attachment regions of human plasma fibronectin with - Semantic Scholar

J. Cell Sci. 81, 125-141 (1986)

125

Printed in Great Britain © The Company of Biologists Limited 1986

IMMUNOCHEMICAL AND ULTRASTRUCTURAL MAPPING OF THE GELATIN-BINDING AND CELLATTACHMENT REGIONS OF HUMAN PLASMA FIBRONECTIN WITH MONOCLONAL ANTIBODIES DAVID L. 34 HASTY1*, HARRY S. COURTNEY34, W. ANDREW SIMPSON , JOHN A. McDONALD5 AND EDWIN H. BEACHEY 234 Departments of ^Anatomy, 2Medicine, and ^Microbiology and Immunology, University of Tennessee Center for the Health Sciences, ^Veterans Administration Medical Center, Memphis, Tennessee 38163, U.SA. and 5The Pulmonary Division, Department of Medicine, Washington University School of Medicine, St Louis, Missouri 63 J JO, U.SA.

SUMMARY Monoclonal antibodies against fibronectin were used to locate the gelatin-binding and cellattachment regions of plasma fibronectin at an ultrastructural level. A total of 23 hybridomas were generated using mice immunized with either intact fibronectin or a 4O0O0Afr gelatin-binding fibronectin fragment. One of these antibodies (D9b) strongly inhibited the interaction of radiolabelled fibronectin with gelatin. Another antibody (IB10) inhibited the attachment of Chinese hamster ovary (CHO) cells to a fibronectin substratum by 99%. Both of these antibodies were purified by affinity chromatography on columns of fibronectin-Sepharose and were then incubated with soluble fibronectin to form antigen-antibody complexes. The complexes were separated from free antibody on a column of Sephadex G-200 and were prepared for electron-microscopic examination by spraying on mica discs and rotary shadowing with platinum. As determined by this method, the fibronectin molecules measured 124 ± l-7nm in length. Monoclonal antibody IB10 was visualized as a globular projection 40 ± l-4nm from one end of the fibronectin filament. Monoclonal antibody D9b, on the other hand, was visualized as a globular projection at or near one or both ends of the molecule. These data provide the first morphological localization of the gelatinbinding and cell-attachment regions of fibronectin and indicate that further studies using monoclonal antibodies directed toward other epitopes should shed light not only on function but also on the tertiary and quaternary structure of the fibronectin molecule.

INTRODUCTION

The fibronectins comprise a group of large, adhesive glycoproteins found in blood and several body fluids (Babu, Simpson & Beachey, 1983; Crouch et al. 1978; Gressner & Wallraff, 1980; Mosesson & Umfleet, 1970; Simpson, Courtney & Beachey, 1982) and associated with the pericellular matrix components of various tissues and cultured cells (Hedman, Vaheri & Wartiovaara, 1978; Linder, Vaheri, Ruoslahti & Wartiovaara, 1975; Stenman & Vaheri, 1978). The most studied members of this group are plasma fibronectin (Mosesson & Umfleet, 1970), amniotic fluid fibronectin (Chen, Mosesson & Solish, 1976; Crouch et al. 1978) and cell •Author for correspondence. Key words:fibronectin,monoclonal antibodies, gelatin-binding regions, cell attachment regions.

126

D. L. Hasty and others

surface fibronectin (Gahmberg & Hakomori, 1973; Hynes, 1973; Ruoslahti & Vaheri, 1974: Yamada & Weston, 1974). Although there are certain differences, these fibronectins are all quite similar to each other (Pande, Corkill, Sailor & Shively, 1981; Yamada & Kennedy, 1979), being composed of two 220X103 to 230xl03Afr polypeptide chains (designated A and B chains; Hayashi & Yamada, 1983) joined near their COOH-terminal ends by a disulphide bond. Numerous studies have demonstrated that fibronectin interacts specifically with several ligands, including gelatin and collagen (Engvall & Ruoslahti, 1977; Engvall, Ruoslahti & Miller, 1978; Kleinman, Wilkes & Martin, 1981), Staphylococcus aureus cells (Kuusela, 1978; Mosher & Proctor, 1980), fibrinogen and fibrin (Engvall et al. 1978; Stathakis & Mosesson, 1977), heparin (Stathakis & Mosesson, 1977), hyaluronic acid (Isemura, Yosizawa, Koide & Ono, 1982; Laterra & Culp, 1982; Yamada, Kennedy, Kimata & Pratt, 1980), proteoglycans (Perkins, Ji & Hynes, 1979) and possibly cell surface glycolipids (Kleinman, Martin & Fishman, 1979; Yamada, Kennedy, Grotendorst & Momoi, 1981). We have recently shown that fibronectin also binds to cells of Streptococcuspyogenes (Simpson, Hasty, Mason & Beachey, 1981), probably via the fatty acid chains of lipoteichoic acid exposed on the surfaces of these organisms (Courtney, Simpson & Beachey, 1983). Fibronectin has been implicated in a number of different cellular interactions (see Hynes & Yamada, 1982, for a recent review), and it is likely that its function in these interactions is due to its capacity to bind various ligands. Structure—function analyses of fibronectin are complicated by the overall size of the molecule and the complexity of its interactions with other molecules. Nevertheless, much progress has been made in understanding the relationships between the structure of the fibronectin molecule and its varied functional properties. Several peptide domains within the fibronectin molecule resist proteolytic digestion and have been isolated from limited digests in forms that retain certain activities of the uncleaved fibronectin molecule (Balian, Click & Bornstein, 1980; Hahn & Yamada, 1979a,6; Hayashi & Yamada, 1981, 1983; McDonald & Kelley, 1980; Pierschbacher, Hayman & Ruoslahti, 1981; Ruoslahti et al. 1979, 1981). Models developed from such studies agree with the physicochemical properties of fibronectin in solution (Alexander et al. 1978; Williams, Janmey, Ferry & Mosher, 1982) and suggest that fibronectin is composed of several ordered (globular) domains linked by flexible polypeptide regions (Hynes & Yamada, 1982). Recently a number of investigators have studied structure-function relationships of fibronectin using hybridoma antibodies (Kohler & Milstein, 1975) directed against different segments of the molecules (Atherton & Hynes, 1981; Hasty & Mainardi, 1982; Kavinsky, Clark & Garber, 1982; Koteliansky et al. 1982; Pierschbacher et al. 1981; Schoen, Bentley & Klebe, 1982; Sekiguchi, Patterson, Ishigami & Hakomori, 1982). In this paper we describe the development of a set of monoclonal antibodies against different domains of fibronectin. We show that one monoclonal antibody is directed against a gelatin-binding peptide fragment of fibronectin and inhibits binding of intact fibronectin to gelatin, whereas another is directed against other peptide fragments and potently blocks attachment of Chinese hamster ovary cells to

Monoclonal antibody mapping offibronectin domains

127

fibronectin-coated substrata. Using a rotary shadowing technique, we have morphologically mapped the binding of these two antibody probes to single molecules of intact fibronectin. Our data suggest that the ultrastructural location of certain reactive regions may not coincide with the position that would have been predicted from previous models of the molecule. Further studies using monoclonal antibodies directed toward other reactive sites should shed light not only on function but also on the tertiary and quaternary stucture of the fibronectin molecule. MATERIALS AND METHODS

Purification offibronectin Fibronectin was isolated from pooled human plasma separated from outdated blood obtained from the blood bank of the City of Memphis Hospitals. Fibronectin was purified by affinity chromatography on columns of gelatin and arginine covalently coupled to cyanogen bromideactivated Sepharose (March, Parikh & Cuatrecasas, 1974) according to the method of Vuento & Vaheri (1979) as previously described (Simpson et al. 1981). Purity was assessed by sodium dodecyl sulphate/polyacrylamide gel electrophoresis (SDS/PAGE; Laemmli, 1970).

Preparation of peptide fragments Purified human plasma fibronectin (lmgrn)" 1 in enzyme buffer: 0-OSM-Tris, 0-lM-NaCl, 0-5 mM-MgCl2, 2-7mM-CaCl2, 2-7mM-KCl, 0-02% NaN 3 , pH7-4) was cleaved by thermolysin (30/igml ; Worthington) digestion for 3h at 37°C. The reaction was terminated by the addition of 0 ' 2 M - E D T A in a volume equal to 0-25 of the reaction mixture. The 40X103M, gelatin-binding peptide fragment of the digested fibronectin was purified by affinity chromatography on a gelatin-Sepharose column as described for intact fibronectin. The gelatin-non-binding and the gelatin-binding fragments were dialysed against distilled water and lyophilized. The gelatin-nonbinding fragments were further fractionated on a heparin-Sepharose affinity column equilibrated with the enzyme digestion buffer, and bound fragments were eluted with 0-5 M-NaCl. The purity of the peptides was assessed by SDS/PAGE on 10% gels (Laemmli, 1970).

Monoclonal antibody production The non-immunoglobulin (Ig) secreting myeloma cell line Sp2/0-Agl4 (Schulman, Wilde & Kohler, 1978) used in these experiments was maintained as previously described (Hasty & Mainardi, 1982; Hasty, Beachey, Simpson & Dale, 1982). Female Balb/c mice were immunized subcutaneously with 50 /ig of either intact fibronectin or the 40X 103A/r gelatin-binding fragment in 0-02M-phosphate, 0-15M-NaCl, pH7-4 (PBS), emulsified in an equal volume of complete Freund's adjuvant. The mice were boosted after 1 month with the same dose of antigen emulsified in incomplete adjuvant. One to three months later, mice were injected intravenously with 50 fig of antigen in PBS via the tail vein and the animals were sacrificed 3 days later. Hybridomas were produced and cloned as described in detail previously (Hasty & Mainardi, 1982; Hasty et al. 1982). Production of antibodies was assayed using an enzyme-linked immunosorbent assay (ELISA; Engvall & Perlman, 1972) as previously described (Hasty & Mainardi, 1982), except that the complete culture medium was replaced with medium containing fibronectin-depleted serum 48 h before assay.

Immunoelectroblot Identical samples of the thermolysin-derived peptide fragments of fibronectin were run on a 10 % SDS/polyacrylamide gel and the separated peptides were then transferred electrophoretically onto nitrocellulose paper according to the method of Towbin (Towbin, Staehelin & Gordon, 1979). After transfer, gel lanes were cut into strips and incubated in 0-05M-Tris buffer (pH7-4) containing 0-15M-NaCl and 3 % bovine serum albumin (BSA). The strips were then incubated

128


overnight in separate wells containing one of the monoclonal antibodies. The antibodies were diluted in the above buffer so that their titres were equal as measured in ELISA. The strips were washed thoroughly and incubated in the same buffer containing 125I-labelled goat anti-mouse IgG (New England Nuclear Corp.). Finally, the nitrocellulose strips were thoroughly washed, dried and exposed to Kodak X-ray film.

Purification of monoclonal antibodies Antibodies were purified by affinity chromatography using immobilized fibronectin. Purified plasma fibronectin (20 mg) was coupled to cyanogen bromide-activated Sepharose (20 ml) and equilibrated with 0-05 M-Tris-HC1 (pH7-4) containing 1 rnM-phenylmethylsulphonyl fluoride (PMSF), 5 mM-benzamidine • HC1 and 0-02% NaN 3 . Ascites fluids were clarified by centrifugation at 10000g for 30min, filtered through a 0-4^m Millipore filter and passed over the fibronectin-Sepharose column. The column was washed extensively with 0-05 M-Tris (pH 7-4) and 0-05 M-Tris, 1-OM-NaCl (pH7-4) and the bound antibodies were eluted with 0-2M-glycine-HCl, 0-5M-NaCl (pH2-3) into tubes containing 0-1 vol. of 2M-Tris base, p H l l - 0 .

Radiolabelling of fibronectin Purified human plasma fibronectin was dialysed extensively against PBS and labelled with 12SI using a lactoperoxidase kit according to the specifications of the manufacturer (New England Nuclear Corp.). The radiolabelled fibronectin was repurified on a gelatin-Sepharose column as described above. The specific activity of the repurified material ranged from 200 to 600 Ci mmol" 1 .

Effect of monoclonal antibodies on fibronectin-gelatin interaction Gelatin-binding assays were performed according to methods described previously (McDonald & Kelley, 1980). Briefly, rat tail tendon collagen was denatured and adsorbed to polystyrene tubes. Radiolabelled fibronectin was preincubated overnight at 4CC in the presence of various concentrations of purified hybridoma IgG and then added to gelatin-coated tubes for 4h at 37°C. The wells were washed and the bound fibronectin was quantified using an autogamma spectrometer. Cell attachment assay Chinese hamster ovary (CHO) cells (American Type Culture Collection) were maintained in F12 medium supplemented with 10% foetal bovine serum; 96-well El A plates (Costar) were coated with fibronectin (5/igml" 1 in 0-lM-Na2CC>3, pH 9-6), washed with PBS and incubated for 2 h at ambient temperature with purified hybridoma IgG in PBS containing 0-5% BSA. Concurrently, CHO cells were trypsinized and washed three times with PBS containing soybean trypsin inhibitor. After washing the antibody-treated wells three times with PBS, 0-1 ml of cells in serum-free F12 medium (2xl0 6 cells ml" 1 ) was added to each well and incubated for 1 h at 37°C. After carefully washing again with PBS, attached cells were fixed to the wells with methanol, stained with crystal violet and quantified using a Dynatech Microelisa Reader.

Rotary shadow casting and electron microscopy Fibronectin—antibody complexes were allowed to form by mixing 0-5 mg fibronectin in 0-2 Mammonium formate (pH7-0) with an equivalent amount of monoclonal antibody that had been purified on a fibronectin affinity column. In order to separate antigen-antibody complexes from free antibody, the mixture was chromatographed on a lcmXlOOcm column of Sephadex G-200 equilibrated with 0-2M-ammonium formate. The material in the first peak was collected and diluted with glycerol to a final concentration of 40%. The preparations were sprayed onto freshly cleaved mica sheets, which were then placed into a Balzers BAF 301 freeze-fracture apparatus and dried for 15-20min at 10~s torr. The adsorbed proteins were shadowed with approximately 2nm of platinum at 7° while rotating at 50 rev.min" 1 and were then coated with carbon at 90°. Replicas were removed from the mica by floating onto hydrofluoric acid, washed three times by floating on distilled water, and small pieces were then picked up on uncoated grids. Electron

Monoclonal antibody mapping offibmnectin

domains

129

micrographs were made with a Philips EM 201 or an AEI EM6B at an initial magnification of X30000 as calibrated with a diffraction grating replica, and enlarged to X240000 for measurements. The lengths of fibronectin molecules and the relative positions of the different antibodies were measured using Hewlett-Packard 9845S microcomputer and 9874A digitizing screen.

RESULTS

Hybridoma production and characterization Twenty-one stable hybridomas were produced in initial fusions and 19 of these secreted antibodies that reacted with intact fibronectin in immunoelectroblot experiments. All 19 antibodies were of the IgGl isotype. To determine the regions of the fibronectin molecule recognized by the monoclonal antibodies, ELISAs were performed using various peptide fragments of fibronectin as the solid-phase antigens. The unfractionated thermolysin digest contained approximately 15 major and numerous less-distinct bands (Fig. 1A, lane 1). From this mixture a 40xl0 3 M r gelatin-binding fragment was purified by affinity chromatography on a gelatin— Sepharose column (Fig. 1A, lane 3). Peptides that failed to bind to the gelatin column (Fig. 1A, lane 2; Fig. 1B, lane 1) were then chromatographed on a heparin— Sepharose column. The majority of these peptides failed to bind, but three prominent fragments were eluted from the column with 0-5 M-NaCl (Fig. 1B, lane 3). In ELISAs, all of the hybridoma antibodies reacted in high dilution (>1:100000) with both intact fibronectin and the unfractionated thermolysin digestion mixture. Reaction of most of the antibodies with the various peptide fractions differed significantly, as expected. Interestingly, none of the 19 antibodies reacted with the 40xl0 3 M r gelatinbinding fragment. This finding supports previous observations (Ruoslahti et al. 1979) that this region of the molecule is much less immunogenic than other regions. Thus, a second fusion was performed by using spleen cells from animals hyperimmunized with the purified gelatin-binding fragment. This fusion resulted in the production of four additional stable hybridomas that produced antibodies reacting in the ELISA with both intact fibronectin and the 40x 103Mr fragment. None of the 23 clones produced antibodies that formed a precipitin line with either purified fibronectin or fresh human serum in double immunodiffusion tests. Indirect immunofluorescence experiments showed that each antibody stained cellular fibronectin within the pericellular matrix of human fibroblasts grown in medium containing fibronectin-depleted foetal bovine serum. The antibodies also reacted with fibronectin from human amniotic fluid and saliva. There were considerable differences, however, between the reactivities of individual monoclonal antibodies with plasma fibronectins isolated from different species of birds and mammals. Effect of monoclonal antibodies on cell attachment and fibronectin-gelatin interaction Further tests were performed to determine the ability of the monoclonal antibodies to inhibit either cell attachment to fibronectin-coated substrata or the interaction of

D. L. Hasty and others fibronectin with gelatin. One of the initial group of antibodies, IB 10, was found to inhibit completely the attachment of CHO cells to fibronectin at concentrations of ZS^gml" 1 (Fig. 2). Attachment of normal rat kidney (NRK) cells to human fibronectin was also inhibited by IB10 (assay kindly performed by Dr E. Ruoslahti). These results imply that the epitope on fibronectin for IB 10 should be at or near the cell attachment region recognized by 3E3 (Pierschbacher et al. 1981), a monoclonal antibody that also inhibits cell attachment. Immunoelectroblots were performed to compare the reactivity of IB 10 and 3E3 with thermolysin-derived fibronectin fragments (Fig. 3). The pattern of reactivity of the two antibodies is somewhat similar, but IB 10 reacts with two additional peptides. Thus, the epitopes recognized by these antibodies appear to be distinct. This was substantiated by the lack of reactivity of IB 10 with a 3E3 reactive ll-5xl0 3 M r peptic fragment containing the cell attachment site (Pierschbacher et al. 1981; kindly provided by Dr L. T. Furcht).

1

2

3

1 2 3

-94

-68 -43 -43 -31

-21-5

-21-5

B Fig. 1. SDS/PAGE of fractions obtained from affinity chromatography of the thermolysin digest of fibronectin on columns of gelatin-Sepharose (A) and heparin-Sepharose (B). A. Original digest (lane 1); peptides not bound to the gelatin-Sepharose column (lane 2) and the40xl0 3 .Af r peptide eluted from gelatin-Sepharose with 4M-urea (lane 3). B. The peptides not bound to the gelatin-Sepharose column (lane 1; from A, lane 2); peptides not bound to the heparin affinity column (lane 2); and peptides eluted from the heparin column with 0'5 M-NaCl (lane 3). Molecular weight markers are given (X 10~3).


131

100 r -

H

60

20

D9b

IB10

Fig. 2. Cell attachment assay. Values given represent the mean±S.E.M. (n = 32). Preincubation of the fibronectin substrate with 25/igml" 1 monoclonal antibody IB10 almost completely inhibits cell attachment. At the same concentrations, D9b, the monoclonal antibody that reacts with the 40xl0 3 Af r gelatin-binding fragment, inhibits very slightly, whereas D6, a monoclonal antibody directed against type I Escherichia colt fimbriae, exhibits negligible inhibition. Below each bar is a micrograph from a representative well of the cell attachment assay plates. Cells were stained with Crystal Violet and number of cells attached was quantified as described in Materials and Methods.

A gelatin-binding assay was performed to determine the ability of the monoclonal antibodies directed against the gelatin-binding peptide of fibronectin to inhibit fibronectin—gelatin interaction. D9b, the anti-40xl0 MT gelatin-binding peptide antibody tested reacted only with the 4OXlO3Mr peptide in immunoelectroblots of the unfractionated thermolysin digest (Fig. 3). This antibody also inhibited the binding of fibronectin to gelatin by 75% at 3ngml~ 1 (Fig. 4). The control monoclonal antibody directed against a region of fibronectin near the COOHterminal end (J. A. McDonald, unpublished observations) inhibited fibronectin binding to gelatin slightly, but only at the highest concentration tested. Ultrastructural mapping of monoclonal antibodies IBl 0 and D9b bound tofibronectin Affinity-purified monoclonal antibodies IB 10 and D9b were complexed with intact fibronectin and rotary shadowed with platinum to obtain a morphological image. Such images should allow us to map the position of the cell attachment and gelatinbinding domains on individual molecules of fibronectin. While these domains are

132


well characterized in models of fibronectin primary and secondary structure, less is known about the molecule's tertiary structure. The appearance of individual antibody molecules varied from oval to slightly triangular structures approximately 10-12nm in diameter (Fig. 5, row A). Under the conditions used in our experiments, nbronectin appeared as a highly extended, rod-shaped molecule 124 ± l-75nm in length (Fig. 5, row B; X± S.E.M.; N— 100). Many, but certainly not all, of the molecules we observed exhibited the distinct kink in the centre of the molecule noted by Engel et al. (1981); however, we did not observe a consistent angle between the two arms. The molecules observed by electron microscopy probably represent the two fibronectin subunits joined at their COOH-termini by disulphide bonds, with the NH2 termini being located at the ends of the rod (Erickson, Carrell & McDonagh, 1981; Engel et al. 1981). Each of the subunits would then be roughly 62nm in length. Rotary shadowing of fibronectin-antibody complexes revealed uncomplexed antibody and fibronectin molecules, individual

1 2

3

4 10"3

-40

Fig. 3. Immunoelectroblot of reaction of monoclonal antibodies with thermolysinderived nbronectin fragments. Equal amounts of unfractionated thermolysin digest (see Fig. 1A, lane 1) were analysed by electrophoreais in lanes of a 10% SDS/polyacrylamide gel and transferred electrophoretically to nitrocellulose sheets. Incubations with antibodies were carried out as described in Materials and Methods. Antibodies used were: H8, control ascites fluid (lane 1); 3E3, anti-cell attachment monoclonal antibody from Pierschbacher et al. (1981) (lane 2); IB10, anti-cell attachment monoclonal antibody (lane 3); and D9b, anti-40x!0 3 M r gelatin-binding fragment monoclonal (lane 4).


133

80

20

0

1 2 Antibody (log, ngml"1)

Fig. 4. Effects of monoclonal antibodies on binding of 125I-labelled fibronectin to gelatin. Total cts min~' of radiolabelled fibronectin added = 68000. Maximum cts min radiolabelled fibronectin bound = 6400. The monoclonal antibodies used as inhibitors of this interaction were: D9b (•), the anti^X^A/r-gelatin-binding peptide antibody; and 868 (O), an anti-fibronectin monoclonal directed against a COOH-terminal 40xlO3Afr region (J. A. McDonald, unpublished observations).

complexes and small aggregates. Aggregate formation like this was not totally unexpected. The antibodies are bivalent and the fibronectin molecule is dimeric, thus individual epitopes are most probably repeated at least once within the molecule. Nevertheless, aggregate formation was somewhat surprising, since we never observed formation of precipitin lines in double-immunodiffusion assays of the antibodies versus either fibronectin or fresh plasma. In addition to the small aggregates, many of the fibronectin molecules or fibronectin-antibody complexes were folded into such compact forms that they were impossible to analyse in terms of antibody binding. Fibronectin-antibody complexes were observed in which IB 10, which is directed against the cell attachment region, was visualized as a 10— 12nm diameter globular projection approximately 40 ± l-4nm (X± s.E.M.; n = 31) from the end of one or both chains (Fig. 5, row C). In every case of well-spread individual D9b-fibronectin complexes, this antibody, which binds to the 40x 103Mr gelatin-binding fragment, was visualized as a roughly 10—12 nm globular projection at the extreme tip of one or both ends of the fibronectin molecule (Fig. 5, row D). The antibodies were almost never inset any distance from the end of the molecule. Measurements were made on a small number of fibronectin molecules with a D9b antibody attached to one end only and the distance from the

134


free end to the antibody was always a minimum of 110-115 nm. Aggregates also formed in complexes of D9b with fibronectin, but, interestingly, a large number of

I & t> «•

6 * i ?

Fig. 5. Representative electron micrographs of rotary-shadowed, affinity-purified monoclonal antibody molecules (row A), fibronectin molecules (row B), affinity-purified IB10 monoclonal bound to fibronectin (row C), and affinity-purified D9b monoclonal bound to fibronectin (row D). D9b-fibronectin complexes formed into a ring are also illustrated (row E). A schematic interpretation of observations is shown below each micrograph of fibronectin or fibronectin-antibody complexes. Electron micrograph negatives were photographically reversed before printing. X180000.


135

circular molecules were seen in which it appeared that single antibody molecules had linked the two ends of the molecule, the NHZ termini of the A and B chains, together (Fig. 5, row E).

DISCUSSION Structure-function studies of fibronectin have been facilitated because several functional domains within the molecule resist proteolytic digestion and can be purified. Numerous models of domain alignment have been developed from studies of these peptide fragments (Fig. 6; see also Hayashi & Yamada 1981, 1983). Of the seven domains that span the 220X 103Mr monomer chain, those of principal interest in the context of this report are the amino-terminal, the gelatin-binding and the cellattachment domains. The peptide isolated from the amino-terminal domain varies slightly depending upon the enzyme used, but it is roughly 25-30X 103Mr whether the enzyme be thermolysin, trypsin or several others (Hayashi & Yamada, 1981, 1983; Mosher & Proctor, 1980; Sekiguchi & Hakomori, 1983). The gelatin-binding domain immediately C-terminal to the amino-terminal domain is released as a 40-45 X 103Mr fragment that is also relatively constant in size and position, whether trypsin or thermolysin is used to digest the parent molecule (Hayashi & Yamada, 1981, 1983; Sekiguchi & Hakomori, 1983). A 75xlO 3 M r cell attachment domain is roughly in the centre of the polypeptide chain and is separated from the gelatinbinding domain by a 20x 103Mr domain of unknown function (Hayashi & Yamada, 1983). Monoclonal antibody D9b reacts in immunoblots only with the 40 X 103Mr gelatinbinding fragment. Several investigators have previously noted that this region of the molecule appears to be relatively less immunogenic than other domains. Thus, it was not surprising that, using mice immunized with intact fibronectin, none of the 19 hybridoma clones obtained produced antibodies that reacted with the 40xl0 3 M r A

[3 11 40 | |

75

|3B|34"T) Ij24| 3B |

75 | | 40 | JT\

Fig. 6. Schematic illustrations of: A, a model of the position of tryptic peptides of fibronectin within the A (left) and B (right) chains (after Hayashi & Yamada, 1983). The numbers are the approximate molecular weights of peptides (XlO~ 3 ). The 40xl0 3 Af r peptide is the gelatin-binding peptide and the 7SXlO3A/r peptide is the fragment containing the cell-attachment region near its COOH end. B. A scale model of fibronectin illustrating the positions of D9b molecules bound to both the A and B chains. C. Scale model of fibronectin molecule illustrating the positions of IB 10 molecules bound to both the A and the B chains.

136


gelatin-binding peptide. Ruoslahti and co-workers calculated that considerably less than 1 % of the total antibody in anti-fibronectin sera was directed against a 30xl0 3 M r gelatin-binding fragment and they also found that this peptide elicited a poor immunogenic response in rabbits (Ruoslahti et al. 1979). McDonald, Broekelmann, Kelley & Villiger (1981), however, were able to generate antisera specific for a 60x10 Mt elastase-derived gelatin-binding fragment of fibronectin, which specifically inhibited fibronectin-gelatin interaction. McDonald, Kelley & Broekelmann (1982) further reported the intriguing finding that maintenance of human fibroblasts in the presence of anti-gelatin-binding fragment antibodies (or Fab' fragments) led to a marked alteration in the organization of both the fibronectin filaments and the collagen fibrils in the cells' extracellular matrix. D9b is monoclonal, reacts only with the 40x10 Mr gelatin-binding peptide and specifically inhibits fibronectin-gelatin interaction at low concentrations. However, we cannot infer that D9b reacts with the gelatin-binding site without additional extensive studies using Fab fragments and, or, much smaller fibronectin-gelatin-binding fragments than we or others have been able to obtain. Indeed, we suspect that D9b inhibits fibronectin—gelatin interaction by steric hindrance. Alternatively, binding of the antibody could cause a conformational change in the fibronectin molecule, so as to obscure the binding site. The actual gelatin-binding site, as well as the cell attachment site or other functional sites, are highly conserved regions of the molecule that should fail to elicit a strong immune response. Nevertheless, D9b should be a useful probe for analysing fibronectin-gelatin and fibronectin-collagen interactions in vitro and in vivo. A cell attachment site has been well characterized by Pierschbacher et al. (1981, 1982) and Pierschbacher & Ruoslahti (1984). A tetrapeptide sequence (R G D S) in the cell binding region apparently accounts for most, if not all, of fibronectin's activity for attachment of several cell types. Relatively high concentrations of a monoclonal antibody directed against a larger 11-5X 103Mr fragment containing the R G D S sequence inhibited attachment of NRK cells to fibronectin. Monoclonal antibody IB 10 is also a potent inhibitor of the attachment of both CHO and NRK cells to human fibronectin substrata. The fact that IB 10 does not react with the 11-5X 103Mr fragment suggests that its inhibitory effect on cell attachment must be due to steric hindrance or effects on fibronectin's conformation. Although we do not have an explanation for the greater activity of IB 10, differences in affinities of the two antibodies could obviously affect the activities in a biological assay. Nevertheless, since neither the binding site for 3E3 within the ll-5xl0 3 M r fragment nor the conformation of this region of fibronectin is known, the binding site for IB 10 could be closer to the cell attachment site than is the binding site for 3E3. Electron microscopy has proved to be an extremely valuable counterpart to physicochemical procedures in the analysis of the structure of certain proteins. Various biophysical procedures have suggested that fibronectin is aflexiblemolecule, existing as either an elongated strand or a compact, folded molecule. The conformation of fibronectin appears to depend upon such variables as pH and salt concentration (Erickson & Carrell, 1983; Tooney et al. 1983; Rocco et al. 1983).


137

Ultrastructural studies have confirmed these concepts and have further suggested that the substrate used for support can also affect molecular conformation (Erickson etal. 1981; EngeleJa/. 1981; Erickson & Carrell, 1983; Tooney etal. 1983). Rotaryshadowed fibronectin molecules deposited onto freshly cleaved mica from a 0-2 M salt solution containing high concentrations of glycerol are very highly extended strands measuring 120-160 nm in length (Ericksonet al. 1981; Engelef al. 1981; this study). Scanning transmission electron microscopy (STEM) of molecules deposited onto a carbon support film from a more physiological ionic strength buffer without glycerol are much more compact (Tooney et al. 1983). In the work described here we have used immunochemical and ultrastructural techniques to map the gelatin-binding and cell attachment regions of human plasma fibronectin; 10—12 nm globular projections were observed on the 124 nm fibronectin strands only in the presence of affinitypurified antibody. We never observed such projections or any other evidence of regular nodularity on fibronectin in the absence of antibody. Although we are convinced that the 10-12 nm globular molecules are antibody, direct evidence awaits development of suitable techniques to label the antibodies directly with an electrondense probe that will not compound the problem of formation of uninterpretible aggregates. Similar immunomapping of several other molecules, including types IV and V collagens (Mayne et al. 1984), has already yielded valuable structural information that was not immediately made clear from biochemical analyses. The immunomap of monoclonal IB 10 shows that this anti-cell-attachment region antibody binds to a position on the fibronectin molecule corresponding to the position that would be predicted from current molecular models (see Fig. 6), assuming that the epitope is near the COOH terminus of the 75XlO3Mr cell attachment domain and also assuming a simple relationship between number of amino acid residues and length along the strand observed in the electron microscope. Perhaps the most interesting observation we have made, however, is the unexpected finding that the epitope recognized by monoclonal antibody D9b appears to be at the extreme tips of the molecule as visualized in the electron microscope. That this was so even in the highly extended forms that existed under our experimental conditions suggests either that the 30X 103Mr amino-terminal domain is more tightly compacted by folds of secondary structure than much of the rest of the molecule or it is folded back so that the gelatin-binding region of the molecule is actually closer to the apparent end of the chain. We know from primary structure data (Peterson et al. 1983) that the 30xl0 3 M r amino-terminal domain contains five regions of type I homology, the so called 'finger domains' where disulphide-bonded loops are formed. Taking a very simplistic view of protein structure, a polypeptide the size of this 30x 103Mr fragment could be as much as 100nm in length if it were fully extended. Quite simplistically, again, the disulphide-bonded loops found in this domain would alone reduce the effective length of the polypeptide chain by at least 60%. Clearly, other aspects of fibronectin secondary structure must reduce the length at least 10fold more, since our methods should have resolved a fragment of 3-5 nm extending from the D9b binding site toward the tip. Our data do not allow us to speculate as to how this might occur, and we cannot speculate as to how the conformation of

138


the amino-terminal end of the molecule might affect the function of fibronectin. However, unfolding of the fibronectin molecule occurs upon binding collagen, collagen peptides or heparin (Williams et al. 1982). Furthermore, Johansson & Hook (1984) have presented evidence for a functional activation of fibronectin by interaction with collagen, collagen peptides or heparin sulphate and have suggested that these functional changes are due to conformational changes in the molecule. In conclusion, we have used monoclonal antibodies to map reactive regions of the fibronectin molecule at both the chemical and the ultrastructural levels. Our data suggest that the ultrastructural location of certain reactive regions may not coincide with those that might have been predicted from previous models of the molecule. Further studies using monoclonal antibodies directed toward other reactive sites should shed light not only on function but also on the tertiary and quaternary structure of the fibronectin molecule. The authors are indebted to Ms Ellen Looney, Ms Sandra Tsiu, Ms Loretta Hatmaker and Ms Susan Martin for their excellent technical assistance. We are also grateful to Dr E. Ruoslahti for providing samples of monoclonal antibody 3E3, for performing cell attachment assays with NRK cells using monoclonal antibody IB 10 and also for reading and giving suggestions on portions of this manuscript. This work was supported by research funds from the U.S. Veterans Administration and by the U.S. Public Health Service Research grants AI-10085, AI-13550, DE07218 and HL-26009. H.S.C. is the recipient of a predoctoral traineeship award (5 T32 AI-07238) from the U.S.P.H.S.; J.A.M. is an Established Investigator of the American Heart Association.

REFERENCES ALEXANDER, S. S., COLONNA, G., YAMADA, K. M., PASTAN, I. & EDELHOCH, H. (1978).

Molecular properties of a major cell surface protein from chick embryo fibroblasts. J. biol. Chem. 253, 5820-5824. ATHERTON, B. T. & HYNES, R. O. (1981). A difference between plasma and cellular fibronectin located with monoclonal antibodies. Cell 25, 133-141. BABU, J. P., SIMPSON, W. A. & BEACHEY, E. H. (1983). Interaction of human plasma fibronectin with cariogenic and non-cariogenic oral streptococci. Infect. Intmun. 41, 162-168. BALIAN, G., CLICK, E. M. & BORNSTETN, P. (1980). Location of a collagen binding domain in fibronectin. J. biol. Chem. 255, 3234-3236. CHEN, A. B., MOSESSON, M. W. & SOLISH, G. (1976). Identification of the cold-insoluble globulin

of plasma in amniotic fluid. Am.J. Obstet. Gynec. 125, 961-968. COURTNEY, H. S., SIMPSON, W. A. & BEACHEY, E. H. (1983). Binding of streptococcal lipoteichoic

acid to fatty acid binding sites on human plasma fibronectin. J. Bad. 153, 763-770. CROUCH, E., BALIAN, G., HOLBROOK, K., DUSKTN, D. & BORNSTEIN, P. (1978). Amniotic fluid

fibronectin. Characterization and synthesis by cells in culture. J. Cell Biol. 78, 701-715. ENGEL, J., ODERMATT, E., ENGEL, A., MADRI, J. A., FURTHMAYR, H., RHODE, H. & TIMPL, R.

(1981). Shapes, domain organizations and flexibility of laminin and fibronectin, two multifunctional proteins of the extracellular matrix. J. molec. Biol. 150, 97-121. ENGVALL, E. & PERLMAN, P. (1972). Enzyme-linked immunosorbent assay (ELISA). Quantitative assay of immunoglobulin G..7. Immun. 109, 129-135. ENGVALL, E. & RUOSLAHTI, E. (1977). Binding of soluble form of fibroblast surface protein, fibronectin, to collagen. Int. J. Cancer 20, 1-15. ENGVALL, E. RUOSLAHTI, E. & MJLLER, E. J. (1978). Affinity of fibronectins to collagens of

different genetic types and to fibrinogen._7. exp. Med. 147, 1584-1595. ERICKSON, H. P. & CARRELL, N. (1983). Fibronectin in extended and compact conformations. Electron microscopy and sedimentation analysis. .7. biol. Chem. 258, 14539-14 544.


139

ERICKSON, H. P., CARRELL, N. & MCDONAGH, J. (1981). Fibronectin molecule visualized in

electron microscopy: a long, thin, flexible strand. J. CellBiol. 91, 673-678. GAHMBERG, C. G. & HAKOMORI, S. (1973). Altered growth behaviour of malignant cells associated with changes in externally labeled glycoprotein and glycolipid. Proc. natn. Acad. Set. U.SA. 70, 3329-3333. GRESSNER, V. A. M. & WALLRAFF, P. (1980). Der cinsatz der lasernephelometrie zur bestimmung und rechnerunterstutzten auswertung der Fibronectin konzentration in verschiedenen koperflussigkeiten.7. din. Chem. din. Biochem. 18, 797-805. HAHN, L.-H. E. & YAMADA, K. M. (1979a). Identification and isolation of a collagen binding fragment of the adhesive glycoprotein fibronectin. Proc. natn. Acad. Sci. U.SA. 76, 1160—1163. HAHN, L.-H. E. & YAMADA, K. M. (19796). Isolation and characterization of active fragments of the adhesive glycoprotein fibronectin. Cell 18, 1043-1051. HASTY, D. L., BEACHEY, E. H., SIMPSON, W. A. & DALE, J. B. (1982). Hybridoma antibodies

against protective and nonprotective antigenic determinants of a structurally defined polypeptide fragment of streptococcal M protein. J. exp. Med. 155, 1010-1018. HASTY, D. L. & MAINARDI, C. L. (1982). Production and preliminary characterization of monoclonal antibodies to rat plasma fibronectin. Biockim biophys. Ada 709, 318—324. HAYASHI, M. & YAMADA, K. M. (1981). Differences in domain structure between plasma and cellular fibronectins. J. biol. Chem. 256, 11 292-11.300. HAYASHI, M. & YAMADA, K. M. (1983). Domain structure of the carboxyl-terminal half of human fibronectin. J . biol. Chem. 258, 3332-3340. HEDMAN, K., VAHERI, A. & WARTIOVAARA, J. (1978). External fibronectin of cultured human fibroblasts is predominantly a matrix protein. J. Cell. Biol. 76, 748—760. HYNES, R. O. (1973). Alteration of cell-surface proteins by viral transformation and proteolysis.

Proc. natn. Acad. Sci. U.SA. 70, 3170-3174. HYNES, R. O. & YAMADA, K. M. (1982). Fibronectins: Multifunctional modular glycoproteins. J. Cell. Biol. 95, 369-378. ISEMURA, M., YosiZAWA, Z., KOIDE, T. & ONO, T. (1982). Interaction of fibronectin and its proteolytic fragments with hyaluronic acid. J. Biochem. 91, 731-734. JOHANSSON, S. & HOOK, M. (1984). Substrate adhesion of rat hepatocytes: On the mechanism of attachment to fibronectin. J. Cell Biol. 98, 810-817. KAVTNSKY, C. J., CLARK, W. A. JR. & GARBER, B. B. (1982). Immunochemical analysis of

fibronectin using monoclonal antibodies. Biochim. biophys. Acta. 705, 330—340. KLEINMAN, H. K., MARTIN, G. R. & FISHMAN, P. H. (1979). Ganglioside inhibition of

fibronectin-mediated cell adhesion to collagen. Pmc. natn. Acad. Sci. U.SA. 76, 3367-3371. KLEINMAN, H. K., WILKES, C. M. & MARTIN, G. R. (1981). Interaction of fibronectin with

collagen fibrils. Biochemistry 20, 2325-2330. KOHLER, G. & MILSTEIN, C. (1975). Continuous cultures of fused cells secreting antibody of predefined specificity. Nature, hand. 256, 495—497. KOTELIANSKY, V. E., ARSENYEVA, E. L., BOGACHEVA, G. T., CHERNENSOV, M. A., GLUKOVA, M. A., IBRAGHIMOV, A. R., METSIS, M. L., PETROSYAN, M. N. & ROKHLIN, 0 . V. (1982).

Identification of the species-specific antigenic determinant(s) of human plasma fibronectin by monoclonal antibodies. FEBS Lett. 142, 199-202. KUUSELA, P. (1978). Fibronectin binds to Staphylococcus aureus. Nature, Land. 276, 718-720. LAEMMLI, U. K. (1970). Cleavage of structural proteins during assembly of the head of bacteriophage T4. Nature, Land. 227, 680-685. LATERRA, J. & CULP, L. A. (1982). Differences in hyaluronate binding to plasma and cell surface fibronectins. J. biol. Chem. 257, 719-726. LINDER, E. VAHERI, A., RUOSLAHTI, E. & WARTIOVAARA, J. (1975). Distribution of fibroblast

surface antigen in the developing chick embryo. J. exp. Med. 142, 41-49. MARCH, S. C , PARIKH, I. & CUATRECASAS, P. (1974). A simplified method for cyanogen bromide activation of agarose for affinity chromatography. Analyt. Biochem. 60, 149-152. MAYNE, R., WIEDEMANN, H., IRWIN, M. H., SANDERSON, R. D., FITCH, J. M., LINSENMAYER,

T. F. & KUHN, K. (1984). Monoclonal antibodies against chicken type IV and V collagens: Electron microscopic mapping of the epitopes after rotary shadowing. J. Cell Biol. 98, 1637-1644.

140


MCDONALD, J. A., BROEKELMANN, T. J., KELLEY, D. G. & VILUGER, B. (1981). Gelatin-binding

domain-specific anti-human plasma fibronectin Fab' inhibits fibronectin-mediated gelatin binding but not cell spreading. J . biol. Chem. 256, 5583-5587. MCDONALD, J. A. & KELLEY, D. G. (1980). Degradation of fibronectin by human leukocyte elastase. Release of biologically active fragments. J . biol. Chem. 255, 8848-8858. MCDONALD, J. A., KELLY, D. G. & BROEKELMANN, T. J. (1982). Role of fibronectin in collagen

deposition: Fab' to the gelatin-binding domain of fibronectin inhibits both fibronectin and collagen organization in fibroblast extracellular matrix. J . Cell Biol. 92, 485-492. MOSESSON, M. W. & UMFLEET, R. A. (1970). The cold-insoluble globulin of human plasma. I. Purification, primary characterization and relationship to fibrinogen and other cold insoluble fraction components. J . biol. Chem. 245, 5728-5736. MOSHER, D. F. & PROCTOR, R. A. (1980). Binding and factor XHIa-mediated cross-linking of a 27-kilodalton fragment of fibronectin to Staphylococcus aureus. Science, N.Y. 209, 927-929. PANDE, H., CORKILL, J., SAILOR, R. & SHTVELY, J. E. (1981). Comparative structural studies of

human plasma and amniotic fluid fibronectins. Biochem. biophys. Res. Commun. 101, 265-272. PERKINS, M. E., J I , T . H. & HYNES, R. O. (1979). Cross-linking of fibronectin to sulfated proteoglycans at the cell surface. Cell 16, 941-952. PETERSON, T . E., THOGERSEN, H. C , SKORSTENGAARD, K., VIBE-PEDERSON, K., SAHL, P.,

SOTTRUP-jENSEN, L. & MAGNUSSON, S. (1983). Partial primary structure of bovine plasma fibronectin: Three types of internal homology. Proc. natn. Acad. Set. U.SA. 80, 137-141. PIERSCHBACHER, M. D., HAYMAN, E. G. & RUOSLAHTI, E. (1981). Location of the cell-attachment

site in fibronectin with monoclonal antibodies and proteolytic fragments of the molecule. Cell 26, 259-267. PIERSCHBACHER, M. D. & RUOSLAHTI, E. (1984). Cell attachment activity of fibronectin can be duplicated by small synthetic fragments of the molecule. Nature, Lond. 309, 30-33. PIERSCHBACHER, M. D., RUOSLAHTI, E., SUNDELIN, J., LIND, P. & PETERSON, P. A. (1982). The

cell attachment domain of fibronectin. Determination of the primary structure..?, biol. Chem. 257, 9593-9597. Rocco, M., CARSON, M., HANTGAN, R., MCDONAGH, J. & HERMANS, J. (1983). Dependence of

the shape of the plasma fibronectin molecule on solvent composition. Ionic strength and glycerol content. 7. biol. Chem. 258, 14545-14549. RUOSLAHTI, E., HAYMAN, E. G., KUUSELA, P., SHTVELY, J. E. & ENGVALL, E. (1979). Isolation of

a tryptic fragment containing the collagen-binding site of plasma fibronectin. J. biol. Chem. 254, 6054-6059. RUOSLAHTI, E., HAYMAN, E. G., ENGVALL, E., COTHRAN, W. C. & BUTLER, W. T . (1981).

Alignment of biologically active domains in the fibronectin molecule. J. biol. Chem. 256, 7277-7281. RUOSLAHTI, E. & VAHERI, A. (1974). Novel human serum protein from fibroblast plasma membrane. Nature, Lond 248, 789. SCHOEN, R. C , BENTLEY, K. L. & KLEBE, R. J. (1982). Monoclonal antibody against human fibronectin which inhibits cell attachment. Hybridoma 1, 99-108. SCHULMAN, M., WILDE, C. D. & KOHLER, G. (1978). A better cell line for making hybridomas secreting specific antibodies. Nature, Lond. 276, 269. SEKIGUCHI, K. & HAKOMORI, S. (1983). Domain structure of human plasma fibronectin. Differences and similarities between human and hamster fibronectins. J. biol. Chem. 258, 3967-3973. SEKIGUCHI, K., PATTERSON, C. M., ISHIGAMI, F. & HAKOMORI, S. (1982). Monoclonal antibodies

directed against two different domains of human plasma fibronectin; their specificities. FEBS Lett. 142, 243-246. SIMPSON, W. A., COURTNEY, H. S. & BEACHEY, E. H. (1982). Fibronectin - a modulator of the

oropharyngeal flora. Microbiology 1982, 346-347. SIMPSON, W. A., HASTY, D. L., MASON, J. M. & BEACHEY, E. H. (1981). Fibronectin-mediated

binding of group A streptococci to human polymorphonuclear leukocytes. Infect. Immun. 37, 805-810. STATHAKIS, N. E. & MOSESSON, M. W. (1977). Interaction among heparin, cold-insoluble globulins, and fibrinogen in formation of the heparin-precipitable fraction of plasma. J. din. Invest. 30, 855-865.

Monoclonal antibody mapping offibronecttn domains

141

STENMAN, S. & VAHERI, A. (1978). Distribution of a major connective tissue protein, fibronectin, in normal human tissues. J. exp. Med. 145, 1054-1064. TOONEY, N. M., MOSESSON, M. W., AMRANI, D. L., HAINFELD, J. F. & WALL, J. S. (1983).

Solution and surface effects on plasma fibronectin structure. ,7. Cell Biol. 97, 1686—1692. TOWBIN, H. T., STAEHELIN, T. & GORDON, J. (1979). Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and applications. Proc. natn. Acad. Sci. U.SA. 76, 4350-4354. VuENTO, M. & VAHERI, A. (1979). Purification of fibronectin from human plasma by affinity chromatography under non-denaturing conditions. Biochem.J. 183, 331—337. WILLIAMS, E. C., JANMEY, P. A., FERRY, J. D. & MOSHER, D. F. (1982). Conformational states

of fibronectin. Effects of pH, ionic strength and collagen binding. J. biol. Chem. 257, 14973-14978. YAMADA, K. M. & KENNEDY, D. W. (1979). Fibroblast cellular and plasma fibronectins are similar but not identical, J. Cell Biol. 80, 492-498. YAMADA, K. M., KENNEDY, D. W., GROTENDORST, G. R. & MOMOI, T . (1981). Glycolipids:

Receptors for fibronectin? J . cell. Physiol. 109, 343-351. YAMADA, K. M., KENNEDY, D. W., KIMATA, K. & PRATT, R. M. (1980). Characterization

of fibronectin interactions with glycosaminoglycans and identification of active proteolytic fragments. J . biol. Chem. 255, 6055-6063. YAMADA, K. M. & WESTON, J. A. (1974). Isolation of a major cell surface glycoprotein from fibroblasts. Prvc. natn. Acad. Sci. U.SA. 71, 3492-3496.

(Received 31 May 1985 -Accepted, in revised form, 10 September 1985)