A New Cell Surface, Detergent-insoluble Glycoprotein Matrix of Human and Hamster Fibroblasts THE ROLE OF DISULFIDE BONDS IN STABILIZATION OF THE MATRIX*

Extraction of cell surface labeled human fibroblasts

GP140 cross-reacts with antisera prepared against fibronectin and are therefore not derived from fibronectin by proteolytic processing.
Extraction of the Empigen BB-insoluble matrix with sodium dodecyl sulfate solubilized all of the GP250, GP170, GP140, and approximately half of the fibronectin. Chromatography of the sodium dodecyl sulfate extract on Sepharose ZB under nonreducing conditions resolved a high molecular weight protein matrix that was subsequently resolved into the subunits: GP140, GP260, and fibronectin under reducing conditions. This new type of disulfide-dependent glycoprotein matrix at the cell surface is composed of fibronectin multimers, GP250 multimers, and GP140 multimers. The GP140 and GP2SO are not disulfide bonded to fibronectin since immune precipitation of the sodium dodecyl sulfate extract with anti-fibronectin antibodies specifically precipitates only fibronectin. (However, evidence is presented suggesting that GP250 and GP140 may form disulfide-dependent heteromultirners.) Since extraction with hot sodium dodecyl sulfate would destroy all noncovalent interactions, the possibility that GP140 and GP250 may form noncovalent interactions with fibronectin in the intact cell is discussed.
In studies of the mechanisms and components involved in cell attachment and spreading (for reviews see Refs. 1-3), as well as cell-cell interactions (4-6), FN,' a major transforma-* This work was supported by National Institutes of Health Grant CA23907. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
f Supported by a fellowship from the National Institutes of Health (5 F32 GM06588-02) and a New Investigator Research Award from tht National Institute of Health (R23 CA29172).
Abbreviations used: FN, fibronectin; BNS, 50 mM sodium borate tion-sensitive cell surface glycoprotein, has been of particular interest (for reviews see . FN stimulates the cell attachment and spreading process of both normal and transformed cells when coated on plastic surfaces or when attached to collagen-coated surfaces (for reviews see Refs. 2 and 9). Studies (3,IO) have also indicated that the attachment of cells to FN layers requires the involvement of the cytoskeleton as well as membrane receptors of FN. Recent studies (10, 11) have also suggested that the formation of a stable interaction of FN with the cell surface during cell attachment requires a polyvalent FN surface, as well as polyvalent receptors functioning in a cooperative manner. Attempts have been initiated to define the cell surface components involved in the interaction with FN. Inhibition studies (12) have indicated that polysialogangliosides may be involved in the interaction of the cell surface with FN-coated surfaces. However, the fact that polysialogangliosides inhibit cell attachment nonspecifically on various surfaces, and the fact that polysialogangliosides are absent from many cells that are stimulated to attach to FN-coated surfaces (13), have suggested that alternative receptors may be involved in the interaction of FN with the cell surface. Cross-linking studies have suggested that soluble FN may interact with sulfated proteoglycan (14) on the cell surface.
Additional information concerning the components involved in the cell attachment process have been gained by differential extraction experiments. Characteristically, extraction of cultured fibroblasts with nonionic or zwitterionic detergents does not disrupt the FN, cytoskeletal, or nuclear matrix organization (15-21), suggesting that these components may play a cooperative role in maintaining the organization of the attached cells. The pericellular matrix of human fibroblasts has also been reported to contain hyaluronic acid, heparan sulfate, and procollagen (22) in addition to myosin and actin. The possibility that FN may be associated with the cytoskeleton was suggested by the co-isolation of FN and actin from hamster fibroblasts (15) and by the observation that FN wiU bind to insolubilized actin in vitro (23). Double label immunofluorescence studies utilizing antibodies to both FN and actin detected coincident staining of the two fibrillar structures and suggested a transmembrane relationship bebuffer, pH 7.8 containing 25 mM NaCI; BSA, bovine serum albumin; DME, Dulbecco's modified Eagle's medium: EDTA, ethylenediaminetetraacetic acid M,, relative molecular mass as estimated by electrophoresis on polyacrylamide gels in the presence of sodium dodecyl sulfate; NIL, hamster embryo fibroblasts; PMSF, phenylrnethylsulfonyl fluoride; salt/P,, 137 mM NaCI, 2.7 mM KC1, 0.7 mM CaCL, 0.5 mM MgClz, 8.1 mM NaZHP04, 1.5 mM KH2POd; SDS, sodium dodecyl sulfate.

6953
tween microfilament bundles and fibronectin (24). A possible connection between external fibronectin-containing fibers and cytoplasmic actin microfilaments was also observed by transmission electron microscopy (25). However, recent studies by Bradley et al. (26) and by Birchmeier et al. (27) have indicated that most fibronectin structures are unrelated in their distribution to focal adhesions or stress fibers. Clearly, the relationship of FN to the cytoskeleton has not been resolved.
We have investigated the cell surface glycoprotein components present in the detergent-insoluble matrix of cultured human and hamster fibroblasts. Our experiments were undertaken to examine the possibility that new glycoproteins may be involved in localizing FN at the ceU surface and possibly connecting the predominantly pericellular FN matrix (28) to the cell surface or the cytoskeleton. Our results indicate that FN and an additional major trypsin-resistant, noncollagenous cell surface glycoprotein, termed GP140, are the major components of the detergent-insoluble matrix. Like FN, GP140 can form an extensive disulfide cross-linked structure possibly with GP250 a second glycoprotein component of the detergent-insoluble matrix. The possibility that the GP140-GP250 matrix may involve FN through covalent and/or noncovalent bonds has been investigated and discussed.

EXPERIMENTAL PROCEDURES
Cells a n d Cell Cultures-Primary cultures of human fibroblasts were obtained from skin biopsies of normal donors in the laboratory of C. Ronald Scott (Dept. of Pediatrics, University of Washington). Routinely, stock cells were frozen at the fifth passage and then used for experimentation up to passage 12. Cells were grown in Dulbecco's modified Eagle's medium supplemented with penicillin and streptomycin (lo2 units of penicillin G/ml and 0.1 mg of streptomycin/ml), 10% v/v new born calf serum, and 1% v/v fetal calf serum in a 5% CO? atmosphere.
Hamster embryo fibroblasts (NIL cells) were maintained in culture as previously described (17).
Alternatively, released form human FN was purified from conditioned culture media of human fibroblasts by affinity chromatography on gelatin-Sepharose (29).
Antisera-Antisera against FN were from two sources: goat antihuman plasma FN was purchased from Calbiochem-Behring Corp., La Jolla, CA, and formed a single precipitin line in an Ouchterlony test against human plasma. Rabbit anti-hamster plasma FN was prepared by Dr. Kiyotoshi Sekiguchi against affinity-purified (29) hamster plasma FN and formed a single precipitin line against both hamster and human plasma in an Ouchterlony test.
Cell Surface and Metabolic Labeling-Human fibroblasts were metabolically labeled with V'HJglucosamine by culturing in DME (0.25 mM glucose, 0.1 mg of pyruvate/ml supplemented with 10% fetal calf serum, and [JH]glucosamine (91 mCi/mg), New England Nuclear, 2 pCi/ml of media) for 48 h (30). Cells were washed with salt/P,, scraped with a rubber policeman, and utilized as described under "Results." For cell surface labeling, cells were washed, scraped with a rubber policeman, and labeled by the galactose o~idase-NaB [~HJ~ method (31), or the sodium peri~date-NaB[~H]~ method (32).
Polyacrylamide Gel Electrophoresis-Polyacrylamide slab gels containing 0.1% SDS were prepared following the basic stacking SDS gel technique of Laemmli (33). Samples were dissolved in a sample buffer containing 2% w/v SDS, 5% v/v 2-mercaptoethanol and heated in a boiling water bath for 5 min. Where indicated, nonreduced samples were treated as above without 2-mercaptoethanol. Slab gels were stained with Coomassie blue R-250 (34). Fluorography of slab gels followed the procedure of Bonner and Laskey (35). Protein standards for relative molecular mass (M,) estimation in polyacrylamide gels were as follows: hamster skeletal muscle myosin, 200,oOa, bovine serum albumin, 68,ooO, hamster skeletal muscle actin, 43,000; Dolichos biflorus lectin subunit, 27,000. Protein was determined by the fluorescamine method (36).
Scanning Electron Microscopy-NIL cells were grown on glass cover slips, then incubated with 2% v/v Empigen BB plus 0.1% w/v SDS as described in Fig. 2. The adherent cells and extracted cell matrix were fured in Karnovsky soiution for 1.5 h at room temperature, dehydrated with increasing concentrations of ethanol, and then critical point dried. The samples were then shadow cast and examined with a Joel scanning electron microscope.

RESULTS
Extraction of Adherent Cultures of Human and Hamster Fibroblasts with Empigen BB Detergent-Extraction of confluent cultures of human or hamster embryo fibroblasts with the zwitterionic detergent, Empigen BB, did not detach the cells from the culture plates and did not disrupt the adhesion mechanisms or components. The attached Empigen BB-insoluble matrix ( Fig. 1) possessed numerous fibrillar meshwork structures that crossed over and under the prominent convoluted nuclei. These results suggest that the components present in the Empigen BB-insoluble matrix may be interacting with each other.
Glycoprotein and Cytoskeletal Components of the Empigen BB-insoluble Matrix of Human Fibroblasts-Confluent cultures of human fibroblasts were cell surface labeled utilizing the sodium ~eriodate-NaB[~H]* method (32) in order to label glycoconjugates containing terminal sialic acid residues. The labeled cells were subject to sequential extraction with buffer containing the following. 1) Empigen BB (2%, v/v, a zwitterionic detergent) and SDS (0.1%, w/v). The addition of 0.1% w/v SDS to the Empigen BB solution decreased the quantity of minor proteins detected in the Empigen BB-insoluble matrix without affecting the major components. 2) SDS (0.5%, w/v). 3) SDS (0.5%, w/v) and dithiothreitol (25 mM). Routinely, the first extraction with Empigen BB solubilized 70 to 80% of the cell protein and 60 to 70% of the radioactivity. As seen in Fig. 2B3, the Empigen BB-insoluble matrix contained three major labeled glycoproteins with relative molecular masses of 250,000 to 230,000, 170,000 (GP170),2 and 140,000 (GPl40). Two glycoproteins were subsequently distinguished in the diffuse, high molecular mass band FN with M , = 230,000 and an additional glycoprotein with a reduced subunit molecular mass of 250,000 (GP250) (see later results). FN and GP250 were not clearly resolved on various gel systems tested. In addition, at least two other major proteins co-migrating with skeletal muscle actin (M, = 43,000) and myosin (Ad, = 200,000) were detected predominantly in the Empigen BBinsoluble matrix. In contrast, intermediate Nament protein ' The similarity and dissimilarities of GP170 found in human fibroblasts to the transformation sensitive 170,000-dalton glycoprotein previously described (Carter, W. G., and Hakomori, S. (1978) 3. Bid. Chem. 253.2867-2874) (37) will be the subject of a detailed report to be presented elsewhere. Essentially all of the cell surface-labeled glycoproteins, GP250, GP170, and GP140 were extracted from the Empigen BB-insoluble matrix with SDS solution (Fig. 2B4). In contrast, 60% of the labeled FN was solubilized under the same conditions. Complete solubilization of FN required both SDS and reducing agents (Fig. 2B5).
In the absence of 2-mercaptoethanol ( Fig. 2B6), the glycoprotein components extracted with SDS exhibited much reduced mobility on the polyacrylamide gel, suggesting extensive disulfide-dependent subunit interactions. Treatment of the cells with 2 mM N-ethylmaleimide, prior to and during the detergent extractions, failed to reduce the extent of the intersubunit disulfide bonding (results not shown). This indicated that the disulfide-dependent interaction of SDS-extractable glycoproteins was not an artifact induced by the gel electrophoresis or extraction conditions. In addition, when cells were metabolically labeled with [:'H]glucosamine and subsequently extracted, the extensive disulfide-dependent interaction was still present, indicating that the cell surface labeling procedure did not induce the disulfide bonding (results not shown) and that the glycoproteins were synthesized by the cells.
Sensitivity of FN, GP250, GP170, and GP140 to Hydrolysis by Trypsin, Collagenase, and Thrombin-As seen in Fig. 3, a time course study of the release of FN, GP250, GP170, and GP140 from the cell surface by trypsin indicates that GP170 was removed first, followed by FN. In contrast, GP250 was slowly removed by trypsin and GP140 was resistant to even prolonged digestion (30 min using 10 pg of trypsin/ml, results not shown). Similar results were obtained when samples of the Empigen BB-insoluble matrix were digested with trypsin, followed by analysis of the insoluble residue by polyacrylamide gel electrophoresis (results not shown). Trypsin digestion did not increase the ability of Empigen BB to solubilize either GP140 or to alter its apparent extensive disulfide bonding under nonreducing conditions (Fig. 3, gel 7 ) . These results also suggest that the detergent insolubility of GP250 and GP140 are independent of either FN or GP170 since removal of GP170 and FN with trypsin did not increase the extractability of GP250 and GP140 with Empigen BB.
Digestion of the Empigen BB-insoluble matrix with bovine thrombin cleaved the FN, resulting in formation of a major 205,000 molecular mass glycopeptide (Fig. 4). Thrombin has previously been reported to cleave plasma fibronectin (381, resulting in the formation of high molecular mass fragments. Similarly, GP170 was degraded by thrombin. In contrast, neither GP250 or GP140 were affected by the incubation with thrombin. Digestion of the Empigen BB-insoluble matrix ( Fig. 4) with highly purified bacterial collagenase in the presence of Nethylmaleimide and PMSF did not affect any of the glycoproteins. In parallel control experiments, collagen was completely digested by the collagenase (Fig. 4). The results indicate that none of the cell surface labeled glycoproteins are collagen-like by the criterion of collagenase sensitivity. However, purification and further analysis of the components will be required to confirm this point.

Disulfide-dependent Interaction of GP140 and GP250: Possible Presence of the Heteromultimer as the Component of Cell Surface
Matrix"GP250 and GP140 possessed similar sensitivities to proteolytic digestion, detergent extractability, and disulfide-dependent subunit aggregation, which suggested the presence of an S-S-dependent heteromultimer structure containing these two glycoproteins. To test this possibility, an SDS extract was prepared from the Empigen BB-insoluble matrix (Fig. 2, gel 5 ) . The extract was then chromatographed on a Sepharose CL 2B column under nonreducing conditions (Fig. 5A). The distribution of labeled glycoproteins in the elution profile indicated that GP250 and GP140 existed as very high molecular mass homo-or heteromultimers under nonreducing conditions (Fig. 5B). Essentially identical results were obtained when cells were treated with 2 mM N-ethylmaleimide before and during the sample preparation, suggesting that the aggregates existed before the extraction began and are not due to artifactual disulfide interchange reactions. In Fig. 5C, fractions from a similar column elution profile are presented and in which FN and GP170 were resolved under both reducing and nonreducing conditions. These results suggest that GP170 does not form a disulfide-dependent heteromultimer with FN.
We also investigated the possibility that either GP250,  2 (left). Differential extraction of cell surface-labeled human fibroblasts, followed by electrophoresis on polyacrylamide gels in the presence of SDS. Cells were surface-labeled with sodium periodate-NaB[:'HI4 and then suspended in 10 volumes of 50 mM sodium borate buffer, pH 7.8, containing 25 mM NaCl (BNS buffer) and 21, v/v, Empigen BB, a zwitterionic detergent, 0.1% w/v SDS, and 1 mM PMSF. The cell suspension was incubated on ice for 30 min and then centrifuged a t 35,000 X g for 20 min. The supernatant was removed and labeled "Empigen BB extract" and the insoluble residue was labeled "Empigen BB-insoluble matrix." The matrix was dissolved in the same volume of 0.5%, w/v, SDS in 25 mM sodium borate buffer, pH 7.8, containing 1 mM PMSF. The suspension was heated in a boiling water bath for 5 min and then recentrifuged as above. The supernatant was removed and labeled "SDS extract" and the insoluble pellet was re-extracted with 0.5% w/v SDS in 25 mM sodium borate buffer, pH 7.8, containing 25 mM dithiothreitol(100 "C, 5 min). The supernatant was labeled "SDS-dithiothreitol extract. GP170, or GP140 might hold antigenic determinants in common with FN and that any of these glycoproteins might be disulfide-bonded to FN. The SDS extract of the Empigen BBinsoluble matrix was incubated with a rabbit antiserum prepared against hamster plasma FN. The resulting immune complexes were precipitated with Staphylococcus aureus (39). Under nonreducing conditions (Fig. 6), anti-FN precipitated FN, GP250, and GP140. However, the preimmune serum or Staphylococcus aureus alone also precipitated GP250 and GP140 to the same extent indicating that the precipitation of GP250 and GP140 was due to nonspecific trapping of the glycoproteins. We obtained the same results when goat antisera prepared against human plasma FN was utilized or when  . 4 (right). Digestion of Empigen BB-insoluble matrix with collagenase and thrombin. Human fibroblasts were cell surface-labeled with the sodium periodate-NaB["H], method and extracted with Empigen BB detergent as described in the legend to Fig.  2. The Empigen BB-insoluble residue was resuspended in 50 mM sodium borate, pH 7.8, and centrifuged a t 35,000 X g for 20 min. The washed pellet was resuspended in 50 mM sodium borate, pH 7.8, containing 1 mM CaCI?. Aliquots of the Empigen BB-insoluble residue were then incubated with or without highly purified bacterial collagenase (650 units/ml containing 1 mM PMSF and 2 mM N-ethylmaleimide) or thrombin (125 pg/ml) for 2 h a t room temperature. The reaction was terminated by addition of an equal volume of 0.5% w/v SDS and heating in a boiling water bath for 5 min. In a control experiment, collagen (1.4 mg/ml was incubated with and without collagenase under the same conditions as described above. Equal aliquots of each sample were subjected to electrophoresis on a polyacrylamide gel (8%), stained for protein, and subjected to fluorography. Gel: I , fluorograph of Empigen BB-insoluble residue incubated without enzyme addition; 2, Same asgel I , incubated with collagenase; 3, Same as gel 1, incubated with thrombin; 4, protein stain of collagen incubated without collagenase; 5, collagen incubated with collagenase. Protein A-Sepharose was utilized to precipitate the antigenantibody complex. These results suggest that neither GP250 nor GP140 is disulfide-bonded to FN. When the immune precipitation was repeated after reduction of the SDS extract, the immune serum specifically precipitated only FN (Fig. 6,  gel 4 ) and the preimmune sera did not precipitate any of the glycoproteins. The reducing conditions utilized in the immune precipitation, and as described in the legend of Fig. 6, were designed to expose the SDS extract to harsh reducing conditions prior to the addition of the anti-FN serum. The immune precipitation itself was carried out in a weak reducing environment. This procedure eliminated the nonspecific precipitation without affecting the specific precipitation of FN. Al-ternatively, FN, GP250, GP170, and GP140 were solubilized from the Empigen BB-insoluble matrix by extraction with 8 M urea containing 20 mM dithiothreitol, followed by dialysis to remove the urea and reducing agent. As seen in Fig. 6, gels 6, 7, and 8, immune precipitation of the prereduced, soluble glycoproteins with anti-FN serum precipitated only FN. No nonspecific trapping of prereduced GP250 or GP140 was detected in the preimmune fraction even in the absence of reducing agent (Fig. 6, gel 7).
These results indicate that neither GP250, GP170, nor GP140 have any antigenic determinants in common with FN.

GP140-
In addition, the stickiness exhibited by W250 and GP140 for the Protein A-immunoglobulin complex or Staphylococcus aureus alone can be eliminated by the reduction of the high molecular mass complex. These results also suggest that the disulfide-dependent matrix of GP140-GP250 does not include covalently attached FN, although noncovalent interactions are possible. Density-dependent Accumulation of GPI4O and GPl70 in NIL Cells-It is possible that GP250, GP170, and GP140 are unique proteins present only in primary human fibroblasts. In order to eliminate this possibility, we examined hamster embryo fibroblasts (NIL cells) for the presence of similar glycoproteins. NIL cells were cultured at sparse, touching, and confluent cell densities, labeled by the galactose oxidase-NaB['HI4 method (31), and extracted with Empigen BB as described above. As seen in Fig. 7, increasing cell density resulted in a corresponding increase in the quantity of GP170, GP140, and FN present in the Empigen BB-insoluble matrix. We were unable to detect any increase in GP250. GP250 may have been masked by the accumulation of FN. It should also be pointed out that in contrast to the results obtained with human fibroblasts, the major quantities of glycoproteins comigrating with GP140 and GP170, in NIL cells, were efficiently extracted with Empigen BB. This may be due to the presence of additional glycoproteins with similar molecular masses or +2"E -2-ME +2"E C -FNX -  300,000 cpm-labeled matrix glycoprotein) was placed in glass tubes that had been presoaked with a solution of BSA (1 mg/ml). The samples were then heated in a boiling water bath for 5 min with or without the addition of 5% (final concentration) 2-mercaptoethanol. The samples were then diluted to 1.0 ml total volume with 0.5% Triton X-100 in BNS buffer containing 1 mg of BSA/ml. Rabbit preimrnune serum (15 pl) or rabbit antiserum prepared against hamster plasma form FN (15 pl) was added to the tubes and incubated on ice for 30 min. An aliquot (150 pl) of IgGsorb (10% v/v Staphylococcus aureus suspension) in 0.5% Triton X-l00/BNS/BSA was added, mixed, and incubated an additional 30 min on ice. The pellet was collected by centrifugation at 4000 X g for 5 min followed by washing once with 1 ml of Triton X-l00/BNS/BSA and twice with 1 ml of Triton X-l00/BNS. The pellet was suspended in 500 pI of 0.5% SDS, heated in a boiling water bath for 3 min, centrifuged, and the supernatant was lyophilized and electrophoresed on a polyacrylamide gel (8%) followed by fluorography. Gel: 1, "SDS extract" sample; 2, anti-FN, antiserum; 3, preimmune serum; 4, anti-FN antiserum reducing conditions; 5, preimmune serum, reducing conditions. The Empigen BB-insoluble matrix of sodium peri~date-NaB[:'H]~-labeled human fibroblasts was extracted with 8 M urea at room temperature for 60 min, and the insoluble residue was collected by centrifugation at 3 X lo4 X g for 20 min. The pellet was incubated with 8 M urea in 20 r n~ sodium borate, pH 7.8, containing 20 mM dithiothreitol and incubated 60 min at room temperature. The supernatant was collected by centrifugation as described above and dialyzed against 20 mM sodium borate pH 7.8. The soluble protein solution (urea-dithiothreitol extract) contains essentially all of the cellular FN, GP140, and GP250. The details of this extraction procedure are to be presented elsewhere." The urea-dithiothreitol extract (100 pl containing 225,000 cpm of labeled matrix glycoproteins) was then subject to immune precipitation as described above without the addition of 2-mercaptoethanol. Gel: 6, "urea-dithiothreitol extract" sample; 7, preimmune serum; 8, anti-FN antiserum.
incomplete formation of intermolecular disulfide bpnds. The accumulation of both GP170 and GP140 in the Empigen BBinsoluble form in NIL cells after cell-cell contact suggests that these components may play a role in an undefined cell-cell contact response. The accumulated GP140 and FN of NIL cells also exhibited extensive intermolecular disulfide bonding on polyacrylamide gels (results not shown).
' W. G. Carter, manuscript in preparation. NIL cells were plated at sparse (0.5 X IOfi cells/l5-cm plate), touching (3.0 X IO" cells/l5-cm plate), and confluent (18.0 X IO" cells/l5-cm plate) cell densities and grown for 29 h. The three cell populations were surface-labeled utilizing the galactose oxidase-NaB[:'HI4 method and then each cell sample was extracted with Empigen BB as described in Fig. 2. The detergent suspension (Gel l ), the detergentsoluble material (Gel 2), and the detergent-insoluble matrix (Gel 3 ) were electrophoresed on polyacrylamide gels (5-14%) in the presence of SDS, followed by fluorography. Numbers in the left margin (200, 6 8 , etc.) represent migration of standard proteins with known molecular masses. Gels labeled 1 received 1.2 X 10" tritium cpm. Gels labeled 2 and 3 received extracts from an equivalent number of cells as used in gels labeled 1.

DISCUSSION
Treatment of cell monolayers with a dilute aqueous solution of nonionic detergent, such as Triton X-100 or NP-40, solubilizes most cytoplasmic and membrane components, leaving a meshwork matrix and nuclei termed the "nuclear monolayer" (18, 40, 41). The composition and/or organization of the nuclear monolayer has been described by various investigators with the aim of defining the mechanisms involved in attachment of the nucleus and of the cells to the substratum (16-21,42). Our own study, as well as work by others, is based on two assumptions; f i t , that at least some of the components present in the nuclear monolayer are involved in the cell adhesion process, and second, that the insolubility of the components in nonionic detergents reflect specific intermolecular interactions that stabilize the matrix and are representative of the in uiuo organization of the matrix components. Characteristically, the nonnuclear, detergent-insoluble matrix components detected in the detergent-insoluble matrix (15-22,42, [44][45][46] include FN and the cytoskeletal proteins: actin, myosin, and intermediate filament protein, as well as numerous minor protein components. The detergent-insoluble matrix prepared from human fibroblasts has also been reported to contain hyaluronic acid, heparan sulfate, and procollagen (22). However, collagens have not been detected as compo-Glycoprotein Matrix of Fibroblast Surface 6959 nents of the detergent-insoluble matrix of chicken fibroblasts (42) or the substrate adhesion sites of murine fibroblasts (43), suggesting that collagen may not be required for stabilization of the detergent-insoluble matrix. Studies utilizing scanning and transmission electron microscopy or immunofluorescence microscopy have suggested that actin is present in the matrix in the form of bundles of microfilaments (18-22, 42, 44, 46). The FN is also filamentous (22,42,46) while procollagen appeared to co-distribute with FN (22). Intermediate fiiaments also appear as a f i e fibrillar network (19,44) (20) concluded that the insolubility of the adherent cytoskeleton in Triton X-100 was acquired either concurrently with cell adhesion or very closely with it. The authors also concluded with cell adhesion or very closely with it. The authors also concluded that the majority (90%) of the cell surface FN could be removed by digestion with trypsin without affecting the relative amount or composition of the anchored cytoskeletal proteins. Apparently most of the cell surface FN and procollagens present in the matrix may not be necessary for maintenance of nuclear or cytoskeletal adhesion.
FN and collagen have been the only cell surface glycoproteins previously detected in the detergent-insoluble matrices (15-22, 42, 45). We have examined the Empigen BB-insoluble matrix of confluent cell monolayers and suspended human and hamster fibroblasts with the specific aim of detecting other cell surface glycoproteins in addition to FN or procollagen and defining the basis for their insolubility in Empigen BB. After extraction of cultured human fibroblasts in suspension, or in monolayer cultures with Empigen BB, an insoluble matrix composed of nuclei, cytoskeletal components, and four major cell surface glycoproteins remain. The four glycoproteins termed FN, GP250, GP170, and GP140, are all insensitive to digestion by purified bacterial collagenase suggesting that they are not procollagens. Both GP170 and FN in human fibroblasts were readily digested with trypsin or bovine thrombin. In contrast, GP250 was slowly released by trypsin digestion and GP140 was resistant to prolonged incubation with trypsin. The resistance of GP140 and GP250 to digestion by trypsin is of interest since removal of the major portion of cell surface FN with trypsin does not alter cell attachment to FNcoated surfaces (13) or the composition of the anchored cytoskeleton (20). These reports suggest that a cell surface receptor for FN can rapidly reappear on the cell surface after initial removal by trypsin, or that it is resistant to digestion by trypsin.
Extraction of the Empigen BB-insoluble matrix with SDS solubilizes all of the cell surface-labeled GP170, GP140, GP250, and approximately 60% of the FN. Complete solubilization of FN required the addition of reducing agents. As previously reported (47,48), FN forms an extensive disulfidedependent matrix on the cell surface. However, molecular sieve chromatography of the SDS extract of the glycoproteins indicated that both GP250 and GP140 were also able to form an extensive disulfide-dependent matrix. However, immune precipitation of the SDS extract with antisera to FN specifically precipitated only FN. These results indicated that neither GP140 nor GP250 appeared to be disulfide-bonded to the FN matrix and also indicated that neither GP250, GP170, nor GP140 had any antigenic determinants shared with FN. Therefore, GP250, GP170, and GP140 are not derived from FN by proteolytic degradation. This conclusion was also sup-ported by the differential extraction results (Fig. 2) and the differential sensitivity of the glycoproteins to proteolytic digestion (Fig. 3). Although FN was not disulfide-bonded to either GP250 or GP140, the co-elution of GP250 and GP140 from the Sepharose 2BC1 column (Fig. 5) suggests that the GP250 and GP140 may be disulfide-bonded to each other in a heteromultimer complex. The formation of a disulfide-dependent heterocomplex would account at least partially for the resistance of both GP140 and GP250 to extraction by Empigen BB. We conclude that GP140 and GP250 exist as a new cell surface matrix, consisting of disulfide stabilized homoor heteromultimer(s) that are not covalently associated with the FN matrix.
The possibility that the FN matrix possesses a noncovalent interaction with GP140, GP250 or GP170 is intriguing. However, the complete removal of FN from the cell surface with trypsin did not remove GP250 or GP140, suggesting that GP250 and GP140 are not bound to the outside of the FN matrix. Rather, GP140 and GP250 would have to be located between the FN matrix and some other structure, such as the cell surface or the cytoskeleton. This possibility is currently under investigation.
Empigen BB-insoluble GP140 and GP170 were also detected in confluent cultures of NIL cells, suggesting that these glycoproteins are not unique to human fibroblasts. However, in NIL cells the accumulation of Empigen-insoluble GP140 and GP170 depended on increasing cell density. This suggests that cell-cell contact was instrumental in stimulating the accumulation, either as a result of increased synthesis or increased disulfide bonding which stabilized the glycoproteins to detergent extraction. The quantity of FN detectable on the cell surface has previously been reported to depend both on the cell cycle and cell density (49)(50)(51)(52). Conceivably, the disulfide-dependent polymerization of GP140 could depend on tbe cell density of the culture. GP250 was not clearly detected in NIL cells, either as a result of overlap with FN or its absence.
GP170 and its relation to the previously reported transformation-sensitive 170,000-dalton glycoprotein (37) will be reported el~ewhere.~ The presence of cell surface GP140, in addition to FN as cellular components of the Empigen BB-insoluble matrix of both attached and suspended cells, suggests that GP140 may play some role in locating FN at the cell surface particularly in areas of cell-cell contact, where FN has been reported to accumulate (46,53). FN exists predominantly as a pericellular glycoprotein (28). The mechanism by which the FN matrix associates with the surface of cells is under intensive investigation. The presence of a trypsin-insensitive, cell surface glycoprotein that is present with the FN matrix after detergent extraction, suggests a possible direct interaction between the two disulfide-stabilized glycoproteins on the cell surface. We have previously examined and discussed the importance of cooperativity in the interaction of polyvalent FN surfaces with the polyvalent cell surface receptors in inducing stable cell attachment and cell spreading (10, 13). It is possible that the presence of polyvalent GP140 may provide sufficient stabilization of even low affinity interaction with FN to resist detergent extraction. This possibility is currently under investigation. After completion of this work, Lehto et al. (54) described a trypsin-resistant, 140,000-dalton glycoprotein present in Triton X-100-extracted cell skeletons. These authors suggested that GP140 may play a significant role in anchorage of the cytoskeleton to the substratum. Also, Wylie et al. (55) prepared antisera to cell surface glycoproteins that will inhibit cell attachment to plastic surfaces. A cell surface glycoprotein ' W. G. Carter and S. Hakomori, manuscript in preparation.