Vascular Endothelial Cells Synthesize a Plasma Membrane Protein Indistinguishable from the Platelet Membrane Glycoprotein Ira*

the mobility of the The isolated under the


From the Department of Blood Coagulation, Central Laboratory of the Netherlands Red Cross Blood Transfusion Service, Amsterdam, The Netherlands
To define the role of membrane components that function in endothelial cell physiology and to characterize them biochemically, we have attempted to prepare monoclonal antibodies specific for endothelial cells. Several clones were obtained producing antibodies which bound to endothelial cells and also to platelets. The antibody of one of these clones, CLB-HEC 75, was studied in more detail. This antibody is directed against a single protein which is synthesized constitutively by endothelial cells and is expressed on the surface of both endothelial cells and platelets. The CLB-HEC 75 antigen was isolated from Nonidet P-40-solubilized endothelial cells and platelets by immunoprecipitation and exhibited an apparent molecular weight by sodium dodecyl sulfate-polyacrylamide gel electrophoresis of approximately 145,000 in the presence of 2-mercaptoethanol. Two-dimensional polyacrylamide gel electrophoresis and crossed immunoelectrophoresis revealed that the mobility of the CLB-HEC 75 antigen relative to platelet glycoproteins Ib, IIa, IIb, and IIIa fits previously defined criteria for platelet membrane glycoprotein IIa. The CLB-HEC 75 antigen isolated from endothelial cells co-migrated under all conditions tested with the antigen from platelets. These results indicate that endothelial cells share a plasma membrane protein indistinguishable from platelet membrane glycoprotein IIa. This protein may be a component involved in the interaction of endothelial cells with their environment including coagulation factors, platelets, and the subendothelial matrix. CLB-HEC 75 may serve as a useful tool for studying these processes.
The endothelium is a continuous sheet of mesenchymal cells that forms the antithrombogenic lining of the vascular system. It is widely held that the non-thrombogenic nature of these cells can be ascribed to the properties of a subset of endothelial cell-derived substances, either associated with the cell surface or secreted by the cells. For instance, endothelial cells may secrete plasminogen activators (1,2), prostacyclin (3, 4), and antithrombin I11 (5), all components that possess the capacity to modulate hemostatic and thrombotic events. Similarly, the synthesis of cell-associated components such as heparan sulfate (6,7), a glycosaminoglycan that accelerates the inactivation of thrombin by antithrombin 111 (8,9), and thrombomodulin, a factor that enhances the thrombin-catalyzed activation of the anticoagulant protein C (10, l l ) , suggests an active role for endothelial cells in the modulation of hemostasis and thrombosis.
On the other hand, a possible role of the endothelium in initiating and localizing coagulation should also be considered.
Recent studies indicate that endothelial cells also can bind several coagulation factors, including factors IX and X (12)(13)(14) and, when stimulated, promote the activation of these factors (15). Thus, the role of the endothelium in maintaining compatibility between blood and the vessel wall is more complex than was previously anticipated. Several of the above-mentioned catalytic processes take place at the surface of the endothelial cell, and the existence of an array of membrane receptors for coagulation factors, including fibrinogen, thrombin, and factors IX and X, which mediate these processes has been proposed (9,10,(12)(13)(14)(15)(16)(17)(18)(19). In this respect, endothelial cells resemble blood platelets which also provide a catalytic surface for coagulation activation once they are activated (20)(21)(22).
So far, limited information is available about the chemical composition and architecture of the endothelial plasma membrane and the chemical nature of the membrane constituents that could contribute to the molecular events associated with both the anticoagulant and procoagulant properties of the endothelial membrane. In order to better define the role of membrane components that function in endothelial cell physiology and to characterize them biochemically, we have prepared endothelial cell-specific monoclonal antibodies. Several of the clones obtained, however, were found to react with platelets as well. We have characterized the antigen recognized by one of these antibodies. Our studies indicate that, in terms of its electrophoretic mobility under various physicochemical conditions, this protein is indistinguishable from platelet GP' IIa. We suggest that GP IIa-like molecules produced by endothelial cells could be involved in the interaction of these cells with blood components and the subendothelial matrix.

EXPERIMENTAL PROCEDURES
Cell Culture and Labeling-Human vascular endothelial cells were isolated from umbilical veins and cultured in 75-cm2 flasks (Corning MT, Corning, NY) according to the method originally described by Jaffe et al. (23) with some minor modifications (24). The cells were lands Utrecht, Utrecht, The Netherlands. dium dodecyl sulfate-polyacrylamide gel electrophoresis. § Present address: Department of Hematology, University Hospital The abbreviations used are: GP, glycoprotein; SDS-PAGE, so-identified by their typical characteristics (23). For metabolic and surface labeling, cells from the second passage were cultured on a microcarrier support system of negatively charged spherical plastic beads (Biosilon, A/S Nunc, Kamstrup, Denmark) precoated with fibronectin, essentially as described by Davies (25). The endothelial cells readily grew to a density of about 40 cells/bead under static conditions. After the cells reached confluency, beads were washed twice with phosphate-buffered saline, pH 7.4, and cells were incubated for 24 h in methionine-depleted Ham's F-10 medium (Flow Laboratories, Irvine, United Kingdom) supplemented with human serum albumin (10 mg/ml, prepared in our laboratory), transferrin (20 pg/ml, Sigma), bovine insulin (10 pg/ml, Sigma), and [%]methionine (30 pCi/ml, The Radiochemical Centre, Amersham, United Kingdom). The beads were then washed three times with TENA buffer (10 mM Tris, 10 mM Na2EDTA, 0.15 M NaCI, pH 7.41, and the cells were extracted for immunochemical analysis (see below). For membrane labeling, beads were washed three times with TENA buffer and resuspended in this buffer a t a concentration of 1.5 X lo5 beads/ ml (about 6.5 X IO6 endothelial cells/ml). Cell-surface iodination with NaIZ5I (Amersham) was catalyzed by 1,3,4,6-tetrachloro-3~~,6a-diphenylglycouril (Iodogen, Pierce Chemical Co., Rotterdam, The Netherlands) (26). Subsequently, the beads were washed three times with TENA buffer and processed for immunochemical studies (see below). Platelets were isolated and iodinated by the Iodogen procedure as described previously (27). 42.5 ml of blood from normal human donors was collected in 7.5 mi of acid citrate (23.4 mM sodium citrate, p H 4.5). Platelets from platelet-rich plasma, prepared by centrifugation of the citrated blood a t 160 X g for 15 min, were washed once in a buffer containing 0.12 M NaC1, 0.13 M sodium citrate, 0.03 M glucose, 10 mM Na2EDTA, pH 6.5, and then twice with TENA buffer and resuspended in this buffer a t a concentration of 1-2 X 109/ml. After labeling, cells were washed three times with TENA buffer and extracted for immunochemical analysis (see below). T o identify the plasma membrane glycoproteins IIb and IIIa, similar labeling experiments and immunochemical analyses were performed with platelets from a patient with Glanzmann's thrombasthenia, a disorder known to be associated with a deficiency of these glycoproteins (28)(29)(30). This patient (R. E.) had a history of a life-long bleeding tendency, a prolonged bleeding time, and absent platelet aggregation in response to ADP and collagen. His platelets did not express the plateletspecific alloantigen Zw" (PIA*), an antigen localized on G P IIIa (31), and did not bind to a monoclonal antibody (coded C17) known to be directed to G P IIIa (27) as well as several other monoclonal antibodies directed to either GP IIIa or the GP IIb. IIIa complex? To identify G P Ib, platelets of a .patient with the Bernard-Soulier syndrome, a disorder known to be associated with a deficiency of this plasma membrane component (32,33), were surface-iodinated and analyzed.
This patient (H. K.) had a life-long bleeding tendency, a prolonged bleeding time, unusually large platelets, moderate thrombocytopenia, defective ristocetin-induced platelet aggregation, normal ADP and collagen-induced aggregation, and a normal plasma factor VIII-von Willebrand factor concentration. The platelets failed to bind a monoclonal antibody (AN 51, provided by Dr. A. J. McMichael, John be directed to GP Ib (34).
Radcliffe Hospital, Oxford, United Kingdom) previously described to Monoclonal Antibody Production-Monoclonal antibodies against cultured human umbilical vein endothelial cells were generated from BALB/c mice immunized with trypsinized cells from the third passage. Lymphocyte hybridization was performed as described (35). Hybridomas were tested for production of antibodies to endothelial cells by means of an indirect enzyme-linked immunosorbent assay using endothelial cells cultured and subsequently fixed with glutaraldehyde on 96-well plates (Costar). Clones secreting antibodies with an affinity for endothelial cells were subcloned twice by limited dilution and injected into mice to obtain ascites fluid. The specificity of the monoclonal antibodies was assessed on wells coated with paraformaldehyde-fixed purified erythrocytes, granulocytes, lymphocytes, monocytes, platelets, and methanol-fixed cultured human fibroblasts and smooth muscle cells. Monoclonal IgG was purified from ascites by ammonium sulfate precipitation and DEAE-Sephadex chromatography.
Immunoprecipitation-Labeled cells (platelets and endothelial cells) were analyzed essentially as described by Tetteroo et al. (27). After washing with TENA buffer, cells were solubilized a t 4 "C for 1 * P. Modderman, H. Huisman, A. E. G. Kr. von dem Borne, and J.
A. van Mourik, unpublished observations. h in 0.01 M Tris, pH 7.8, containing 1% Nonidet P-40 (Sigma), 0.15 M NaCI, and the following protease inhibitors: 1 mM phenylmethylsulfonyl fluoride, 10 mM Na2EDTA, 0.02 mg/ml soybean trypsin inhibitor, aprotinin (100 units/ml), 10 mM benzamidine, 1 mM diisopropylfluorophosphate, and 5 mM N-ethylmaleimide (lysate buffer). The lysate was centrifuged a t 13,000 X g for 15 min a t room temperature to remove undissolved material. The supernatant was centrifuged for 60 min a t 100,000 X g at 4 "C and precleared three times with a preformed complex of a control monoclonal antibody (antipollen ascites) and goat anti-mouse IgG, once for 18 h at 4 "C, and then twice for 1 h a t 4 "C. Precleared lysates were incubated for 16 h with a preformed complex of the monoclonal antibody and goat antimouse IgG a t 4 "C. The precipitate was resuspended in lysate buffer, containing 0.5% deoxycholate, and washed in a discontinuous gradient consisting of one layer of 10% sucrose containing 0.5% Nonidet P-40 and 0.5% deoxycholate in lysate buffer and one layer of 20% sucrose without detergents. For one-dimensional electrophoresis under reducing conditions, the precipitate was resuspended in 40 pl of SDS sample buffer (0.25 M Tris-HC1, 8% SDS, 40% glycerol, 0.004% bromphenol blue, and 15% 0-mercaptoethanol, pH 6.8) and heated for 5 min a t 100 "C. This material was subjected to SDS-PAGE as described by Laemmli (36). For nonreduced-reduced two-dimensional slab gel SDS-PAGE, the method of Phillips and Poh Agin (37) was employed. Radioactive bands or spots were visualized by exposing dried gels to Kodak X-Omat R P x-ray films in a cassette containing an intensifier screen. Molecular weight standards included myosin (200,000), &galactosidase (116,250), phosphorylase b (92,500), bovine serum albumin (66,200), ovalbumin (45,000), carbonic anhydrase (31,000), and soybean trypsin inhibitor (21,500) (Bio-Rad).
Crossed Immunoelectrophoresis-Crossed immunoelectrophoresis of solubilized (unlabeled) platelets and endothelial cells was performed as described previously (38), except that labeled monoclonal IgG was incorporated into the second dimension along with the unlabeled anti-platelet antiserum (39). After staining with Coomassie Brilliant Blue, radioactive precipitation lines were visualized by autoradiography.
Immunofluorescence-Immunofluorescence studies were carried out on various tissue sections and confluent endothelial cells grown on glass coverslips precoated with fibronectin. Slides were fixed in methanol for 5 min. Staining was carried out at room temperature. Slides were incubated with monoclonal antibodies (ascites, 1:500) for 30 min, washed, stained with fluorescein isothiocyanate-conjugated polyvalent goat anti-mouse immunoglobulin, and examined by immunofluorescence microscopy after mounting in a solution containing 9 parts 87% glycerol, 1 part phosphate-buffered saline, p H 7.4, and 0.1% (w/v) p-phenylenediamine in 1 mM sodium phosphate, pH 8.6.

RESULTS
Antigen Specificity of Monoclonal Antibody CLB-HEC 75-Of the 18 hybridomas tested that had antibodies that reacted with endothelial cells, three of these cloned lines produced antibodies which also reacted with platelets. One of these clones, CLB-HEC 75 was studied in more detail. The antigen recognized by this monoclonal antibody was distinct from any of the known platelet-specific alloantigens; the antibody reacted equally strong with typed platelets from six donors, all having a different platelet alloantigen make-up  (62). Magnification of cell cultures was X 540; that of tissue sections was X 320. nous distribution of the CLB-HEC 75 antigen (Fig. 1A). The intensity of the staining was variable. Although this was not systematically studied, it appeared that the intensity of the staining was dependent on the cell density; particularly those cells that were in close contact or growth-arrested stained more brightly. Preliminary experiments also indicated a matrix location for the CLB-HEC 75 antigen (not shown). Immunofluorescent staining of tissue sections of various organs revealed that only the endothelium in all tissues examined, including pancreas (Fig. IC), spleen (Fig. lD), skin (Fig. l E ) , and liver (not shown), stained positively.
Identification of the CLB-HEC 75 Antigen from Endothelial Cells and Platelets-Several approaches were used to characterize the endothelial and platelet antigens recognized by CLB-HEC 75. We first asked ourselves whether the antigen is actually synthesized by endothelial cells. Cells may internalize or bind serum proteins, including platelet remnants or release products during culturing (41)(42)(43). To see if the endothelial antigen can be metabolically radiolabeled, we immunoprecipitated the antigen from endothelial cells that were cultured in the presence of [3sS]methionine. Fig. 2 shows an autoradiograph of the radiolabeled endothelial cell proteins (lune I ) and the material precipitated with preformed complexes of CLB-HEC 75 and goat anti-mouse immunoglobulins (lune 2 ) after electrophoresis on SDS-PAGE gel under reduced conditions. A labeled band, migrating as a protein with an apparent M , of 145,000, was retained by the CLB-HEC 75 anti-mouse immunoglobulin complex. The band revealed by CLB-HEC 75 is specific; this protein band was not seen with monoclonal antibody controls (Fig. 2, lane 3).
Analysis of a CLB-HEC 75 insolubilized antigen preparation from Nonidet P-40 extracts of endothelial cells, iodinated by the Iodogen procedure, a surface labeling method (26), also showed a single band migrating with the same mobility as the metabolically labeled antigen (Fig. 2, lane 4). In view of the cross-reactivity with platelets, also the protein from solubilized, surface radioiodinated platelets was immunoprecipitated with this antibody. The protein recognized by CLB-HEC 75 was radiolabeled and co-migrated with its endothelial cell counterpart (lane 6). For comparison, lune 7 shows glycoproteins IIb and IIIa, precipitated with CLB-C17, a monoclonal antibody previously shown by us to be directed to GP IIIa (27). The apparent M , of the CLB-HEC 75 antigen is clearly different from that of GP IIIa and GP IIb (the membrane glycoprotein that occasionally coprecipitates with GP IIIa). In fact, electrophoretic analysis performed with the unreduced molecule rather suggests that the CLB-HEC 75 antigen is similar or identical to GP IIa. The unreduced protein (Fig. 2, lane 9) migrates slightly faster than the molecule in its reduced form and also faster than unreduced  (26). Lysates of these cells were incubated with preformed complexes of monoclonal antibodies and goat anti-mouse immunoglobuolins (27). Washed ternary complexes were subsequently reduced with 15% 2-mercaptoethanol and analyzed by SDS-PAGE on 9% gels (36). Visualization was by autoradiography. GP IIb (lane 8). This change in mobility on SDS gels relative to GP IIb is similar to that previously described for GP IIa (37).
As one-dimensional SDS-PAGE does not allow a precise identification of the antigen recognized by CLB-HEC 75, we next employed nonreduced-reduced two-dimensional gel electrophoresis to identify the CLB-HEC 75 membrane antigen (37). Fig. 3A shows an autoradiograph of a two-dimensional slab gel of solubilized radioiodinated platelets. Some of the membrane glycoproteins could be identified on the basis of their position using previously published platelet membrane maps as a reference (33,37,44). In addition, the position of GP Ib was established when the platelets of a patient with the Bernard-Soulier syndrome were examined by this technique (Fig. 3B). Similarly, GP IIb and IIIa could be unequivocally identified using platelets from a patient with Glanzmann's thrombasthenia (Fig. 3C) .   Fig. 3, D and E, shows the position of the proteins immunoprecipitated with CLB-HEC 75 from surface-radiolabeled platelets and metabolically radiolabeled endothelial cells, respectively. A mixture of these samples is shown in Fig. 3 F and indicates that the antigen from respectively endothelial cells and platelets behaves identically under these conditions. Two-dimensional analysis of a mixture of the CLB-HEC 75 antigen, either isolated from platelets or endothelial cells, with immunoprecipitated GP IIb. IIIa revealed that the position of the CLB-HEC 75 antigen is clearly different from that of GP IIb or IIIa (Fig. 3, G and H). Its position relative to GP IIb and IIIa supports our view that the antigen isolated from both platelets and endothelial cells is very similar or identical to platelet GP IIa (33, 37, 44).
We finally employed crossed immunoelectrophoresis (38,39) to substantiate this view. Fig. 4A shows the Coomassie Brilliant Blue-stained pattern obtained when solubilized platelets were electrophoresed against polyclonal rabbit antiplatelet antibodies. As suggested by previous reports, the predominant immunoprecipitate observed represents the GP 1Ib.IIIa complex (45,46). This could be confirmed when radiolabeled IgG of the anti-GP IIIa monoclonal CLB-C17 was incorporated into the second dimension along with the unlabeled polyclonal antiserum (Fig. 4B). When either radioiodinated CLB-HEC 75 IgG or a mixture of CLB-HEC 75 and CLB-C17 was incorporated in the second dimension, the corresponding autoradiograph demonstrated (Fig. 4, C and D) that the mobility of the CLB-HEC 75 antigen was distinct from GP IIb. IIIa but similar to that of GP IIa reported previously (38). Identical results were obtained when instead of platelets, solubilized endothelial cells were examined by crossed immunoelectrophoresis with CLB-HEC 75 incorporated in the second dimension (Fig. 4E) or when a mixture of platelet and endothelial cells lysate was analyzed in this way (Fig. 4F). Again, these results strongly suggest that endothelial cells synthesize a protein closely related or identical to platelet GP IIa.

DISCUSSION
During the past few years, it has become increasingly apparent that endothelial cells act as more than passive spectators in the events implicated in hemostasis and thrombosis. For instance, endothelial cells may secrete several components, including protease inhibitors (5,47,48)

49, 50)
, which are potentially capable in modulating hemostatic and thrombotic processes. The potential of endothelial cells to control these processes is corroborated by the findings that these cells also possess the capacity to bind certain coagulation factors (9, 10,12-19) and may serve as a catalytic surface for the coagulation cascade (14, 15). In this respect, endothelial cells behave similarly to platelets which also may assemble coagulation factors at their surface and catalyze activation of the coagulation system (20-22). Although it becomes more clear now that several molecular events related to hemostasis and thrombosis may occur at the endothelial cell membrane level, no information is available concerning the exposure or expression of receptors and enzymatic activities which are unique to the endothelial cell membrane sur-face. To identify important structures on the endothelial cell membrane, we have undertaken a study to prepare monoclonal antibodies specific for endothelial cells. Several clones were obtained, however, that reacted with endothelial cells as well as platelets.
In this study, we have characterized the antigen recognized by one of these antibodies, CLB-HEC 75. The accumulated data suggest that the CLB-HEC 75 antigen has not been previously identified. In the first place, the determinant described here can be distinguished on the basis of cell specificity and apparent molecular weight from previously described (51, 52) endothelial cell-surface antigens. Similarly, the antigen is distinct with regard to its cellular distribution and molecular weight from von Willebrand factor, thrombospondin, and fibronectin which are present in both endothelial cells and platelets (43,50,(53)(54)(55)(56)(57)(58). In fact, the following arguments led us to conclude that the antigen recognized by CLB-HEC 75 is similar or identical to platelet GP IIa. The molecule isolated by immunoprecipitation with CLB-HEC 75 from '251-labeled patients and [35S]methionine-labeled endothelial cells fits the criteria established for GP IIa with two-dimensional nonreduced-reduced SDS-PAGE (33,37,44). Our data clearly indicate that the CLB-HEC 75 antigen isolated from both endothelial cells and platelets does not co-migrate with GP Ib, GP IIb, or GP IIIa (Fig. 3). This view is supported by the finding that, as visualized by crossed immunoelectrophoresis, another technique frequently used to identify platelet membrane glycoproteins, the CLB-HEC 75 antigen from platelets as well as endothelial cells migrates with the mobility ascribed to GP IIa (38) and is clearly different from GP IIb, IIIa, or Ib (Fig. 4).
Taking these findings together, it seems justifiable to conclude that endothelial cells synthesize a protein structurally related to or identical with GP IIa. It is of course possible that the observed cross-reactivity is due to the fortuitous presence of a similar epitope on the two cellular proteins involved (59). In view of the close similarity between platelet GP IIa and its endothelial counterpart in terms of their physicochemical properties plus the fact that up until now three other monoclonal antibodies were obtained (CLB-HEC 65, CLB-HEC 170, and CLB-6G10) that exhibit this crossreactivity, this proposition seems unlikely. Our data suggest that, as has been proposed for platelet GP IIa, that the GP IIa-related protein of endothelial cells is a membranous component as well. This conclusion is based on the finding that, as in platelets, endothelial cell GP IIa is readily iodinated by the Iodogen procedure, a surface labeling method (26). Although it has been shown that with this technique predominantly surface proteins are labeled, it cannot, however, be excluded that intracellular proteins are radioiodinated as well.
On the other hand, as demonstrated by immunofluorescent analysis, only the plasma membrane of the endothelial cell is clearly stained with the three anti-GP IIa antibodies available (cf. Fig. l), supporting our view that the GP IIa-related protein is an endothelial cell membrane protein that is expressed on the cell surface. This finding raises an important question: what is the function of endothelial cell GP IIa? As, unlike the other major membrane glycoproteins including GP Ib and the GP IIb . IIIa complex, the functional role of (platelet) GP IIa is unknown, this question is difficult to answer. The antigen recognized by CLB-HEC 75 probably differs from the antigen recognized by two recently described antibodies specific for thrombin-activated platelets (60, 61), one of which was proposed to be directed to GP IIa (60). In preliminary experiments, we have observed no increase of binding of CLB-HEC 75 antibodies to platelets upon thrombin treatment. Further-more, the CLB-HEC 75 antibody and its Fab' fragment induce platelet aggregation (not shown).
Whatever the functional significance of endothelial G P IIa may be, the antibody described herein seems a potential tool to determine the function of not only platelet GP IIa but also its endothelial counterpart. Such studies should increase our understanding, at the molecular level, of the role of the endothelial cell membrane in the array of processes related to both thrombosis and hemostasis.