Differences in laminin fragment interactions of normal and transformed endothelial cells

EXPERIMENTAL CELL RESEARCH

196, 177-183(1991)

Differences in Laminin Fragment Interactions of Normal and Transformed Endothelial Cells MONIQUEAUMAILLEY,’RUPERTTIMPL, ANDWERNERRISAU* Max-Planck-Znstitute fiir Biockemie and *Psychiutrie, D-8033, Martirtsried, Federal Republic of Germany

Bovine aortic and microvascular endothelial cells showed good adhesion with spreading on fibronectin or collagen IV and to a lower extent on laminin. Recognition of native laminin was due to its long arm fragment ES and was mediated by a6 integrins as demonstrated by antibody inhibition. A considerably stronger, RGDdependent interaction was observed with the isolated laminin short arm fragment P 1 previously shown to represent a cryptic cell-binding site. No adhesion was observed with the heparin-binding fragment E3. In contrast, murine microvascular endothelial cells transformed by the polyoma middle T oncogene showed preferential adherence and spreading on laminin via its ES cell-binding site and also showed adhesion to fragment E3. Attachment to laminin fragment Pl and to collagen IV was low or negative and was never followed by spreading. These data show that the transformation of microvascular endothelial cells, which give them the property to form hemangiomas, also leads to changes in cell adhesion to extracellular matrix proteins, particularly to laminin fragments. 0 1991 Academic PKSS, h.

INTRODUCTION Adhesion of endothelial cells to components of the extracellular matrix has been shown to modulate their proliferation, migration, and differentiation [l-3]. These effects are considered to be relevant for angiogenesis and maturation of the vascular system in uiuo [2, 41. Endothelial cells apparently need a versatile receptor repertoire for extracellular recognitions. During the formation of new capillaries they initially migrate into a fibronectin-rich matrix which is later replaced by basement membrane structures [4]. Regression of blood vessels is also accompanied by extracellular matrix changes in developing cartilage [5] and the ductus arteriosus [6] and in model systems of angiogenesis inhibition [ 71.Important observations in this context are the ability of endothelial cells to form capillary tube-like structures in vitro in response to collagen IV, laminin, and collagen 1To whom correspondence dressed.

and reprint

request

should he ad-

matrices [l, 8-101. Another aspect so far not studied with endothelial cells relates to the possibility of changes in cell-matrix interactions and integrin expression upon transformation such as observed in other oncogenically transformed cells [ 111. In endothelial cells, expression of the polyoma middle T oncogene induces rapid proliferation in vitro and hemangioma formation in uiuo [12,13]. These tumors consist of endothelial celllined cystic cavities implicating an atypical production and recognition of extracellular matrix constituents. The binding of various integrin receptors to extracellular ligands appears to be a major initial event in cellmatrix interaction [ 14,151. Endothelial cells attach and spread on a large variety of extracellular matrix components including interstitial collagens, basement membrane collagen IV, fibronectin, vitronectin, and laminin [l, 3, 16-181. They also express several integrin receptors including primarily the heterodimers (u2/31,~y3@1, a5fi1, (~~$33, aIIbp3, and to a lower extent &I1 and (u6/31 with some variations observed among cells obtained from large and small vessels [17-201. Studies using affinity chromatography and antibody inhibition implicated a2Bl [17] and in addition @l, ~y6/?1,and &V/33 [20] as laminin receptors. Several of these integrins are also known to bind to other, structurally unrelated extracellular substrates [ 141. Examination of Arg-GlyAsp (RGD) synthetic peptides which bind to many integrins have so far provided conflicting results for endothelial cell-laminin interactions [l, 9, 16, 201. The large number of endothelial cell integrins binding laminin have not yet been correlated with distinct domains of the protein. Studies using other cell types have localized major cell-binding sites to laminin fragments E8 and E3 originating from its long arm and to fragment Pl, a short arm structure [21-231. The E8 cell-binding site appears to be the major site recognized on laminin [21,24], is RGD-independent but dependent on protein conformation [ 25,261, and primarily recognized by (~6/31 integrin [23, 251. The Pl cell-binding site is apparently cryptic [24], depends on an RGD sequence [27], and is recognized by several ,f31and 83 integrins [23]. Fragment E3 possesses also the major heparin-binding site of laminin [28] and may interact with integrin and nonintegrin receptors [23].

177 Copyright 8 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.

178

AUMAILLEY,

a

TIMPL,

b

1.0-

Substrate added Iyg/welll

FIG. 1. Attachment profiles of MEC (a) and of eEnd.2 (b) cells to laminin, laminin fragments, and collagen IV. The substrates used were laminin-nidogen (0); fragments E8 (m), Pl (A), and E3 (A); and collagen IV (Cl). Attached cells were analyzed by a calorimetric assay with background values on serum albumin (about 0.1) already subtracted.

The precise binding sites of laminin for endothelial cells have so far not been identified except that they may include an RGD sequence and possibly an YIGSR sequence [29] both being exclusively localized in fragment Pl. In the present study we have analyzed the adhesion patterns for defined laminin fragments and studied their specificity and relevance. In addition, we compared normal and transformed microvascular endothelial cells and found distinct differences in their recognition patterns. MATERIALS

AND METHODS

Cells and ceU cultures. Bovine aortic endothelial cells (AEC) were prepared from fresh adult aortas as described previously [30] and used after 3-12 passages. Microvascular endothelial cells (MEC) from bovine adrenocortical glands were kindly supplied by Dr. D. Gospodarowicz, San Francisco. The origin and maintenance of the mouse endothelioma cell lines bEnd.3 and eEnd.2 has been described [12,13]; eEnd.2 cells were kindly provided by Dr. E. Wagner, Vienna. Cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) with low glucose (1 g/liter) and 5% fetal calf serum. Confluent cells were then harvested for adhesion assays. Proteins, synthetic peptides, and antibodies. The laminin-nidogen complex was purified from the mouse Engelbreth-Holm-Swarm (EHS) tumor and used after elastase digestion to prepare the cellbinding fragments E8 and E3 [28,31]. Pepsin digestion in 0.1 M glytine . HCl, pH 1.9, was used to obtain fragment Pl [27]. Collagen IV from the EHS tumor grown in lathyritic mice was solubilized by partial reduction [32]. The preparation of neutral salt-soluble collagen I from rat skin [33] and pepsin-solubilized collagen VI from human placenta [34] followed standard protocols. Fibronectin from human plasma (Behringwerke Marburg) was further purified by chromatography on heparin-Sepharose. Synthetic peptides GRGDS (Promega, Switzerland) and RGES (Peninsula, St. Helens) were purchased from commercial sources. Peptides YIGSR-amide and its cyclic variant and CQAGTFALRGDNPQGCSPamide corresponding to mouse laminin Bl and A chain sequences, respectively, were synthesized and kindly provided by Drs. S. Henke and J. Knolle, Hoechst AG.

AND

RISAU

Polyclonal antisera specific for laminin fragments Pl or ES were raised in rabbits [21, 281. A rat monoclonal antibody (GoH3) against the (~6 subunit of mouse integrin [35] was kindly given by Dr. A. Sonnenberg, Amsterdam. Cell attachment assay. Cell attachment was performed on 96-well tissue culture plates (Costar, Cambridge, MA) as previously described [36]. Briefly, the wells were coated by overnight adsorption at 4°C of protein solutions in distilled water ranging from 0.6 to 40 fig/ ml followed by blocking (2-4 h) with 1% bovine serum albumin. Control wells were only blocked with albumin and showed absorbance readings < 0.2 which were subtracted from the actual adhesion readings. Confluent cells were suspended by 0.05% trypsin, 0.02% EDTA in phosphate-buffered saline (PBS); were pelleted by low speed centrifugation; resuspended in serum-free DMEM (l-4.10’ cells/ml), and were seeded on the coated wells (100 wl/well). After 30-60 min incubation at 37°C nonattached cells were washed off with PBS. Attached cells were fixed for 10 min with 70% ethanol and stained with 1% crystal violet. After washing the color was resuspended in 0.2% Triton X-100 and the absorbance measured in an Elisa reader (Dynatech) at 570 nm. In inhibition assays the inhibitors (synthetic peptides or antibodies) were mixed with the cells before seeding. When using anti-substrate antibodies the coated wells were incubated before the attachment assay with serial dilutions of antibodies in PBS (60 min, 25’C) which were removed prior to adding cells.

RESULTS

Adhesion of Endothelial Cells to Laminin and CellBinding Laminin Fragments (E8, PI, E3) Cell adhesion to laminin, laminin fragments, collagens, and fibronectin substrates was analyzed using bovine aortic endothelial cells (AEC) and microvascular endothelial cells (MEC) from bovine adrenocortical glands serving as prototypes of normal cells from large and small vessels, respectively. These cells were compared with two endothelioma cell lines bEnd.3 and eEnd.2 which have been transformed with the middle T oncogene (Fig. 1, Table 1). The examination of dose-response profiles with MEC (Fig. la) and AEC (not TABLE

1

Adhesion of Various Endothelial ments, Fibronectin, and Collagen to Laminin-Nidogen (100%)

Cells to Laminin FragSubstrates in Comparison

% Adhesion

to

Laminin fragments

Collagens

Cells

ES

Pl

E3

AEC MEC bEnd.3 eEnd.2

84 76 127 113

148 142 68 18

9 0 62 82

Fibronectin 144 128 78 96

IV

I

VI

128 167 106 48

nd nd 78 14

51 92 nd nd

Note. Dose dependence of adhesion was determined as shown in Fig. 1 and plateau values obtained are recorded in comparison to laminin-nidogen substrates. Average of three to seven independent determinations. nd, not determined.

LAMININ

INTERACTIONS

OF NORMAL

AND

shown) demonstrated similar attachment to lamininnidogen and fragment E8 which, however, were relatively inefficient both in terms of the plateau levels achieved (Table 1) and the concentration required for a maximal response when compared to more efficient substrates (i.e., collagen IV). Yet, the same cells adhered much stronger to fragment Pl when compared to laminin but did not bind to fragment E3. Adhesion patterns of the endothelioma cell lines (Fig. lb, Table 1) were completely different with laminin-nidogen and fragment E8 being now the most active substrates. The cells also adhered distinctly to fragment E3 but only moderately or weakly to fragment Pl. Previous data with several endothelial cells [ 1,16,17] have shown that fibronectin and basement membrane collagen IV are more active adhesion substrates than laminin. We could confirm these data for AEC and MEC but not for the transformed cells (Table 1). For the latter, laminin-nidogen and fibronectin were equally efficient substrates while in one case (eEnd.2) the activity of collagen IV was distinctly reduced. Yet, collagen IV was for all four cells still the better substrate when compared to the interstitial tissue collagens I or VI (Table 1). A further parameter examined was spreading of cells exposed to optimal substrate concentrations for 60 min with some examples shown in Fig. 2. For each type of cells, substrates which induced the highest adhesion also induced spreading. AEC and MEC spread well on fibronectin, collagen IV, and fragment Pl while no or only little spreading was observed on laminin-nidogen and fragment E8. On the contrary, both endothelioma cell lines, eEnd.2 and bEnd.3, spread on fibronectin, laminin-nidogen and fragment E8 but did not spread on fragment Pl or collagen IV. They exhibited partial spreading on fragment E3.

Inhibition of Adhesion by Antibodies to Laminin Domains and Integrins The substrate specificity of individual laminin fragments to the adhesion of cells on laminin was studied by using polyclonal antibodies to fragments Pl and E8 (Fig. 3). The antibodies were shown by Elisa tests to bind as strongly to laminin as to the fragments and to be domain-specific as shown previously [21,28]. The adhesion of AEC (Fig. 3a) and of MEC (not shown) to Pl substrates could be completely inhibited by an antiserum against fragment Pl while adhesion to lamininnidogen was not affected by the same antiserum. Conversely, a comparable inhibition of endothelioma cell adhesion to both fragment E8 and laminin-nidogen was achieved with an antiserum against E8 (Fig. 3b). All the inhibitions were specific as shown by controls with normal serum or antisera against nidogen. Since these data implicated that recognition of lami-

TRANSFORMED

ENDOTHELIAL

CELLS

179

nin by endothelial cells occured by its E8 cell-binding site we examined also the inhibition by a monoclonal antibody to the a6 integrin subunit. This antibody was previously shown to block completely adhesion of a large number of cells to fragment E8 but to a variable degree adhesion to laminin-nidogen [23,25]. This antibody could completely block adhesion of MEC, bEnd.3 (Fig. 4), and AEC (not shown) to fragment E8 but exhibited only partial inhibition (up to 50%) when tested against laminin-nidogen over the same concentration range. The eEnd.2 cell line was exceptional in being completely blocked for laminin but only partially for E8 adhesion by the same antibody.

Inhibition

of Adhesion by RGD Peptides

We have previously shown that the cell-binding site recognized on fragment Pl can be competed for by synthetic peptides possessing the RGD sequences of fibronectin or laminin [27]. Similar inhibitions were performed for MEC adhesion on fragment Pl (Fig. 5). Both the fibronectin sequence (GRGDS) and a longer laminin peptide sequence (CQAGTFALRGDNPQGCSPamide) used in linear or cyclic, disulfide-bonded form were highly efficient and of comparable activity in blocking adhesion. Control peptides including RGES and YIGSR-amide were inactive. Our latter observation is at variance with observations for umbilical vein endothelial cells [29] since their adhesion to laminin was reported to be sensitive against YIGSR peptides. The inhibitory activity of GRGDS was then compared for all four endothelial cells using different substrates (Table 2). Most efficient inhibition was observed for the Pl substrate with the bEnd.3 cell line being less sensitive than AEC or MEC. The peptide also inhibited adhesion to fibronectin of AEC and MEC but not of the transformed cells. No inhibitory effects were observed for laminin-nidogen and collagen IV substrates except for a very low one in the case of MEC. We, however, also noticed a large variation in different experiments with MEC and laminin-nidogen showing IC, values for GRGDS in the range 40 to >lOOO PLM. Peptide RGES was inactive up to 1000 PM when tested in all the combinations shown in Table 2. DISCUSSION

Basic observations of our study are significant differences between two normal and transformed endothelial cell lines in the recognition of laminin and other extracellular matrix substrates and of established cell-binding fragments of laminin in adhesion and spreading patterns. Normal cells obtained from large or small bovine vessels displayed a relatively low binding to laminin but a stronger one for its fragment Pl comparable to fibronectin and collagen IV, the latter data being in agree-

180

AUMAILLEY,

eEnd.2

TIMPL,

AND

RISAU

IUlEC

endothelial cells (MEC) and endothelioma cells (eEnd.2). Cells were seeded on the indicated FIG. 2. Spreading patterns of microvascular substrates. After 60 min. unattached cells were removed by PBS washings. Attached cells were fixed and stained (see Materials and Methods) and photographs were taken under phase contrast microscopy (X200).

LAMININ

INTERACTIONS

OF NORMAL

AND

TRANSFORMED

ENDOTHELIAL

181

CELLS

0

I 0

10

100 Inhibitor

Antiserum

dilution

FIG. 3.

Inhibition of cell adhesion of AEC (a) and eEnd.2 (b) on various laminin substrates by domain-specific anti-laminin antibodies. The substrates used were laminin-nidogen (0, 0), fragment Pl (A, A), and fragment ES (W, 0). Inhibitors were in (a) an antiserum against Pl (0, A) and normal rabbit serum (0, A) and in (b) antisera against E8 (0, n ) and nidogen (0, q ).

ment with most previous observations [l, 16, 171. Transformed murine microvascular endothelial cells which express the polyoma middle T oncogene [13] show a stronger preference for laminin and two different fragments, ES and E3. Using several normal and tumor cells all three laminin fragments implicated have been previously shown to provide different cell-binding sites which can be distinguished by their binding properties to the cognate integrin receptors [21-271. We are aware of the problem that we compare normal (bovine) and transformed (mouse) cells of different origin, but the large numbers of cells required for adhesion assays preclude a comparison with normal primary mouse endothelial cells. However, normal endothelial cells from different species and organs have been found to display similar adhesion properties to extracellular matrix pro-

100

[JJMI

FIG. 5. Inhibition of adhesion of MEC cells on laminin fragment Pl by various synthetic peptides. Inhibitors used were the fibronectin sequence GRGDS (0), the laminin sequence CQAGTFALRGDNPQGCSP-amide either in linear (A) or cyclic form (A), and the peptides RGES (Cl) and YIGSR-amide (w). A cyclic form of the latter peptide was also inactive over the same concentration range (not shown).

teins [3, 81. Thus, we presume normal mouse endothelial cells would show similar adhesion characteristics as their bovine counterparts. The E8 cell-binding site is located at the end of the long arm of laminin, is of high affinity (& about nM), and represents a major site on laminin recognized by cells as indicated by inhibition experiments using domain-specific antibodies and radioligand binding assays [21, 22, 241. Most but not all cells recognize this site by the (~6/ll integrin [23, 251 including as shown here several endothelial cells. The E8 cell-binding site is conformation-dependent and primarily contributed by the laminin A chain [26]. It was also shown that recognition of the E8 site by a661 integrin is an essential step in establishing epithelial cell polarity during renal tubulogenesis which correlates with the new expression of the (~6integrin subunit and the laminin A chain at this stage [37-391. Yet, it was not possible to detect A chain ex-

100

TABLE

2

Inhibition of Endothelial Cell Adhesion to Various Substrates by Synthetic GRGDS Peptide IC,

0” 0.1

1 Antibody

O-03 concentration

1 [pe/mil

FIG. 4. Inhibition of adhesion of microvascular endothelial cells (a) and endothelioma cells bEnd.3 (b) by a monoclonal antibody (GoH3) to the integrin 016subunit. Substrates used were fragment E8 (0) and laminin-nidogen complex (w). Values are recorded in comparison to controls without antibody.

Cells AEC MEC bEnd.3 eEnd.2

Fragment Pl <20 <20 108 nt”

(pM) for adhesion to

Lamininnidogen >lOOO 440* >lOOO >lOoo

Fibronectin

Collagen

150 45 >looo >looO

a Not tested because of low adhesion. * Range 40 to >lOC@ when tested in different

experiments.

>630 500 >looO >looO

IV

182

AUMAILLEY,

TIMPL,

pression in embryonic blood vessels by in situ hybridization and immunofluorescence [37,40]. This indicates no or only low amounts of A chain in their basement membranes which as shown here may reflect the low potential of normal endothelial cells to recognize the fragment E8 cell-binding site. Whether endothelial cells may recognize isoforms of laminin different to the EHS tumor prototype [41] remains an open possibility. The major Pl cell-binding site was found to be cryptic in EHS tumor laminin [24] but highly sensitive to inhibition by RGD-containing peptides [27]. This agrees with the lack of blocking the site on laminin but not on Pl by Pl-specific antibodies (Fig. 3a) and RGD-peptides (Table 2). Synthetic RGD peptides designed according to fibronectin or laminin A chain sequences were of equal inhibitory activity as previously shown for other cells [27]. Given the low abundance of laminin A chain in vessels it is also not clear whether the potential receptors on normal endothelial cells in fact interact in situ with the Pl structure. The cryptic nature of the Pl site on mouse laminin also implicates a possibly proteolytic activation mechanism if this site is being used in duo. So far no evidence exists for such a mechanism. It is, however, of interest in this context that MEC exhibited a variable sensitivity for inhibiting their adhesion to laminin by RGD peptides. Such cells may therefore be useful for analyzing whether they are capable of unmasking the Pl site. Cellular recognition of the heparin-binding domain E3 of laminin A chain was recently demonstrated by adhesion [23] and radioligand assays [42]. Receptors for this site are expressed by a large variety of cells (M. Aumailley, unpublished) but apparently not by normal endothelial cells. Also, the E3 site seems to play no major role in laminin adhesion of the endothelioma cell lines although they possess the receptors. The nature of the E3 receptors is unclear but they may include integrins [23], cell surface heparan sulfate proteoglycans, and sulfatides [42]. The integrin ~~201was recently shown to be a laminin receptor on umbilical vein endothelial cells by affinity chromatography and antibody inhibition [17]. Yet, the same integrin from platelets did not bind to laminin [43]. Integrins from skin MEC which bind to a laminin affinity column included a2@1, a6@1, avB3, and small amounts of cul/31[20]. Partial antibody inhibitions demonstrated that both (r6fil and av@3 are the major receptors involved in cell adhesion to laminin [20]. This agrees with our observation on the incomplete inhibition of laminin adhesion but not E8 adhesion by anti-a6 observed at least for three types of endothelial cells. The nature of the laminin domains recognized by the other integrins is not quite clear. Pl recognition by the vitronectin receptor avfl3 and unknown /31 integrins was implicated from inhibition studies with human tumor cell lines [23]. With rat hepatocytes it was shown that a181

AND

RISAU

integrin binds to both the laminin fragments E8 and Pl [44]. No laminin domain which binds a201 integrin is known so far. Changes in cell-substrate adhesion are a hallmark of transformed cells and may be attributed to changes in integrin receptors observed in oncogenically transformed cells [ 111. In endothelial cells, expression of the polyoma middle T oncogene induces hemangioma formation in uiuo [13] and an aberrant hemangioma-like morphogenetic behavior of endothelioma cells in threedimensional fibrin gels in vitro [ 121. These studies have also shown that polyoma middle T oncogene expression in endothelial cells caused an increase in proteolytic activity due to an upregulation of urokinase-type plasminogen activator and a simultaneous decrease of plasminogen activator inhibitor type 1 (PAI-1). Neutralization of excess proteolytic activity by exogenously added protease inhibitors corrected the hemangioma-like behavior and allowed the formation of capillary-like structures in vitro [12]. It is unknown whether the increase in proteolytic activity is necessary or sufficient for hemangioma formation in uiuo. The profound difference in laminin fragment and collagen IV interaction between normal and polyoma middle T oncogene expressing endothelial cells suggests that changes in cell adhesion mechanisms may also be important for normal and pathological vascular development. Laminin is expressed during embryonic vasculogenesis and angiogenesis and may be involved in the maturation of embryonic blood vessels [4]. It is conceivable that the altered interaction of endothelioma cells with laminin prevents the maturation of a capillary network and supports endothelial cell proliferation and hemangioma formation. Alternatively, the excess proteolytic activity may alter adhesion molecules by either destroying important binding sites or by uncovering new binding sites such as the cryptic RGD-site. Altered adhesive mechanisms and the increase in proteolytic activity may act synergistically to induce hemangiomas in vim. In addition, changes in integrin receptors might have a direct effect on the control of the proteolytic activity of endothelial cells. Vitronectin, for example, has been shown to bind and stabilize PAI- [45] and in turn PAI- was reported to stabilize the vitronectin-dependent adhesion of human fibrosarcoma cells [46]. Thus, the expression of the a@3 vitronectin receptor might be critical for the control of PAI- activity. Further studies are needed to investigate these possibilities. We thank Mrs. Hildegard Reiter and Miicella 6calan for skilled technical assistance, Dr. Erwin Wagner for discussions, and the Deutsche Forschungsgesellschaft for financial support (project Ti95/7-1).

REFERENCES 1.

Herb&, T. J., McCarthy, J. B., Tsilibary, (1988) J. Cell Bid. 106, 1365-1373.

E. C., and Furcht, L. T.

LAMININ Ingber, D. E., and Folkman,

INTERACTIONS

OF NORMAL

AND

24.

J. (1989) Cell 58,803~805.

Madri, J. A., Williams, S. K., Wyatt, T., and Mezzio, C. (1983) J. Cell Biol. 97, 153-165. Risau, W., and Lemmon, V. (1988) Deu. Biol. 125,441-450.

6.

7.

26.

de Reeder, E. D., Girard, N., Poelmann, R. E., van Munsteren, J. C., Patterson, D. F., and Gittenberger, G. (1988) Am. J. Pathol. 132,574-585.

27.

Ingber, D. E., Madri, J. A., and Folkman, 119, 1768-1775. B. M., and Madri,

29.

9.

Kubota, Y., Kleinman, H. K., Martin, G. R., and Lawley, T. J. (1988) J. Cell Biol. 107,1589-1598. Montesano, R., Orci, L., and Vasalli, P. (1983) J. Cell Biol. 97, 1648-1652. Plantefarber, L. C., and Hynes, R. 0. (1989) Cell 56,281-290.

J. A. (1986) Lab. Znuest.

12.

Montesano, R., Pepper, M. S., Miihle-Steinlein, U., Risau, W., Wagner, E. F., and Orci, L. (1990) Cell 62,435-445.

13.

Williams, R. L., Risau, W., Zerwes, H. G., Drexler, H., Aguzzi, A., and Wagner, E. F. (1989) Cell 57, 1053-1063. Humphries, M. J. (1990) J. Cell Sci. 97,585-592. Ruoslahti, E., and Pierschbacher, M. D. (1987) Science 238, 491-497.

14. 15. 16.

Basson, C. T., Knowles, W. J., Bell, L., Albelda, S. M., Castronovo, V., Liotta, L. A., and Madri, J. A. (1990) J. Cell Biol. 110, 789-801.

17.

Languino, L. R., Gehlsen, K. R., Wayner, E. A., Carter, W. G., Engvall, E., and Ruoslahti, E. (1989) J. Cell Biol. 109, 24552462. Tarone, G., Stafanuto, G., Mascarello, P., Defilippi, P., Altruda, F., and Silengo, L. (1990) J. Lipid Med. 2,545-553.

18.

28.

J. (1986) Endocrinology

Form, D. M., Pratt, 55,521-530.

11.

25.

Hallmann, R., Feinberg, R. N., Latker, C. H., Sasse, J., and Risau, W. (1987) Differentiation 34,98-105.

8.

10.

TRANSFORMED

19.

Albelda, S. M., Daise, M., Levine, E. M., and Buck, C. A. (1989) J. Clin. Invest. 83, 1992-2002.

20.

Kramer, R. H., Cheng, Y.-F., and Clyman, 111, 1233-1243.

21.

Aumailley, M., Nurcombe, V., Edgar, D., Paulsson, Timpl, R. (1987) J. Biol. &em. 262, 11,532-11,538.

22.

Goodman, S. L., Deutzmann, CellBiol. 105, 589-598.

23.

Sonnenberg, A., Linders, C. J. T., Modderman, P. W., Damsky, C. H., Aumailley, M., and Timpl, R. (1990) J. Cell Biol. 110, 2145-2155.

R. (1990) J. Cell Biol.

30. 31. 32. 33.

34. 35. 36. 37. 38. 39. 40. 41. 42. 43.

M., and 44.

R., and von der Mark, K. (1987) J.

Received February 28,199l Revised version received May 27, 1991

45. 46.

ENDOTHELIAL

CELLS

183

Nurcombe, V., Aumailley, M., Timpl, R., and Edgar, D. (1989) Eur. J. Biochem. 180,9-14. Aumailley, M., Timpl, R., and Sonnenberg, A. (1990) Exp. Cell Res. 188.55-60. Deutzmann, R., Aumailley, M., Wiedemann, H., Pysny, W., Timpl, R., and Edgar, D. (1990) Eur. J. Biocbem. 191,513-522. Aumailley, M., Gerl, M., Sonnenberg, A., Deutzmann, R., and Timpl, R. (1990) FEBS Lett. 262.82-86. Ott, U., Odermatt, E., Engel, J., Furthmayr, H., and Timpl, R. (1982) Eur. J. Biochem. 123,63-72. Grant, D. S., Tashiro, K. I., Segui-Real, B., Yamada, Y., Martin, G. R., and Kleinman, H. K. (1989) Cell 58,933-943. Schwartz, S. M. (1978) In Vitro 14,966-980. Paulsson, M., Aumailley, M., Deutzman, R., Timpl, R., Beck, K., and Engel, J. (1987) Eur. J. Biochem. 166, 11-19. Yurchenco, P. D., and Furthmayr, H. (1984) Biochemistry 23, 1839-1850. Miller, E. J., and Rhodes, R. K. (1982) in Methods in Enzymology (Cunningham, L. W., and Frederiksen, D. W., Eds.), Vol. 82, pp. 33-64, Academic Press, San Diego. Odermatt, E., Risteli, J., van Delden, V., and Timpl, R. (1983) Biochem. J. 211,295-302. Sonnenberg, A., Janssen, H., Hogervorst, F., Calcafat, J., and Hilgers, J. (1987) J. Biol. Chem. 262, 14,030-14,038. Aumailley, M., Mann, K., von der Mark, H., and Timpl, R. (1989) Exp. Cell Res. 181,463-474. Ekblom, M., Klein, G., Mugrauer, G., Fecker, L., Deutzmann, R., Timpl, R., and Ekblom, P. (1990) Cell 60, 337-346. Klein, G., Langegger, M., Timpl, R., and Ekblom, P. (1988) Cell 55,331-341. Sorokin, L., Sonnenberg, A., Aumailley, M., Timpl, R., and Ekblom, P. (1990) J. Cell Biol. 111, 1265-1273. Klein, G., Ekblom, M., Fecker, L., Timpl, R., and Ekblom, P. (1990) Development 110,823-837. Timpl, R. (1989) Eur. J. Biochem. 180,487-502. Taraboletti, D., Rao, C. N., Krutzsch, H. C., Liotta, L. A., and Roberts, D. D. (1990) J. Biol. Chem. 265,12,253-12,258. Kirchhofer, D., Languino, L. R., Ruoslahti, E., and Pierschbather, M. D. (1990) J. Biol. Chem. 265,615-618. Forsberg, E., Paulsson, M., Timpl, R., and Johansson, S. (1990) J. Biol. Chem. 265,6376-6381. Schleef, R. R., Podor, T. J., Dunne, E., Mimuro, J., and Loskutoff, D. J. (1990) J. Cell Biol. 110,155-163. Ciambrone, G. J., and McKeown-Longo, P. J. (1990) J. Cell Biol.

111,2183-2195.