,000 and 105,000) cells, which were similar to human platelet GP IIb (Mr = 125,000) and GP IIIa (Mr = 108,000). The two membrane proteins from HUVE cells and the two from BAE cells cosedimented in sucrose gradients, indicating that they exist as a complex. Unlike the human platelet GP IIb-IIIa complex, the HUVE and BAE membrane protein complexes were not dissociated by chelation of Ca2+. Platelet GP IIb and GP IIIa and the related membrane proteins on both HUVE and BAE cells showed similar changes in electrophoretic mobility upon disulfide reduction. These data demonstrate that human and bovine endothelial cells synthesize membrane proteins that have properties similar to the platelet membrane GP IIb-IIIa complex.


Human umbilical vein endothelial (HUVE) and bovine aortic endothelial (BAE) cells in culture were
The two membrane proteins from HUVE cells and the two from BAE cells cosedimented in sucrose gradients, indicating that they exist as a complex. Unlike the human platelet GP IIb-IIIa complex, the HUVE and BAE membrane protein complexes were not dissociated by chelation of Ca*+. Platelet GP IIb and GP IIIa and the related membrane proteins on both HUVE and BAE cells showed similar changes in electrophoretic mobility upon disulfide reduction. These data demonstrate that human and bovine endothelial cells synthesize membrane proteins that have properties similar to the platelet membrane GP IIb-IIIa complex.
Endothelial cells provide a surface that is in contact with plasma proteins and cells normally found in circulation, as well as malignant cells that metastasize by hematogenous spread (1). Cultured endothelial cells have been shown to bind polymorphonuclear leukocytes (2), lymphocytes (3), monocytes (4), and stimulated platelets (5). Endothelial cells also bind proteins that are involved in coagulation (6) and agents that affect endothelial cell function, such as thrombin (7). However, with the exception of thrombomodulin (8), little is known about endothelial cell surface proteins and their possible receptor functions.
There are two reasons to expect that endothelial cell surface proteins may be similar to the cell surface proteins of platelets.
* This work was supported in part by Grants HL28947, HL32254, and HL23454 from the National Institutes of Health. 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 U.S.C. Section 1734 solely to indicate this fact.

MATERIALS AND METHODS
Preparation of Antibodies-Antisera against the purified G P IIb-IIIa complex from human platelets' was produced in rabbits. The immunoglobulin G (IgG) fraction was purified by affinity chromatography using Protein A-Sepharose (Sigma) (19). Affinity purified G P IIIa antibody was obtained by incubating the G P IIb-IIIa antisera with nitrocellulose blots of G P IIIa (20) and eluting the antibody from the washed nitrocellulose with 0.1 M glycine, p H 2.8. Cell Culture-The HUVE cells were obtained by incubating umbilical veins for 10 min with Medium 199 (pH 7.4) that contained 1 mg/ml collagenase (Type I, Worthington) (21). The HUVE cells were cultured in Medium 199 that was supplemented with 20% (v/v) fetal calf serum, 25 mM Hepes buffer (pH 7.4), 100 units/ml penicillin and streptomycin, 90 pg/ml porcine heparin (Sigma), and 100 ng/ml fibroblast growth factor (kindly provided by Dr. Denis Gospodarowicz, University of California, San Francisco, CA); 60-mm dishes (Falcon) coated with gelatin (Sigma) were used. Primary cultures were plated at 5 X lo5 cells/dish, and the cells were grown to confluency a t 37 "C in 5% CO, and 95% humidity. Cultured HUVE cells exhibited a typical cobblestone morphology (21). Primary and early passage HUVE cells were used for further study. The BAE cells, provided by Dr. George Rodgers (Gladstone Foundation Laboratories), were grown in Dulbecco's modified Eagle's medium (DMEM-H16) supplemented with 10% calf serum, and the cells between passages 5 and 10 were used. Human foreskin fibroblasts were grown in Dulbecco's modified Eagle's medium H21. Subcultures of HUVE, BAE, and fibroblast cells were obtained by detaching cells with 0.2 mg/ml EDTA and 0.5 mg/ml trypsin (4 min, 37 "C) and diluted 1:4 before replating. Tissue culture media and supplements were obtained from Gibco.
Cell Surface Labeling-Confluent monolayers of HUVE and BAE cells were washed five times with serum-free medium and labeled with 1 mCi of ' ' ' I (Amersham) and M lactoperoxidase (Sigma) a t room temperature in 1 ml of serum-free medium. Aliquots (10 pl) of 0.06% HZ02 were added every 1 min for 10 min (22). The cells were again washed five times and lysed with Tris-buffered saline (TBS) (20 mM Tris-HC1 and 150 mM NaCI, pH 7.4) that contained 1% Triton X-100,l mM EDTA, and 1 mM phenylmethylsulfonyl fluoride. The lysates were centrifuged (15,000 X g ) for 5 min.
Metabolic Labeling of Cultured Cells-Confluent monolayers of HUVE, BAE, and human fibroblasts were incubated for 24 to 48 h in methionine-free DMEM-H16 medium supplemented with 10% fetal calf serum and 250 FCi of either [35S]methionine (Amersham) or [3H] leucine (Amersham) per 60-mm dish. Labeled monolayers were lysed as described above. The lysates were centrifuged for 1 h at 4 "C a t 100,000 X g to remove cytoskeletal components (24).
Immunoprecipitation-Cell lysates were diluted to 1 ml with TBS

RESULTS AND DISCUSSION
T h e specificity of antibodies against GP IIh-IIIa and GP IIIa was determined by immunoprecipitation of Triton X-100-soluhilized, ""I-labeled platelets (Fig. I). Glycoproteins llh<, and IIIa were the predominant surface glycoproteins that were laheled by lactoperoxidase-catalyzed iodination of intact platelet,s (Fig. 1, lnnc I ) . T h e GP Ilh-IIla antihody detected three hands (Fig. 1, lane 2 )  lIIa antihody immunoprecipated hoth GP IIh and GP IlIa ( Fig. 1, lnna .?), hecause the glvcoproteins were present as a Ca'+-dependent, heterodimer complex (28). However. when platelets were lysed using EDTA at X "C and pH 8.5, the GI' IlIa antihody detected only GP IlIa (Fig. 1, Inn(, .I), hecarlse. under these conditions, the GP IIh-IIIa complex was dissociated (28). The GP IIla antihodv did not detect the M , = 150,000 protein (GI' IIa). Controls were negative (Fig. 1, lnnr. .5). Thus, the "GP Ilh-IIIa antihodv" includes antihodies against G P Ila, while the affinity-purified ''(;I' 111a antihody" is specific for GP Illa hut also immunoprecipitates GP IIh if the two glycoproteins are complexed. Fig. 2 shows that HUVE (lnnr I ) and RAE (lnnr 5 1 crlls contained about 15 to 20 ""I-srlrface-laheled proteins. Immunoprecipitation of the HUVE cell lysate with either the G P Ilh-Illa (Fig. 2, lanc 2 ) or the GP Illa (Fig. 2, lnnr 3 ) antihody detected hands of M , = 130.000 and 110.000, which are molecular weights similar to those of human platelet G I ' Ilh and GP IIla. In addition, the GI' Ilh-IIIa antihody precipitated a third protein of M. = 150,000. which may he related to human platelet GP IIa. A control immunoprecipitate was negative (Fig. 2, lnnc 4 ) . Lysates of HAP: cells that were immunoprecipitated with either the GP llt)-IIla (Fig. 2, Inn(,  6 ) or the CP Illa (Fig. 2, lnnP 7) antihodv contained laheled proteins of M , = 135,000 and 105.000. These glycoproteins were laheled on confluent monolayers, which indicates that they are on the apical surface of the cell. T h e (;I' Ila-like protein was not detected in RAE cells. Tahle I compares the molecular weights of human and hovine platelet G I ' IIh and GP lIIa and the related memhrane proteins o f Hlh'F: and RAE cells. These molecular weights were determined hy comparing the immunoprecipitates of human platelets, HlIVE. and RAE cells to protein standards a11 electrophoresed on the same gel. The characteristic shifts in electrophoretic mohilitv of human platelet GP Ilh and GI' 11111 upon disulfide reduction (26) also occurred with the corresponding memhrane proteins of hovine platelets and endothelial cells.
Numerous proteins were laheled when H ( W E cells were incuhated with ['"Sjmethionine ( Fig.  A. B.
..-". 110,000 proteins were detected with the GP 11Ia antibody, and these proteins were not precipitated in controls (Fig. 3,  lanw 4 and 6). Immunoprecipitates from [:"S]methioninelaheled RAE cells (Fig.  3 , lane 7) contained two labeled proteins with molecular weights identical to those of the surface-labeled proteins shown in Fig. 2. Control immunoprecipitates of RAE cells were negative (Fig.

, lane X ) . [""SI
Methionine-laheled human fihrohlasts were also negative (data not shown). This experiment demonstrates that endothelial cells synthesize the memhrane proteins that are related to platelet (;P Ilh, G P IIIa, and GP Ila. T h e GP IlIa antihody coimmunoprecipitated the two HUVE and the two RAE proteins, suggesting that these proteins were complexed. Sucrose gradient sedimentation was used to further examine this possihility. Glycoproteins IIh and IIIa from ""I-laheled platelets lysed with Ca"-Triton X-100 co-sedimented as a complex (Fig. 4A, lanes 1 1 -1 3 ) . In contrast, platelets lysed with EDTA-Triton X-100 at 37 "C a n d p H 8.5 contained predominantly monomeric GP llh and GI' Illa. which sedimented independently in the higher gradient fractions (Fig. 4H). Lysates of ""I-laheled HUVE cells prepared with Ca"-Triton X-100 had two proteins of M , = 1:10,000 and 110,000 that co-sedimented with a sedimentation coefficient similar to that of the platelet GP Ilh-IIIa complex (Fig. 4 C ) . To positively identify these two labeled hands, they were immunoprecipitated from the individual sucrose gradient fractions with the GP IIIa ant,ihody (Fig. 41:). In this experi-  ment, an additional protein that had a molecular weight similar to that of human platelet GI' Ilh,, (26) was detected. This protein was not detected in earlier experiments (Figs. ' L and 3 ) because these gels used slightly lower percentages o f acrylamide. The Sedimentation of these HIJVF: proteins did not change when lysates were prepared using EDTA at 3'7 "C and pH 8.5 (data not shown). Thus, the complex o f t he HI JVE GP Ilh-Illa-like proteins is not (?a'+-dependent. T h e M , = 135,000 and 105,000 HAF: proteins also co-sedimented with R sedimentation coefficient similar to that o f t he human platelet G P Ilh-IIla complex (Fig. 4 0 ) and were selectively immunoprecipitated with the GI' IIla antihodv (Fig. 4I.Y. In addition, hovine platelet GP IIh and GP llla co-sedimented as a complex that could not he dissociated hy Ca'" chelation (data not shown). Thus, complex formation appears t o he a common ies against human platelet GI' Illa. It is surprising that these authors did not detect the presence of a GP IIb-related protein, because the data in the present report indicate that these HUVE membrane proteins are complexed. Thus, antibodies that bind to either GP IIb or G P IIIa should detect both proteins. An additional difference is that Thiagarajan et al.
(29) did not detect a change in electrophoretic mobility of the HUVE GP IIIa-like protein upon reduction of the disulfide bonds. These discrepancies could be due to differences in either (i) the number of passages of the cells in culture or (ii) the cell surface labeling methods.
The presence of endothelial cell membrane proteins that are related to platelet membrane proteins may be significant for two reasons. First, the platelet GP IIb-IIIa complex is a receptor for fibrinogen, fibronectin, and von Willebrand factor (30, 31). The demonstration that endothelial cells have a similar membrane protein complex suggests that endothelial cells may also have specific receptors for these three proteins. Indeed, Dejana et al. (14) have demonstrated specific binding of fibrinogen to HUVE cells. Second, because the platelet GP IIb-IIIa complex serves to mediate platelet aggregation (31), the related proteins on endothelial cells could serve as attachment sites for circulating cells. We are currently examining endothelial cells for the presence of additional platelet membrane proteins and other cells of the vessel wall for the presence of GP IIb-and GP IIIa-like proteins.