Experimental Cell Research 149 (1983) 483-497
Factor VIII-related Antigen A Pericellular Matrix Component of Cultured Human Endothelial Cells
M. HORMIA,‘S2,* V.-P. LEHTO’ and I. VIRTANEN’ ‘Department of Pathology and ‘Bacteriology and Immunology, University of Helsinki, SF-00290 Helsinki 29, Finland
We studied the extracellular localization of factor VIII-related antigen (VIIIR : Ag) in cultures of human endothelial cells. The cells deposited both VIIIR : Ag and fibronectin already during their initial adhesion phase and in immunofluorescence microscopy of spread cells extracellular VIIIR : Ag was localized to tibrils coaligning with pericellular tibronectin. When human tibroblasts, which do not synthesize VIIIR : Ag, were cultured in endothelial cell post-culture medium, a fibrillar matrix localization of VIIIR : Ag was seen, comparable to that of endothelial cell cultures. A fibrillar VIIIR : Ag-specific staining was also seen in cell-free pericellular matrices of endothelial cells, produced by deoxycholate treatment. In irnmunoelectron microscopy, VIIIR : Ag was seen in iibrillar extracellular material between and underneath the cells and in cell-free matrices of endothelial cells as well. In immunofluorescence microscopy of cell-free matrices, VIIIR : Ag codistributed with both fibronectin and type III procollagen. Digestion of the matrices with purified bacterial collagenase abolished the type III procollagen-specific fluorescence, whereas the tibrillar VIIIR : Ag-specific staining, codistributing with tibronectin, remained unaffected. In electrophoresis of isolated, metabolically labelled endothelial cell matrices, major polypeptides with M, 220-240; 180; 160; 80 and 45 kD and some minor polypeptides were resolved. In addition, immunoblotting revealed Iibronectin, VIIIR : Ag and type III procollagen as components of cell-free matrices of endothelial cells. Direct overlay of iodinated cellular libronectin on electrophoretically separated polypeptides of cultured endothelial cells, transferred to nitrocelhtlose, suggested that fibronectin binds directly to VIIIR: Ag. Our results indicate that VIIIR : Ag produced by human endothelial cells is a component of the pericellular matrix and is not bound to collagen but may directly associate with iibronectin.
Factor VIII-related antigen (VIIIR : Ag) is a large, multimeric glycoprotein which forms the bulk of the factor VIII-complex circulating in plasma [l-3]. Both endothelial cells and megakaryocytes synthesize VIIIR : Ag and it is also found in platelets [3-6]. Recent studies have shown that in cultured endothelial cells VIIIR : Ag can be found intracellularly in Weibel-Palade bodies [7] and extracellularly in fibrillar material associated with the pericellular matrix [7-91. In vivo, VIIIR: Ag has been shown to be present both in endothelial cells and in the subendothelial basement membrane [6, 10, 111. VIIIR : Ag appears to bind to the surface of activated platelets [12] and to participate in platelet aggregation and adhesion [12-141. Also flbronectin and * To whom offptint requests should be sent. Address: Department of Pathology, University of Helsinki, Haartmaninkatu 3, SF-00290 Helsinki 29, Finland. copyri&t Q 1983 by APress, Inc. All t+ts of reproduction in ally form reserwd 0014-48?7/83 583.4)
484 Hormia, Lehto and Virtanen thrombospondin have been suggested to have a related role in the interaction of platelets with collagen [15, 161and in the adhesion of platelets [12]. The mechanisms of platelet adhesion and the interactions between platelet surface glycoproteins and subendothelial matrix components are, however, still unclear. The pericellular matrix produced in vitro by different types of endothelial cells has been reported to contain fibronectin, laminin, collagens type III, IV and V and a novel type of collagen, designated EC collagen [17-231. On the other hand, Kramer & Nicolson [24] reported that the isolated pericellular matrix of bovine arterial endothelial cells mainly consists of tibronectin. In this study we show that VIIIR : Ag is a component of the pericellular matrix produced by human umbilical vein endothelial cells. We further show that it is not bound to collagen as suggested earlier [25], but that it appears to interact with fibronectin. MATERIALS
AND METHODS
Cell Culture and Matrix Preparations Endothelial cells were harvested from umbilical veins by collagenase treatment [26] and then cultured in Petri dishes coated with human plasma fibronectin (Collaborative Research, Cambridge, UK) i Ham’s FlO medium supplemented with 20 % pooled and inactivated human AB serum (Finnish Red Cross Blood Transfusion Service, Helsinki, Finland), 75 l&ml endothelial cell growth supplement (ECGS, Collab. Res.) and antibiotics as described previously [9, 261. Human embryonal Iibroblasts were obtained from a local source and were cultured in RPM1 1640medium supplemented with 10% fetal calf setum (FCS) (Gibco Biocult, Irvine, Scotland) and antibiotics. For metabolic labelling, the cells were cultured in methionine-free MEM medium supplemented with L-[35S]methionine (50 uCi ml-‘; sp. act. 1000 Ci mmol-‘, Radiochemical Centre, Amersham, UK) for 3 h, in glucose-free MEM medium (puryvate added), supplemented with n-[2-3H]mannose (2 uCi ml-‘, Radiochemical Centre), or in complete MEM medium in the presence of [2-3H]glycine (20 uCi ml-‘, Radiochemical Centre) and 50 ergml-’ sodium ascorbate (Sigma, St Louis, MO.). Cell-free pericellular matrices were prepared from confluent cultures by the deoxycholate method [27,28]. The cell layers were first extracted with 0.5 % sodium deoxycholate in 10 mM Tiis-HCl buffer, pH 8, for 30 min at 22°C. Then the dishes were washed with 10 mM Tiis-HCl, pH 8.0, containing 50 ug ml-’ DNse I (Worthington, Freehold, N.J.), and 1 mM phenylmethyl sulfonyhluoride (PMSF), for 5 min, followed by 2 mM Tris-HCI, pH 8.0, with 1 mM PMSF, for 5 min. For some experiments the pericellular matrix preparations were subsequently digested with purified bacterial collagenase (100 U ml-‘, form III, Advanced Biofactures, Lynbrook, N.Y.) for 1 h at room temperature.
Zmmunojluorescence Microsocpy For fluorescence staining, the cells were grown on glass coverslips coated with human plasma tibronectin. The cells or cell-free pericellular matrix preparations were fixed with 3.5 % paraformaldehyde, in 0.1 M phosphate buffer (pH 7.2), for 10 min. For cytoplasmic staining the cells were subsequently permeabihzed with 0.05% Nonidet P40 (NP 40, BDH Chemicals Ltd Poole, UK). Rabbit antibodies against VIIIR : Ag and fibronectin were obtained from Behringwerke (Marburg, FRG). The monospeciticity of these antibodies has been described earlier [26] and identical results in both indirect immunofluorescence and immunoblotting were obtained with anti-VIIIR : Ag antibodies preabsorbed with purified human plasma tibronectin (Collaborative Research) and with monospecific rabbit antibodies against human VIIIR : Ag kindly provided by Dr L. W. Hoyer (cf [lo]). Tetramethyl rhodamine isothiocyanate (TRITC)-coupled goat immunoglobulins against VIIIR : Ag were from Atlantic Antibodies (Westbrook, Me). Rabbit antibodies against type III procollagen have been described elsewhere [29] and were a generous gift of Dr Rupert Timpl (Martinsried, FRG). TRITCconjugated rabbit anti-Iibronectin antiserum and TRITC-or FITC (fluorescein isothiocyanate)-conjuExp Cell Res 149(1983)
VIIIR : Ag as pericellular matrix component
485
gated sheep anti-rabbit IgG antisera were from Cappel Laboratories (Cochranville, Pa). The specimens were examined in a Zeiss Universal microscope, equipped with an epi-illuminator IIIRS and filters for FITC and TRITC fluorescence.
Electron Microscopy For electron microscopic visualization of VIIIR : Ag, confluent endothelial cell cultures or cell-free matrices (see above) were fixed with 2.5 % glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2, for 30 min and then incubated in NaCl-P buffer (140 mM NaCl, 10 mM sodium phosphate, pH 7.2) containing 0.1 M lysine-HCl, for 2 h. After this, the cells and the pericellular matrix preparations were exposed to anti-VIIIR : Ag antibodies or normal rabbit serum (control specimens) in NaCl-P buffer containing 1% bovine serum albumin (BSA, Sigma) and 10% inactivated normal sheep serum, and then washed overnight in NaCl-P buffer. The samples were further exposed to biotinylated sheep antirabbit IgG and, after washing, to aviditiiotinylated horseradish peroxidase-complex (VectastainTM ABC kit, Vector Laboratories, Burlingame, Calif.). Then the samples were washed and incubated for 20 min in peroxidase substrate solution (0.03 % 3,3-diaminobenzidine-tetrahydrochloride (DAB) in 50 mM ‘B-is-HCl, pH 7.6, with freshly added 0.06% H*O,). The cells and cell-free matrices were postfixed with 1% 0~0~ in 0.1 M phosphate buffer for 90 min and then dehydrated with ethanol and mounted in Epon 812. Thin sections were made either horizontally or vertically to the cell layer. The specimens were studied without post-staining in a JEOL 100 CX transmission electron microscope at an accelerating voltage of 60 kV.
Electrophoresis, Immunoblotting
and Fibronectin Overlay
Polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDS) was done according to Laemmli [30[ using 6.5 % slab gels. For fluorography, the gels were processed according to Bonner & Laskey [31]. 14C-labelled molecular weight (MW) standards were from Radiochemical Centre: myosin (210 kD), phosphorylase B (93 kD), bovine serum albumin (68 kD), ovalbumin (43 kD) and carbonic anhydrase (30 kD). Immunoblotting was done by the technique ot Towbin et al. [32]. Briefly, electrophoretically separated polypeptides were transferred from polyacrylamide gels to nitrocellulose sheets using a commercial blotting apparatus (Bio Rad Laboratories, Richmond, Va). Amido black (0.1%) was used for protein staining. For immunostaining, the sheets were exposed to rabbit anti-VIIIR : Ag, anti-type III procollagen or anti-tibronectin antibodies, whereafter biotinylated sheep anti-rabbit IgG was applied, followed by the avidinibiotinylated horseradish peroxidase complex (VectastainTM, Vector Laboratories). The polypeptides recognized by the primary antibodies were then visualized by exposing the sheets to the peroxidase substrate solution as above. Overlay with iodinated cellular libronectin was used to detect tibronectin-binding polypeptides of endothelial cells. Purified human cellular fibronectin (cFn, Bethesda Research Laboratories, Gaithersburg, Md) was iodinated by the chloramine T-method [33] to a specific activity of 400 Ci mmol-’ (‘251-cFn). Carrier-free “‘1 was from Radiochemical Centre. For overlay experiments the nitrocellulose sheets, with endothelial cell polypeptides (see above), were first incubated overnight in 3 % BSA in NaCl-P buffer. Then, ‘251-tibronectin was applied to the sheets for 3 h. After thorough washing in NaCl-P buffer the sheets were dried and exposed to the X-ray fdm.
RESULTS Distribution of VIIIR : Ag in the Pericellular Matrix In immunofluorescence microscopy of confluent endothelial cell cultures, antibodies to VIIIR: Ag gave in surface staining a fibrillar matrix fluorescence codistributed with tibronectin-specific staining and restricted to cell interphase areas (fig. 1a-c). Instead, in permeabilized cell layers, in which also the underside of the cells was accessible to antibodies, both VIIIR: Ag- and fibronectinspecific fluorescence were seen as a fibrillar pericellular matrix type of staining 32-838339
Exp Cell Res 149 (1983)
486 Hormia, Lehto and Virtanen
1. Double staining with anti-VIIIR: Ag- (a, d, g, J and anti-fibronectin (b, e, h, k) antibodies of cultured endothelial cells. In surface staining of confluent cultures VIIIR : Ag-specific staining codistributes (a) with fibronectin-specific fluorescence and (b) at cell interphase areas. (c) Phase contrast
Fig.
Exp Cell Res 149 (1983)
VIIIR : Ag as pericellular matrix component
487
Fig. 2. Immunofluorescence microscopy of human embryonal tibroblasts cultured in endothelial cellconditioned medium. Anti-VIIIR : Ag antibodies give a cell-surface-associated fibrillar fluorescence (a) which codistributes with pericellular fibronectin (b). Note the granular appearance of VIIIR : Agspecific fluorescence (a) as compared to the fibrillar libronectin-specific staining (b), compare with fig. 1g, h). (c) Phase contrast micrograph of the same field. x600.
extended to the entire growth area (fig. 1d-A. A similar codistribution of VIIIR : Ag and tibronectin could also be seen in subconfluenct cultures in which the fibrillar organization of VIIIR : Ag was even more clearly discernible (fig. 1g-9. In immunofluorescence microscopy of spreading endothelial cell cultures antibodies to both VIIIR : Ag and flbronectin gave, already 2 h after plating, a fibrillar fluorescence underneath the cells, located mainly at the cell edges (fig. 1j-l). In immunofluorescence microscopy of human embryonal fibroblasts, cultured either in standard conditions or in the presence of 20 % human serum, no staining was seen with anti-VIIIR : Ag antibodies. However, if these cells were cultured in endothelial cell post-culture medium for 48 h and were then double-stained for VIIIR : Ag and fibronectin, a librillar cell surface-associated staining was seen with both antibodies. The VIIIR : Ag-specific fluorescence appeared, however, slightly more granular than the fibronectin-specific fluorescence, similarly as in endothelial cell cultures (fig. 2 a-c, compare with fig. 1g, h). In immunofluorescence microscopy of cell-free matrix preparations of conflumicrograph of the same field. After permeabilization of the cell layers VIIIR : Ag (d) coalignes with fibronectin-containing fibrils (e) throughout the growth substratum. (f) Phase contrast micrograph of the same field. In subconfluent cultures the fibrillar localization of VIIIR: Ag (g) in co-distribution with pericellular fibronectln tibrils (h, r) is more clearly discernible. In spreading endothelial cell cultures VIIIR : Ag b’) and fibronectin (k) are seen in fibrlllar material located mainly at the cell edges (0. Note the more coarse appearance of VIIIR: Ag-specific staining (a, d, g, j) as compared with tibronectin-specific fluorescence (b, e, h, k). (a-f)x400; (g-f)x800. Exp Cell Res 149 (1983)
488 Hormia, Lehto and Virtanen
Fig. 3. Double staining of cell-free matrix preparations of (n, b) confluent and (c, d) spreading endothelial cell cultures with (a, c) anti-VIIIR : Ag and (b, d) anti-fibronectin antibodies. VIIIR : Ag and fibronectin are codistributed in a network of fibrillar material in (a, 6) and in thin tibrils apparently located to cell attachment areas in (c, d). (a, 6)x400; (c, d)x800.
ent endothelial cell cultures both anti-VIIIR : Ag and anti-fibronectin antibodies decorated a fibrillar meshwork extending throughout the culture substratum (fig. 3 a, 6). Despite the somewhat granular staining with anti-VIIIR : Ag antibodies, an apparent codistribution was seen between both components. Similarly, in pericellular matrices prepared from spreading endothelial cell cultures, a codistribution of VIIIR : Ag- and fibronectin-specific fluoresence could be observed at regions apparently corresponding to cell-substratum contact areas (fig. 3 c, 6). In immunoelectron microscopy of vertical sections of whole endothelial cells, VIIIR : Ag was located to extracellular fibrillar material, seen both underneath the cell layer and between overlapping cells (fig. 4a, b). In horizontal sections VIIIR : Ag was associated to a meshwork of extracellular tibrils which at some places seemed to be attached to the plasma membrane (fig. c). In cell-free matrices, similarly, the peroxidase reaction product decorated a network of Exp Cell Res 149 (1983)
VIIIR : Ag as pericellular matrix component
489
Fig. 4. Immunoelectron microscopy of extracellular VIIIR : Ag in (a-c) cell layers and in (d) cell-free matrix of endothelial cells. In vertical sections of cell layers the peroxidase reaction can be seen in flbrillar material both underneath the cells (a) and between overlapping cells (b). In (b) staining can also be seen within membrane vesicles. In horizontal sections the peroxidase staining is located to a tibrillar network between adjacent cells (c). Note the close association of VIIIR : Ag-containing fibrils with the plasma membrane in (c). In cell-free matrices peroxidase reaction product is seen throughout a network of librillar material (4. (a) x36000; (b) x42000; (c) x10000; (d) x15000.
fibrillar material (fig. 44. In control specimens only a faint background staining was seen (not shown). The Effect of Collagenase Digestion on Matrix Composition Double immunofluorescence staining of pericellular matrix preparations of confluent endothelical cell cultures with anti-VIIIR : Ag and anti-type III procollagen antibodies revealed a codistribution between the two components (fig. 5 a, b) comparable to that obtained with anti-VIIIR : Ag and anti-fibronectin antibodies. After digestion of pericellular matrices with bacterial collagenase (100 U ml-‘), an unaltered VIIIR : Ag-specific fluorescence was seen (fig. 5 c), whereas type III procollagen-specific fluorescence was completely abolished (fig. 56). The codistribution between VIIIR : Ag and fibronectin appeared, however, unaltered also after collagenase treatment (fig. 5 e, j). Polyacrylamide gel electrophoresis of [2-3H]glycine-labelled endothelial cell Exp Cell Rcs 149 (1983)
4% Hormia, Lehto and Virtanen
Fig. 5. Double-staining of cell-free matrices of endothelial cells for (a, c) VIIIR : Ag and (b, d) type III
procollagen, (a. b) before and (c, d) after collagenase digestion and of a collagenase-digested matrix preparation for (e) VIIIR : Ag and (f) tibronectin. Note the codistribution of VIIIR : Ag and type III procollagen in (a, b) and the unaffected distribution of VIIIR : Ag in the collagenase-digested matrix (c), in which no type III procollagen-specific fluorescence can be seen (6). Double staining of such a preparation for (e) VIIIR : Ag and cf) fibronectin shows a codistribution comparable to that seen in undigested matrices (compare e, fwith fig. 3 a, b). (a, b) x400; (c-B x800. Exp Cell Res 149 (1983)
VIIIR :Ag as pericellular matrix component
491
Fig. 6. Polypeptide analysis of [3H]glycine-labelled endothelial cell matrix preparations 1, before; 2, after collagenase digestion. Before digestion major polypeptides with 1, M, 220; 19&180 and 160 kD can be seen and of these 2, the M, I!%180 kD polypeptides are removed by collagenase digestion. Fig. 7. Polypeptide analysis of I, [35S]methionine-labelled whole endothelial cells, of 2, culture medium, and of 3, cell-free matrices of such cultures or of 4, cultures labelled with [3H]mannose. Note the prominent 45 kD polypeptide in 3, 4, matrix preparations and the M, 190-180 kD polypeptides seen in 3, but absent in 4. (Left) MW standards.
matrix polypeptides revealed major polypeptides with M, 220; 1%180; 160 kD and some minor polypeptides (fig. 6, I). The 220 kD polypeptide which could be identified as fibronectin by immunoblotting (see below) and the 160 kD polypeptide, apparently corresponding to thrombospondin [34,35], where not susceptible to collagenase, whereas the 180-190 kD polypeptide bands were removed by the collagenase digestion (fig. 6, 2). Identification of Endothelial Cell Matrix Components and Fibronectin-binding Polypeptides Numerous polypeptides were seen in polyacrylamide gel electrophoresis of whole endothelial cells metabolically labelled with [35S]methionine (fig. 7, I) or of culture medium of such cells (fig. 7, 2). Cell-free matrix preparations, labelled with [35Slmethionine (fig. 7,3), instead, showed only a few major polypeptides, a broad diffuse polypeptide with M, 220-235 kD and additional polypeptides with Exp Cd Res 149 (1983)
492 Hormia, Lehto and Virtanen
Fig. 8. Immunoblotting of matrix polypeptides from (a) endothelial cells and (b) tibroblasts. (a, b) I, Amido black staining of matrix polypeptides; (a) 2, immunoblotting with anti-type III procollagen antibodies; (a) 3; (b) 2, immunoblotting with anti-fibronectin antibodies and (a) 4; (b) 3, with antiVIIIR : Ag antibodies. (Left) MW standards. Fig. 9. I, Direct overlay of iodinated cellular libronectin on electrophoretically separated endothelial cell polypeptides transferred to nitrocellulose and 2, immunoblotting with anti-VIIIR : Ag antibodies of a similar sheet. Iodinated cellular tibronectin mainly binds to a 1, M, 235-225 kD polypeptide doublet which can be identified as 2, VIIIR : Ag by immunoblotting. (L&i) MW standards.
180-190; 160; 80 kD and a prominent polypeptide with M, of ca 45 kD. On the other hand, in [3H]mannose-labelled matrices distinct polypeptides with M, of 22&235; 160 and ca 45 kD were revealed (fig. 7, 4). In protein staining of endothelial cell matrix polypeptides, transferred to nitrocellulose, major polypeptides with M, of ca 22&230 kD and of 180 kD could be discerned (fig. 8 a, 1). In immunoblotting of the pericellular matrix polypeptides, anti-type III procollagen antibodies recognized a polypeptide band of M, ca 180 kD (fig. Sa, 2) antiVIIIR : Ag antibodies a sharp polypeptide doublet of M, ca 225-235 kD (fig. 8 a, 4) and anti-fibronectin antibodies decorated a more diffuse polypeptide band with M, of ca 220-230 kD (fig. 8a, 3). Immunoblotting of polypeptides from whole endothelial cells showed identical results with both anti-VIIIR : Ag and antifibronectin antibodies (not shown). Fibronectin but not VIIIR: Ag could be identified by immunoblotting in matrix preparations of human fibroblasts (fig. 8 b, Z-3). In direct overlay of electrophoretically separated polypeptides of endothelial Exp Cell Res 149 (1983)
VIIIR :Ag as pericellular matrix component
493
cells ‘%cFn bound to a polypeptide doublet with M, 225-235 kD apparently representing VIIIR : Ag, as judged by immunoblotting of a parallel lane with antiVIIIR : Ag antibodies (fig. 9,1-2). Also some other endothelial cell polypeptides, M, 47 and 110-140 kD, appeared to bind ‘*‘I-cFn. DISCUSSION The extracellular localization of VIIIR: Ag in endothelial cell cultures has recently been described by us [93 and others [7, 81. VIIIR : Ag also appears to be present in the vascular subendothelium in vivo [l 1, 361. The nature and significance of such a localization have, however, so far remained unclarified, but VIIIR : Ag has been suggested to be associated with collagen fibrils [25]. In addition to tibronectin [37] several other non-collagenous glycoproteins have recently been identified and isolated from connective tissues and pericellular matrices of cultured cells [38]. Of these, laminin [39] and entactin [40] appear to be components of basement membranes. Chondronectin [41,52] is a glycoprotein promoting the interaction of chondroblasts with collagen 1411and is apparently secreted by chondroblasts [42]. Osteonectin refers to a phosphoglycoprotein isolated from fetal calf bone which binds to hydroxyapatite and type I collagen [43]. All these glycoproteins appear to mediate cell adhesion, interact with collagens or proteoglycans or to have a role in organizing the extracellular matrix [38]. In this study we show that VIIIR : Ag is a component of the perisellular matrix produced by cultured human endothelial cells. Our results also indicate that pericellular VIIIR : Ag is not associated with collagen but may interact with tibronectin. In immunofluorescence microscopy, endothelial cells adhering to the growth substratum after trypsinization were found to deposit within a few hours typical tibronectin-containing plaques similar to those seen in fibroblast cultures [44,45]. Also VIIIR : Ag was present in such newly deposited plaques indicating concomitant deposition with the formation of a pericellular matrix and not only adsorption to a fully developed matrix. These results are in some contrast to the findings of Wagner et al. [7] who could find extracellular VIIIR : Ag only in cultures older than one week. In confluent endothelial cell cultures the pericellular matrix was mainly located underneath the cell layer and VIIIR : Ag was codistributed with fibronectin fibrils throughout the growth substratum. Also in immunofluorescence microscopy of matrices from both confluent and spreading endothelial cell cultures VIIIR : Ag was seen in codistribution with fibronectin fibrils. The presence of VIIIR : Ag in isolated pericellular matrices was further confirmed by immunoelectron microscopy and immunoblotting. In immunoblotting of both matrix polypeptides and polypeptides of whole endothelial cells anti-VIIIR : Ag antibodies decorated a polypeptide doublet with M, ca 225-235 kD. These two bands probably represent differently processed forms of VIIIR: Ag [46]. In immunoelectron microscopy of cell layers VIIIR : Ag was located to extracellular Erp Cell Res 149 (1983)
494 Hormia, Lehto and Virtanen fibrillar material both underneath and between the cells. In some areas also diffuse association with the plasma membrane and membrane vesicles and invaginations was seen. Such a distribution closely resembles the immunoelectron microscopical localization of both fibronectin and collagen in fibroblast cultures [47, 481. To study the interaction of VIIIR: Ag with other pericellular matrix proteins we cultured human fibroblasts in endothelial cell post-culture medium containing secreted VIIIR : Ag (cf [4]). Double immunofluorescence staining of such cultures revealed a codistribution of VIIIR : Ag and fibronectin fibrils comparable to that seen in endothelial cell cultures. This indicates that the extracellular matrix produced by fibroblasts contains the component(s), probably fibronectin, capable of binding VIIIR : Ag. On the other hand, VIIIR : Ag from normal human serum did not appear to bind to the pericellular matrix of cultured fibroblasts probably indicating that there are some functional differences between plasmatic VIIIR : Ag and VIIIR : Ag produced by endothelial cells. This is supported by the results of Nachman et al. [49] showing that plasmatic VIIIR: Ag differs in its multimeric composition from VIIIR: Ag secreted by endothelial cells. In addition, Sussman & Rand [36] recently reported that, in vivo, only minimal amounts of circulating VIIIR : Ag are deposited to the exposed subendothelium. Some disagreement has existed regarding the synthesis of type III collagen by endothelial cells [21, 23, 501. Sage & Bornstein [231 were unable to demonstrate type III collagen in the culture medium of human umbilical vein endothelial cells. However, as discussed by the authors themselves, this could be due to the culture conditions applied in their study. In the present study synthesis of type III procollagen in human endothelial cell cultures was revealed both by immunoblotting and immunofluorescence microscopy. When cell-free matrices were digested with purified bacterial collagenase, which removed all detectable type III procollagen, no effect on the fibrillar distribution of either VIIIR : Ag or fibronectin or on their codistribution could be seen. This was in agreement with the findings of Rand et al. [I I] who reported that collagenase treatment did not affect the subendothelial localization of VIIIR : Ag in immunofluorescence microscopy of human vessels. It is also known that collagenase treatment of pericellular matrices produced by fibroblasts does not affect the fibrillar structure or the localization of fibronectin in such preparations [27]. Furthermore, the deposition of tibronectin is known to precede that of procollagens during matrix formation in fibroblast cultures [51] indicating that the organization of a pericellular tibronectin network is independent of collagen [27]. In line with this, our results show that both fibronectin and VIIIR : Ag are rapidly deposited into the pericellular matrix of endothelial cells and that their fibrillar codistribution is independent of collagen. The direct protein overlay technique has recently been used for the detection of cellular binding sites such as growth factor receptors [52] and actin-binding proteins in fibroblasts [53]. In the present study, fibronectin-binding endothelial Exp Cell Res 149 (1983)
VIIIR :Ag as pericellular matrix component
495
cell polypeptides were studied by the direct 1251-fibronectin-overlay procedure. The results show that in such experimental conditions VIIIR : Ag is the major libronectin-binding protein of endothelial cells and suggest that the codistribution of VIIIR : Ag with pericellular fibronectin fibrils in endothelial cell cultures and in the fibroblast cultures exposed to endothelial cell-conditioned medium could be due to a direct association between VIIIR: Ag and fibronectin. Proteoglycans have been suggested to have a decisive role in the formation of the pericellular matrix of cultured cells [54]. Proteoglycans apparently may also contribute to the interaction of VIIIR : Ag with fibronectin and other matrix components. Further studies are needed to clarify this aspect in endothelial cell cultures. Electrophoretical analysis of endothelial cell matrix proteins revealed a prominent 160 kD polypeptide apparently representing thrombospondin which is known to be synthesized by endothelial cells [34, 551. This is in agreement with the recent results of Jaffe et al. [35] who showed that thrombospondin is a matrix component of cultured fibroblasts. Thus, it appears that fibronectin, VIIIR: Ag and thrombospondin, which all are synthesized by endothelial cells [3, 18, 19,34, 551 and participate in normal platelet function [12, 56-611, also are matrix glycoproteins of cultured endothelial cells. In contrast to fibronectin [37] and even thrombospondin [35, 551 VIIIR: Ag appears, however, to have unique status in being present only in endothelial cells and platelets [3]. On the basis of the present results and of biochemical [37,49,62], functional [3, 12, 371 and even molecular [ 1, 631 similarities between VIIIR : Ag and the typical adhesive matrix glycoprotein fibronectin, it can be implicated that VIIIR: Ag belongs to the expanding category of non-collagenous structural matrix components. The skilful technical assistance of MS Pipsa Kaipainen and MS Raili Taavela is kindly acknowledged. This study was supported by grants from the Finnish Foundation of Blood and Circulatory Diseases, The Finnish Cultural Foundation, the Finnish Medical Research Council, The Sigrid Juselius Foundation and the Association of Finnish Life Insurance Companies.
REFERENCES 1. Ohmori, K, Fretto, L J, Harrison, R L, Switzer, H E P, Erickson H P & McKee PA, J cell bio195 (1982) 632. 2. Hoyer, L W & Shainoff J R, Blood 55 (1980) 1056. 3. Hoyer, L W, Blood 58 (1981) 1. 4. JatTe, E A, Hoyer, L W L Nackman R L, J clin invest 52 (1973) 2757. 5. Nachman, R, Levine R & Jaffe E A, J clin invest 60 (1977) 914. 6. Bloom, A L, Giddings, J C & Wilks, C J, Nature 241 (1973) 217. 7. Wagner, D D, Olmsted, J B & Marder V J, J cell bio195 (1982) 355. 8. Booyse, F M, Quarfoot, A J, Bell, S, Fass, D N, Lewis, J C, Mahn, K G & Bowie, E J W, Proc natl acad sci US 74 (1977) 5702. 9. Hormia, M, Cell biol int res rep 6 (1982) 1123. 10. Hoyer, L W, de 10s Santos, R & Hoyer, J R, J clin invest 52 (1973) 2737. 11. Rand, J H, Sussman, K K, Gordon, R E, Chu, S V & Solomon, V, Blood 55 (1980) 752. 12. Lahav, J & Hynes, R 0, J supramol struct cell biochem 17 (1981) 299. Exp CellRes 149(1983)
496 Hormia, Lehto and Virtanen 13. Reddick, R L, Griggs, T R, Lamb, M A & Brinkhous, K M, Proc natl acad sci US 79 (1982) 5076. 14. Sakariassen, K S, Bolhuis, P A & Sixma, J J, Nature 279 (1979) 636. 15. Hoffmann P R & Hynes, R 0, Clin res 27 (1979) 297. 16. Zucker, M B, Mosesson, M V, Bockman, M J & Kaplan, K L, Blood 54 (1979) 12. 17. Sage, H, Pritzl, P & Bomstein P, Biochemistry 19 (1980) 5747. 18. Jaffe, E A & Mosher, D F, J exp med 147 (1978) 1779. 19. Macarak, H J, Kirby, E, Kirk, T & Kefalides, N A, Proc natl acad sci US 75 (1978) 2621. 20. Gospodarowicz, D, Greenburg, G, Foidart, J M & Savior, N, J cell physiol 107 (1981) 107. 21. Jaffe, E A, Minick, R, Adelman, B, Necker, C G & Nachman, R, J exp med 144 (1976) 209. 22. Sage, H, Pritzl, P & Bomstein, P, Arteriosclerosis 1 (1981) 427. 23. Sage, H & Bomstein, P, Arteriosclerosis 2 (1982) 27. 24. Kramer, R H, Gonzalez, R & Nicolson, G L, Int j cancer 26 (1980) 639. 25. Rand, J H, Gordon, R E, Sussman, I I, Chu, S V L Solomon, V, Blood 60 (1982) 627. 26. Hormia, M, Linder E, Lehto, V-P, Vartio, T, Badley, R A & Virtanen, I, Exp cell res 138 (1982) 159. 27. Hedman, K, Ku&men, M, Alitalo, K, Vaheri, A, Johansson, S & Hook, M, J cell biol81 (1979) 83. 28. Lehto, V-P, Vartio, T & Virtanen, I, FEBS lett 124 (1981) 289. 29. Timpl, R, Wick, G & Gay, R, J immunol meth 18 (1977) 165. 30. Laemmli, U K, Nature 227 (1979) 680. 31. Bonner, W M & Laskey, R A, Eur j biochem 46 (1974) 83. 32. Towbin, H, Staehelin, T & Gordon, J, Proc natl acad sci US 76 (1979) 4350. 33. Cuatrecasas, P, Biochemistry 12 (1973) 1312. 34. Mosher, D F, Doyle, M J & Jaffe, E A, J cell bio193 (1982) 343. 35. Jaffe, E A, Ruggiero, J T, Leung, L L K, Doyle, M J, McKeown-Longo, P J & Mosher, D F, Proc natl acad sci US 80 (1983) 998. 36. Sussman, I I & Rand, J H, J lab clin med 100 (1982) 526. 37. Mosher, D F, Progr hemost thormb 5 (1980) 111. 38. Aplin, J D & Hughes, R C, Biochim biophys acta 694 (1982) 375. 39. Timpl, R, Rhode, H, Robey, P G, Reennard, S I, Foidart, J-M & Martin, G R, J biol them 254 (1979) 9933. 40. Bender, N J, J&e, R, Carlin, B & Chung, A E, Am j path01 103 (1981) 419. 41. Hewitt, A T, Kleinman, H K, Pennypacker, J P & Martin, G R, Proc natl acad sci US 77 (1980) 385. 42. Hewitt, A T, Vamer, N H, Silver, M H, Dessau, W, Wilkes, C M & Martin, G R, J biol them 257 (1982) 2330. 43. Termine, J D, Kleinman, H K, Whitson, S W, Conn, K M, McGarvey, M C & Martin, G R, Cell 26 (1981) 99. 44. Virtanen, I, Vartio, T, Badley, R A & Lehto, V-P, Nature 298 (1982) 660. 45. Grinnell, F & Feld, M K, Cell 17 (1979) 117. 46. Wagner, D D & Marder, V J, J biol them 258 (1983) 2065. 47. Hedman, K, Vaheri, A & Wartiovaara, J, J cell bio176 (1978) 748. 48. Fursht, L T, Smith, D, Wendelschafter-Crabb, G, Mosher, D F & Foidart, J M, J histochem cytochem 28 (1980) 1319. 49. Nachman, R L, Jaffe, E A & Ferris, B, Biochim biophys acta 667 (1981) 361. 50. Madri, J A, Dreyer, B, Pitlick, F A L Furthmayer, H, Lab invest 43 (1980) 303. 51. Vaheri, A, Kurkinen, M, Lehto, V-P, Linder, E & Timpl, R, Proc natl acad sci US 751 (1978) 4944. C? Femandez-Pol, J A, Klos, D J & Hamilton, P D, Biochem intemat 5 (1982) 213. 53. Snabes, M C, Boyd III, A E & Bryan, J, J cell biol90 (1981) 809. 54. Hedman, K, Johansson, S, Vartio, T, Kjellen, L, Vaheri, A t Hiiiik, M, Cell 28 (1982) 663. 55. Raugi, G J, Mumby, S M, Abbott-Brown, D & Bomstein, P, J cell bio195 (1982) 351. 56. Nachman, R L & Jatfe, E A, J exp med 141 (1975) 1101. 57. Koutts, J, Walsh, P N, Plow, E F, Fenton, J W, Bouma, B N & Zimmerman, T S, J clin invest 62 (1978) 1255. 58. Zucker, M B, Mosesson, M W, Broekman, J M & Kaplan, K L, Blood 54 (1979) 8. 59. Baenzinger, N L, Brodie, G N & Majents, P W, Proc natl acad sci US 68 (1971) 240. db.
Exp Cell Res 149 (1983)
VIIIR : Ag as pericellular matrix component 60. 61. 62. 63.
497
Lawler, J W, Slayter, H S & Coligan, J E, J biol them 253 (1978) 8609. Phillips, D R, Jennings, L K & Prasanna, H R, J biol them 255 (1980) 11629. Nachman, R L, Jaffe, E A & Ferris, B, Biochem biophys res commun 92 (1980) 1208. Erickson, H P, Carrell, N & McDonagh, J, J cell bio191 (1981) 673.
Received April 14, 1983 Revised version received July 25, 1983 Printedin Sweden
Exp Cell Res 149 (1983)